a review of satellite-based rainfall estimation methods - CNR-ISAC

Consiglio Nazionale delle Ricerche Istituto di Scienze dell’Atmosfera e del Clima Via Gobetti 101 I-40129 Bologna Italy Tel: Fax: e-mail: Web:

+39-051-6399578 +39- 051-6399649 [email protected] www.isao.bo.cnr.it/~meteosat

MUSIC – MUltiple-Sensor Precipitation Measurements, Integration, Calibration and Flood Forecasting A Research Project supported by the European Commission under the Fifth Framework Programme and contributing to the implementation of the Key Action “Sustainable Management and Quality of Water” within the Energy, Environment and Sustainable Development Contract n°: EVK1-CT-2000-00058

A Review of Satellite-based Rainfall Estimation Methods by

Vincenzo Levizzani and Roberta Amorati Istituto di Scienze dell’Atmosfera e del Clima-CNR, Bologna

and Francesco Meneguzzo Istituto di Biometeorologia-CNR, Firenze

Deliverable 6.1

WP6 - Implementation of techniques for satellite image derived rainfall estimates

Page left intentionally blank

i

MUSIC – Satellite Rainfall Estimations - A Review

A REVIEW OF SATELLITE-BASED RAINFALL ESTIMATION METHODS Index 1

Introduction _______________________________________________________1

2

Single-instrument satellite rainfall estimations _____________________________3 2.1

VIS and thermal IR ------------------------------------------------------------------------------------------------- 3 2.1.1 2.1.2 2.1.3 2.1.4

2.2

Passive MW ---------------------------------------------------------------------------------------------------------- 7 2.2.1 2.2.2

2.3 2.4

3

Multispectral cloud analyses and MW synergy-------------------------------------------------------------29 Passive MW and PR algorithms -------------------------------------------------------------------------------34 Use of lightning detection----------------------------------------------------------------------------------------36

High image repetition rate and MW synergy----------------------------------------------------------------39 Physical initialization of weather prediction models -------------------------------------------------------42 Climate ---------------------------------------------------------------------------------------------------------------46 Future perspectives – The GPM -------------------------------------------------------------------------------49

Glossary _________________________________________________________51 6.1 6.2

7

Multispectral identification of rain clouds --------------------------------------------------------------------21 Algorithms using information from IR and NIR channels ------------------------------------------------25

Applications and future perspectives ___________________________________39 5.1 5.2 5.3 5.4

6

Active instruments -------------------------------------------------------------------------------------------------14 Known problems and physical constraints ------------------------------------------------------------------15

Combined rainfall estimation techniques ________________________________29 4.1 4.2 4.3

5

Imagers............................................................................................................................................8 Sounders........................................................................................................................................13

Cloud properties from solar and thermal channels ________________________21 3.1 3.2

4

Cloud indexing methods...................................................................................................................3 Bispectral methods...........................................................................................................................4 Life-history methods .........................................................................................................................4 Cloud model-based techniques ........................................................................................................5

Acronyms ------------------------------------------------------------------------------------------------------------51 Physical and mathematical quantities ------------------------------------------------------------------------53

References_______________________________________________________55

MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

ii





1

1 1

INTRODUCTION

____________________________________________________________________________

A longstanding promise of meteorological satellites is the improved identification and quantification of precipitation at time scales consistent with the nature and development of cloud rain bands. Meteorological satellites expand the coverage and time span of conventional ground-based rainfall data for a number of applications, above all hydrology and weather forecasting. The primary scope of satellite rainfall monitoring is to provide information on rainfall occurrence, amount and distribution over the globe for meteorology at all scales, climatology, hydrology, and environmental sciences. The uneven distribution of raingauges and weather radars and the relative lack of rainfall measurements over the oceans have significantly limited the use of global as well local data. In this sense the problem is not different from the determination of wind, pressure, temperature, and humidity fields with the complication that precipitation is one of the most variable quantities in space and time. Precipitation has also a direct impact on human life that other atmospheric phenomena seldom have: an example is represented by heavy rain events and flash floods (Barrett and Michell, 1991). Geostationary weather satellite visible (VIS) and infrared (IR) imagers provide the rapid temporal update cycle needed to capture the growth and decay of precipitating clouds. The swath widths of satellites in tropical orbit such as the Tropical Rainfall Measuring Mission (TRMM) (Kummerow et al., 1998) and of sensors in polar orbits like the Special Sensor Microwave Imager (SSM/I) series leave substantial gaps all over the globe. Only recently the launch of newly conceived platforms and sensors allow us to think in terms of a continuous global monitoring, which becomes reliable enough for operational applications. Operational applications, however, require quantitative rainfall determination from a variety of precipitating systems, which differ both dynamically and microphysically. This fact prompts for non-unique solutions based on the physics of precipitation formation processes. Barrett and Martin (1981) and Kidder and Vonder Haar (1995) give excellent reviews of the available methods. Petty (1995) has examined the status of satellite rainfall estimation over land. Recent reviews by Levizzani (1998) and Levizzani et al. (2001) has covered results and future perspectives from the geostationary orbit. The perspective varies widely from the relatively simple methods used for climatic-scale analyses (e.g. Arkin and Ardanuy, 1989; Arkin and Janowiak, 1991) to the more elaborate instantaneous rainrate estimations for research and nowcasting (Ba and Gruber, 2001; Turk et al., 2000; Vicente et al., 1998).


2


It is difficult to look at satellite precipitation estimates from a unified perspective encompassing all possible applications. Technical limitations presently restrict the launch of microwave (MW) sensors to Low Earth Orbits (LEO) though advancements are studied, which may eventually lead to the first MW sensor at the geostationary level (Savage et al., 1994). Recent technological developments of MW instruments on board polar orbiters have been dramatic while the use of VIS, IR and water vapor (WV) channels of geostationary satellites is by no means over. In particular, the launch of the newest generation of geostationary satellites, the Geostationary Operational Environmental Satellite GOES-I-M series (Menzel and Purdom, 1994) and the upcoming METEOSAT Second Generation (MSG) (Schmetz et al., 2002) with its Spinning Enhanced Visible and Infrared Imager (SEVIRI), adds new channels to the traditional VIS/IR/WV triplet. Some of the new channels have been tested for decades as part of the Advanced Very High Resolution Radiometer (AVHRR) series on board the National Oceanic and Atmospheric Administration (NOAA) polar orbiters or have other heritages. The TRMM payload (Kummerow et al., 1998) has represented a quantum leap with its five instruments: the TRMM Microwave Imager (TMI), the Precipitation Radar (PR), the Visible and Infrared Scanner (VIRS), the Clouds and Earth’s Radiant Energy System (CERES), and the Lightning Imaging System (LIS). The objectives of the three-year mission are to measure rainfall and energy exchange (i.e. latent heat of condensation; see Olson et al., 1999; Tao et al., 1993) of tropical and subtropical regions for a better initialization of global weather and climate models. The mission, however, has proved to be most successful and will also last much more than it was originally planned. The SSM/I will be soon replaced by the Special Sensor Microwave Imager/Sounder (SSMI/S) with its 24 channel radiometer. The challenges deriving from the exploitation of the combination of the channels are unprecedented and the science community is at work to use the available data and be ready for the new ones. An International Precipitation Working Group (IPWG) has been established with this scope in mind by the Coordination Group for Meteorological Satellites (CGMS) of the World Meteorological Organization (WMO). The present report examines the state of the art in satellite rainfall estimations trying to be as exhaustive and up to date as a short document can be, while giving reason of the new advances and potential applications that emerge more or less continuosuly. Note that the field is in constant advancement due to an international cooperation that sees scientists and operational meteorologists join forces to go for products that will necessarily have to be tested, respondent to monitoring requirements, and suitable for assimilation into numerical weather prediction (NWP) models at all scales.



3

2 2

SINGLE-INSTRUMENT SATELLITE RAINFALL ESTIMATIONS

____________________________________________________________________________

VIS and IR techniques were the first to be conceived and are rather simple to apply while at the same time they show a relatively low degree of accuracy. They are mostly used in climatological applications over long time frames. The first two sections of the present chapter provide a fairly complete review of the techniques that use VIS and thermal IR (2.1), and other IR channels (2.2). In section 2.3 the algorithms that rely upon the use of MW data from polar orbiters are described. Finally, the new instruments on board the most recent spacecrafts, including precipitation radars, are examined in section 2.4.

2.1 VIS and thermal IR A complete overview of the early work and physical premises of VIS and thermal IR (10.5 – 12.5 µm) techniques is provided by Barrett and Martin (1981); Kidder and Vonder Haar (1995) present some of the more recent results. Following Barrett and Martin’s (1981) classification the rainfall estimation methods can be divided in the following categories: cloud-indexing, bispectral, life history, and cloud model-based. Each of the categories stresses a particular aspect of the sensing of cloud physics properties using satellite imagery.

2.1.1 Cloud indexing methods Cloud indexing techniques assign a rainrate level to each cloud type identified in the satellite imagery. The simplest and perhaps most widely used is the one developed by Arkin (1979) during the GARP (Global Atmosphere Research Programme) Atlantic Tropical Experiment (GATE) on the basis of a high correlation between radar-estimated precipitation and fraction of the area colder than 235 K in the IR. The scheme, named GOES Precipitation Index (GPI) MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

4


-1

(Arkin and Meisner, 1987), assigns these areas a constant rainrate of 3 mm h , which is appropriate for tropical precipitation over 2.5° × 2.5° areas. The GPI is a standard for climatological rainfall analysis (Arkin and Janowiak, 1991) and is regularly applied and archived for climatological studies. The World Climate Research Programme (WCRP) uses the GPI for estimating global precipitation for periods from five days (pentads) to one month (WCRP, 1986; Arkin et al., 1994; Huffman et al., 1997) under the auspices of the Working Group on Data Management (WGDM) of the Global Precipitation Climatology Project (GPCP) of the Global Energy and Water Cycle Experiment (GEWEX, 1996). For a review on climatic-scale satellite rainfall estimates see Arkin and Ardanuy (1989). During the development of the GPI product for the current METEOSAT satellites a comparison was carried out between the GPI and a set of African raingauges (Turpeinen et al., 1987). An interesting aspect of this study was the use of the upper tropospheric humidity (UTH) in the vicinity of convective clouds as additional predicting parameter, although results were not conclusive. The GPI was also reconstructed from METEOSAT’s Climatological Data Set (CDS) comparing favorably with the actual index derived from IR counts for pixels corresponding to temperatures lower than 235 K (Kerrache and Schmetz, 1988). Ba and Nicholson (1998) have recently analyzed the convective activity and its relationship to rainfall over the Rift Valley lake region of East Africa using an index of convection based on METEOSAT IR data and rainfall measurements: positive correlations are found justifying the use of the index for annual and area-averaged monthly rainfall estimates. A family of cloud indexing algorithms was developed at the University of Bristol, originally for polar-orbiting NOAA satellites and recently adapted to geostationary satellite imagery. “Rain days” are identified from the occurrence of IR brightness temperatures (TB) below a threshold at a given location. The estimated rain days are combined with rain-per-rain day means that are spatially variable to produce rainfall estimations for extended periods (10 or more days). The most recent version of the Bristol technique uses variable IR rain/no-rain TB thresholds and has been applied to the River Nile catchment (Todd et al., 1995, 1999).

2.1.2 Bispectral methods Bi-spectral methods are based on the very simple, although not always true, relationship between cold and bright clouds and high probability of precipitation, which is characteristic of cumulonimbus. Lower probabilities are associated to cold but dull clouds (thin cirrus) or bright but warm (stratus). The RAINSAT technique (Lovejoy and Austin, 1979; Bellon et al., 1980) screens out cold but not highly reflective clouds or those that are highly reflective but have a relatively warm top. The number of false alarms of the pure IR techniques is reduced. The algorithm is based on a supervised classification trained by radar to recognize precipitation from both VIS brightness and IR TB. RAINSAT was applied to METEOSAT and optimized over the UK by Cheng et al. (1993) and Cheng and Brown (1995). Tsonis and Isaac (1985) and Tsonis (1987) developed clustering methods similar to that applied to bi-spectral cloud classifications. Rain areas are determined by classifying pixel clusters in the VIS/IR scene histogram and radar data are used as “ground truth” for the validation of the method. King et al. (1995) have examined the role of VIS data in improving IR rainfall estimates. Their results show a higher correlation with validation data over the IR alone for the case of warm, orographically-induced rainfall. For cold, bright clouds (e.g. cumulonimbus) the correlations are similar. Other approaches have used pattern recognition techniques applied to VIS/IR data sets. O’Sullivan et al. (1990) used brightness and textural characteristics during daytime and IR temperature patterns to estimate rainfall over a 10 × 10 pixel array in three categories: no rain, light rain, and moderate/heavy rain.

2.1.3 Life-history methods A family of techniques that specifically require geostationary satellite imagery are the life-history methods that rely upon a detailed analysis of the cloud’s life cycle, which is particularly relevant


5


FIGURE 2-1. 31 August 1996, 0700 UTC. Left: rainfall estimation using the thermal infrared of METEOSAT and the NAW Technique (Negri et al., 1984); convective heavy rainfall (yellow) is separated from stratiform light rain (light blue). Right: radar reflectivity pattern from the S. Pietro Capofiume Doppler weather radar near Bologna; high reflectivity values are indicated by the yellow, green and red pixels over the purple low-reflectivity background. The ellipsoidal area refers to the radar coverage. The retrieval performs reasonably well in separating the convective and stratiform components. Note the convective shower area on the tip of the V-shaped anvil structure of the supercell on the upper-left corner. The scattered pattern of heavy precipitation as seen by the radar is not identified by the rainfall scheme. (from Amorati et al., 2000; courtesy of the Royal Meteorological Society)

for convective clouds. An example is the Griffith-Woodley technique (Griffith et al., 1978). A major problem arises in the presence of cirrus anvils from neighboring clouds: they often screen the cloud life cycle underneath leading to possible underestimates early in the day and overestimates towards the evening. Negri et al. (1984) have simplified the Griffith-Woodley technique eliminating cloud tracking and producing a precipitation scheme that treats each cloud as if existing only in one image. The resulting Negri-Adler-Wetzel (NAW) scheme has been proved to perform at the same level of the Griffith-Woodley’s for the tropical environment. Advantages and limitations of the technique when applied to frontal precipitation leading to flood episodes were examined by Levizzani et al. (1990). The NOAA-NESDIS (National Environmental Satellite Data and Information Service) technique (Scofield and Oliver, 1977; Scofield, 1987) involves direct interaction by a meteorologist and is currently used for operational nowcasting of heavy rainfall and flash floods (Scofield and Naimeng, 1994; Robinson and Scofield, 1994; Vicente and Scofield, 1996). Reasonable performances of this type of methods are obtained for deep convective storms while contradictory results arise from their application to stratiform systems or weak convection (Amorati et al., 1999). Porcù et al. (1999) have combined the NAW with radar data for nowcasting purposes. An application of the NAW to deep convection is shown in Fig. 2-1.

2.1.4 Cloud model-based techniques Cloud model techniques aim at introducing the cloud physics into the retrieval process for a quantitative improvement deriving from the overall better physical description of the rain formation processes. Gruber (1973) first introduced a cumulus convection parameterization to relate fractional cloud cover to rainrate. Wylie (1979) used a cloud model to adjust calibration coefficients. A one-dimensional cloud model relates cloud top temperature to rainrate and rain area in the Convective Stratiform Technique (CST) (Adler and Negri, 1988; Anagnostou et al., 1999). Local minima in the IR TB. are sought and screened to eliminate thin, non-precipitating cirrus. To do so a slope parameter is calculated for each temperature minimum Tmin; the parameter is defined as

S = T1−6 − Tmin where

(2.1)

T1−6 is the average temperature of the six closest pixels. If the Tmin is located at (i,j),

T1−6 = (Ti−2, j + Ti−1, j + Ti+1, j + Ti+2, j + Ti , j +1 + Ti , j −1 ) / 6

(2.2)


6


FIGURE 2-2. Processing and adjustment scheme of the CST technique (Adler and Negri, 1988) as developed by Bendix (1997). (from Bendix, 1997; courtesy of Taylor & Francis)

Adler and Negri (1988) have established an empirical discrimination of thin cirrus in the temperature/slope plane using radar and visible imagery data. If the Tmin and its slope fall to the left of the discrimination line, the Tmin location is classified as thin cirrus (non-raining). A larger slope implies a more clearly defined minimum, that is a thunderstorm. For Florida during the Florida Area Cumulus Experiment (FACE) the discrimination line was defined as

Slope = 0.568 (Tmin − 217)

(2.3)

Reudenbach et al. (2001) have modified the CST using numerical model data (1D cloud model and mesoscale model) and their Enhanced CST (ECST) is better adjusted to the actual meteorological conditions in Western Europe, not anymore relying upon vertical profiles from the tropics (see Fig. 2-2). Once the locations of the convective cells have been identified, the rain parameters are assigned based on a 1-D cloud model (e.g. Adler and Mack, 1984) that calculates maximum rainrate and maximum volume rainrate from a sequence of model runs as a function of maximum cloud height (or minimum cloud model temperature, Tc). The convective rain area (Ar) is assumed to be five times the model updraft area (on the basis of observations). Therefore

Ar = 5π r 2

(2.4)

The average rainrate (Rmean) over the raining area of the cell is

Rmean = VRR / Ar

(2.5)

where VRR is the instantaneous volume rainfall rate calculated from the cloud model results. A linear fit of Tc and Rmean for FACE yields

Rmean = 74.89 − 0.266Tc


(2.6)

7


FIGURE 2-3. Rainfall map from the CST technique (Adler and Negri, 1988) for the El Niño event 1991-92 over Ecuador and northern Peru based on 45 days of heavy precipitation. (from Bendix, 2000; courtesy of Taylor & Francis)

while a similar log-linear fit of Tc and Ar yields

Ar = exp(15.27 − 0.0465 Tc )

(2.7) -1

To every other element colder than the stratiform threshold a fixed rainrate of 2 mm h is assigned. Bendix (1997, 2000) has successfully used the CST to investigate the formation, dynamics, and spatial distribution of heavy precipitation during the 1991-92 El Niño in Ecuador and Northern Peru (Fig. 2-3). The adaptation of these methods, originally developed for tropical deep convection, to other areas of the globe or climate regimes is by no means trivial (e.g. Levizzani et al., 1990; Marrocu et al., 1993; Negri and Adler, 1993; Pompei et al., 1995).

2.2 Passive MW Clouds are opaque in the VIS and IR spectral range and precipitation is inferred from cloud top structure. At passive MW frequencies precipitation particles are the main source of attenuation of the upwelling radiation. MW techniques are thus physically more direct than those based on VIS/IR radiation. The emission of radiation from atmospheric particles results in an increase of the signal received by the satellite sensor while at the same time the scattering due to hydrometeors reduces the radiation stream. Type and size of the detected hydrometeors depend upon the frequency of the upwelling radiation. Above 60 GHz ice scattering dominates and the radiometers can only sense ice while rain is not detected. Below about 22 GHz absorption is the primary mechanism affecting the transfer of MW radiation and ice above the rain layer is virtually transparent. Between 19.3 and 85.5 GHz, the common passive MW imagers’ frequency range, radiation interacts with the main types of hydrometeors, water particles or droplets (liquid or frozen). Scattering and emission happen at the same time with radiation undergoing multiple transformations within the cloud column in the sensor’s field of view (FOV). At different frequencies the radiometers observe different parts of the rain column. As for other parts of the spectrum, MW radiation is absorbed (but not scattered) by cloud


8


droplets, water vapor and oxygen thus making precipitation estimates based on absorption potentially difficult. Precipitation drops strongly interact with MW radiation and are detected by radiometers without the IR strong biases. The biggest disadvantage is the poor spatial and temporal resolution, the first due to diffraction, which limits the ground resolution for a given satellite MW antenna, and the latter to the fact that MW sensors are consequently only mounted on polar orbiters. The matter is further complicated by the different radiative characteristics of sea and land surfaces underneath. Sea surface has a relatively constant and low emissivity (ε = 0.4) so that the radiation emitted from it is small and precipitation (emissivity around ε = 0.8) will increase the amount of radiation detected by the sensor through emission. The high sea surface polarization also contrasts very much with the low polarization of rain. Land surfaces have a high and variable emissivity (ε = 0.7 - 0.9), close to that of precipitation, and low polarization. The emissivity is dependent upon the characteristics of the surface including vegetation and moisture content. Rainfall over land will increase the upwelling radiation stream but at the same time will absorb radiation introducing considerable difficulties in the identification of rain areas. Scattering is thus the key to the MW rainfall estimation techniques over land and the 85.5 GHz channel of the SSM/I is very sensitive to scattering from small particles. Several algorithms exist that make variable use of MW channels and polarizations. Among others, Alishouse et al. (1990), Aonashi and Liu (2000), Berg and Chase (1992), Berg et al. (1998), Ferraro and Marks (1995), Ferraro et al. (1986), Ferriday and Avery (1994), Grody (1984, 1991), Hinton et al. (1992), Kniveton et al. (1994), Kummerow and Giglio (1994a,b), Kummerow et al. (1996), Liu and Curry (1992, 1993), Petty (1994a,b), Spencer et al. (1989), Wentz (1997), Wilheit et al. (1977). An overview of passive MW methods is given by Wilheit et al. (1994) while Petty (1995) concentrates on rainfall estimation over land. Kidd et al. (1998) have examined advantages and disadvantages of statistically derived, empirically calibrated MW algorithms. Smith et al. (1994) discuss the emergence of inversion-type, physically-based algorithms and Smith et al. (1998) present the recent results of the WetNet Precipitation Intercomparison Project (PIP-2). Ebert and Manton (1998) do the same for the exercise during the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA COARE). The major instruments used for MW-based rainfall estimations are the SSM/I, a scanning-type instrument that measures MW radiation over a 1400-km wide swath at four separate frequencies, 19.35, 22.235, 37.0 and 85.5 GHz, the latter extending the spectral range of previous instruments into the strong scattering regime (as regards to precipitation-size particles). The radiometer operates in dual polarization (both vertical and horizontal) at each frequency except the water vapor channel at 22.235 GHz, where only vertical polarization is measured. The effective ground resolutions at these different frequencies are approximately 69 × 43, 60 × 40, 37 × 28 and 15 × 13 km (in order of lowest to highest frequency). The TMI on board TRMM represents an evolution of the SSM/I with a new 10.7 GHz channel vertically and horizontally polarized and the shift of the water vapor channel from 22.235 to 21.3 GHz. The SSMI/S will take over the SSM/I with 24 channels between 19 and 183 GHz in a coincidentviewing, conical scanning system (current SSM/I, SSM/T2 and SSM/T are separate instruments), an upper-air sounding capability up to 70 km, and a wider swath (1700 km).

2.2.1 Imagers Weinman and Guetter (1977) described a method for reducing the effects of the background emissivities at MW frequencies, thus making it possible to extract rainfall information from over water bodies and land surfaces using the same algorithm. Using model calculations of 37 GHz TB (at both vertical and horizontal polarizations) over land and sea they first noted that -1 background surfaces only affect the satellite measurements at rainfall rates less than 4 mm h , and that the measurements could be defined in a single equation which would effectively remove the problem of double-valued TBs to rainfall rate conversion. Areas whose temperature was less than 285 K represented rain areas that were similar to areas of precipitation identified


9


by the coincident scanning of a radar. A technique was proposed by Grody (1984) to enhance the precipitation signal by minimizing the effects of surface emissivity on MW measurements. This equation utilized the 37 GHz vertical and horizontal channels, and a constant, to enable the contrasting background emissivities to be normalized. The derived equation was:

PCT = Tv − βTh / (1 − β )

(2.8)

where PCT is the polarization corrected temperature and β is a constant. Changes due to surface temperature are on a larger scale than those related to surface emissivities, and could usually be neglected when determining β. One possible method to determine the constant is to compare low and high emissivity backgrounds to obtain the value

Θ = β / (1 − β ) = (Tvl − Tvs ) / ( poll − pols )

(2.9)

where Tvl, Tvs, poll and pols are the vertical TBs and polarizations over land and sea, respectively. Thus

PCT = Tv + Θ (Tv − Th )

(2.10)

The variable nature of the constant was suggested by Spencer et al. (1989), who suggested both theoretical and empirical methods to determine it, dependent upon climatic regions. The value of Θ needs to be set in relation to the meteorological conditions prevailing at the time, and a threshold chosen to delineate rain from no-rain. The values of Θ and the rain/no-rain threshold are not unrelated: as the first increases the rain/no-rain threshold also increases. As noted by Kidd (1998), the use of the PCT at 85, 37, and 19 GHz will provide information about height distribution of the hydrometers within the cloud system. Although the 85 GHz channels, and hence the PCT at 85 GHz, will be strongly affected by thick cirrus, the sensitivity to light precipitation makes it ideal for rain/no-rain delineation. The application of the PCT algorithm with different Θ values is illustrated in Fig. 2-4, which shows the effect of increasing

FIGURE 2-4. Application of the PCT at 85 GHz using Θ (see text) values of 0.0 ( 85 GHz vertical channel ), 0.4, 0.77, and 1.0 on SSM/I data for 27 August 1992. (from Kidd, 1998; courtesy of Taylor & Francis)


10


the Θ value upon the imagery. At low Θ values, the PCTs are close to the 85 GHz vertical channel TB, characterized by high land and low sea TBs. As the Θ value increases, the amount of polarization added to the vertical channel increases so that when Θ equals one, the PCT is the polarization added to the vertical TB. At a certain value of Θ, the TBs of the land surfaces will match those of the sea surfaces. Any rain will appear as lower PCT values due to the scattering of the upwelling radiation stream. In Fig. 2-4, Θ = 0.77 is probably best over southern England and northern France. At this value of Θ the green/blue area (< 272 K) is rain related. However, some misclassification occurs over Scotland where a slightly lower value of Θ would be needed to equalize the land and sea background emissions. One more application over the UK can be found in Todd and Bailey (1995). The selection of the Θ values can be quite accurately determined by interactive visualization of the data, whilst thresholds may be derived by comparing the PCT values with the outputs of a frequency difference algorithm (37 GHz vertical minus 85 GHz vertical TBs) over land to determine the rain/no-rain boundaries. Thus it is clearly central to the success of these technique the appropriate assignment of values to two key variables: one detecting the calculation of the PCT, the other the threshold of the rain/no-rain boundary. Kidd (1998) results indicate that the technique is good at delineating rainfall and adjusting for seasonal fluctuations in the background surface temperature. More complex approaches are based on time-dependent cloud-radiation models that take full account of precipitation microphysics. Several examples can be found in the literature. Mugnai et al. (1990) have simulated the physics of an intense hailstorm in the MW characterizing the vertical sources of radiation that contribute to the top-of-the-atmosphere (TOA) MW TBs measured by satellite radiometers. A vertically, angularly, and spectrally detailed radiative transfer model was applied to the highly resolved thermodynamic and microphysical output of the Regional Atmospheric Modeling System (RAMS). Weighting functions are found that are essentially vertically resolved radiative structure functions describing the process by which radiation originates and reaches the satellite radiometer. The functions are then subdivided into individual contributions by the various hydrometeor species generated by the cloud model. Smith et al. (1992) and Mugnai et al. (1993) have successively expanded the model using the University of Wisconsin RAMS (UWRAMS) and proposing a time-dependent cloud-radiation model that establishes the microphysical settings as a base of precipitation retrieval from passive MW. The most important problem while using these highly-detailed and physically

FIGURE 2-5. Color isosurfaces in the 3D space of MW brightness temperatures (K) at 19, 37, and 85 GHz, of

the Valtellina Flood measurement manifold, and COHMEX and HURRICANE model manifolds calculated over land. The model manifolds are those obtained with the modified particle size distribution and the T–q profiles for the Valtellina event. Two different perspective views of the manifolds are given. The range limits of all axes are 50–350 K. (from Panegrossi et al., 1998; courtesy of American Meteorological Society)



11

based retrieval schemes is the cloud-radiation database, which needs necessarily to be as close as possible to the real conditions observed by the satellite. In very plain words, it is for example impossible to obtain correct results from a retrieval in a mid-latitude frontal convection if using a cloud-radiation database conceived for tropical hurricanes. Panegrossi et al. (1998) have developed and analyzed cloud–radiation databases from three different storm simulations involving two different mesoscale models run at cloud scales. Each database relates a set of microphysical profile realizations describing the space–time properties of a given precipitating storm to multifrequency TBs associated to a measuring radiometer. The various cloud–radiation databases are also used with a simplified profile retrieval algorithm to examine the sensitivity of the retrieved hydrometeor profiles and surface rainrates to the different microphysical, macrophysical, and environmental factors of the simulated storms. In Fig. 2-5 an example is shown that demonstrates how different are the behaviors of different systems in the frequency space with respect to cloud microphysics. The results emphasize the need for physical retrieval algorithms to account for a number of these factors, thus preventing biased interpretation of the rain properties of precipitating storms, and minimizing root mean square (rms) uncertainties in the retrieved quantities. The choice of state-of-the-art cloud models is crucial for a thorough representation of cloud microphysical processes (Khain et al., 2000). The importance of the separation between convective and stratiform components of precipitation in the algorithms comes out from the results of Kummerow et al. (2001) who examined the problems of the at-launch version of the Goddard Profiling (GPROF) algorithm. A substantial overestimation in the inter-tropical convergence zone was noted. This was traced back to the algorithm’s poor separation between convective and stratiform components coupled with a poor separation between stratiform and transition regions in the a-priori cloud model database. Prabhakara et al. (2000) have contributed a discrimination between convective and stratiform rainfall in their algorithm. Convective precipitation is accompanied by extensive stratiform systems both in the tropics and at midlatitudes. Petty (2001a,b) presents a highly simplified, yet meteorologically realistic and flexible, parametric model for generating hydrometeor profiles and other environmental properties relevant to MW radiative transfer calculations in quasi-stratiform rain clouds. With this model, it is possible to vary cloud and environmental properties, including hydrometeor size distributions and densities, in a continuous, self-consistent, fashion and to assess the impact of these changes on computed multichannel MW brightness radiances. It is also possible to utilize gradient descent methods to find plausible combinations of cloud properties that explain observed multichannel MW radiances in rain clouds. Bauer et al. (2000) have simulated the explicit particle spectra during cloud evolution by a two-dimensional spectral cloud model to investigate the response of MW radiative transfer to particle spectra development with special focus on the radiative effects of melting particles below the freezing level in stratiform clouds. The inclusion of the radiative properties of the melting layer in the cloud model has been considered by Bauer (2001a). Using two cloud models and considering the large variety of observed rain systems, he found a good agreement at lower frequencies between observations and simulations concerning both TB distributions and TB offsets due to bright bands. The melting layer model cannot compensate for insufficiencies of cloud model ice microphysics, so that significant differences between simulations and observations at high MW frequencies are noted. Another recent work on the effects of the melting layer at MW frequencies was carried out by Olson et al. (2001a,b). Fig. 2-6 and 2-7 show, respectively, the effects of the bright band on the MW channel response as well as on the 13.8 GHz TRMM PR radar measurements as they come out of model simulations. Observational and modeling studies have revealed the relationships between convective– stratiform rain proportion and the vertical distributions of vertical motion, latent heating, and moistening in mesoscale convective systems. Therefore, remote sensing techniques that can be used to quantify the area coverage of convective or stratiform rainfall could provide useful information regarding the dynamic and thermodynamic processes in these systems. Anagnostou and Kummerow (1997) have proposed a discrimination method based upon 85 GHz TB observations. Hong et al. (1999) separate the two cloud types and precipitation regimes MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

12


FIGURE 2-6. Radiance difference fields between 180-min cloud model simulation including mixed-phase, melting precipitation and the same simulation without mixed-phase precipitation, both at model grid resolution. Positive differences indicate increased radiances due to the presence of mixed-phase precipitation. Plan views of the difference fields at (a) 10.65, (b) 19.35, (c) 37.0, and (d) 85.5 GHz. (from Olson et al., 2001b; courtesy of American Meteorological Society)

FIGURE 2-7. Vertical cross sections from the 180-min simulation of Fig. 2-6 including mixed-phase, melting precipitation. (a),(b) The vertical cross sections of precipitating liquid and ice, respectively. (c),(d) Simulated fields of 13.8-GHz extinction coefficient and radar reflectivity, respectively, at model grid resolution. (e),(f ) Extinction coefficient and radar reflectivity differences from simulations with and without mixed-phase precipitation, respectively. (from Olson et al., 2001b; courtesy of American Meteorological Society)

using both emission channels (19 and 37 GHz) and the scattering from the 85 GHz channel MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002


13

FIGURE 2-8. Imagery of a mesoscale convective system near the Cape Verde Islands in the North Atlantic on 12 Sep 1999. (a), (b) The convective area fractions within 85.5-GHz footprints based upon TMI texture and polarization information, respectively. (c) TMI 85.5-GHz imagery. (d) PR-derived convective area fractions at a resolution comparable to the TMI. (from Olson et al., 2001c; courtesy of American Meteorological Society)

using also a probability matching method to relate the convective-stratiform index to a convective fraction of precipitation area that overcomes the problem of the low-resolution pixels that contain mixtures of both types. In the method of Olson et al. (2001c), if sufficient MW scattering by ice-phase precipitation is detected, the method relies mainly on the degree of polarization in oblique-view, 85.5 GHz radiances to estimate the fraction of the radiometer footprint covered by convection. In situations where ice scattering is minimal, the method draws mostly on texture information in radiometer imagery at lower MW frequencies to estimate the convective area fraction. In Fig. 2-8 an example is given of the convective-stratiform partition using the Olson et al. (2001c) method. Finally, Mohr et al. (1999) have quantified the contribution of deep convective systems to tropical rainfall by analyzing their type, size and intensity over a calendar year using the 85 GHz channel.

2.2.2 Sounders MW sounders were flown on board polar orbiting satellites since the early 70-ies and continuously evolved from the original few-channel, poor resolution configurations to the actual Advanced Microwave Sounding Unit (AMSU-B). This latter has twelve channels within the 5 - 60 GHz oxygen band, four channels around the 183 GHz water vapor line, and four window channels at 23.8, 31.4, 50.3 and 89 GHz. The AMSU series is devoted to temperature soundings, but the channels can be applied to measure precipitation, water vapor, cloud liquid water, snow cover, sea ice concentration, surface wetness. Other sounders are flown on board


14


the Defense Meteorological Satellite Program (DMSP) satellite series SSM/T and SSM/T2, mainly conceived for temperature soundings. A major caveat comes from the scanning geometry: for the SSM/I series it is conical and thus it maintains a constant footprint over the whole imagery, while for the other instruments the geometry is across-track so that the footprint can increase by more than a factor of two as the instrument scans from nadir to limb position. For large-scale features such as atmospheric water vapor and temperature the smoothing due to larger footprints is often very limited. This is in generally not true when deriving precipitation and cloud liquid water and the measurements have to be normalized to a common resolution. Grody et al. (1999) have proposed an algorithm that is based on scattering indices that vary according to the instrument characteristics. For AMSU-A the two scattering indices for land and oceans are

SI land = TB (23) − TB (89) SI ocean = −113.2 + [2.41 − 0.0049 TB (23)] TB (23) + 0.454 TB (31) − TB (89)

(2.11) (2.12)

where SIland ≥ 3 and SIocean ≥ 9. The indices were obtained using scatter-free AMSU data over the corresponding surface and regressing the high frequency channel 89 GHz against the lower frequency channels. They provide a measure of the scattering at high frequencies relative to low frequency channels. Calibration against radar rainfall data must be adopted to retrieve precipitation. For example, over land the relationship was found to be

R = 0.005[(SI land + 18) / 1.3]

(2.13)

where R is the rainrate. For AMSU-B the channels at 89 and 150 GHz are used to derive rainrates at a higher resolution (16 km at nadir compared to the 48 km of AMSU-A). Experimental algorithms use the following scattering indices

SI land = [42.72 + 0.85 TB (91)] − TB (150) SI ocean = 0.013{TB (91) + 33.58 ln[300 − TB (150)] − 341.17}

(2.14) (2.15)

Validation of the AMSU-A algorithms was carried out by Zhao et al. (2000) who verified the good ability to capture the gross feature and produce spatial patterns of rainfall in various cases. Also over land a reasonable agreement was found by separating convective and stratiform components. Improvements are generally observed in weak precipitation areas.

2.3 Active instruments The most important novelty of the actual panorama of precipitation measuring instruments from space is the PR, a precipitation radar operating at 13.8 GHz on board TRMM, the first of its kind to be flown on board a spacecraft. The instrument aims at providing the vertical distribution of rainfall for the investigation of its three-dimensional structure, obtaining quantitative measurements over land and oceans, and improving the overall retrieval accuracy by the combined use of the radar, and the TMI and VIRS instruments. The radar is expected to add information not only on the intensity and spatial/temporal distribution of rain but also on rain type, storm structure, melting layer and latent heat release at different heights. The instrument, though, brings to space the same shortcomings of ground-based weather radars, above all the difficult calibration in terms of absolute rainrate values. The day-1 TRMM 2A-25 algorithm is described by Iguchi et al. (2000) and is akin to the Hitschfeld-Bordan inversion that uses PR-based climatology of cloud-free surface radar reflectance to assign total path attenuation and underlying ground radar-based Z-R relationship for defining allowable rain microphysics, where initial top-down attenuation path estimated from unified Z-R / Z-A relationships, are adjusted so the final R profile gives rise to A estimated from MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

15


FIGURE 2-9. Hurricane Brett, horizontal cross section of reflectivity fields at 3.2-km height: (a) NOAA P3 aircraft radar reflectivity ZP3 , corrected for path-attenuation and radar calibration error; (b) TRMM PR attenuationcorrected reflectivity ZTRMM , from the version-5 2A-25 algorithm. In (a), the P3 radar field is corrected for advection and averaged at the PR beam resolution, and the actual flight track of the P3-42 aircraft is indicated (two smallradius loops performed within the hurricane eye are not drawn). (c) The rain-type classification index derived from the PR, with S (C) for stratiform (convective) rain. The box delineates the domain used for point-to-point comparisons. (from Ferreira et al., 2001; courtesy of American Meteorological Society)

surface reflectance methods (R is the rainrate, and A and Z are the radar attenuation and reflectivity, respectively). The rainfall estimates are calculated from the Z profiles by using a b power law R = aZ in which the parameters a and b are both functions of the rain type and the heights of the 0ºC isotherm and storm top. Effects of rain type, presence or absence of a bright band, the phase state, the temperature, and the difference in terminal velocity from changes in the air density are taken into account. Moreover, the initial values of a and b are modified so that there is consistency with the assumed drop size distribution model. The coefficient a is further modified by the index of non-uniformity. Ferreira et al. (2001) have re-examined the algorithm correcting for underestimations of the day-1 version (see Fig. 2-9).

2.4 Known problems and physical constraints There are several constraints that limit the accuracy and applicability of MW-derived rain retrievals. One of the most prominent problems concerns the definition of the characteristics of the underlying surface (Ferraro et al., 1994). Ferraro et al. (1998) point out that most authors concentrate on improving the physics of their algorithms while forgetting that the inability to identify and eliminate false rain regions, or conversely to fail to identify all valid raining regions, can equally lead to substantial errors and failures. Mis-identified pixels create first-order errors in the statistical comparison of the algorithms with ground-based measurements or with other algorithms. An idea of how TBs vary with respect to land/sea surface types can be gathered from Fig. 2-10. Although the most complex, physically based algorithms may include the screening step as an integral part of the precipitation retrieval algorithm, most current SSM/I algorithms are designed as a two-step process, particularly those that perform over-land retrievals (see Smith et al. MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

16


FIGURE 2-10. Left. Brightness temperature variations as a function of frequency for various land surface types. (a) Vertical polarization variations and (b) polarization differences. Right. Same as left but for ocean surfaces. Wind conditions are for a speed of 30 m s-1. (from Ferraro et al., 1998; courtesy of American Meteorological Society)

1998). The first step is the rain detection (or screening step); the second is the TB to rain-rate transformation (or conversion step). The reason that it is useful to discriminate between the screening and conversion steps is that it is not uncommon for the algorithms to use different input channels, as input to the screening module, from those used for input to the conversion module. Grody (1991) has clearly demonstrated that the process of screening is at least as important for the algorithm’s success as is that of the retrieval itself. The paper of Ferraro et al. (1998) addresses, besides the issue of surface screens, the problem of coastline identification and methods to validate the quality of TB measurements. Although many of the screens appear simple in formulation and stem from empirical analysis, one cannot underestimate their importance nor their physical significance. The failure to properly identify and remove surface artifacts from the retrievals of a given algorithm will ultimately lead to the discrediting of that algorithm, especially when used to generate global monthly rainfall estimates, particularly those deemed suitable for climatic studies. The complicated interaction of earth-emitted MW radiation with various surface types and atmospheric variables makes the development of surface screens that work everywhere very difficult. Hence, the designs of global screens are conservative in nature and may involve incorrectly removing some raining pixels in certain regions during certain meteorological situations. The detection of light rain and the entire question of the threshold of detectability of light rain remains a pressing research problem. Screens can be developed for a variety of effects such as: quality control screens, data corrections, geographic screens, land surface screens (desert, semi-arid land, snow cover,…), and ocean surface screens (sea ice, winds,…). For example a simple screen for discriminating desert areas Grody (1991) uses the following threshold MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

17


TB (19V ) − Tb (19 H ) > 20

(2.16)

where V and H denote vertical and horizontal polarization, respectively. For intense convective rain, which contains large ice particles, and snow cover scattering occurs at low frequencies for both rain and snow. This can cause the TB(22V) to fall below 264 K and the threshold can be applied. A second condition is necessary (Grody, 1991)

TB (22V ) > 175 + 0.49 TB (85V )

(2.17)

An example of rain discrimination technique application is given in Fig. 2-11. Grecu and Anagnostou (2001) have carried out a comprehensive investigation of overland precipitation estimation from TRMM passive MW observations. Three aspects relevant to MW rainfall estimation — the rain/no-rain discrimination, convective/stratiform (C/S) rain-regime classification, and the estimation of vertically integrated hydrometeor content (VIHC) — were considered. For rain/no-rain and C/S classification the authors used artificial neural network (ANN) schemes; for VIHC estimation both linear and non-linear functional forms were investigated. The three components of the algorithm were evaluated based on 1 yr of TRMM observations over the continental US. The reference precipitation datasets were derived from PR observations using a variational retrieval technique and TRMM PR rain detection and C/S classification products. Results showed that accurate (less than 10% increase in rain estimation errors) rain/no-rain discrimination is achieved from TMI observations based on the proposed ANN scheme and multifrequency MW–based descriptors. The C/S classification performance was not as high (60% agreement); nevertheless, it was proven to improve considerably the

FIGURE 2-11. Examples of rain/no rain discrimination techniques, without any screening logic, for PIP-2 case 2, overpass 1 (25 November 1987, 0044 UTC). Rain areas are denoted by the black regions and the SSM/I swath in gray. (a) Use of the scattering index with a threshold of 5 K, (b) TB(22V) 2 TB(85V) ≥ 3 K, and (c) TB(85H) < 253 K. (from Ferraro et al., 1998; courtesy of American Meteorological Society)


18


precipitation estimation. Furthermore, the technique was shown to properly detect the major convective cores. The VIHC estimation indicated better performance than did vertically averaged rainrate estimation. Also, the VIHC estimation is expected to be more consistent in converting to rain accumulation than are surface rain fluxes. Another algorithm that is designed to work overland is the one of Prabhakara et al. (1999). The characterization of the uncertainty inherent in the passive MW retrievals was conducted by Coppens et al. (2000) using TMI data. The authors concentrated on understanding the problems connected to linking the vertical distribution of hydrometeors within the cloud, which has obviously a large impact on upwelling MW TBs. As previously seen this means that a database need to be constructed that relates cloud species to MW signatures: the problem is finding the right scenario in the database whose associated radiances are closest to the MW measurements. The study shows that the single most crucial variable characterizing the rain profile is the vertically averaged rain rate, followed by the difference between the high-altitude subfreezing-level rain and the precipitation closer to the surface. The remaining variables have very small variances and thus they can safely be considered constant (equal to their respective means). The measurements of the passive MW channels can similarly be described using two linear combinations of the TBs. Kummerow (1998) examined the largest uncertainty in many of the physical retrieval models that use passive MW data, i.e. the homogeneous rainfall assumption. Four months of TOGA COARE shipborne radar data was used to describe the horizontal characteristics of rain. The vertical hydrometeor structures needed to simulate the upwelling TB were taken from a dynamical cloud model and radiative transfer computations were performed using a fully threedimensional Monte Carlo solution in order to test all aspects of the beamfilling problem. Results show that biases as well as random errors depend upon the assumed vertical structure of hydrometeors, the manner in which inhomogeneity is modeled in the retrieval, and the manner in which the radiative transfer problem is handled. The author explored the impact of inhomogeneous rainfall upon the predicted TBs so that these relations may eventually be used to develop a physically based error model for MW precipitation retrievals. In this study, the spatial inhomogeneity of rainfall within the satellite FOV is designated by σFOV: 1/ 2

σ FOV

N FOV  1 (R − Ri ) =  ∑  N FOV − 1 i 

(2.18)

FIGURE 2-12. The TB to rainfall relations at 10 GHz, horizontal polarization for various values of rainfall inhomogeneity. σ is the inhomogeneity and R the mean rainfall in the pixel. A lognormal distribution is used. (from Kummerow, 1998; courtesy of American Meteorological Society) MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002


19

where R is the field-of-view (FOV) averaged rainfall rate and NFOV represents the number of high resolution radar pixels in the satellite FOV. The parameter σFOV is a function of both the resolution of the radar data as well as the satellite FOV dimensions. In Fig. 2-12 the 10.7 GHz TB values as a function of mean footprint rainfall with different inhomogeneities. For example, a -1 TB of 240 K implies rainfall rates of approximately 12 mm h for a homogeneous rainfall but 60 -1 mm h if σFOV is four times the mean rainfall in the pixel. If the characteristics of rainfall are not described, then no assumptions may be made about spatial inhomogeneities, and results on any space- and timescale will be as uncertain as any individual measurement. This condition leads to unrealistically high uncertainty estimates for monthly mean rainfall or large area averages. While the horizontal inhomogeneity is perhaps the greatest single source of error in retrieval algorithms, it is not the only source. Kummerow (1998) has considered only the horizontal component, but other components such as the vertical structure and background fields need to be examined. The impact of inhomogeneous rainfall upon specific algorithms must still be investigated within the context of that specific algorithm.


20




21


3 3

CLOUD PROPERTIES FROM SOLAR AND THERMAL CHANNELS

____________________________________________________________________________

Multispectral data have long since been available both from polar orbiting and geostationary satellite sensors and used for retrieving cloud properties. The relevant channels for cloud characterization were part of the payload of the polar satellites, while the sensors at geosynchronous altitude were almost exclusively devoted to VIS-IR operational monitoring of precipitation system displacements. This has considerably changed in the past few years since more and more sophisticated sensors have been conceived for the GEO orbits that will allow for global real-time cloud characterization. It is in this changed framework that new applications are possible that will considerably improve rainfall estimations from space. A brief overview of multispectral methods for cloud characterization is given in 3.1 as the necessary background information for understanding the methods that estimate precipitation using IR and NIR channels (3.2).

3.1 Multispectral identification of rain clouds Cloud radiative properties at VIS, near IR (NIR) and IR wavelengths have long since been studied and documented (among others Arking and Childs, 1985; Hunt, 1973; Kleespies, 1995; Liou, 1992; Saunders and Kriebel, 1988; Slingo and Schrecker, 1982). They can be summarized as follows: In the thermal IR the radiative properties are sensitive to the size distribution of the hydrometeors. In particular, an increase in the particle size increases the transmissivity, decreasing the reflectivity and increasing the emissivity of the cloud layer. This latter dominates at these wavelengths. The emissivity of ice clouds is less than that of water clouds.


22


In the NIR (e.g. the 3.9 µm MSG channel) the emissivity of a cloud layer is lower than in the thermal IR window: there is a large contribution of reflected radiation at the cloud top. Clouds with small hydrometeors scatter and reflect much of the 3.9 µm radiance. An increase in cloud particle size or the presence of large drops or ice crystals near the cloud top reduces the 3.9 µm reflectance from the cloud. Clouds containing more ice reflect less solar radiation in the 3.7 - 3.9 µm range as ice strongly absorbs at these wavelengths and ice crystals are generally larger than cloud droplets at cloud top. NIR reflectance mostly refers to cloud particles effective radius (re). VIS reflectance is primarily due to cloud optical depth.

Several methods have been proposed for the retrieval of cloud parameters from various cloud types. Pioneering studies were conducted by Arking and Childs (1985) and Nakajima and King (1990). At the Swedish Meteorological and Hydrological Institute (SMHI) a thorough classification of the radiative cloud properties in the IR, NIR and VIS over Scandinavia resulted in a operational model for cloud and precipitation classification (Liljas, 1986; Karlsson and Liljas, 1990) that makes full use of all the multispectral signatures of the AVHRR channels. VIS and NIR AVHRR channels were used by Masuda and Takashima (1990) for the discrimination of optical characteristics of cirrus clouds over the ocean using a multiple scattering approach. The AVHRR split window channels were investigated by Parol et al. (1991) for extracting re information content from the difference of their TBs. Ou et al. (1993) developed a method for the remote sensing of cirrus cloud temperature, optical depth and mean re using thermal IR and NIR channels. Setvák and Doswell (1991) used the AVHRR channel 3 (3.7 µm) for investigating cumulonimbus cloud top reflectance and Levizzani and Setvák (1996) found very high reflectivity values on top of severe storms related to the emission of plumes of small ice crystals above the storm top. The removal of the solar component from the AVHRR channel 3 for the retrieval of cirrus cloud parameters in daytime was introduced by Rao et al. (1995). Recently Baran et al. (1999) have used the 3.7 and 10.8 µm channels of the Along Track Scanning Radiometer (ATSR) on board the European Remote Sensing Satellite (ERS-1/2) of the European Space Agency (ESA) to derive the thermal optical depth, crystal size and shape of tropical cirrus clouds during night-time. Cloud properties over polar snow-covered areas were retrieved using AVHRR measurements by Han et al. (1999) who also propose a comprehensive radiative transfer model coupling the atmosphere to the snowpack. Turk et al. (1998c) have developed a method for using NIR cloud reflectances from GOES to characterize stratus and fog microphysics during the day. Finally, King et al. (1997) have conceived the official operational algorithm for optical thickness and effective radius of the Moderate Resolution Imaging Spectroradiometer (MODIS). Fundamental studies on the remote sensing of cloud phase and composition were conducted by Pilewskie and Twomey (1987) who used a ground-based NIR sensor for the discrimination of ice from water in cumulus congestus and cumulonimbus clouds. Platnick and Twomey (1994) have tackled the problem of the susceptibility of cloud albedo to changes in drop concentration. Rosenfeld and Gutman (1994) were among the first to concentrate on cloud top microphysical properties in view of remote sensing of potential rain clouds. They used AVHRR channel 1 (0.65 µm), 3 (3.7 µm), 4 (10.8 µm), and 5 (12.0 µm) to develop a quantitative methodology for the retrieval of cloud top properties that are relevant to the precipitation potential of clouds. The idea is that the tops of raining clouds, that is clouds with larger water droplets and ice particles, reflect very little solar radiation in the 3.7 - 3.9 µm region. In particular the most crucial parameter seems to be the effective radius of cloud particles defined as

re


∫ = ∫

∞

0 ∞ 0

r 3n (r ) dr r n (r ) dr 2

(3.1)

23


where n(r) is the number size distribution as a function of the particle radius r. The authors showed that optically thick clouds with retrieved re > 14 µm correspond to well defined areas with radar echoes that indicate the existence of precipitation size particles. The cloud liquid water content (LWC) is also widely used to characterize cloud microphysics and defined as follows

LWC =

4πρ w 3

∞

∫ n(r ) r 0

3

dr

(3.2)

where ρw is the density of water. re can be parameterized in terms of the single scattering albedo a, the volume extinction coefficient k, and the asymmetry factor g (Slingo and Schrecker, 1982). The cloud optical depth is in turn defined as z2

τ = ∫ k dz z1

(3.3)

where z1 and z2 are the height of cloud base and cloud top, respectively, and the liquid water path (LWP) simply is z2

LWP = ∫ LWC ( z )dz z1

(3.4)

The LWC cannot be explicitly derived from remote sensing measurements. τ is proportional to LWP and inversely proportional to re in the VIS. Since τ and re can be derived from satellite data, LWP can be in turn estimated under the assumption that re is vertically constant throughout the cloud column (for parameterizations and tables see Hu and Stamnes, 1993; Espinoza and Harshvardhan, 1996). Lensky and Rosenfeld (1997) have conceived a multispectral rainfall estimation technique based on the method of Rosenfeld and Gutman (1994). They concentrated on areas of around 2 2000 km that Rosenfeld and Gagin (1989) showed to be the critical limit beyond which a further increase in cloud cluster area does not result in higher rainrates. The probability matching theory was used to compare with radar data in Israel (Rosenfeld et al., 1994). The authors also introduced an index addressing the possible presence of multi-layered clouds and a parameter of cloud “convectivity” based on cloud top texture analysis. Precipitation areas and rainfall amounts are quantitatively determined taking into account the significant microphysical and dynamical differences behind rain formation processes in convective and stratiform clouds that lead to large differences between cloud top properties and rain intensities. Rosenfeld and Lensky (1998) have applied the technique to continental and maritime convective clouds on AVHRR imagery calculating the evolution of re with temperature and inferring from it information on precipitation forming processes. The concept of distinguishing between clouds of maritime and continental origin with respect to their original aerosol content is perhaps the most striking novelty in cloud physics of the last decade and is bound to reach the importance of that between convective and stratiform types. The authors link in-cloud microphysical processes such as diffusional growth, rainout, mixed-phase precipitation, and glaciation to the particular cloud system under observation since not all processes appear in a single cloud system. Differences are observed between clouds forming from different air masses referring to the modification that the pristine air mass (e.g. maritime) is undergoing while moving (e.g. inland). Other transformations are documented for those air masses moving towards areas affected by massive aerosol loads (e.g. biomass burning, urban air pollution) (Rosenfeld, 1999, 2000). Suitable look-up-tables and color codes are constructed that allow for a relatively straightforward identification of precipitating clouds on the satellite imagery. The authors also provide some aircraft validation of the satellite observations over Israel, Thailand and Indonesia. By the same token the presence of supercooled droplets down to –37.5°C is confirmed by aircraft penetrations in deep convective clouds over South America (Rosenfeld and Woodley, 2000), multispectral satellite observations, and numerical modeling (Khain et al., 2001). In other MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

24


-30 IN D O EX p ollute d Australia urb an T ha i p re -m o ns Am azon sm o ke Israel dust

-10

o

T [ C]

-20

0 10 20 0

5

10

r

15 eff

20

25

30

35

[µ m]

FIGURE 3-1. Satellite retrieved median effective radius of particles near the top of deep convective clouds at various stages of their vertical development, as a function of the cloud top temperature, which serves as a surrogate for cloud top height. This is shown for clouds forming in polluted (solid lines) and pristine air (broken lines) during: INDOEX, over southern India and the southern Indian Ocean (red); Australia pollution tracks (blue); Thailand pre-monsoon clouds with suppressed coalescence (purple); Smoke over the Amazon (green); desert dust over Israel (black). The vertical green line denotes the 14 µm precipitation threshold. (from Ramanathan et al., 2001; courtesy of American Association for the Advancement of Science)

FIGURE 3-2. 1 March 1998, 0250 UTC. Top. TRMM VIRS image of fires, smoke and clouds over Kalimantan, Indonesia. The lookup table consists of red for visible reflectance, green for 3.7 µm reflectance (approximating re), and blue for the inverse of 10.8 µm TB. The northwest coast of the island is denoted by the yellow line. The small orange areas on the upper right (east) corner are hot spots indicating the fires. The smoke, streaming from the hot spots south-westward, is indicated by the fuzzy purple color of the background. The smoke-free background is blue. This color scheme shows clouds with small droplets (re15 µm) are colored pink, and cold ice clouds appear red. The black hatching marks the areas in which the TRMM PR detected precipitation. Bottom. Vertical cross section along the AB transect (left to right). The gray area is the clouds, as measured by their top temperature. The colors represent the precipitation reflectivity, in dBZ, as measured by the PR. The white line is the TB of the TMI 85 GHz vertical polarization, plotted at the altitude of that temperature. (from Rosenfeld and Lensky, 1998; courtesy of American Meteorological Society)

words this kind of technique gives reason of cloud microphysical composition and is highly instrumental for precipitation estimations together with passive MW and radar techniques. Fig. 3-1 shows the effect of various air masses on the formation of clouds in different parts of the world documented by the Rosenfeld and Lensky (1998) technique: such effects may have a substantial impact onto the global water cycle (Ramanathan et al., 2001). In Fig. 3-2 the direct influence of forest fire smoke in suppressing precipitation is shown for Indonesia (Rosenfeld and


25


Lensky, 1998). Finally, the suppression of precipitation by desert dust as a possible desertification feedback mechanism is discussed by (Rosenfed et al., 2001). Ellrod (1994) has explored the potential of the TB difference (BTD) between the 3.9 and the 11 µm channels of the GOES-I for nighttime detection of precipitation. Due to the lower emissivity of 3.9 µm versus 11 µm, significant BTD values were observed within cloud systems that could be related to both cloud thickness and cloud phase (liquid or ice) as well as the thermal structure of the atmosphere. For example, the TB of the inactive outer portion of the cloud cirrus anvil is usually significantly higher (> 5°C) at 3.9 µm than at 11 µm, whereas convectively active or thick multilayered regions that produce rainfall are either slightly warmer or cooler in the 3.9 µm channel. A potential thus stems for using the channels in nighttime monitoring of precipitation. Recently Key and Intrieri (2000) have used the AVHRR 3.7, 11 and 12 µm channels for cloud particle phase determination for daytime, nighttime, cold cloud, and warm cloud conditions. They demonstrate that BTD at the three wavelengths and 3.7 µm reflectance provide necessary, but not sufficient, information for phase determination, and that the relationship between the cloud and surface temperatures must also be considered. The phase determination for homogeneous phase cloud cases were shown to be accurate while the spectral resolution of the AVHRR is insufficient for cloud phase determinations in multi-layer, multiphase conditions. These techniques are well in line with the potential of the new MSG channels as indicated by the recent study of Watts et al. (1998): the authors have conducted a thorough simulation of the response of the SEVIRI instrument and identified the potential for the retrieval of cloud properties. Their retrieval method, called the Optimal Estimation (OE), could in principle provide first-hand data for a microphysically-based MSG rainfall algorithm.

3.2 Algorithms using information from IR and NIR channels IR and NIR channels other than the thermal IR window show some potential for application to rainfall estimations. Techniques for the instantaneous delineation of convective rainfall areas using split window data were initially conceived for the NOAA AVHRR (Inoue, 1987a,b, 1997) and are instrumental for the detection of semi-transparent cirrus clouds (Inoue, 1985). These techniques rely upon the detection of non-precipitating cirrus and low-level cumulus clouds using the two window channels at 10.5 - 11.5 and 11.5 - 12.5 µm (the so-called split window). The information content of the split window channels partially corrects erroneous rainfall area delineation (and consequent frequent rainfall overestimate) of simple IR techniques producing better false alarm ratios (FAR). Kurino (1997) has applied a split window technique to data from the Japanese Geostationary Meteorological Satellite (GMS) (Fig. 3-3). He used three parameters: the 11 µm TB, the BTD between 11 and 12 µm, and that between 11 and 6.7 µm. The technique was statistically characterized against digital radar data over Okinawa islands using a 6-hourly GMS-5 data set during the typhoon season July-September, 1995. Heavy rainfall was examined and the technique seems to recommend itself as a nowcasting tool on a 1° × 1°-area average. The potential of using the WV channel of geostationary satellites as done by Kurino (1997) is indicated by observations of “warm water vapor pixels” (Tjemkes et al., 1997) over deep convective clouds. The brightness temperature in the water vapor (TWV) over such clouds is often higher than in the IR (TIR) and is strictly related to the presence of stratospheric water vapor and its amount. A correlation between water vapor structure above these storms and rainfall is to be more closely investigated. Amorati et al. (2000) have found a qualitative correspondence between the occurrence of (TWV - TIR) > 0 and rainfall amount above deep convective storms in Northern Italy. The NIR 3.9 µm channel of GOES-8/9 satellites includes spectral features suitable for applications to rainfall detection and estimation. This channel was included for a long time in NOAA/AVHRR instruments (centered at 3.7 µm) for a variety of purposes including ice


26


FIGURE 3-3. 23 September, 1995, 0000UTC. Top left. GMS-5 false color image of IR TB at 11 µm (TB11) for the case

of typhoon Ryan (T9514). Top right. BTD between 11 and 12 µm (TB11-12). Middle left. BTD between 11 and 6.7 µm (TB11-6.7). Middle right. Radar image of digitized echo intensity for 16 levels from the Japan Meteorological Agency (JMA) radar network. Bottom left. Comparison of satellite rain estimation and radar ground truth for 1°×1° squares centered at “a” in the previous panels from 1200 UTC, 22 September, through 1200 UTC, 23 September. a) Hourly and b) accumulated rainfall. Bottom right. Same as in previous graphs but for square centered at “b”. LUT: 3-D technique by Kurino (1997). CST: Convective stratiform technique (Adler and Negri, 1988). GPI: GOES Precipitation Index (Arkin and Meisner, 1987). (from Kurino, 1997; courtesy of Copernicus Gesellschaft e. V.)

discrimination and sunglint detection. Vicente (1996) developed a simple and fast algorithm for rainfall retrieval using the 11 and 3.9 µm channels with the obvious advantage of nighttime use and sensitivity to the presence of ice and water vapor. The different sensitivity of the two



27

channels to the two microphysical categories can be used as a tool to select cloud areas associated to a higher probability of producing rain. Inoue and Aonashi (2000) compared cloud information from TRMM VIRS with rain measurements from the PR for rain cases during June 1998 over a frontal zone in east Asia (see an example in Fig. 3-4). The authors selected the following four parameters: 1) radiance ratio of 0.6 and 1.6 µm [channel 1/channel 2 (Ch1/Ch2)], 2) BTD between 11 and 12 µm (BTD45), 3) BTD between 3.8 and 11 µm (BTD34), and 4) TB in channel 4 (Ch4) as the cloud information. The flags of “rain certain” stratiform rain, bright band existence, and convective rain observed by the PR, and integrated rain rate from the rain bottom to rain top were used as the rainfall information (Fig. 3-4). From the comparison between rain–no-rain information by the PR and the four cloud parameters, they found that values of the radiance ratio of Ch1/Ch2 larger than 25, BTD45 smaller than 1.5 K, and BTD34 smaller than 8 K are effective in delineating rain area. The authors computed the probability of detection (POD), FAR and skill score (SS), and compared for several rain and no-rain algorithms The BTD34 scores better in FAR than the BTD45 and is better than BTD45 in delineating the thicker part of cirrus clouds. The use of the second channel shows better scores than does the use of the single IR threshold algorithm.

FIGURE 3-4. a) 11 µm TB map, b) and c) satellite-derived rain-expected cloud map, and d) rain certain index map by PR for orbit number 3116 on 13 June 1998. (from Inoue and Aonashi, 2000; courtesy of American Meteorological Society)


28




29


4 4

COMBINED RAINFALL

ESTIMATION TECHNIQUES ____________________________________________________________________________

Low-earth orbit MW data have been proposed by several authors for a combined use with geosynchronous IR radiances using different approaches that vary widely from the monthly to the instantaneous scale. The approaches are very variable and depend on the instruments to be combined as described in chapter 4.1. Recent techniques draw upon the availability of TRMM PR data and are discussed in chapter 4.2. A brief section (4.3) is finally devoted to the emerging use of lightning detection both from ground-based networks and space sensors like the TRMM LIS.

4.1 Multispectral cloud analyses and MW synergy With the advent of passive MW measurements several of the existing VIS/IR techniques were re-examined and synergies sought that could help adjusting some of the well known problems of the top-down approach of these methods, which generally infer precipitation only from cloud top information (see chapter 2). Recent international algorithm intercomparison projects (e.g. Arkin and Xie, 1994; Barrett et al., 1994; Smith et al., 1998) show, as a general result, that passive MW techniques perform superiorly for instantaneous applications over oceans, while IR or combined IR/MW techniques show improved monthly rainfall accumulations, mainly due to the high temporal sampling made available by geosynchronous observations. Adler et al. (1991, 1993, 1994) proposed the Adjusted GPI (AGPI) method that corrects the GPI monthly rainfall estimates using an adjustment factor based on MW and IR data. Berg (1994) used the algorithm of Berg and Chase (1992) to combine IR and MW TB over pentads and 2.5° boxes. Xu et al. (1999a) have used MW and IR data to obtain the new Universally Adjusted GPI (UAGPI) method providing stable estimates of monthly rainfall at various spatial scales. This latter method addresses two MW-related errors: a) the sampling error caused by insufficient sampling rate, and b) the measurement error of instantaneous rainrate.


30


The first work on mixed MW and IR rainfall estimation strategies for instantaneous estimations was conducted by Vicente, (1994) and Vicente and Anderson (1993). The need for such an approach has long since been recognized for effective combining MW-based precision and IR geostationary time repetition and coverage for estimates that go from 1 day to the instantaneous scale (Jobard and Desbois, 1994; Levizzani et al., 1996; Sheu et al., 1995). In particular, a precipitation retrieval by Liu and Curry (1992) provides the foundation of a novel cloud classification scheme of both thin and deep cloud systems thuse meeting the need to use IR data for cloud top characterization in addition to MW emission and scattering signals (Liu et al., 1995). The attempt of calibrating IR geosynchronous data using SSM/I retrievals over the Pacific Ocean by Vicente and Anderson (1994) involves two multilinear regressions a day allowing a twice-a-day calibration between MW rainfall rate and IR TB. Merging MW, VIS, and IR imagery data available on the same satellite such as TMI and VIRS on board TRMM provides further potential for the improvement of instantaneous retrievals (Bauer et al., 1998). Simultaneous VIS and IR data may contribute mainly to better rain-regime classification, in particular when sophisticated cloud identification techniques and cloud parameter retrievals are incorporated. Apart from the retrieval of rain profiles, the PR can contribute to improved beam filling corrections on an instantaneous basis. In contrast, the role of VIRS is twofold: 1) to provide more information about cloud structure and development state using its full spectral and spatial information content, and 2) to serve as a part of a calibration standard for the combination of TRMM and geosynchronous observations. The latter point aims more at climatological investigations, that is, diurnal variations of tropical rainfall and monthly rain accumulations neither limited by the retrieval accuracy inherent to IR methods nor by the sampling accuracy inherent to MW retrievals. Refined cloud detection and classification methods as well as the retrieval of cloud parameters from VIS, NIR, and IR data support rainregime discrimination. Automatic cloud structure and texture analyses using the comparably high spatial resolution of VIRS provide multidimensional weighting measures for MW measurements of bulk cloud parameters as well as improved coastline treatment. Comprehensive studies of combined VIS, IR, and MW retrievals for instantaneous applications require more efforts in the future (Bauer et al., 1998). Turk et al. (1998a, 2000a) have proposed a hybrid MW-IR technique that is based on a statistical probability matching (see for example Rosenfeld et al. 1984) between precipitation levels (R) from SSM/I + TMI algorithms (now being upgraded to include AMSU-A and B data) and TBs from geostationary satellites. Quasi-real time MW rainfall data are used to constantly renew the R(TB) relationship. The idea is shown in Fig. 4-1. The technique takes advantage from the more precise rain identification of MW techniques and the time repetition of IR imagery. The critical point is the duration of MW cloud characterization that depends on the specific precipitation system. The more MW sensor overpasses the better for the probability matching validity. The technique is presently used within an operational time frame with maximum

FIGURE 4-1. Scheme of the algorithm of Turk et al. (1998, 2000a). Cyan circles represent equally-spaced geostationary IR TB observations while gold circles denote non-routine, non-equally spaced MW-based rainfall estimations from SSM/I, TMI, AMSU-B and other satellites. Shaded box represents the previous-time “window” prior to t0 (courtesy of J. F. Turk, Naval Research Laboratory, Monterey, CA)


31


FIGURE 4-2. 9 August 2001, 1500 UTC. Top. Global IR TB from geostationary meteorological satellites. Bottom. Example of global 3-hourly accumulated precipitation product from the technique of Turk et al. (1998a, 2000a) (courtesy of J. F. Turk, Naval Research Laboratory, Monterey, CA)

FIGURE 4-3. 6 October, 1998, 1200 UTC. Left. IR image from METEOSAT over Central Mediterranean. Right. Instantaneous rainfall estimation from the technique of Turk et al. (1998a, 2000a) with the latest MW overpass 2 hours earlier. (courtesy of J. F. Turk, Naval Research Laboratory, Monterey, CA)

emphasis on the identification and tracking of rapidly developing rain storms particularly those that form over the ocean and head inland. Absolute rainfall amounts are secondary with respect to relative values. The other major goal of the method is to develop a 3-hour global rainfall analysis devoid of spatial and temporal gaps, which are a characteristic of LEO satellites; these analyses are meant for assimilation into numerical weather prediction models. The technique statistically blends together SSM/I and TMI data with any geostationary instrument’s 11 µm IR data in a near real time fashion for retrieving instantaneous rainfall rates and accumulations at the geostationary update cycle. An example of the 3-hourly global product is shown in Fig. 4-2 and an instantaneous product in Fig. 4-3.


32


There are other approaches that seek IR TB–rainrate relationships, variable in space and time, from coincident observations of IR TB and passive MW rainrate (accumulated over a calibration domain). Todd et al. (2001) use the probability matching method and their algorithm is named MW IR Rainfall Algorithm (MIRA). The IR TB–rainrate relationship is then applied to IR imagery at full temporal resolution. The authors investigate MIRA estimates of rainfall over a range of spatial and temporal scales. Over the global Tropics and subtropics, optimum IR thresholds and IR TB–rainrate relationships are highly variable, reflecting the complexity of dominant cloud microphysical processes. As a result, MIRA shows sensitivity to these variations, resulting in potentially useful improvements in estimate accuracy at small scales in comparison to the GPI and the MW-calibrated UAGPI. Unlike some existing PMW/IR techniques, MIRA can successfully capture variability in rain rates at the smallest possible scales. At larger scales MIRA and UAGPI produce very similar improvements over the GPI. Miller et al. (2000) have proposed the Microwave/Infrared Rain Rate Algorithm (MIRRA), which was tested and developed using data from the TOGA-COARE campaign, with shipboard radar rainrate estimates used as truth. Results indicate enhanced performance in bias, correlation and rms error for MIRRA compared with other IR and combined algorithms at the instantaneous scale, while retaining the good performance of geostationary algorithms at daily and monthly scales. PERSIANN, an automated system for Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks, was been developed for the estimation of rainfall from geosynchronous satellite longwave IR imagery (GOES-IR) at a resolution of 0.25°×0.25° every half-hour (Hsu et al., 1997; Sorooshian et al., 2000). The accuracy of the rainfall product is improved by adaptively adjusting the network parameters using the instantaneous rain-rate

FIGURE 4-4. Monthly rainfall distribution from TRMM 3A11 (5°×5°), 3B31 (5°× 5°), 3B42 (1°×1°), 3B43 (1°×1°), and PERSIANN-GT (1°×1°) products. (from Sorooshian et al., 2000; courtesy of American Meteorological Society)



33

estimates from the TMI product 2A12 (Kummerow et al., 2001), and the random errors are further reduced by accumulation to a resolution of 1°×1° daily. The authors’ current GOES-IR– TRMM TMI based product, named PERSIANN-GT, was evaluated over the region 30°S–30°N, 90°E–30°W, which includes the tropical Pacific Ocean and parts of Asia, Australia, and the Americas. The resulting rainrate estimates agree well with the National Climatic Data Center (NCEP) radar-gauge composite data over Florida and Texas (correlation coefficient ρ > 0.7). The product also compares well (ρ ∼ 0.77–0.90) with the monthly WMO gauge measurements for 5°×5° grid locations having high gauge densities. The PERSIANN-GT product was evaluated further by comparing it with current TRMM products (3A11, 3B31, 3B42, 3B43) over the entire study region. The estimates compare well with the TRMM 3B43 1°×1° monthly product, but the PERSIANN-GT products indicate higher rainfall over the western Pacific Ocean when compared to the adjusted geosynchronous precipitation index–based TRMM 3B42 product. A comparison of the performance of PERSIANN against TRMM algorithms for monthly rainfall distribution is given in Fig. 4-4. The Auto-Estimator technique proposed by Vicente et al. (1998) (Fig. 4-5) follows another concept making use of IR 11 µm GOES data and radar data from the US network with applications to flash flood forecasting, numerical modeling, and operational hydrology. The rainfall retrieval is performed through a statistical analysis between surface radar-derived instantaneous rainfall estimates and satellite-derived IR cloud top temperatures collocated in space and time. A power law regression is computed between IR cloud top temperature and radar-derived rainfall estimates at the ground as first suggested by Gagin et al. (1985). Rainfall

FIGURE 4-5. 24-hour cumulated rainfall over the continental US from the Auto Estimator technique (Vicente et al., 1998) for 19, 20 and 22 July, 1998. In the occasion floods in Iowa and northern US happened. (courtesy of NOAA-NESDIS and G. A. Vicente, NASA GSFC, Greenbelt, MD)


34


estimates are adjusted for different moisture regimes using precipitable water and relative humidity fields from the NCEP Eta Model and SSM/I measurements. In this sense the approach is the other way around with respect to physical initialization of numerical models. An application for real time estimations in the Mediterranean basin has been done by Meneguzzo et al. (1998). Calibration is under investigation and a neural network approach has been added. An entire new research field opens up with the launch of new multispectral sensors like the SEVIRI on board MSG or the GOES Imager and are the object of present international projects such as EURAINSAT (Levizzani et al., 2000, 2001a,b). The best possible match of MW-based rainfall retrieval algorithms with the geostationary multispectral and tracking capabilities needs to be ensured. Note that quantitative MW algorithms are those that train the geostationary rapid update cycle. Physical initialization and consistency tests such as that of Panegrossi et al. (1998) appear of the maximum importance together with specialized MSG studies oriented towards the full exploitation of SEVIRI’s spectral signatures. Multispectral cloud top analysis comes again into play and its relationships to MW scattering signatures needs to be closely investigated in order to explain the very large rainfall variability over land and oceans. An open problem concerns the presence within active convection systems (Houze, 1997) of stratiform regions that introduce potential problems in satellite rain estimates. Texture analysis seems a promising way to get around the problem (Lensky and Rosenfeld, 1997; Olson et al, 2001c). Significant changes in rainfall regimes and drop size distributions from convective to stratiform type were documented by Tokay and Short (1996) by examining tropical raindrop spectra. Hong et al. (1999) have used MW emission (19 and 37 GHz) and scattering (85 GHz) for the separation of convective and stratiform precipitation areas over oceans in the tropics: results are still preliminary and studies are conducted using SSM/I and TMI data. The removal of norain clouds in IR imagery has recently been considered using ancillary MW data by Xu et al. (1999b) who used a cloud-patch algorithm to describe cloud top structural properties taking into account the varying patch features of raining and non-raining clouds.

4.2 Passive MW and PR algorithms The prototype of hybrid MW-PR rain algorithms was the TRMM “day-1” or 2B-31 (Haddad et al., 1997), which uses a rain profiling approach that gives equal importance to measurements from TMI and PR. The algorithm is based on estimating the rain profile using PR reflectivities, while constraining the inversion to be consistent with the radiometer-derived estimate of the total attenuation. Data fusion is performed expressing the problem in terms of drop size distribution (DSD) variables. The authors start with an a priori probability density function (PDF) for these variables and use a Bayesian procedure to condition the PDF successively on the radar and radiometer measurements. Viltard et al. (2000) have checked the consistency between TMI and PR by varying the DSD: results suggest that a crude consistency may be achieved if a different DSD is used for the radiometer and the radar. Although they offer no independent validation of their conclusions, the authors demonstrate that rainfall validation need not be confined to surface rainfall measurements, which are only loosely related to the volumetric observations made by most sensors. The issue of spatial resolution of present MW sensor channels cannot be overlooked since the low-resolution channels are sampled at a spatial resolution, which is several times larger than the scale at which rainfall is generated in typical convective rainbands. Turk et al. (1998b) intentionally degraded the native fine scale precipitation measurements from aircraft radiometers and radar to that of current and future MW space borne sensors during TOGA COARE in the western Pacific Ocean in order to verify the impact of sensor resolution upon a combined radiometer-radar vertical profiling rain retrieval algorithm. Retrieved values of the columnar graupel content were more influenced by the addition of the radar profile than was the columnar rain content. The retrieved values of columnar graupel were also significantly smaller than previously published results for land-based rainfall. The results show that the general trend of the rain structure is maintained but fine scale details are lost once the observations are reduced to resolutions of 15 km.



35

FIGURE 4-6. 10 February, 1998, Cyclone Anacelle in the Indian Ocean. Distributions of retrieved LWC from (a) TMI-only (BAY-7) algorithm, (b) PR estimates averaged to TMI reference resolution, (c) calibrated LWC, and (d) 2A12 V.5 algorithm. (from Bauer et al., 2001; courtesy of American Meteorological Society)

Multispectral data from GOES and TRMM PR data were used by Bellerby et al. (2000) to generate rainfall estimates at high spatial and temporal resolution. Coincident PR measurements were matched with four-band GOES image data to form the training dataset for an artificial neural network. Statistical information derived from multiple GOES pixels was matched with each precipitation measurement to incorporate information on cloud texture and rates of change into the estimation process. The ANN was trained for a region of Brazil and used to produce half-hourly precipitation estimates at a spatial resolution of 0.12°. These products were validated using PR and gauge data. Instantaneous precipitation estimates demonstrated correlations of ∼ 0.47 with independent validation data, exceeding those of an optimized GPI method locally calibrated using PR data. A combination of PR and GOES data thus may be used to generate precipitation estimates at high spatial and temporal resolutions with extensive spatial and temporal coverage, independent of any surface instrumentation. A new methodology for the combination of active and passive MW measurements for nearsurface precipitation retrieval from TRMM data was developed by Bauer (2001b) and Bauer et al. (2001) using a stand-alone passive MW algorithm calibrated by collocated PR estimates. The passive MW technique was based on combined cloud model–radiative transfer simulations including varying surface conditions, a melting layer parameterization, and approximate threedimensional radiative transfer (Bauer, 2001a; Bauer et al., 2000) (Fig. 4-6). The authors tested several inversion techniques: multistage regressions and ANN (Tsintikidis et al., 1997) as well as Bayesian estimators (Evans et al., 1995). They found that both Bayesian and ANN


36


techniques perform equally well against PR estimates if all TMI channels were used. However, not using the 85.5 GHz channels produced consistently better results. Generally, regressions performed worse; thus they seem less suited for general application due to the insufficient representation of the non-linearities of the TB-rainrate relation. It is concluded that the databases represent the most sensitive part of rainfall algorithm development. The TMI estimates showed -3 a slight underestimation of rainfall at low rain liquid water contents (< 0.1 g m ) as well as at -3 very high rainfall intensities (> 0.8 g m ) and excellent agreement in between. The biases were found not to depend on beamfilling with a strong correlation to rain liquid water for stratiform clouds that may point to melting layer effects.

4.3 Use of lightning detection Deep cumulonimbus clouds of maritime origin may well extend into the troposphere without being associated with thunderstorms and this is frequently observed in tropical oceanic regions with copious convective rainfall, but infrequent lightning as documented by remote sensing platforms. Observations show that lightning is by far concentrated over continents and islands and very rare over the oceans (Fig. 4-7) (e.g. Boccippio et al., 2000). Since lightning is a manifestation of the growth rate of large ice hydrometeors, a positive correlation is reasonably expected with ice scattering intensity in high frequency MW radiometry, for example the 85.5 GHz channel. The amount of lightning per given rain amount varies greatly with rain regimes. This implies that rainfall algorithms over land, based primarily

FIGURE 4-7. May 1995–Apr 1999. a) Mean Optical Transient Detector (OTD) area rate density. Data are normalized by total sensor view time. No detection efficiency adjustment was applied. b) Climatological map of OTD flashes per area (i.e., regional flash rate density divided by regional area rate density). Again no detection efficiency adjustment was applied. (from Boccippio et al., 2000; courtesy of American Meteorological Society)


37


on ice scattering, do suffer from considerable biases in the various rain regimes. Petersen and Rutledge (1998) have quantified the amount of rainfall per lightning flash for various rainfall 7 regimes. They have observed something like 5.7×10 kg of rainfall per lightning flash in the arid 8 8 SW US, increasing to 1.3×10 kg in the SE humid US, 3.4×10 kg in tropical continental 10 Australia, and as much as 1-2×10 kg over tropical oceans. Ba et al. (1998) were among the first to show that the biases of MW-derived rainfall estimates are very much related to the glaciation temperature, which can thus be used for improving the algorithms’ performances. Their results show that lightning strikes may be used as a proxy (because they are more frequent) of MW scattering signature to study the variability of the different rain regimes. Modeling results illustrate the relationship between lightning and cloud microphysical properties. Relationships between lightning and microphysical parameters such as the maximum updraft velocity at the charging zone boundary and the peak liquid water flux into the charging zone (Solomon and Baker, 1998) and with the ice crystal concentration in the charging zone show a strong dependence of lightning flash rate to ice content, updraft velocity and liquid water, which can be useful in the retrieval of cloud profiles from remote sensing sources. Dietrich et al. (2001) have combined LIS and PR data (specifically those provided from the 2A25 TRMM data product from the Goddard Space Flight Center-Distributed Active Archive Center, GSFC-DAAC) by collocating individual events and flashes with the corresponding PR attenuation corrected reflectivity profile for data between May and September 2000. Convective and stratiform rainfall events are determined from the 2A25 data product. In Fig. 4-8 the vertical profiles have been divided for different event rates. They show changes in cloud structure for different values of event rates. With increasing event rate, the maximum reflectivity tends to increase (somewhat in magnitude but also in the vertical extent of higher reflectivity) for both convective and stratiform clouds while the two regimes remain quite distinct from each other, providing a useful means of quantifying their respective cloud structure. There is a strong

FIGURE 4-8. May-September, 2000. Average profiles of reflectivity divided into different event rates (number of events min-1 km-2) for Central Mediterranean. (from Dietrich et al., 2001; courtesy of EUMETSAT)


38


FIGURE 4-9. 15 Jul, 1997, 0014 UTC. GPROF-based (top), CST-based (middle), and combined IR–lightning-based (bottom) rain estimates. (from Grecu et al., 2000; courtesy of American Meteorological Society)

relationship between microphysical properties and lightning in both convective and stratiform clouds. Although it is well known that there are differences between convective and stratiform regimes, there is a lack of direction when choosing between microphysical profiles using MW retrieval algorithms. The preliminary results illustrate that lightning may help by distinguishing between different microphysical characteristics. Grecu et al. (2000) have examined the combined use of cloud-to-ground lightning and satellite IR data for rainfall estimation. Based on the analysis of the correlation between satellite MW and IR rainfall estimates and on the number of strikes in ‘‘contiguous’’ areas with lightning, where the contiguity is defined as a function of the distance between strikes, the authors developed an empirical algorithm for convective rainfall estimation. The rainfall in areas not associated with lightning is determined using a modified version of the IR-based CST algorithm (Adler and Negri, 1988; Anagnostou et al. 1999). The combined lightning and CST technique is evaluated based on 15 days of data in July 1997 provided by geostationary and polar-orbiting satellites and the US National Lightning Detection Network (an example is shown in Fig. 4-9). The general conclusion is that lightning data contain useful information for IR rainfall estimation. Results show a reduction of about 15% in the rms error of the estimates of rain volumes defined by convective areas associated with lightning. It is shown that the benefit of using lightning information extends to the whole rain domain, since the error caused by missing convective areas because of the absence of lightning is smaller than that caused by overestimating the convective rain areas because of cirrus that obscure underlying convective storms when only satellite IR data are used.


39


5 5

APPLICATIONS AND

FUTURE PERSPECTIVES ____________________________________________________________________________

Local, regional and global satellite rainfall data are of the maximum value for a wealth of different applications that can draw upon the repetition frequency and coverage of weather satellites. Rainfall data are essential for NWP, nowcasting, agrometeorology, transportation, traffic control, communications, and, last but not least, leisure and the everyday life. It might seem rather frivolous but this latter is the most promising market for tailored weather forecasts: days are coming when each one of us will have his/her own 24-h link to find out how the weather is going to be in the next 15-30 min or so! Rainfall information, it goes without saying, are on top of economically valuable weather reports. On the opposite side of the time-space scale are climate-related weather phenomena. Recent results show that precipitation is by far the most prominent parameter upon which man is exerting a substantial influence causing global changes that will affect the global water cycle for the centuries to come (Ramanathan et al. 2001). Therefore, climatological monitoring of precipitation is a compelling necessity for our survival. It is very difficult to give an exhaustive account of applications of satellite rainfall data: we will therefore limit ourselves to examine applications of high image repetition and MW synergies (5.1), initialization of NWP models (5.2), climate (5.3). In chapter 5.4 we will finally catch a glimpse of the upcoming future with the Global Precipitation Mission (GPM).

5.1 High image repetition rate and MW synergy Unequivocal needs for rapid updates of rainfall estimates over land and ocean exist for quantitative precipitation forecasting, numerical weather prediction, hydrology, and Earth-space Ka-band communications (Turk et al., 1999). Golding et al. (2001) have set a list of requirements for precipitation estimates for nowcasting and Very Short Range Forecasting (VSRF) in view of the MSG follow-on program in the 2015-2025 time span. In summary: Reliable detection (10 km, 1 h) for general forecasting – to remove current problems with radar spurious echoes and missed light rain at long range. -1 Accurate high rates (within 10 mm h at 2 km, 15 min) for flooding – to improve on radar accuracy. MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

40


-1

Information over the sea (within 5 mm h at 10 km, 1 h) for general forecasting – to provide upwind information beyond radar range. -1 Snow detection (within 0.2 mm h at 10 km, 1 h) for transport warnings. Large hail detection (10 km, 15 min) for protection of life and property. Freezing rain detection (10 km, 1 h) for protection of life. Satellites are the key platform to maintain routine observations along coastlines and over the oceans. Existing SSM/I radiometers on board the DMSP satellites and the TMI on board TRMM allow for a limited coverage of precipitation systems along polar or tropical orbits. Rapidly developing cyclones in midlatitudes are characterized by large cloud shields and abundant condensation, which is captured by IR and VIS satellite imagery. Using passive MW data the rapidly developing cyclones can be quantitatively characterized in terms of integrated water vapor and precipitation intensity. McMurdie and Katsaros (1996) have examined fields of integrated water vapor, water vapor anomaly (that is observed water vapor content minus the monthly mean at the particular location), and precipitation intensity from the SSM/I for 12 North Atlantic rapidly deepening cyclones and 11 North Atlantic non-rapidly deepening cyclones during the winter in 1988 and 1989. Maximum water vapor anomaly and average precipitation index have correlations with the concurrent 6-h deepening rates of 0.56 and 0.55, respectively. Removing outliers, the correlation improves to 0.68 and 0.70, respectively. Results indicate that, although most rapidly deepening cyclones have high water vapor anomaly and stronger precipitation index than non-rapidly deepening cyclones, there are storms that deepen rapidly in absence of high water vapor anomaly or heavy precipitation, or the contrary is true. Results were obtained using the presently available low repetition rate precipitation estimates, while an increase of the observational repetition rate would indeed be beneficial. The Auto-Estimator technique (Vicente et al., 1998) has been used experimentally for almost 6 years to provide real-time instantaneous rainfall rate estimates, average hourly estimates, and 3-, 6-, and 24-h accumulations over the conterminous 48 US and nearby ocean areas. The NOAA/NESDIS Satellite Analyses Branch (SAB) has examined the accuracy of the rainfall estimates daily for a variety of storm systems. They have determined that the algorithm produces useful 1–6-h estimates for flash flood monitoring but exaggerates the area of precipitation causing overestimation of 24-h rainfall total associated with slow-moving, coldtopped mesoscale convective systems (MCS). The SAB analyses have also shown a tendency for underestimation of rainfall rates in warm-top stratiform cloud systems. Until further improvements, the use of this technique for stratiform events should be considered with caution. Validation of the hourly rainfall rates of the auto-estimator using gauge-adjusted radar precipitation products (with radar bias removed) is conducted. Results show that the autoestimator has modest skill at 1-h time resolution for a spatial resolution of 12 km. Results 2 improve with larger grid sizes (48 × 48 km or larger). Applications to devastating floods (Robinson and Scofield, 1994; Scofield and Naimeng, 1994) and flash floods (Vicente and Scofield, 1996) have been investigated. One more application to nowcasting was tempted by Porcù et al. (1999) in the context of a realtime flood warning scheme. The NAW technique (Negri et al., 1984) is used together with radar data as a calibration source. The radar calibration consistently reduces both bias and variance 2 of original NAW estimates even for an integration area as small as 2000 km indicating applicability for nowcasting over medium-sized river basins. Unequivocal needs for rapid updates of rainfall estimates over land and ocean exist for quantitative precipitation forecasting, numerical weather prediction, hydrology, and Earth-space Ka-band communications (Turk et al., 1999). Satellites are the key platform to maintain routine observations along coastlines and over the oceans. Existing SSM/I radiometers on board the DMSP satellites and the TMI on board TRMM allow for a limited coverage of precipitation systems along polar or tropical orbits. Over the last few years there has been an increasing request of large-bandwidth information services coupled with high availability and low-fade margin communication systems (Watson MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

41


and Hu, 1994). This scenario has prompted for exploring channel frequencies at Ka band and above and developing sophisticated countermeasure techniques to mitigate outage periods (Jones and Watson, 1993). The implementation of most advanced adaptive countermeasure techniques is related to the possibility of monitoring in quasi real-time the beacon attenuation in a given region and period. Spatial and frequency diversity methods, power link control, data rate and error correction on the downlink/uplink can be effectively adopted only if the propagation conditions are known in real time. Exploitation of remote sensors and their products represents a natural way to optimize the performances of a satellite communication system with low power margins, specifically while applying fade mitigation techniques. MW signatures of precipitation, as given by a space-borne multi-frequency radiometer, have been shown to be the base for estimating the path attenuation in K-band satellite communications (Marzano et al., 2000). A general approach should attempt to estimate rainfall intensity and attenuation by polar-orbiting MW radiometers and temporally track the rain system by means of geostationary IR radiometers. A statistical approach can be used to derive a prediction model of path attenuation from MW TB and surface rainrate (Crone et al., 1996). Surface rainrate is generally related to total path attenuation by means of a powerlaw relation, adopting a suitable cloud-radiation database

A f = αR β

(5.1)

where f indicates for example the channel frequency of Olympus and Italsat telecommunication satellite beacons and α and β are coefficients depending on channel frequency, observation angle and precipitation type (e.g. from nimbostratus or cumulonimbus). The rainfall contribution to the total path attenuation is being separated from the contribution of the entire cloud. Indeed, total attenuation as well as rainfall rate can be directly estimated from TB measurements. Assuming a quadratic relationship between Af and TB we can write the follows equation

(

A f = a0 + ∑ f a1 f TBfp + a 2 f TBf2

)

(5.2)

FIGURE 5-1. Attenuation case study, 6 Oct., 1998. Upper left: METEOSAT IR image, 0600 UTC. Upper right: 85GHz image, 0540 UTC. Lower left: 24-h accumulated rain (mm). Lower right: 20-GHz attenuation (dB) map, 0600 UTC. (from Levizzani et al., 2001b; courtesy of EUMETSAT)


42


(

R = b0 + ∑ f b1 f TBfp + b2 f TBf2

)

(5.3)

The coefficient set is different for over-land and over-ocean estimates. Since the land surface model does not include polarization, the average between the two orthogonally-polarized TB’s is used for all four SSM/I frequencies over land. Over the ocean all seven polarized channels are employed. An example is given in Fig. 5-1.

5.2 Physical initialization of weather prediction models The initialization of numerical models through sparse surface and upper-air observations has always represented a problem due to the inadequate description of the initial state of the atmosphere by the scattered observations of the synoptic stations. The description of the fourdimensional atmospheric continuum lacks both space and time resolution over land and is very deficient over the oceans. The assimilation of satellite data into numerical models is now operationally done via variational assimilations of clear-sky radiances (e.g. Anderson et al., 1994). Simulations and model runs (e.g. Matricardi et al., 1996) indicate that model output fields are improved by these assimilation schemes also allowing for dynamical information to be extracted. On the other hand, initialization of numerical models with physical variables that are not primary variables of the model is very complex and can produce instabilities and poor model performances depending on many factors and the type of model. Early works of assimilation of satellite-derived precipitation for improving tropical forecasts were conducted by Mathur et al. (1992). Mathur (1995) continued on trying to ameliorate convective and non-convective condensation with the assimilation of rainfall for improving initial conditions. Manobianco et al. (1994) have dynamically assimilated satellite-derived precipitation into a regional scale model by scaling the internally generated model profiles of latent heating for the simulation of tropical cyclones. At points where the model does not produce precipitation the vertical distribution of total latent heating resulting from satellite precipitation was specified from instantaneous model-based profiles at adjacent points using a search algorithm. Simulations showed that a) satellite precipitation does not induce noise during or after the assimilation period, b) it forces the model to reproduce the magnitude and distribution of satellite rainfall patterns, and (c) it improves the simulated mean sea level pressure (MSLP) minima, frontal positions and the low-level vertical-motion patterns. The model retains information from the assimilation up to 8 hours after the end of the assimilation itself. SSM/I-retrieved rainfall rates were assimilated into a limited area model (LAM) by Peng and Chang (1996) for the same reasons. The assimilation was done via a reversed cumulus parameterization scheme, that is adjusting the moisture field in the model so that the model is forced to generate the SSM/Iobserved rainfall rates. For the test cases the assimilation reduced the average 48-hour forecast distance error from 239 km in the control runs down to 81 km in the assimilation experiments. The assimilation of satellite rainrates during the initialization of the numerical simulation of an extra-tropical cyclone was conducted by Alexander et al. (1999). Data were used from passive MW, IR and lightning detectors to produce a continuous time series of rainrates suitable to be assimilated into the mesoscale model. Also in this extra-tropical case a significant improvement in the forecasts of precipitation patterns, MSLP fields, and geopotential height fields was found. A method to assimilate observed rain rates in the Tropics for improving initial fields in forecast models was proposed by Falcovich et al. (2000). It consists of a 6-h integration of a numerical forecast model; the specific humidity at every time step at each grid point is modified (nudged) in such a way that the total model precipitation accumulated during this integration becomes very close to the observed one. An increase in the model precipitation is achieved by moistening the lower troposphere above a grid point with prescribed supersaturation; a decrease in the model rainfall is brought about by decreasing the specific humidity in the lower MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002


43

troposphere in proportion to the difference between the model and reference specific humidity profiles. The modified values depend on the difference between the model and target precipitation. The depth of the atmospheric column in which the humidity is changed is proportional to the target rain rate. Physical initialization was used by Krishnamurti et al. (1994) to produce, in a diagnostic sense, a thermodynamic consistency between the humidity variable, the surface fluxes, rainfall distributions, diabatic heating and clouds. It modifies the initial state variables via an incorporation of tropical rainrates. The correlation between the observed rainfall and the modelbased rainfall is significantly improved. The initialization is completed by the assimilation of raingauge records over land. A marked improvement in the nowcasting and one day forecast skills for tropical rainfall was noted. Krishnamurti et al. (2001) address real-time precipitation forecasts from the Florida State University (FSU) superensemble forecast model, a multianalysis–multimodel superensemble. A multianalysis component based on the FSU global spectral model that utilizes TRMM and SSM/I datasets and a number of rain-rate algorithms is also included. The difference in the analysis arises from the use of these rain rates within a physical initialization that produces distinct differences among these components in the divergence, heating, moisture, and rain-rate descriptions. A total of 11 models, of which 5 represent global operational models and 6 represent multianalysis forecasts from the FSU model initialized by different rain-rate algorithms, are included in the multianalysis–multimodel system. ‘‘Multimodel’’ refers to different models whose forecasts are being assimilated for the construction of the superensemble and ‘‘multianalysis’’ to different initial analysis contributing to forecasts from the same model. The term superensemble is being used to denote the biascorrected forecasts based on the products derived from the multimodel and the multianalysis. The training period is covered by nearly 120 forecast experiments prior to 1 January 2000 for each of the multimodels. These are all 3-day forecasts. The statistical bias of the models is determined from multiple linear regression of these forecasts against a ‘‘best’’ rainfall analysis field that is based on TRMM and SSM/I datasets and using the rain-rate algorithms recently developed at NASA GSFC. The main results of this study are that the multianalysis–multimodel superensemble has a much higher skill than the participating member models. The skill of this system is higher than those of the ensemble mean that assigns a weight of 1.0 to all including the poorer models and the ensemble mean of bias-removed individual models. The selective weights for the entire multianalysis–multimodel superensemble forecast system make it superior to individual models and the above mean representations. The skill of precipitation forecasts is addressed in several ways. The skill of the superensemble-based rain rates is shown to be higher than the following: a) individual model’s skills with and without physical initialization, b) skill of the ensemble mean, and c) skill of the ensemble mean of individually bias-removed models. Rainfall 1-, 2- and 3-day forecasts from the superensemble and improvements of the skills over member models are shown in Fig. 5-2 and 5-3, respectively. A multivariate optimal interpolation analysis was applied by Turk et al. (1997, 2000b) to the physical initialization of the Naval Operational Global Atmospheric Prediction System (NOGAPS) for the nowcasting of tropical precipitation. SSM/I-derived rainrates were incorporated into files at 6-hour intervals and the initialization was done during the integration to produce the 6-hour forecast (first guess). SSM/I oceanic rainrates were also used to calibrate a 30-km resolution geostationary-based rainrate retrieval algorithm and the results assimilated as well. Added accuracy, better location and intensity of convective precipitation were found to lead to an improvement in the NOGAPS assimilation rainrates as verified against satellite observations. Turk et al. (1999), in particular, have shown a positive impact in the 24- and 36hour forecast positioning of hurricane Georges (28 September 1998) by 12 to 16%, while virtually no impact (3.5%) was found at the 72-hour forecast level. At the European Centre for Medium-Range Weather Forecasts (ECMWF) a one dimensional variational (1D-Var) was applied to retrieve simultaneously the specific humidity profile, the sea surface wind speed (SSWS) and the LWP from SSM/I TB over oceans in non-rainy areas (Phalippou, 1996). These parameters are derived using optimal estimation theory and are therefore the best set of parameters that explains the observed TB while being consistent with the available a priori information provided by a short-term model forecast. Total column water


44


vapor (TCWV) and SSWS retrievals from the 1D-Var have been assimilated operationally in the ECMWF four-dimensional variational (4-D Var) analysis system since June 1998 and October 1999, respectively. Gérard and Saunders (1999) showed that the impact of the 1D-Var TCWV

FIGURE 5-2. 6 June 2000. Days 1, 2, and 3 of forecast rain (mm day-1) from the superensemble forecast is compared with the observed rainfall estimate from the TMI-2A12 and SSM/I-GPROF algorithm. (from Krishnamurti et al., 2001; courtesy of American Meteorological Society)

FIGURE 5-3. Percentage improvement (based on correlation) of the superensemble forecasts over the ensemble mean, the best, and the poorest models. (from Krishnamurti et al., 2001; courtesy of American Meteorological Society)



45

assimilation on forecast scores is positive or neutral depending on the season and on the region. Marécal and Mahfouf (2000) have successively examined the performance of a 1-D Var assimilation of TMI-derived surface rainfall rates in the ECMWF model. Temperature and specific humidity profiles are retrieved that are consistent with both observed and model shortrange forecast rain rates. Atmospheric situations encompassing tropical cyclones, frontal bands, and mesoscale convective systems were considered. Results show that 1D-Var is generally able to find modified profiles within the range of forecast errors (specified from the operational ECMWF statistics) that provide a precipitation field close to observations. Increments of temperature with respect to the background state are small indicating that 1D-Var essentially adjusts specific humidity to modify precipitation amounts. Observation errors are increased from 25% of the observed value to 50%. It is shown that 1D-Var is still able to assimilate some information from observations. Marécal et al. (2000) have studied the impact on analysis and forecasts of using TMI TB instead of the SSM/I ones in the ECMWF 4D-Var assimilation system.

Chang et al. (2001) have evaluated the impact of several newly available sources of meteorological data on mesoscale model forecasts of the extratropical cyclone that struck Florida on 2 February 1998 (the Groundhog day storm). Intermittent measurements of precipitation and integrated water vapor (IWV) distributions were obtained from the SSM/I and TMI observations. The TMI also provided sea surface temperature (SST) with structural detail of the Loop Current and Gulf Stream. Continuous lightning distributions were measured with a network of very low frequency radio receivers. Lightning data were tuned with intermittent spaceborne MW radiometer data through a probability matching technique to continuously estimate convective rainfall rates. A series of experiments were undertaken to evaluate the effect of those data on mesoscale model forecasts produced after assimilating processed rainfall and IWV for 6 h. Assimilating processed rainfall, IWV, and SSTs from TMI measurements in the model yielded improved forecasts of precipitation distributions and vertical motion fields. Assimilating those data also produced an improved 9-h forecast of the radar reflectivity cross section that was validated with a coincident observation from the TRMM PR. Sensitivity experiments showed that processed rainfall information had greater impact on the rainfall forecast than IWV and SST information. Assimilating latent heating in the correct location of the forecast model was found to be more important than an accurate determination of the rainfall intensity. A statistical objective analysis (SOA) scheme was developed by Pereira Fo. et al. (1999a) to adjust estimates of rainfall accumulation from the WSR-88D radar in central Oklahoma using rain gauge measurements from the Oklahoma mesonetwork. Their results were applied in the context of a hydrometeorological forecast system that uses the analyzed rainfall field to adjust rainfall rates for nowcasting (0 – 3 h), to improve rainfall forecasts (0 – 12 h) via its assimilation into mesoscale models, and to verify the accuracy of these rainfall forecasts. New schemes to analyze precipitation and to adjust radar rainfall rates were proposed (Pereira Fo. et al., 1999b) to improve the quantitative precipitation forecast (QPF) for hydrologic purposes. Adjusted WSR88D rainfall rates were advected by the 700-mb wind field to produce an extrapolation QPF. Several experiments were conducted to evaluate the effect of the rainfall adjustment and wind field upon the extrapolation QPF. In parallel to precipitation assimilation, cloud physics parameters such as the above described effective radius were introduced into a fast radiation scheme of the High Resolution Limited Area Model (HIRLAM) by Wyser et al. (1999). While the authors derive re from the cloud condensate content and temperature of the model, satellite retrievals can be used instead. The parameterization mainly affects the short-wave radiation of thin clouds and its effects seem small on the synoptic scale, whereas local significant changes are found in the model’s 2 m temperature.


46


5.3 Climate Long time series of fine-scale observation-based global precipitation are needed to support a variety of studies, including global change, surface hydrology, and numerical weather and climate model initialization and validation. WCRP established the GPCP with the initial goal of producing precipitation estimates on a monthly 2.5º × 2.5º lat/lon grid (WCRP, 1986) and the goal is well reached. However, because of the lack of finer-scale precipitation data, numerous applications cannot use satellite data: for example the validation of streamflow models by forcing them with observed data that resolve individual storms and catchments. One more problem concerns monthly scale events that are difficult to study with calendar month-based datasets (e.g. extra tropical blockings). Once again the fundamental obstacle to finer-scale global estimates is the lack of accurate, dense global data, either from in situ or remote sensing instruments. Such datasets are still way ahead especially when the accuracy is at stake. However, some attempts have been made such as the one of Janowiak et al. (2001) that have set up a real-time, global, half-hourly general purpose IR dataset from geostationary sensors tackling problems such as intercalibration, and corrections for parallax and viewing angles. The dataset has already been used to produce precipitation and an example is given in Fig. 5-4.

FIGURE 5-4. Examples of estimated precipitation from various sources at 0.5° × 0.5° grid resolution: (top left) MW observations only (Ferraro 1997); (bottom left) MIRRA, (Miller et al. 2000); (top right) the GPI (IR-only); (bottom right) rain gauge analysis. All estimates are valid for the 24-h period beginning at 5 Mar 2000, 0600 UTC. Units are mm. Inset in rain gauge analysis shows distribution of rain gauge sites. (from Janowiak et al., 2001; courtesy of American Meteorological Society)



47

The issue of building a dataset for climatological purposes is by no means trivial. Gruber et al. (2000) have compared two different datasets, the GPCP and the Climate Prediction Center Merged Analysis of Precipitation (CMAP) of the US National Weather Service (NWS). Both datasets blend together satellite and gauge estimates of precipitation. Despite good spatial and temporal correlations between the two fields some of the observed differences were significant at the 95% level. The reasons were traced to the use of some different input data such as the use by CMAP of atoll gauges in the tropical Pacific Ocean and gauges which were uncorrected for wetting evaporation and aerodynamic effects. The former impacts the tropical oceanic amounts while the latter is particularly evident in the Northern Hemisphere. Nine years (1986–94) of tropical and subtropical precipitation estimates based on the GPI were examined by Joyce and Arkin (1997). The GPI, based on the results of studies relating fractional coverage of cold cloud to convective rainfall, uses IR observations gathered by geostationary and polar-orbiting satellites (Arkin and Meisner, 1987; Arkin et al., 1994). Longitudinal discontinuities in mean GPI coincident with the boundaries of satellite coverage led to a comparison of GPI derived from each geostationary satellite in overlap regions. This study revealed both inter-satellite calibration differences and satellite zenith angle dependence. The goals were to remove these sources of systematic error within the GPI, investigate the climatology of the corrected GPI, and compare against other estimated rainfall datasets. To correct calibration differences, GPCP geostationary satellite IR data were standardized to one satellite by temperature adjustments deduced by the International Satellite Cloud Climatology Project (ISCCP). The resulting GPI values were corrected for zenith angle dependence based on a comparison between GOES-7 and METEOSAT-3 that found a systematic increase in GPI

FIGURE 5-5. 1DD images for (top) 1 and 2 January, 1998 (bottom) in mm day-1. (from Huffman et al., 2001; courtesy of American Meteorological Society)


48


of 9% for every 10° of zenith angle beyond 25°. The corrections remove noticeable -1 discontinuities in time-averaged GPI and are largest (> 2 mm day ) over the eastern Indian Ocean, the equatorial Pacific near the date line, and South America. The spatial correlation between corrected GPI and rainfall derived from rain gauges is greater than 0.8 in tropical regions with adequate gauge density. Huffman et al. (2001) have conceived a 1 Degree Daily (1DD) technique for producing globally complete daily estimates of precipitation on a 1º × 1º lat/lon grid from currently available observational data, the so-called Threshold-Matched Precipitation Index (TMPI). The dataset is available within the 40ºS - 40ºN latitude range: 3-hourly IR TB values are compared with a threshold and all “cold” pixels are assigned a single precipitation rate. The threshold and conditional rainrate are set locally by month from SSM/I precipitation frequency (Kummerow et al., 1996) and the GPCP satellite-gauge combined monthly precipitation frequency (Huffman et al., 1997), respectively. The 1DD at higher latitudes features a rescaled precipitation from the Television and Infrared Observation Satellite Operational Vertical Sounder (TOVS). Data have been produced from 1997 through 1999 with production continuing a few months behind real time (to allow access to monthly input data). An example is given in Fig. 5-5. The GPCP datasets have already allowed for the identification of large scale phenomena such as the discrepancy between gauges and satellite estimates in equatorial Africa (McCollum et al., 2000). GPCP estimates have approximately twice the magnitude of estimates produced from the rain gauges used by the GPCP in central equatorial Africa. Different possible explanations were identified and investigated by the authors. The first is that there may not be enough GPCP rain gauges in the area to provide accurate estimates of rainfall for comparisons with satellite estimates. A comparison of the time-averaged GPCP rain gauge estimate with a long-term (over 40 yr) climatology indicated that the GPCP gauge estimates are similar to long-term rainfall averages, suggesting that the GPCP rain gauge analysis is not underestimating rainfall. Two other possible explanations related to the physical properties of the air masses in this region were studied. Evidence from the literature and from estimates of the effective radii of cloud droplets suggests that there may be an abundance of aerosols in central Africa, resulting in an abundance of cloud condensation nuclei, small drops, and inefficient rain processes. The second explanation is that convective clouds forming under dry conditions generally have cloud bases considerably higher than those of clouds forming in moist environments. This leads to an increase in the evaporation rate of the falling rain, resulting in less precipitation reaching the ground. Analysis of the moisture distributions from both the NCEP NWP model reanalysis data and the NASA’s Water Vapor Project global moisture dataset reveals that the lower troposphere in this region of Africa is relatively dry, which suggests that cloud bases are high. Laing et al. (1999) studied mesoscale convective complexes (MCC) in Africa deriving a relationship between SSM/I-derived precipitation characteristics and METEOSAT IR data. Results show that these systems have characteristics, such as rain area and volume, which are of the same order of magnitude as those of systems in the US. In addition they provide a significant fraction of the rainfall in Sahelian Africa. Rodgers et al. (2000, 2001) have conducted studies on tropical cyclone rainfall climatology in order to identify the contribution of tropical cyclones to North Atlantic and North Pacific rainfall climatology as observed from satellites. They used a 11-yr period of MW data within a 444-km radius of the center of those North Pacific and North Atlantic tropical cyclones that reached storm stage and greater. The main results for North Pacific indicate that 1) tropical cyclones contribute 7% of the rainfall to the entire domain of the North Pacific during the tropical cyclone season and 12%, 3%, and 4% when the study area is limited to, respectively, the western, central, and eastern third of the ocean; 2) the maximum tropical cyclone rainfall is poleward (5º – 10º latitude depending on longitude) of the maximum non-tropical cyclone rainfall; 3) tropical cyclones contribute a maximum of 30% northeast of the Philippine Islands and 40% off the lower Baja California coast; 4) in the western North Pacific, the tropical cyclone rainfall lags the total rainfall by approximately two months and shows seasonal latitudinal variation following the Intertropical Convergence Zone (ITCZ); and 5) in general, tropical cyclone rainfall is enhanced during the El Niño years by warm SSTs in the eastern North Pacific and by the MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002


49

monsoon trough in the western and central North Pacific. The main results for North Atlantic indicate 1) that tropical cyclones contribute, respectively, 4%, 3%, and 4% to the western, eastern, and entire North Atlantic; 2) similar to that observed in the North Pacific, the maximum in North Atlantic tropical cyclone rainfall is approximately 5º – 10º poleward (depending on longitude) of the maximum non-tropical cyclone rainfall; 3) tropical cyclones contribute regionally a maximum of 30% of the total rainfall northeast of Puerto Rico, within a region near 15ºN, 55ºW, and off the west coast of Africa; 4) there is no lag between the months with maximum tropical cyclone rainfall and non-tropical cyclone rainfall in the western North Atlantic, whereas in the eastern North Atlantic, maximum tropical cyclone rainfall precedes maximum non-tropical cyclone rainfall; 5) like the North Pacific, North Atlantic tropical cyclones of hurricane intensity generate the greatest amount of rainfall in the higher latitudes; and 6) warm El Niño –Southern Oscillation (ENSO) events inhibit tropical cyclone rainfall. Adler et al. (1993) estimated monthly rainfall over Japan from the SSM/I and geostationary IR data from the GMS. Their results show that in the areas where the MW technique performs well (small bias) the combined MW-IR monthly total estimates have better error statistics than either the MW or IR techniques individually. Berg and Avery (1995) evaluated monthly rainfall estimates derived from the SSM/I over the tropical Pacific. Eight years of MW data from the SSM/I were analyzed by Ferraro (1997) to depict seasonal, annual, and interannual variability in global rainfall. The primary algorithm is a 85 GHz scattering-based algorithm over land, and a combined 85 GHz scattering and 19/37 GHz emission over the ocean, both of which calibrated with ground-based radar data. Rainfall climatologies derived from both methods were shown to describe annual and interannual rainfall variability over the globe. The interannual differences were shown to exhibit the same spatial fields and magnitudes. Moreover, the correlation between the two techniques was in excess of 0.9 when comparing the annual mean global rainfall fields, with the poorest results occurring at high latitudes. Major differences were found between the two algorithms over the oceans as well as a lack of sensitivity to light rain of the 37 GHz algorithm. A complete time series for the same period (1987-1994) was also generated including, rainfall, clouds, water vapor, snow cover and sea ice by Ferraro et al. (1996).

5.4 Future perspectives – The GPM NASA and NASDA have recently approved a new joint mission, called the Global Precipitation Mission, aiming at measuring precipitation on a global basis with sufficient quality, Earth coverage, and sampling to improve prediction of the Earth’s climate, weather, and specific components of the global water cycle. GPM will consist of two components (Fig. 5-6): an improved TRMM-like satellite (called core) inclined at about 65°, and a constellation of several (up to eight) drone satellites in sunsynchronous orbits, carrying passive MW radiometers to provide global rainfall coverage at 3-h intervals (depending on latitude). The mission, which should be launched in 2007, is in fact a multinational satellite project for NASA and NASDA will provide the core satellite and two drones, but are envisioning international cooperation for other drones. The objectives of the mission will be accomplished by making substantive improvements - with respect with the present-day combined TRMM and SSM/I observations - in global precipitation observations: specifically, improvements in measurement accuracy, precision, sampling frequency, spatial coverage, and spatial resolution. Thus, the main requirements of the mission can be summarized as follows: Global spatial coverage with a rainfall accuracy estimation better than that available by using current SSM/I radiometers aboard DMSP satellites. Temporal sampling less or equal to 3 hours in order to improve NWP models, data assimilation models and hydrological models and, possibly, to help flash flood forecast.


50


Lifetime sufficiently long in order to monitor and understand potential long-term changes in the global distribution and frequency of precipitation and related latent heating.

In addition, efficient down-link data transmission is foreseen in order to provide flash flood forecast centers with near-real time products. The mainstream scientific objectives of GPM are the following: For Climate: to accurately measure the global-regional variability of rainfall, relate those variations to concomitant variations in global-regional temperature, detect the presence or absence of a speculated acceleration in the global water cycle due to global temperature change, and improve global climate datasets and climate prediction through data assimilation of global rainfall measurements into global climate models (i.e., global climate re-analyses and simulation experiments). For Weather: to improve the accuracy of global and regional NWP models through data assimilation of precipitation measurements, with emphasis on improving predictability of hurricanes and severe local storms, and verification of such models with globally continuous and consistent rainfall measurements. For Global Water Cycle: to improve the understanding and predictability of relevant components of the Earth’s water cycle – which includes water in the atmosphere, within and on the land surface, in the oceans, and in the cryosphere – by achieving substantive accuracy improvement in basin-scale water balance across the relevant space-time scales, with particular emphasis on improving the prediction of damaging floods and the availability of freshwater resources.

FIGURE 5-6. The Global Precipitation Mission (GPM) concept. (courtesy of NASA)


51


6 6

GLOSSARY

____________________________________________________________________________

6.1 Acronyms AIP AGPI AMSU ANN ASI ATSR AVHRR CDS CERES CGMS CMAP CST DAAC DMSP ECMWF ECST ENSO ERS ESA EUMETSAT FACE FSU GARP GATE GEO GEWEX GMS GOES GPCP GPI

Algorithm Intercomparison Project Adjusted GOES Precipitation Index Advanced Microwave Sounding Unit Artificial Neural Network Agenzia Spaziale Italiana (Italian Space Agency) Along Track Scanning Radiometer Advanced Very High Resolution Radiometer Climatological Data Set Clouds and Earth’s Radiant Energy System Coordination Group for Meteorological Satellites Climate prediction center Merged Analysis Precipitation Convective-Stratiform Technique Distributed Active Archive Center Defense Meteorological Satellite Program European Centre for Medium-Range Weather Forecasts Enhanced Convective-Stratiform Technique El Niño-Southern Oscillation European Remote sensing Satellite European Space Agency European Organization for the Exploitation of Meteorological Satellites Florida Area Cumulus Experiment Florida State University Global Atmosphere Research Programme GARP Atlantic Tropical Experiment Geostationary Global Energy and Water cycle Experiment Geostationary Meteorological Satellite Geostationary Operational Environmental Satellite Global Precipitation Climatology Project GOES Precipitation Index


52

GPM GPROF GSFC HIRLAM IR IPWG ISCCP ITCZ JMA LAM LEO LIS LUT MCC MCS MIRA MIRRA MODIS MSG MW NASA NASDA NCEP NESDIS NIR NOAA NOGAPS NWP NWS OE OTD PIP PERSIANN PR QPF RAMS SAB SAF SEVIRI SMHI SOA SSM/I SSMI/S SSM/T TMI TMPI TOGA COARE TOVS TRMM UAGPI UWRAMS VIRS VIS VSRF WCRP WGDM


Global Precipitation Mission Goddard Profiling Algorithm Goddard Space Flight Center High Resolution Limited Area Model Infrared International Precipitation Working Group International Satellite Cloud Climatology Project Inter Tropical Convergence Zone Japan Meteorological Agency Limited Area Model Low Earth Orbit Lightning Imaging Sensor Look Up Table Mesoscale Convective Complex Mesoscale Convective System MW IR Rainfall Algorithm Microwave/Infrared Rain Rate Algorithm MODerate resolution Imaging Spectroradiometer METEOSAT Second Generation Microwave National Aeronautics and Space Administration National Space Development Agency National Center for Environmental Predictions National Environmental Satellite Data and Information Service Near IR National Oceanic and Atmospheric Administration Naval Operational Global Atmospheric Prediction System Numerical Weather Prediction US National Weather Service Optimal Estimation Optical Transient Detector Precipitation Intercomparison Project Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Precipitation Radar Quantitative Precipitation Forecast Regional Atmospheric Modeling System Satellite Analyses Branch (NOAA/NESDIS) EUMETSAT’s Satellite Application Facility Spinning Enhanced Visible and Infrared Imager Swedish Meteorological and Hydrological Institute Statistical Objective Analysis Special Sensor Microwave/Imager Special Sensor Microwave Imager/Sounder Special Sensor Microwave/Temperature sounder TRMM Microwave Imager Threshold-Matched Precipitation Index Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment Television and infrared Observation Satellite Operational Vertical Sounder Tropical Rainfall Measuring Mission Universally Adjusted GOES Precipitation Index University of Wisconsin Regional Atmospheric Modeling System Visible and InfraRed Scanner Visible Very Short Range weather Forecast World Climate Research Program Working Group on Data Management of the GPCP


53


WMO WV 1DD 1-D Var 4-D Var

World Meteorological Organization Water Vapor 1 Degree Daily technique One-dimensional variational assimilation Four-dimensional variational assimilation

6.2 Physical and mathematical quantities A BTD CCN DSD FAR FOV IN IWV LWC LWP MSLP n(r) PCT PDF POD re rms SS SST SSWS VIHC TB TCWV Z ρw σFOV τ

Radar attenuation Brightness Temperature Difference Cloud Condensation Nuclei Drop Size Distribution False Alarm Ratio Field Of View Ice Nuclei Integrated Water Vapor Liquid Water Content Liquid Water Path Mean Sea Level Pressure Particle size distribution Polarization Corrected Temperature Probability Density Function Probability Of Detection Cloud particle effective radius Root mean square Skill Score Sea Surface Temperature Sea Surface Wind Speed Vertically Integrated Hydrometeor Content Brightness temperature Total Column Water Vapor Radar reflectivity factor Density of water Spatial inhomogeneity of rainfall within the satellite field of view Cloud optical thickness


54




55


7 7

REFERENCES

____________________________________________________________________________

Adler, R. F., and R. A. Mack, 1984: Thunderstorm cloud height-rainfall rate relations for use with satellite rainfall estimation techniques. J. Clim. Appl. Meteorol., 23, 280-296. Adler, R. F., and A. J. Negri, 1988: A satellite infrared technique to estimate tropical convective and stratiform rainfall. J. Appl. Meteorol., 27, 30-51. Adler, R. F., A. J. Negri, and I. M. Hakkarinen, 1991: Rain estimation from combining geosynchronous infrared and low-orbit microwave data. Paleogeogr. Paleoclimatol. Paleoecol., 90, 87–92. Adler, R. F., A. J. Negri, P. R. Keehn, and I. M. Hakkarinen, 1993: Estimation of monthly rainfall over Japan and surrounding waters from a combination of low-orbit microwave and geosynchronous IR data. J. Appl. Meteorol., 32, 335-356. Adler, R. F., G. J. Huffman, and P. R. Keehn, 1994: Global tropical rain estimates from microwave-adjusted geosynchronous IR data. Remote Sens. Rev., 11, 125-152. Alexander, G. D., J. A. Weinman, V. M. Karyampudi, W. S. Olson, and A. C. L. Lee, 1999: The effect of assimilating rain rates derived from satellites and lightning on forecasts of the 1993 superstorm. Mon. Wea. Rev., 127, 1433-1457. Alishouse, J. C., S. Snyder, and R. R. Ferraro, 1990: Determination of oceanic precipitable water from the SSM/I. IEEE Trans. Geo-sci. Remote Sens., 26, 811–816. Amorati, R., P. P. Alberoni, V. Levizzani, and S. Nanni, 2000: IR-based satellite and radar rainfall estimates of convective storms over northern Italy. Meteor. Appl., 7, 1-18. Anagnostou, E. N., and C. D. Kummerow, 1997: Stratiform and convective classification of rainfall using SSM/I 85-GHz brightness temperature observations. J. Atmos. Oceanic Technol., 14, 570-575. Anagnostou, E. N., A. J. Negri, and R. F. Adler, 1999: A satellite infrared technique for diurnal rainfall variability studies. J. Geophys. Res., 104, 31477–31488. Anderson, E., J. Pailleux, J. Thépaut, J. R. Eyre, A. McNally, G. A. Kelly, and P. Courtier, 1994: Use of cloud-cleared radiances in three/four-dimensional variational data assimilation. Q. J. R. Meteorol. Soc., 120, 627-653. Aonashi, K., and G. Liu, 2000: Passive microwave precipitation retrievals using TMI during the Baiu period of 1998. Part I: Algorithm description and validation. J. Appl. Meteorol., 39, 2024–2037. Arkin, P. A., 1979: The relationship between fractional coverage of high cloud and rainfall accumulations during GATE over the B-scale array. Mon. Wea. Rev., 106, 1153-1171.


56


Arkin, P. A., and B. N. Meisner, 1987: The relationship between large-scale convective rainfall and cold cloud over the Western Hemisphere during 1982-84. Mon. Wea. Rev., 115, 51-74. Arkin, P. A., and P. E. Ardanuy, 1989: Estimating climatic-scale precipitation from space: A review. J. Climate, 2, 1229-1238. Arkin, P. A., and J. Janowiak, 1991: Analysis of the global distribution of precipitation. Dyn. Atmos Oceans, 16, 5-16. Arkin, P. A., and P. Xie, 1994: The Global Precipitation Climatology Project: First algorithm intercomparison project. Bull. Amer. Meteor. Soc., 75, 401–419. Arkin, P. A., R. Joyce, and J. E. Janowiak, 1994: The estimation of global monthly mean rainfall using infrared satellite data: The GOES Precipitation Index (GPI). Rem. Sens. Rev., 11, 107124. Arking, A., and J. D. Childs, 1985: Retrieval of cloud cover parameters from multispectral satellite images. J. Climate Appl. Meteorol., 24, 322-333. Ba, M. B., and S. E. Nicholson: 1998:. Analysis of convective activity and its relationship to the rainfall over the Rift Valley lakes of East Africa during 1983-90 using the Meteosat infrared channel. J. Appl. Meteorol., 37, 1250-1264. Ba, M. B., and A. Gruber, 2001: GOES Multispectral Rainfall Algorithm (GMSRA). J. Appl. Meteorol., 40, 1500-1514. Ba, M. B., D. Rosenfeld, and A. Gruber: 1998: AVHRR multispectral derived cloud parameters: th relationship to microwave scattering signature and to cloud-to-ground lightning. Prepr. 9 Conf. Satellite Meteorology and Oceanography, AMS, 408-411. Baran, A. J., S. J. Brown, J. S. Foot, J. S., and D. L. Mitchell, 1999: Retrieval of tropical cirrus thermal optical depth, crystal size, and shape using a dual-view instrument at 3.7 and 10.8 µm. J. Atmos. Sci., 56, 92-110. Barrett, E. C., and D. W. Martin, 1981: The Use of Satellite Data in Rainfall Monitoring. Academic Press, 340 pp. Barrett, E. C., and J. Michell, 1991: Satellite remote sensing of natural hazards and disasters in the Mediterranean. In Current Topics in Remote Sensing, 2, E. C. Barrett, K. Brown, and A. Michallef, A., Eds., Gordon and Breach, 51-67. Barrett, E. C., and Coauthors, 1994: The First WetNet Precipitation Intercomparison Project: Interpretation of results. Remote Sens. Rev., 11, 303–373. Bauer, P., 2001a: Microwave radiative transfer simulation in clouds: Including a melting layer in cloud model bulky hydrometeor distributions. Atmos. Res., 157, 9-30. Bauer, P., 2001b: Over-ocean rainfall retrieval from multisensor data of the Tropical Rainfall Measuring Mission. Part I: Design and evaluation of inversion databases. J. Atmos. Oceanic Technol., 18, 1315-1330,. Bauer, P., L. Schanz, R. Bennartz, and P. Schlüssel, 1998: Outlook for combined TMI-VIRS algorithms for TRMM: lessons learned from the PIP and AIP projects. J. Atmos. Sci., 55, 1714-1729. Bauer, P., A. Khain, A. Pokrovsky, R. Meneghini, C. D. Kummerow, F. S. Marzano, and J. P. V. Poiares Baptista, 2000: Combined cloud–microwave radiative transfer modeling of stratiform rainfall. J. Atmos. Sci., 57, 1082-1104. Bauer, P., P.Amayenc, C. D. Kummerow, and E. A. Smith, 2001: Over-ocean rainfall retrieval from multisensor data of the Tropical Rainfall Measuring Mission. Part II: Algorithm implementation. J. Atmos. Oceanic Technol., 18, 1838-1855. Bellerby, T., M. Todd, D. Kniveton, and C. Kidd, 2000: Rainfall estimation from a combination of TRMM Precipitation Radar and GOES Multispectral Satellite Imagery through the use of an artificial neural network. J. Appl. Meteorol., 39, 2115–2128. Bellon, A., S. Lovejoy, and G. L. Austin, 1980: Combining satellite and radar data for the shortrange forecasting of precipitation. Mon. Wea. Rev., 108, 1554-1556. Bendix, J., 1997: Adjustment of the Convective-Stratiform Technique (CST) to estimate 1991/93 El Niño rainfall distribution in Ecuador and Peru by means of Meteosat-3 data. Int. J. Remote Sensing, 18, 1387-1394. Bendix, J., 2000: Precipitation dynamics in Ecuador and Northern Peru during the 1991/92 El Niño: a remote sensing perspective. Int. J. Remote Sensing, 21, 533-548. Berg, W., 1994: Precipitation retrieval during TOGA-COARE using a combination of SSM/I and GMS data with application to climate studies. Preprints, Seventh Conf. on Satellite Meteorology and Oceanography, Monterey, CA, Amer. Meteor. Soc., 67–70. MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002


57

Berg, W., and R. Chase, 1992: Determination of mean rainfall from the Special Sensor Microwave/Imager (SSM/I) using a mixed lognormal distribution. J. Atmos. Oceanic Technol., 9, 129–141. Berg, W., and S. K. Avery, 1995: Evaluation of monthly rainfall estimates derived from the special sensor microwave/imager (SSM/I) over the tropical Pacific. J. Geophys. Res., 100, 1295–1315. Berg, W., W. Olson, R. Ferraro, S. J. Goodman, and F. J. LaFontaine, 1998: An assessment of the first- and second-generation Navy operational precipitation retrieval algorithms. J. Atmos. Sci., 55, 1558-1575. Boccippio, D. J., S. J. Goodman, and S. Heckman, 2000: Regional differences in tropical lightning distributions. J. Appl. Meteorol., 39, 2231–2248. Chang, D.-E., J. A. Weinman, C. A. Morales, and W. S. Olson, 2001: The effect of spaceborne microwave and ground-based continuous lightning measurements on forecasts of the 1998 Groundhog Day Storm. Mon. Wea. Rev., 129, 1809-1833. Cheng, M., and R. Brown, 1995: Delineation of precipitation areas by correlation of METEOSAT visible and infrared data with radar data. Mon. Wea. Rev., 123, 2743-2757. Cheng, M., R. Brown, and C. G. Collier, 1993: Delineation of precipitation areas using METEOSAT infrared and visible data in the region of the United Kingdom. J. Appl. Meteorol., 32, 884-898. Coppens, D., Z. S. Haddad, and E. Im, 2000: Estimating the uncertainty in passive-microwave rain retrievals. J. Atmos. Oceanic Technol., 17, 1618–1629. Crone, L. J., L. M. McMillin, and D. S. Crosby, 1996: Constrained regression in satellite meteorology. J. Appl. Meteorol., 35, 2023-2039. Dietrich, S., R. Solomon, C. Adamo, and A. Mugnai, 2001: Rainfall monitoring at geostationary scale: Potential of lightning data in a rapid update approach. Proc. The 2001 EUMETSAT Meteorological Data Users’ Conf., Antalya, 393-398. Ebert, E. E., and M. J. Manton, 1998: Performance of satellite rainfall estimation algorithms during TOGA-COARE. J. Atmos. Sci., 55, 1537-1557. Ellrod, G. P., 1994: Potential use of GOES-I multispectral, infrared imagery for nighttime th detection of precipitation. Prepr. 7 Conf. Satellite Meteorol. Oceanography, AMS, Monterey, CA, 164-167. Espinoza, R. C., Jr., and Harshvardhan, 1996: Parameterization of solar near-infrared radiative properties of cloudy layers. J. Atmos. Sci., 53, 1559-1568. Evans, F., J. Turk, T. Wong, and G. Stephens, 1995: A Bayesian approach to microwave precipitation retrieval. J. Appl. Meteorol., 34, 260–279. Falcovich, A., E. Kalnay, S. Lord, and M. B. Matur, 2000: A new method of observed rainfall assimilation in forecast models. J. Appl. Meteorol., 39, 1,282-1,298. Ferraro, R. R., 1997: Special sensor microwave imager derived global rainfall estimates for climatological applications. J. Geophys. Res., 102 (D14), 16715-16735. Ferraro, R. R., and G. F. Marks, 1995: The development of SSM/I rain-rate retrieval algorithms using ground-based radar mea-surements. J. Atmos. Oceanic Technol., 12, 755–770. Ferraro, R. R., N. C. Grody, and J. Kogut, 1986: Classification of geophysical parameters using passive microwave satellite measurements. IEEE Trans. Geosci. Remote Sens., 24, 1008– 1013. Ferraro, R. R., N. C. Grody, and G. F. Marks, 1994: Effects of surface conditions on rain identification using the DMSP-SSM/I. Remote Sens. Rev., 11, 195–210. Ferraro, R. R., E. A. Smith, W. Berg, and G. J. Huffman, 1998: A screening methodology for passive microwave precipitation retrieval algorithms. J. Atmos. Sci., 55, 1583-1600. Ferraro, R. R., F. Weng, N. C. Grody, and A. Basist, 1996: An eight-year (1987-1994) time series of rainfall, clouds, water vapor, snow cover, and sea ice derived from SSM/I measurements. Bull. Am. Meteorol. Soc., 77, 891-905. Ferreira, F., P. Amayenc, S. Oury, and J. Testud, 2001: Study and test of improved rain estimates from the TRMM Precipitation Radar. J. Appl. Meteorol., 40, 1878-1899. Ferriday, J. G., and S. K. Avery, 1994: Passive microwave remote sensing of rainfall with SSM/I: Algorithm development and im-plementation. J. Appl. Meteorol., 33, 1587–1596. Gagin, A., D. Rosenfeld, and R. E. Lopez, 1985: The relationship between height and precipitation characteristics of summertime convective cells in South Florida. J. Atmos. Sci., 42, 84-94.


58


Gérard, É., and R. W. Saunders, 1999: 4D-var assimilation of SSM/I total colums water vapour in the ECMWF model. Q. J. R. Meteorol. Soc., 125, 3077-3101. th GEWEX, 1996: GPCP Working Group on Data Management, 10 session. GEWEX News, August 1996, 8–9. Golding. B. W., S. Senesi, K. A. Browning, B. Bizzarri, W. Benesch, D. Rosenfeld, V. Levizzani, H. P. Rösli, U. Platt, T. E. Nordeng, J. T. Carmona, P. Ambrosetti, P. Pagano, M. Kurz, 2001: EUMETSAT position paper on observation requirements for nowcasting and very short range forecasting in 2015-2025. EUMETSAT, Darmstadt, 62 pp. Grecu, M., and E. N. Anagnostou, 2001: Overland precipitation estimation from TRMM passive microwave observations. J. Appl. Meteorol., 40, 1367-1380. Grecu, M., E. N. Anagnostou, and R. F. Adler, 2000: Assessment of the use of lightning information in satellite infrared rainfall estimation. J. Hydrometerol., 1, 211-221. Griffith, C. G., W. L. Woodley, P. G. Grube, D. W. Martin, J. Stout, and D. N. Sikdar, 1978: Rain estimation from geosynchronous satellite imagery - Visible and infrared studies. Mon. Wea. Rev., 106, 1153-1171. Grody, N. C., 1984: Precipitation monitoring over land from satellites by microwave radiometry. Proc. IGARSS’ 84 Symposium, Strasbourg, 27-30 August. Grody, N. C., 1991: Classification of snow cover and precipitation using the Special Sensor Microwave/Imager (SSM/I). J. Geophys. Res., 96, 7423–7435. Grody, N. C., F. Weng, and R. R. Ferraro, 1999: Application of AMSU for obtaining hydrological parameters. In: Microwave Radiometry and Remote Sensing of the Environment, P. Pampaloni and S. Paloscia Ed., VSP Int. Sci. Publisher, Utrecht (The Netherlands), 339-351. Gruber, A.,1973: Estimating rainfall in regions of active convection. J. Appl. Meteorol., 12, 110118. Gruber, A., X. Su, M. Kanamitsu, and J, Schemm, 2000: The comparison of two merged rain gauge-satellite precipitation datasets. Bull. Am. Meteorol. Soc., 81, 2631-2644. Haddad, Z., E. A. Smith, C. D. Kummerow, T. Iguchi, M. Farrar, S. Darden, M. Alves, and W. Olson, 1997: The TRMM ‘Day-1’ radar/radiometer combined rain-profile algorithm. J. Meteor. Soc. Japan, 75, 799-808. Han, W., K. Stamnes, and D. Lubin, 1999: Remote sensing of surface and cloud properties in the Arctic from AVHRR measurements. J. Appl. Meteorol., 38, 989-1012. Hinton, B. B., W. S. Olson, D. W. Martin, and B. Auvine, 1992: A passive microwave algorithm for tropical ocean rainfall. J. Appl. Meteorol., 31, 1379–1395. Hong, Y., C. D. Kummerow, and W. S. Olson, 1999: Separation of convective and stratiform precipitation using microwave brightness temperature. J. Appl. Meteorol., 38, 1195-1213. Houze, R. A., 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Am. Meteorol. Soc., 78, 2179-2196. Hsu, K.-L., X. Gao, S. Sorooshian, and H.V. Gupta, 1997: Precipitation estimation from remotely sensed information using artificial neural networks. J. Appl. Meteorol., 36, 1176-1190. Hu, Y. X., and K. Stamnes, 1993: An accurate parameterization of the radiative properties of water clouds suitable for use in climate models. J. Climate, 6, 70-83. Huffman, G. J., R. F. Adler, P. Arkin, A. Chang, R. Ferraro, A. Gruber, J. Janowiak, A. McNab, B. Rudolf, and U. Schneider, 1997: The Global Precipitation Climatology Project (GPCP) combined precipitation dataset. Bull. Am. Meteorol. Soc., 78, 5-20. Huffman, G. J., R. F. Adler, M. M. Morrissey, D. T. Bolvin, S. Curtis, R. Joyce, B. McGavock, and J. Susskind, 2001: Global precipitation at one-degree daily resolution from multisatellite observations. J. Hydrometeorol., 2, 36-50. Hunt, G. E., 1973:. Radiative properties of terrestrial clouds at visible and infra-red thermal window wavelengths. Quart. J. R. Met. Soc., 99, 346-369. Iguchi, T., T. Kozu, R. Meneghini, J. Awaka, and K. Okamoto, 2000: Rain-profiling algorithm for the TRMM precipitation radar. J. Appl. Meteorol., 39, 2038-2052. Inoue, T., 1985: On the temperature and effective emissivity determination of semi-transparent cirrus clouds by bi-spectral measurements in the 10 µm window region. J. Meteorol. Soc. Japan, 63, 88-98. Inoue, T., 1987a: A cloud type classification with NOAA 7 split-window measurements. J. Geophys. Res., 92 D, 3991-4000. Inoue, T., 1987b: An instantaneous delineation of convective rainfall area using split window data of NOAA-7 AVHRR. J. Meteorol. Soc. Japan, 65, 469-481. MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002


59

Inoue, T., 1997: Detection of rain-bearing convective systems using cloud information derived from split window. Proc. The 1997 Meteorological Satellite Data Users’ Conf., EUMETSAT, 173-182. Inoue, T., and K. Aonashi, 2000: A comparison of cloud and rainfall information from instantaneous Visible and InfraRed Scanner and Precipitation Radar observations over a frontal zone in East Asia during June 1998. J. Appl. Meteorol., 39, 2292-2301. Janowiak, J. E., R. J. Joyce, and Y. Yarosh, 2001: A real-time global half-hourly pixel-resolution infrared dataset and its applications. Bull. Am. Meteor. Soc., 82, 205-217. Jobard, I., and M. Desbois, 1994: Satellite estimation of the tropical precipitation using the Meteosat and SSM/I data. Atmos. Res., 34, 285-298. Jones, S. M. R., and P. A. Watson, 1993: Attenuation and countermeasures in millimeter-wave point-to-multipoint networks. Radio Sci., 28, 1057-1069. Joyce, R., and P. A. Arkin, 1997: Improved estimates of tropical and subtropical precipitation using the GOES Precipitation Index. J. Atmos. Oceanic Technol., 14, 997-1011. Karlsson, K. G., and E. Liljas, 1990: The SHMI model for cloud and precipitation analysis from multispectral AVHRR data. Swedish Meteorological and Hydrological Institute, Norrköping, SMHI PROMIS Rep., 10, 74 pp. Kerrache, M., and J. Schmetz, 1988: A precipitation index from the ESOC climatological data set. ESA J., 12, 379-383. Key, J. R., and J. M. Intrieri, 2000: Cloud particle phase determination with the AVHRR. J. Appl. Meteorol., 39, 1797-1804. Khain, A., M. Ovtchinnikov, M. Pinsky, A. Pokrovsky, and H. Krugliak, 2000: Notes on the stateof-the-art numerical modeling of cloud microphysics. Atmos. Res., 55, 159-224. Khain, A. P., D. Rosenfeld, and A. Pokrovsky, 2001: Simulating convective clouds with sustained supercooled liquid water down to -37.5 degrees C using a spectral microphysics model. Geophys. Res. Lett., 28, 3887-3890. Kidd, C., 1998: On rainfall retrieval using polarization-corrected temperatures. Int. J. Remote Sensing, 19, 981-996. Kidd, C., D. Kniveton, and E. C. Barrett, 1998: The advantages and disadvantages of statistically derived-empirically calibrated passive microwave algorithms for rainfall estimation. J. Atmos. Sci., 55, 1576-1582. Kidder, S. Q., and T. H. Vonder Haar, 1995: Satellite Meteorology: An Introduction. Academic Press, 466 pp. King, M. D., S.-C. Tsay, S. E. Platnick, M. Wang, and K.-N. Liou, 1997: Cloud retrieval algorithms for MODIS: optical thickness, effective particle radius, and thermodynamic phase. MODIS Algorithm Theoretical Basis Doc., ATBD-MOD-05 and MOD-06. King, P. W. S., W. D. Hogg, and P. A. Arkin, 1995: The role of visible data in improving satellite rainrate estimates. J. Appl. Meteorol., 34, 1608-1621. Kleespies, T. J., 1995: The retrieval of marine stratiform cloud properties from multiple observations in the 3.9-µm window under conditions of varying solar illumination. J. Appl. Meteorol., 34, 1512-1524. Kniveton, D. R., and E. C. Barrett, 1994: Composite algorithms for rainfall estimation using data th from the DMSP SSM/I. Prepr. 7 Conf. Satellite Meteorology and Oceanography, Monterey, CA, Am. Meteor. Soc., 140–143. Krishnamurti, T. N., G. D. Rohaly, and H. S. Bedi, 1994: On the improvement of precipitation forecast skill from physical initialization. Tellus, 46a, 598-614. Krishnamurti, T. N., S. Surendran, D. W. Shin, R. J. Correa-Torres, T. S. V. Vijaya Kumar, E. Williford, C. D. Kummerow, R. F. Adler, J. Simpson, R. Kakar, W. S. Olson, and F. J. Turk, 2001: Real-time multianalysis-multimodel supernsemble forecasts of precipitation using TRMM and SSM/I products. Mon. Wea. Rev., 129, 2861-2883. Kummerow, C. D., 1998: Beamfilling errors in passive microwave rainfall retrievals. J. Appl. Meteorol., 37, 356–370. Kummerow, C. D., and L. Giglio 1994a: A passive microwave technique for estimating rainfall and vertical structure information from space. Part I: Algorithm description. J. Appl. Meteorol., 33, 3–18. Kummerow, C. D., and L. Giglio, 1994b: A passive microwave technique for estimating rainfall and vertical structure information from space. Part II: Applications to SSM/I data. J. Appl. Meteorol., 33, 19–34. MUSIC – EVK1-CT-2000-00058 Deliverable 6.1 04/02/2002

60


Kummerow, C. D., W. S. Olson, and L. Giglio, 1996: A simplified scheme for obtaining precipitation and vertical hydrometeor profiles from passive microwave sensors. IEEE Trans. on Geosci. and Remote Sensing, 34, 1213-1232. Kummerow, C. D., W: Barnes, T. Kozu, J. Shiue, and J. Simpson, 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809-817. Kummerow, C. D., Y. Hong, W. S. Olson, S. Yang, R. F. Adler, J. McCollum, R. Ferraro, G. Petty, D.-B. Shin, and T. T. Wilheit, 2001: The evolution of the Goddard Profiling Algorithm (GPROF) for rainfall estimation from passive microwave sensors. J. Appl. Meteorol., 40, 1801-1820. Kurino, T., 1997: A satellite infrared technique for estimating “deep/shallow” convective and stratiform precipitation. Adv. Space Res., 19, 511-514. Laing, A. G., J. M. Fritsch, and A. J. Negri, 1999: Contribution of mesoscale convective complexes to rainfall in Sahelian Africa: estimates from geostationary infrared and passive microwave data. J. Appl. Meteorol., 38, 957-964. Lensky, I. M., and D. Rosenfeld, 1997: Estimation of precipitation area and rain intensity based on the microphysical properties retrieved from NOAA AVHRR data J. Appl. Meteorol., 36, 234-242. th Levizzani, V., 1998: Intense rainfall monitoring from geostationary satellites. Prepr. 9 Conf. Satellite Meteorology and Oceanography, AMS, 327-330. Levizzani, V., and M. Setvák, 1996: Multispectral, high-resolution satellite observations of plumes on top of convective storms. J. Atmos. Sci., 53, 361-369. Levizzani, V., F. Porcù, and F. Prodi, 1990: Operational rainfall estimation using METEOSAT infrared imagery: An application in Italy’s Arno river basin – Its potential and drawbacks. ESA J., 14, 313-323. Levizzani, V., F. Porcù, F. S. Marzano, A. Mugnai, E. A. Smith, and F. Prodi, 1996: Investigating a SSM/I microwave algorithm to calibrate METEOSAT infrared instantaneous rainrate estimates. Meteorol. Appl., 3, 5-17. Levizzani, V., J. Schmetz, H. J. Lutz, J. Kerkmann, P. P. Alberoni, and M. Cervino, 2001: Precipitation estimations from geostationary orbit and prospects for METEOSAT Second Generation. Meteorol. Appl., 8, 23-41. Levizzani, V., P. P. Alberoni, P. Bauer, L. Bottai, A. Buzzi, E. Cattani, M. Cervino, P. Ciotti, M. J. Costa, S. Dietrich, B. Gozzini, A. Khain, C. Kidd, F. S. Marzano, F. Meneguzzo, S. Migliorini, A. Mugnai, F. Porcù, F. Prodi, R. Rizzi, D. Rosenfeld, L. Schanz, E. A. Smith, F. Tampieri, F. Torricella, F. J. Turk, G. A. Vicente, and G. Zipoli, 2000: Use of the MSG SEVIRI channels in a combined SSM/I, TRMM and geostationary IR method for rapid updates of rainfall. Proc. st 1 MSG-RAO Workshop, ESA SP-452, 63-66. Levizzani, V., P. Bauer, A. Buzzi, D. Hinsman, A. Khain, C. Kidd, F. S. Marzano, F. Meneguzzo, A. Mugnai, J. P. V. Poiares Baptista, F. Prodi, J. F. W. Purdom, D. Rosenfeld, J. Schmetz, E. A. Smith, F. Tampieri, F. J. Turk, G. A. Vicente, 2001a: EURAINSAT: European Satellite th Rainfall Analysis and Monitoring at the Geostationary Scale. Prepr. 11 Conf. Satellite Meteorol. Oceanography, AMS, Madison, WI, 15-18 Oct., 650-653. Levizzani, V., P. Bauer, A. Buzzi, S. Davolio, D. E. Hinsman, C. Kidd, F. S. Marzano, F. Meneguzzo, A. Mugnai, J. P. V. Poiares Baptista, F. Porcù, F. Prodi, J. F. W. Purdom, D. Rosenfeld, J. Schmetz, E. A. Smith, F. Tampieri, F. J. Turk, and G. A. Vicente, 2001b: EURAINSAT - Looking into the future of satellite rainfall estimations. Proc. The 2001 EUMETSAT Meteorological Satellite Data Users' Conf., EUMETSAT, Antalya, 1-5 Oct., 375384. Liljas, E., 1986: Use of the AVHRR 3.7 micrometer channel in multispectral cloud classification. Swedish Meteorological and Hydrological Institute, Norrköping, SMHI Promis Rep., 2, 23 pp. Liou, K. N., 1992: Radiation and cloud processes in the atmosphere. Oxford Univ. Press, 487 pp. Liu, G., and J. A. Curry, 1992: Retrieval of precipitation from satellite microwave measurement using both emission and scattering. J. Geophys. Res., 97 D, 9,959-9974. Liu, G., and J. A. Curry, 1993: Determination of characteristic features of cloud liquid water from satellite microwave measurement. J. Geophys. Res., 98, 5069–5092. Liu, G., J. A. Curry, and R.-S. Sheu, 1995: Classification of clouds over the western equatorial Pacific Ocean using combined infrared and microwave satellite data. J. Geophys. Res., 100 D, 13,811-13,826.



61

Lovejoy, S., and G. L. Austin, 1979: The delineation of rain areas from visible and IR satellite data from GATE and mid-latitudes. Atmos.-Ocean, 17, 77-92. Manobianco, J., S. Koch, V. M. Karyampudi, and A. J. Negri, 1994: The impact of assimilating satellite-derived precipitation rates on numerical simulations of the ERICA IOP 4 cyclone. Mon. Wea. Rev., 122, 341-365. Marécal, V., and J.-F. Mahfouf, 2000: Variational retrieval of temperature and humidity profiles from TRMM precipitation data. Mon. Wea. Rev., 128, 3853–3866. Marécal, V., É. Gérard, J.-F. Mahfouf, and P. Bauer, 2000: The comparative impact of the assimilation of SSM/I and TMI brightness temperatures in the ECMWF 4D-Var system. Q. J. R. Meteorol. Soc., 127, 1123-1142 Marrocu, M., A. Pompei, G. Dalu, G. L. Liberti, and A. J. Negri, 1993: Precipitation estimation over Sardinia from satellite infrared data. Int. J. Remote Sensing, 14, 115-134. Marzano, F. S., J. F. Turk, P. Ciotti, S. Di Michele, and N. Pierdicca, 2000: Combined use of spaceborne polar-orbiting microwave and geo-stationary infrared radiometers for estimating st rainfall attenuation along earth-satellite links. Proc. 1 Int. Workshop on Radiowave Propagation Modelling for SatCom Services at Ku-band and above, ESA WPP-146, 189196. Masuda, K., and T. Takashima, 1990: Deriving cirrus information using the visible and near-IR channels of the future NOAA-AVHRR radiometer. Remote Sens. Environ., 31, 65-81. Mathur, M. B., 1995: Amelioration of convective and nonconvective condensation with the assimilation of rainfall for improving initial conditions. Mon. Wea. Rev., 123, 1851–1861. Mathur, M. B., H. S. Bedi, T. N. Krishnamurti, M. Kanamitsu, and J. S. Woollen, 1992: Use of satellite-derived rainfall for improving tropical forecasts. Mon. Wea. Rev., 120, 2540–2560. Matricardi, M., G. A. Kelly, and M. Tomassini, 1996: Simulations of METEOSAT water vapour imagery using the ECMWF forecast model and its application to variational analysis. Proc. 1996 Meteorological Satellite Data Users’ Conf., EUMETSAT, 239-246. McCollum, J. R., A.Gruber, and M. B. Ba, 2000: Discrepancy between gauges and satellite estimates of rainfall in equatorial Africa. J. Appl. Meteorol., 39, 666-679. McMurdie, L. A., and K. B. Katsaros, 1996: Satellite-derived integrated water vapor and rain intensity patterns: Indicators for rapid cyclogenesis. Wea. Forecasting, 11, 230-245 Meneguzzo, F., S. Migliorini, G. A. Vicente, L. Bottai, B. Gozzini, and G. Zipoli, 1998: Satellite rainfall estimates in real time in the Mediterranean countries. Verifications in Tuscany. Proc. th ICAM-98, 25 Int. Conf. Alpine Meteor., Turin, 14-19 Jul, 2-7. Menzel, W. P., and J. F. W. Purdom, 1994: Introducing GOES-I: the first of new generation of Geostationary Operational Environmental Satellites. Bull. Am. Meteor. Soc., 75, 757-781. Miller, S. W., P. A. Arkin, and R. J. Joyce, 2000: A combined microwave/infrared rain rate algorithm. Int. J. Remote Sens., 22, 3285-3307. Mohr, K. I., J. S. Famiglietti, and E. J. Zipser, 1999: The contribution to tropical rainfall with respect to convective system type, size, and intensity estimated from the 85-GHz icescattering signature. J. Appl. Meteorol., 38, 596-606. Mugnai, A., H. J. Cooper, E. A. Smith, and G. J. Tripoli, 1990: Simulation of microwave brightness temperatures of an evolving hailstorm at SSM/I frequencies. Bull. Am. Meteor. Soc., 71, 2-13. Mugnai, A., E. A. Smith, and G. J. Tripoli, 1993: Foundations for statistical-physical precipitation retrieval from passive microwave satellite measurements. Part II: Emission-source and generalized weighting-function properties of a time-dependent cloud-radiation model.J. Appl. Meteorol., 32, 17-39. Nakajima, T., and M. D. King, 1990: Determination of the optical thickness and effective particle radius of clouds from reflected solar radiation measurements. Part I: theory. J. Atmos. Sci., 47, 1878-1893. Negri, A. J., and R. F. Adler, 1993: An intercomparison of three satellite infrared rainfall techniques over Japan and surrounding waters. J. Appl. Meteorol., 32, 357-373. Negri, A. J., R. F. Adler, and P. J. Wetzel, 1984: Rain estimation from satellite: An examination of the Griffith-Woodley technique. J. Climate Appl. Meteorol., 23, 102-116. Olson, W. S., P. Bauer, N. E. Viltard, D. E. Johnson, W.-K. Tao, R. Meneghini, and L. Liao, 2001a: A melting-layer model for passive/active microwave remote sensing applications. Part I: Model formulation and comparison with observations. J. Appl. Meteorol., 40, 11451163.


62


Olson, W. S., P. Bauer, C. D. Kummerow, Y. Hong, and W.-K. Tao, 2001b: A melting-layer model for passive/active microwave remote sensing applications. Part II: Simulation of TRMM observations. J. Appl. Meteorol., 40, 1164-1179. Olson, W. S., Y. Hong, C. D. Kummerow, and J. F. Turk, 2001c: A texture-polarization method for estimating convective-stratiform precipitation area coverage from passive microwave radiometer data. J. Appl. Meteorol., 40, 1577-1591. O’Sullivan, F., C. H. Wash, M. Stewart, and C. E. Motell, 1990: Rain estimation from infrared and visible GOES satellite data. J. Appl. Meteorol., 29, 209-223. Olson, W. S., C. D. Kummerow, Y. Hong, and W.-K. Tao, 1999: Atmospheric latent heating distributions in the tropics derived from satellite passive microwave radiometer measurements. J. Appl. Meteorol., 38, 633-664. Ou, S. C., K. N. Liou, W. M. Gooch, and Y. Takano, 1993: Remote sensing of cirrus cloud parameters using advanced very-high-resolution radiometer 3.7- and 10.9-µm channels. Appl. Optics, 32, 2171-2180. Panegrossi, G., S. Dietrich, F. S. Marzano, A. Mugnai, E. A. Smith, X. Xiang, G. J. Tripoli, P. K. Wang, and J. P. V. Poiares Baptista, 1998: Use of cloud model microphysics for passive microwave-based precipitation retrieval: significance of consistency between model and measurement manifolds. J. Atmos. Sci., 55, 1644-1672. Parol, F., J. C. Buriez, G. Brogniez, and Y. Fouquart, 1991: Information content of AVHRR channels 4 and 5 with respect to the effective radius of cirrus cloud particles. J. Appl. Meteorol., 30, 973-984. Peng, M. S., and S. W. Chang, 1996: Impacts of SSM/I retrieved rainfall rates on numerical prediction of a tropical cyclone. Mon. Wea. Rev., 124, 1181-1198. Pereira Fo., A. J., and K. C. Crawford, 1999a: Mesoscale precipitation fields. Part I: Statistical analysis and hydrologic response. J. Appl. Meteorol., 38, 82-101. Pereira Fo., A. J., K. C. Crawford, and D. J. Stensrud, 1999b: Mesoscale precipitation fields. Part II: Hydrometeorological modeling. J. Appl. Meteorol., 38, 102-125. Petersen, W. A., and S. A. Rutledge, 1998: On the relationship between cloud-to-ground lightning and convective rainfall. J. Geophys. Res., 103, 14025-14040. Petty, G. W., 1994a: Physical retrievals of over-ocean rain rate from multichannel microwave imaging. Part I: Theoretical characteristics of normalized polarization and scattering indices. Meteor. Atmos. Phys., 54, 79–100. Petty, G. W., 1994b: Physical retrievals of over-ocean rain rate from mul-tichannel microwave imaging. Part II: Algorithm implementation. Meteor. Atmos. Phys., 54, 101–122. Petty, G. W., 1995: The status of satellite-based rainfall estimation over land. Remote Sens. Environ., 51, 125-137. Petty, G. W., 2001a: Physical and microwave radiative properties of precipitating clouds. Part I: Principal component analysis of observed multichannel microwave radiances in tropical stratiform rainfall. J. Appl. Meteorol., 40, 2105-2114. Petty, G. W., 2001b: Physical and microwave radiative properties of precipitating clouds. Part II: A parametric 1D rain-cloud model for use in microwave radiative transfer simulations. J. Appl. Meteorol., 40, 2115-2129. Phalippou, L., 1996: Variational retrieval of humidity profile, wind speed and cloud liquid water path with the SSM/I: Potential for numerical weather prediction. Q. J. R. Meteorol. Soc., 122, 327-355. Pilewskie, P., and S. Twomey, 1987: Discrimination of ice from water in clouds by optical remote sensing . Atmos. Res., 21, 113-122. Platnick, S., and S. Twomey, 1994: Remote sensing the susceptibility of cloud albedo to changes in drop concentration. Atmos. Res., 34, 85-98. Pompei, A., M. Marrocu, P. Boi, and G. Dalu, 1995: Validation of retrieval algorithms for the infrared remote sensing of precipitation with the Sardinian gauge network data. Il Nuovo Cimento, 18 C, 483-496. Porcù, F., M. Borga, and F. Prodi, 1999: Rainfall estimation by combining radar and infrared satellite data for nowcasting purposes. Meteorol. Appl., 6, 289-300. Prabhakara, C., R. Iacovazzi, R. Oki, and J.A. Weinman, 1999: A microwave radiometer rain retrieval method applicable to land areas. J. Meteorol. Soc. Japan, 77, 859-871.



63

Prabhakara, C., R. Iacovazzi, J. A. Weinman, and G. Dalu, 2000: A TRMM microwave radiometer rain rates estimation method with convective and stratiform discrimination. J. Meteorol. Soc. Japan, 78, 241-258. Ramanathan, V., P. J. Crutzen, J. T. Kiehl, and D. Rosenfeld, 2001: Aerosols, climate and the global environment. A hazy future for the blue planet? Science, 294, 2119-2124. Rao, N. X., S. C. Ou, and K. N. Liou, 1995: Removal of the solar component in AVHRR 3.7-µm radiances for the retrieval of cirrus cloud parameters. J. Appl. Meteorol., 34, 482-499. Reudenbach, C., G. Heinemann, E. Heuel, J. Bendix, and M. Winiger, 2001: Investigation of summertime convective rainfall in Western Europe based on a synergy of remote sensing data and numerical models. Meteorol. Atmos. Phys., 76, 23-41. Robinson, J. M., and R. A. Scofield, 1994: Using satellite imagery to analyze the devastating th flash floods in the western region between December 1992 and February 1993. Prepr. 7 Conf. Satellite Meteorology and Oceanography, AMS, 417-420. Rodgers, E. B., R. F. Adler, and H. F. Pierce, 2000: Contribution of tropical cyclones to the North Pacific climatological rainfall as observed from satellite. J. Appl. Meteorol., 39, 16581678. Rodgers, E. B., R. F. Adler, and H. F. Pierce, 2001: Contribution of tropical cyclones to the North Atlantic climatological rainfall as observed from satellite. J. Appl. Meteorol., 40, 17851800. Rosenfeld, D., 1999: TRMM observed first direct evidence of smoke from forest fires inhibiting rainfall. Geophys. Res. Lett., 26, 3105-3108. Rosenfeld, D., 2000: Suppression of rain and snow by urban and industrial air pollution. Science, 287, 1793-1796. Rosenfeld, D., and A. Gagin, 1989: Factors governing the total rainfall yield of continental convective clouds. J Appl. Meteorol., 28, 1015-1030. Rosenfeld, D., and G. Gutman, 1994: Retrieving microphysical properties near the tops of potential rain clouds by multispectral analysis of AVHRR data. Atmos. Res., 34, 259-283. Rosenfeld, D., and I. M. Lensky, 1998: Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull. Am. Meteor. Soc., 79, 24572476. Rosenfeld, D., and W. L. Woodley, 2000: Deep convective clouds with sustained supercooled liquid water down to-37.5 degrees C. Nature, 405, 440-442. Rosenfeld, D., D. B. Wolff, and E. Amitai, 1994: The window probability matching method for rainfall measurements with radar. J. Appl. Meteorol., 33, 683-693. Rosenfeld, D., Y. Rudich, and R. Lahav, 2001: Desert dust suppressing precipitation: a possible desertification feedback loop. PNAS, 98, 5975-5980. Saunders, R. W., and K. T. Kriebel, 1988: An improved method for detecting clear sky and cloudy radiances from AVHRR data. Int. J. Remote Sensing, 9, 123-150. Savage, R. C., E. A. Smith, and A. Mugnai, 1994: Concepts for a geostationary microwave th imaging sounder (GeoMIS). Prepr. 7 Conf. Satellite Meteorology and Oceanography, AMS, 516-519. Schmetz, J., P. Pili, S. A. Tjemkes, D. Just, J. Kerkmann, S. Rota, and A. Ratier, 2002: An introduction to Meteosat Second Generation (MSG). Bull. Amer. Meteor. Soc., in press. Scofield, R. A., 1987: The NESDIS operational convective precipitation technique. Mon. Wea. Rev., 115, 1773-1792. Scofield, R. A., and V. J. Oliver, 1977: A scheme for estimating convective rainfall from satellite imagery. NOAA Tech. Memo. NESS, 86, Dept. of Commerce, Washington, D.C., 47pp. Scofield, R. A., and L. Naimeng, 1994: The use of satellite imagery during the great floods of 1993. Prepr. 7th Conf. Satellite Meteorology and Oceanography, AMS, 345-350. Setvák, M., and C. A. Doswell, III, 1991: The AVHRR channel 3 cloud top reflectivity of convective storms. Mon. Wea. Rev., 119, 841-847. Sheu, R.-S., J. A. Curry, and G. Liu, 1995: Satellite retrieval of tropical precipitation using combined International Satellite Cloud Climatology Project DX and SSM/I data. J. Geophys. Res., 101, 21 291–21 301. Slingo, A., and H. M. Schrecker, 1982: On the shortwave radiative properties of stratiform water clouds. Q. J. R. Meteorol. Soc., 108, 407-426.


64


Smith, E. A., C. D. Kummerow, and A. Mugnai, 1994: The emergence of in-version-type precipitation profile algorithms for estimation of precipitation from satellite microwave measurements. Remote Sens. Rev., 11, 211–242. Smith, E. A., A. Mugnai, H. J. Cooper, G. J. Tripoli, and X. Xiang, 1992: Foundations for statistical-physical precipitation retrieval from passive microwave satellite measurements. Part I: Brightness-temperature properties of a time-dependent cloud-radiation model. J. Appl. Meteorol., 31, 506-531. Smith, E. A., J. E. Lamm, R. F. Adler, J. Alishouse, K. Aonashi, E. C. Barrett, P. Bauer, W. Berg, A. Chang, R. Ferraro, J. Ferriday, S. Goodman, N. Grody, C. Kidd, D. R. Kniveton, C. D. Kummerow, G. Liu, F. S. Marzano, A. Mugnai, W. Olson, G. Petty, A. Shibata, R. Spencer, F. Wentz, T. Wilheit, and E. Zipser, 1998: Results of the WetNet PIP-2 project. J. Atmos. Sci., 55, 1483-1536. Solomon, R., and M. B. Baker, 1998: Lightning flash rate and type in convective storms. J. Geophys. Res., 103, 14041-14057. Sorooshian, S., K.-L. Hsu, X. Gao, H. V. Gupta, B. Imam, and D. Braithwaite, 2000: Evaluation of PERSIANN system satellite–based estimates of tropical rainfall. Bull. Am. Meteor. Soc., 81, 2035–2046. Spencer, R. W., H. M. Goodman, and R. E. Hood, 1989: Precipitation retrieval over land and ocean with SSM/I. Part I: Identification and characteristics of the scattering signal. J. Atmos. Oceanic Technol., 6, 254-273. Tao, W.-K., J. Simpson, C. H. Sui, S. Lang, and R. F. Adler, 1993: Retrieval algorithms for estimating the vertical profiles of latent heat release: their applications for TRMM. J. Meteor. Soc. Japan, 71, 685-700. Tjemkes, S. A., L. van de Berg, and J. Schmetz, J., 1997: Warm water vapour pixels over high clouds as observed by METEOSAT. Contr. Atmos. Phys., 70, 15-21. Todd, M. C., and J. O. Bailey, 1995: Estimates of rainfall over the United Kingdom and surrounding seas from the SSM/I using the polarization corrected temperature algorithm. J. Appl. Meteorol., 34, 1254–1265. Todd, M. C., E: C. Barrett, M. J. Beaumont, and J. L. Green, 1995: Satellite identification of rain days over the upper Nile river basin using an optimum infrared rain/no-rain threshold temperature model. J. Appl. Meteorol., 34, 2600-2611. Todd, M. C., E: C. Barrett, M. J. Beaumont, and T. J. Bellerby, 1999: Estimation of daily rainfall over the upper Nile river basin using a continuously calibrated satellite infrared technique. Meteorol. Appl., 6, 201-210. Todd, M. C., C. Kidd, D. Kniveton, and T. J. Bellerby, 2001: A combined satellite infrared and passive microwave technique for estimation of small-scale rainfall. J. Atmos. Oceanic Technol., 18, 742-755. Tokay, A., and D. A. Short, 1996: Evidence from tropical raindrop spectra of the origin of rain from stratiform versus convective clouds. J. Appl. Meteorol., 35, 355-371. Tsintikidis, D., J. Haferman, E. Anagnostou, W. Krajewski, and T. Smith, 1997: A neural network approach to estimating rainfall from spaceborne microwave data. IEEE Trans. Geosci. Remote Sens., 35, 1079–1093. Tsonis, A. A., 1987: Determining rainfall intensity and type from GOES imagery in the midlatitudes. Remote Sens. Environ., 21, 29-36. Tsonis, A. A., and G. A. Isaac, 1985: On a new approach for instantaneous rain area delineation in the midlatitudes using GOES data. J. Climate Appl. Meteorol., 24, 1208-1218. Turk, F. J., G. D. Rohaly, and P. Arkin, 1997: Utilization of satellite-derived tropical rainfall for nd analysis and assimilation into a numerical weather prediction model. Prepr. 22 Conf. Hurricanes and Tropical Meteor., AMS, 310-311. Turk, F. J., F. S. Marzano, and E. A. Smith, 1998a: Combining geostationary and SSM/I data for th rapid rain rate estimation and accumulation. Prepr. 9 Conf. Satellite Meteorology and Oceanography, AMS, 462-465. Turk, F. J., F. S. Marzano, and A. Mugnai, 1998b: Effects of degraded sensor resolution upon passive microwave precipitation retrievals of tropical rainfall. J. Atmos. Sci., 55, 1689-1705. Turk, F. J., J. Vivekanandan, T. Lee, P. Durkee, and K. Nielsen, 1998c: Derivation and applications of near-infrared cloud reflectances from GOES-8 and GOES-9. J. Appl. Meteorol., 37, 819-831.



65

Turk, F. J., G. D. Rohaly, J. Hawkins, E. A. Smith, F. S. Marzano, A. Mugnai, and V. Levizzani, 1999: Meteorological applications of precipitation estimation from combined SSM/I, TRMM and infrared geostationary satellite data. In Microwave Radiometry and Remote Sensing of the Earth’s Surface and Atmosphere, P. Pampaloni and S. Paloscia Eds., VSP Int. Sci. Publ., 353-363. Turk, F. J., J. Hawkins, E. A. Smith, F. S. Marzano, A. Mugnai, and V. Levizzani, 2000a: Combining SSM/I, TRMM and infrared geostationary satellite data in a near-realtime fashion for rapid precipitation updates: advantages and limitations. Proc. The 2000 EUMETSAT Meteorological Satellite Data Users' Conference, 452-459. Turk,F. J., G. Rohaly, J. Hawkins, E. A. Smith, A. Grose, F. S. Marzano, A. Mugnai, and V. Levizzani, 2000b: Analysis and assimilation of rainfall from blended SSM/I, TRMM and th geostationary satellite data. Prepr. 10 Conf. on Satellite Meteorol. and Oceanography, AMS, 66-69. Turpeinen, O. M., A. Abidi, and W. Belhouane, 1987: Determination of rainfall with the ESOC precipitation index. Mon. Wea. Rev., 115, 2699-2706. Vicente, G. A., 1994: Hourly retrieval of precipitation rate from the combination of passive microwave and infrared satellite radiometric measurements. Ph.D. Thesis, Dept. of Atmospheric and Oceanic Sciences, Univ. of Wisconsin - Madison, WI, USA. Vicente, G. A., 1996: Algorithm for rainfall rate estimation using a combination of GOES-8 11.0 th and 3.9 micron measurements. Prepr. 8 Conf. Satellite Meteorology and Oceanography, AMS, 274-278. Vicente, G. A., and J. R. Anderson, 1993: Retrieval of rainfall rates from the combination of th passive microwave radiometric measurements and infrared measurements. Prepr. 20 Conf. Hurricane and Tropical Meteorol., AMS, 151-154. Vicente, G. A., and J. R. Anderson, 1994: A new rain retrieval technique that combines geosynchronous IR and MW polar orbit data for hourly rainfall estimates. Case Study: Kwajalein and TOGA-COARE. Prepr. 7th Conf. Satellite Meteorology and Oceanography, AMS, 34-37. Vicente, G. A., and R. A. Scofield, 1996: Experimental GOES-8/9 derived rainfall estimates for flash flood and hydrological applications. Proc. The 1996 EUMETSAT Meteorological Satellite Data Users’ Conf., EUMETSAT, 89-96. Vicente, G. A., R. A. Scofield, and W. P. Menzel, 1998: The operational GOES infrared rainfall estimation technique. Bull. Am. Meteor. Soc., 79, 1883-1898. Viltard, N., C. D. Kummerow, W. S. Olson, and Y. Hong, 2000: Combined use of the radar and radiometer of TRMM to estimate the influence of drop size distribution on rain retrievals. J. Appl. Meteorol., 39, 2103–2114. Watson, P. A., and Y. F. Hu, 1994: Prediction of attenuation on satellite-earth links for systems operating with low fade margins. IEE Proc. - Microw. Antennas Propagat., 141, 467-472. Watts, P. D., C. T. Mutlow, A. J. Baran, and A. M. Zavody, 1998: Study on cloud properties derived from Meteosat Second Generation observations. Final Rep. ITT No. 97/181, EUMETSAT, Darmstadt, Germany. Weinman, J. A., and P. J. Guetter, 1977: Determination of rainfall distributions from microwave radiation measured by the Numbus-7 ESMR. J. Appl. Meteorol., 16, 437-442. Wentz, F., 1997: A well-calibrated ocean algorithm for Special Sensor Microwave/Imager. J. Geophys. Res., 102, 8703–8718. Wilheit, T. T., A. T. C. Chang, M. S. V. Rao, E. B. Rodgers, and J. S. Theon, 1977: A satellite technique for quantitatively mapping rainfall rates over the oceans. J. Appl. Meteorol., 16, 551–560. Wilheit, T. T., R. F. Adler, S. Avery, E. Barrett, P. Bauer, W. Berg, A. Chang, J. Ferriday, N. Grody, S. Goodman, C. Kidd, D. Kniveton, C. D. Kummerow, A. Mugnai, W. Olson, G. Petty, A. Shibata, E. A. Smith, and R. Spencer, 1994: Algorithms for the retrieval of rainfall from passive microwave measurements. Remote Sens. Rev., 11, 163-194. WCRP, 1986: Global large-scale precipitation data sets for the WCRP. WCP, 111, WMO/TD no. 94. Wylie, D. P., 1979: An application of a geostationary satellite rain estimation technique to an extra-tropical area. J. Appl. Meteorol., 18, 1640-1648. Wyser, K., L. Rontu, and H. Savijärvi, 1999: Introducing the effective radius into a fast radiation scheme of a mesoscale model. Contr. Atmos. Phys., 72, 205-218.


66


Xu, L., X. Gao, S. Sorooshian, P. A. Arkin, and B. Imam, 1999a: A microwave infrared threshold technique to improve the GOES precipitation index. J. Appl. Meteorol., 38, 569-579. Xu, L., S. Sorooshian, X. Gao, and H. V. Gupta, 1999b: A cloud-patch technique for identification and removal of no-rain clouds from satellite infrared imagery. J. Appl. Meteorol., 38, 1170-1181. Zhao, L., R. R. Ferraro, and D. Moore, 2000: Validation of NOAA-15 AMSU-A rain rate th algorithms. Prepr. 10 Conf. on Satellite Meteorol. Oceanography, AMS, 192-195.
