demonstration of operational ice

1 downloads 0 Views 1MB Size Report
This report contain a brief model description (2), followed by a presentation of the SSMII data (3). The operational model set up is described (4), and .a few ...
The Nansen Environm.ental and ReDlote Sensing Center

NERSC Edvard Griegsvei 3a, N-5037 Solheimsvik, Norway

A non-profit environmental

research institute ,affiliated with the University of Bergen

Special Report No. 10

TO

HOV:

DEMONSTRATION OF OPERATIONAL ICE FORECASTING IN THE BARENTS SEA by (Z)ystein Skagseth

With Contribution from Peter M. Haugan, Stein Sandven. Ola M. Johannessen. Kjell Kloster and Age R. Nilsen

BERGEN, APRIL

1991

Nansen Environmental and Rem.ote Sensing Center - Special Report No. 10.

A REGIONAL ICE MODEL FOR THE BARENTS SEA HAS BEEN IMPLEMENTED AND TESTED FOR OPERATIONAL USE IN HOV. THE MODEL USES DATA FROM THE SPECIAL SENSOR MICROWAVE/IMAGER (SSM/I) ONBOARD THE SATEWTES OF THE

U. S. AIR FORCE DEFENSE METEOROLOGICAL SATELLITE PROGRAM (DMSP). THESE DATA, WHICH ARE INDEPENDENT OF CLOUD CONDITIONS, PROVIDE ICE EDGE AND ICE CONCENTRATION IN REAL-TIME EVERY DAY USED TO INITIALIZE THE MODEL WIND FORECAST FROM THE NORWEGIAN METEOROLOGICAL INSTITUTE IS USED AS INPUT FORCING TO THE MODEL THE MODEL ALSO INCLUDES THE EFFECT OF OCEAN CURRENT AND HEAT FWX AS WELL AS INTERNAL ICE STRESS. OPERATIONAL USE OF THE MODEL HAS BEEN DEMONSTRATED IN A TWO-WEEK PERIOD IN APRIL

1991. THE

REULTS FROM THE DEMONSTRATION ARE PRESENTED IN THIS REPORT.

1. 2. 3. 4. 5. 6. 7.

Introduction ......................................................................... page Model description ................................................................. page SSM/I data ........................................................................... page Operational scheme for the ice model.. .................................. page Case studies ......................................................................... page Recommendations ................................ ................................ page References ............................................................................ page

1 1 2 3 3 3 4

FigS. 1 - 3

Speciai Report no. 10

This report is made under contract for HOV (Ocean Monitoring and Forecasting Programme)

Bergen, April 29, 1991

o~~ Director

Nansen Environmental and Remote Sensing Center, . . Edvard Griegsvei 3Ak, N-5037 SOLHEIMSVI NORWAY , Telephone: +47 5 29 72 88 . r::?-.~M~\J""G

NI\\.J0

OGfJ-,

1. Introduction.

A regional ice forecasting model for the Barents Sea has been implemented an tested at the Nansen Environmental and Remote Sensing Center (NERSC). The study have been focused on; 1) real time use of SSMII data for initializing the model ice concentration field and, 2) using the atmospheric model operated by The Norwegian Meteorological Institutte (DNMI), Lamb 50, as model wind forcing. This report contain a brief model description (2), followed by a presentation of the SSMII data (3). The operational model set up is described (4), and .a few examples of model results are presented (5). Finally some model improvements are suggested (6). 2. Model description The ice model used in this study is based on the original code first described by Hibler (1979). The essential idea in the model is to couple the dynamics to the ice thickness characteristics by allowing the ice interaction to become stronger as the ice becomes thicker and/or contains a lower percentage of thin ice. The prognostic model parameters are l)ice concentration, 2) ice thickness and 3) vector ice drift. The basic elements of the sea ice model are:

Newton's second"law: 'ta

+

aIr stress

'tw

water stress

1:F=ma

mg gradH + sea surface tilt

F

mfkxu

internal ice force

Coriolis force

=

mDtu acceleration

In addition the model contains a thermodynamic code specifying the ice growth and decay. The equations describing the evolution of the ice thickness (h) due to dynamic + thermodynamic effects are: (dhldt)

=- grad (uh) +Sh

(dNdt)

=- grad (uA) +SA

1

where A is the ice concentration and SA and Sh thermodynamic source and sink terms. The internal ice force term is calculated relating the ice stress to ice deformation and strength. Figure 1 show the approximate model area. The model resolution is 20 km and the dimension of the model array is 70 * 88. The principle of the boundary conditions is to; 1) allow inflow of ice from the artic basin into the model area, whereas, 2) no ice from the Norwegian Sea should advect into the model area. 3. SSMII data The SSMII is a passive mIcrowave. radiometer measuring at four frequencies; 19,22,37 and 85 GHz. The 19 GHz frequency on the SSMII is

used together with the 85GHz to compute ice parameters as ice edge and ice concentration. The absolute resolution of the measurements are approximately 20km at the highest frequency. A continuos and relatively sharply defined ice edge, defined as e.g. 30% ice concentration, can be found with spatial accuracy of 10-20km (Svendsen et.al 1983). As mentioned in part 2 the ice concentration is one of the prognostic variables in the model. Therefore it is of invaluable importance to have good information about the initial ice concentration field with respect to ice forecasting on an operational base. Due to the weather independency of the SSMJI sensor regular ice information can be retrived. For the test period, 9 - 23 April 1991, six model test run initializing with SSMII ice concentration have been performed. In Figure 2 the initial ice concentration fields are shown. The scale indicating % of Ice concentration is shown in Figure 3. White arrows indicate the vector wind from the Lamb 50 atmospheric model. Possible data gaps in the new SSMII file is filled out using the prior ice concentration data. The 30% ice edge (yellow) is changing somewhat through the period from . 9 - 23 April 1991. The most obvious ice edge variability is seen in the inner part of the Barents Sea (upper middle of the images). A 50km southward extension of the ice edge north of Bear Island is observed from the 12th April 1991 to the 23th April 1991. In addition there is also a noteable change in ice concentration. 60-70 % (orange) on 12 April 1991 compared to 80-90% (dark red) on 23 April 1991. ,

.

2

4. Operational scheme for the ice model The satellite passage covering the Barents Sea is at about 00 GMT each day and is accessible for use after 3-4 hours. The SSM/I data file is thereafter transferred by DATAPAK or direct call (modem). The transfer time is on the order of 90 minute, but can potentially be reduced by a factor of 10. The SSMII data are converted to ice concentration at the NERSC using the NORCSEX algorithm (Svendsen et aI., 1983) and then interpolated in space to the ice model grid. The file is thereafter transferred to the CRAY, via telnet, and used as initial ice concentration. The model is run with 2 hour time step from T=O to 36, corresponding to the Lamb 50 forecast. The CRAY computing time is about 10 minute but can be reduced. Consequently the model could be run on a workstation. The model results are transferred via telnet to workstation for graphical presentation. If i.e. the transfer of SSMII data starts at 0600 GMT the presentation of the ice model results could be finished by approximately 0900 GMT.

5. Case studies. To demonstrate the model the 24 hour forecast from April 9 00 GMT is compared with the corresponding ice concentration from the SSMII from 10 April 00 GMT, and the 24 hour forecast from 16 April 00 GMT is compared with the SSMII ice concentration from 17 April 00 GMT (Figure 3). The model is run with the ocean heat flux turned off and therefore no ice melting limit the ice extension., Focusing on the area between Bear Island and Spitzbergen, for both cases, the wind is blowing in off ice direction. The ice edge for both 24 hour forecast (9 and 16 April 1991) extends further seaward within the forecast period. This is also the case for the SSMII retrived ice edge for the corresponding period. 6. RecomendatioDS. 1) To integrate the model for long term forecast a well defined ocean heat flux field need to be defined. Since the predominant variation of this field is an anual cycle, monthly or halfmonthly mean ocean heat flux fields should be sufficient. In addition a climatologocal background ocean current field should be defined. 2) The ERS-l SAR data should be blended with the SSMII data. In particular the ice vectors derived from the ERS-l would be very usefull to use to validate the Hibler model and possibly

3

"tune" this for the Barents Sea. 3) It would also be interesting to compare the ice models run at the NERSC and at DNMI.

References: Hibler, W.D., A Dynamic Thermodynamic Sea Ice Model, J.P.O., pp. 815846, July 1979. Svendsen, E. et aI, Norwegian Remote Sensing Experiment: Evaluation of the Nimbus 7 Scanning Mutichannel Microwave Radiometer for Sea Ice Research, J.G.R., pp. 2781-2791, March 30, 1983.

Figures: Figure 1. Map of the Barents Sea showing the inner model area. Figure 2. The ice concentration fields (%) retrived from the SSMII data for different days used as initial condition for the ice model are shown. The color code for the different ice concentrations are shown in Figure 3. Land is black while blue is open water. The ice concentration in % are given as : yellow 30-50 %, orange 50-70 %, red 70-90 %, pink >90%. White arrows are the vector wlnd. Figure 3. The 24 hour forecast from April 9 00 GMT is shown , together with the measured SSM/I ice concentration field from AprillO 00 GMT. Similarly the 24 hour forecast from April 16 00 GMT is shown together with the measured SSMII ice concentration field from April 17 00 GMT. The color code is the same as in Figure 2. White arrows are' the vector wind.

.•

4

Figure 1. Map of the Barents

8~a

.

showing the inner model area .

April 9 1991 00 GMT

April 10 1991 00 GMT

April 12 1991 00 GMT

April 16 1991 00 GMT

April 17 1991 00 GMT

April 23 1991 00 GMT

Figure 2 ~---------------------------------------------

~~

24 hour forecast from April 9 00 GMT

SSM/I ice edge April 10 00 GMT

24 hour forecast from April 16 00 GMT

SSMII ice edge April 17 00 GMT

Colour scale

Figure 3

~-----------------------------------------

fii~