Spectral signature characters based on measurement ...

3 downloads 886 Views 356KB Size Report
around) characters more sensitive to SSC variation at bands 500-900nm. Remote .... The grey plate in spectralon (diffuse reflector) should be scaled in a special ...
Spectral signature characters based on measurement in situ of high turbid water in Yangtze River estuary Fang SHEN* a, Jie ZHANG a, Zhiguo LIU a, Zhihua MAO b a State Key Laboratory of Estuarine and Coastal Research, East China of Normal University, Shanghai, 200062, China; b State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, State Oceanic Administration, Hangzhou, 310012, China ABSTRACT Estimated total sediment mass in the fluid mud system is about 4.25×108 tons per year as river inputs into Yangtze River estuary. The spatial-temporal distribution of suspended sediment concentration (SSC) is concerned by estuarine and coastal researchers. In this work, in situ measurements were carried out in Yangtze River estuary in Sep.2004, Jul. and Aug.2005 and Feb.2006. An experiment in barrel on land as a complementary measurement was conducted in July 2006. All of the spectral signatures observed show three spectrum-peak (at wavelengths 590nm, 690nm and 810nm around) characters more sensitive to SSC variation at bands 500-900nm. Remote sensing reflectance (Rrs) at 590nm around is bigger than that at 690nm around when SSC is about 1.0g/l. Here a parameter SRP (Sediment Response Parameter) described a changeable rate of ratios among three spectral peaks was defined, which presented a logarithmic decreasing tendency with the SSC increasing. The preliminary result shows that SRP can reflect SSC variation in the estuary. Keywords: high turbid water, suspended sediment concentration (SSC), remote sensing reflectance (Rrs), sediment response parameter (SRP).

1.

INTRODUCTION

Suspended sediment matter is a main component of high turbid water in estuarine and coastal areas. Suspended sediment concentration (SSC) has a great effect on transparency, turbidity and color of water, as well as on depositing and eroding along shores. Therefore, it is significant for aquatic ecosystem, harbor engineering and navigation channels to understand SSC amount and its spatial-temporal distribution in estuarine and coastal areas. Remote sensing technique developments provide a powerful tool for the SSC monitoring. Some retrieval models of SSC from satellite data such as MODIS data, SeaWiFS data, SPOT data and etc. (e.g., Mertes et al., 1993; S.Tassn, 1994; S.Ouillon et al., 1998; D.Doxaran et al., 2002; A.K.Mishra, 2004) were applied in coastal areas through measurement in situ and synchronous water samples collection. These models mostly are empirical relationships established between SSC and spectral signatures, which generally are suitable to local areas. As a contribution to these efforts, in situ measurements and in barrel experiment as a complementary test were carried out in Yangtze River estuary (Fig.1) during these cases, including high and low tidal cycle, flood and ebb tide, flood and dry seasons (in Sep.2004, Jul. and Aug.2005, and Feb.2006).

2.

STUDY

AREA

Yangtze River is the largest river in China and the third large in the world. Yangtze River estuary is an extreme example of high turbid water ─ sediment dominated case 2 waters influenced by river inputs. It was estimated that total sediment mass in the fluid mud system is about 4.25×108 tons per year (statistic from 1995 to 2005 at Datong observation site) (Chen Shenliang, 2001). Because tidal fluid system has a strong impact on the estuary, spatial-temporal distribution of suspended sediment matter frequently varies in the estuary, which is more concerned by hydrodynamic researchers, coastal engineers and geographers. *

[email protected]; phone: 086 021 62233467

Remote Sensing of the Marine Environment, edited by Robert J. Frouin, Vijay K. Agarwal, Hiroshi Kawamura, Shailesh Nayak, Delu Pan, Proc. of SPIE Vol. 6406, 640608, (2006) 0277-786X/06/$15 · doi: 10.1117/12.694207 Proc. of SPIE Vol. 6406 640608-1

Ifl• I5

.

Z,!li,!jig A

I12 I5

121•

31•

CIIOgIniugislaRd

3, W

Heugsha island

Naugaug

A

a

A

:4

-

-.

.

Haighoubay ft

A tide smvey site SSC sample point

SSC in ]ow tide (O.2g

)

SSC in ragh tide (O.2g

45•

3D

Fig.1. Location map of Yangtze River estuary and data samples sites in 2004

It was showed that the amount of SSC increased gradually from the estuarine entry (Xuliujing site —— hydrological observation station) to the estuarine gate and to Hangzhou bay in Fig.1. The SSC from surface water was only 0.1289g/l in Xuliujing site, increased to 0.3580 g/l in Yinshuichuan site (estuarine gate), and yet reached at 1.5558 g/l in Hangzhou bay in Fig.1. Yangtze River estuary has well-developed zone of turbidity maximum, which is near the estuarine gate (Shen Huanting & Pan Ding'an, 2001). This survey in Yangtze River estuary indicated that the SSC variation with tidal cycle is evident, but the influence degree is not same in different sections of estuary. We respectively calculated the average of SSC during high and low tide periods for each observation site (see Fig.1). It was found that tidal cycle impacts on the distribution of SSC evidently from Hengsha island to the outer estuary, especially in Hangzhou bay. However, it has less influence towards the inner estuary (see Fig.1). Moreover, the SSC in Yangtze River estuary usually varies with flood and ebb tide at the time of day (e.g. Fig.2). Fig.2 demonstrated the relationship of the SSC and tidal height in the estuarine gate and Hangzhou bay. It was shown that the amount of SSC is big in flood tide period, and is small in ebb tide period in Fig.2. In addition, it is analyzed that, although the time period from 9:00 to 15:00 is just suitable for optical measurement, the SSC variations are weak at a certain observation station (e.g. K1, H1 or K2 site) in this period, and the maximum of SSC is generally less than 0.5g/l.

Sr (gTh — —

— — Kl-TH Hl-TH TH (cm

Kl.K2-SSC

51

2.5

2.0

_________

____

____

0.5 — _____ —

x

400

21

II

0123 456789l0lll2l3l4l5l6l7l8l9202l2223

0

Time (horn)

Fig.2. Suspended sediment concentration (SSC) and tidal height (TH) at K1, H1 and K2 sites in Yangtze River estuary

Proc. of SPIE Vol. 6406 640608-2

3.

MEASUREMENT AND METHOD

In this optical measurement, optical data are recorded with ASD Field Pro spectroradiometer between 350 nm and 2500 nm (bandwidth: 1.4 nm/sensor in 350-1000nm, 2nm/sensor in 1000-2500nm). Fixed scan time is 0.1 second. Scalar reflectance R (%) can be yielded in a direct measurement with the spectroradiometer, but remote sensing reflectance Rrs (sr-1) can only be estimated approximately with some methods, one of which is followed as (Mobley, 1999)

Rrs ( sr −1 ) =

Lw . E d (0 + )

Here Lw (W m-2 sr-1 nm-1) is leaving-water spectral radiance that also is not directly measurable, which can approximately be estimated with a formula of Lt − ρLs . The total upwelling radiance Lt from the water can be measured when the sensor directs towards the water above the surface, and Ls represents incident sky radiance when the sensor directs towards the sky. Water surface reflectance factor ρ depends on sky conditions, sun zenith angle, viewing geometry and wind speed (Mobley, 1999). E d (W m-2 nm-1) represents the

n -Ls

tc — r —— N \Wt \.\8



Fig.3. Spectroradiometer and sun geometry relation in above-water radiance measurement

downwelling spectral plane irradiance onto the surface, which can approximately be estimated from a grey plate (which has a known diffuse reflectance factor R p ) radiance measurement. It nearly equals to

πLd / R p , where the downwelling radiance Ld

can directly be measured

when the sensor points vertically downwards the grey plate. The grey plate in spectralon (diffuse reflector) should be scaled in a special laboratory before the measurement, so as to obtain an accurate E d . Above-water radiance Lt can be measured while the radiometer is pointed downwards towards the nadir at the direction of θ angle (here 45° is adopted) in the vertical plane transecting the solar azimuthal plane with the angle of φ(from 90°~135°) for minimizing solar glitter effect (see Fig.3). Then, we turned the same sensor pointed upwards towards the sky at the direction of same angle (e.g. 45°) within the same viewing plane for measuring the skylight radiance Ls . Here, the ρ value is regarded as 0.0337 for a clear blue sky (Mobley, 1999). 3.1 In situ measurements Data samples were collected in Yangtze River estuary in Sep. 2004 (SKLEC ‘973’ oceanographic survey), Jul. and Aug. 2005, and Feb.2006, which included hydrological data, oceanographical data and suspended sediment samples. Synchronously, the above-water optical measurements were carried out in the surveys, mainly which included two cruising measurements in Nancao navigation channel (near the estuarine gate) in 2004 (a series of scalar reflectance shown in Fig.4 were directly measured) and in Nanguang channel (between Changxing island and Shanghai land) in 2006 (a series of Rrs shown in Fig.5 were estimated with apparent optical data measured), and fixed five-station measurements in 2005 (but the SSC variations measured are not evident with the time series, e.g. Fig.2 indicated) . From Fig.4 and Fig.5, whatever scalar reflectance and Rrs (sr-1), generally increases with the SSC augment.

Proc. of SPIE Vol. 6406 640608-3

SSC (g/l)

0.57 0.47 0.46 0.12 0.11 0.09 0.08 0.07 0.06

Reflectance

0.25 0.2 0.15 0.1 0.05 0 350 450 550 650 750 850 950

SSC (g/l)

0.05

0.0807

0.04 Rrs (1/sr)

0.3

0.0707

0.03

0.069

0.02

0.069

0.01

0.0637 0.0477

0 350

450

550

650

750

850

Wavelength (nm)

Wavelength (nm)

Fig.4. Scalar reflectance (R) in Nancao channel in Sept.2004

Fig.5. Remote sensing reflectance (Rrs) in Nangang channel in Feb. 2006

However, it is difficult to collect a series of optical data from SSC gradational variations in one cruise or station. The differences exist among multi-cruise or multi-station data due to various sun illuminations, cloud cover, wind speed, hydrodynamic condition and etc. In addition, in situ measurement is in influence since the ship platform is unstable. Therefore, it is necessary to do an experiment as a complementary optical measurement in order to further understand spectral signature characters of SSC variation in Yangtze River estuary. An experiment and measurement thus were designed and adopted in water pool or barrel on land. 3.2 In barrel experiments Sediment samples were collected from Chongming east tidal flat where the suspended sediment matters deposited, and then separated into the gradational sedimentary matters according to particle size with a filter. We chose the sedimentary matter whose particle size is less than 0.063mm in the experiment, for particulate matters at less than this size are easily suspended. Through laboratory analysis, the suspended matter is a mixture of organic and mineral composites, where the organic fraction represents only 0.904% of the total material. The mineral fraction is composed of quartz and feldspar (total 90%), micas (2%) and anti-weathering mineral (2%), while clay phases contain four minerals: montmorillonite (4.17%), illite (66.62%), kaolinite (16.55%) and chlorite (12.67%). The grain-size distribution is: 0.063mm: 0.37%. SSC (g/l)

0.08 0.07 0.06

Rrs (1/sr)

0.05 0.04 0.03 0.02 0.01 0 350

450

550

650

750

850

Wavelength (nm)

950

0.5495 0.505833333 0.483666667 0.389833333 0.302 0.2615 0.260666667 0.150833333 0.105833333 0.0915 0.086833333 0.077333333 0.0665 0.064 0.053166667 0.044666667 0.043833333 0.031333333 0.017166667

Fig.6. Spectral signatures observed with the SSC variations at an experimental barrel in July 2006

Proc. of SPIE Vol. 6406 640608-4

An experimental barrel with black color and dark brightness was placed on an empty ground. Firstly, optical measurements should be taken without water filled in the barrel. Secondly, a group of 25g sediment samples which had been weighted and prepared before the measurement would be thrown into the barrel filled with water one by one. Meanwhile, the water body was continually churned up until close to uniform turbid water, and optical backscatter sensor (OBS) was used to monitor the SSC variation in vertical column. Lastly, above-water radiance measurement was carried out when the distribution of suspended sediments is approximately homogenous in vertical column, with a co-incident water sample for each optical measurement. The experiments and measurements were accomplished from 9:00am to 3:00pm in July 2006 under the conditions of clear blue sky and low wind speed. A series of Rrs (sr-1) with SSC variation were deduced in Fig.6 shown.

4.

DISCUSSION

Through the analysis for measuring and estimating results, it was found that spectral bands from 500 to 900nm can effectively reflect the SSC. The Rrs or R corresponding to each spectral wavelength will increase with the SSC increasing (see Fig.4, Fig.5 and Fig.6). All of the (a) spectral signatures resulted from in situ measurement and in barrel experiment almost show 1 0.95 three-peak character in 500-900nm. The first 09 spectral peak is located at 590nm around, the 0.85 second at 690nm around and the third at 810nm 0.8 around. The Rrs or R at the first peak is usually 0.75 higher than that at two other peaks when the SSC is 0.1 very low (e.g. 1.0g/l), the Rrs or R at the third peak will surpass that at the first and second peak, which situation will occur in high turbid water observed in Yangtze River estuary. 0.9 To discuss the relationship between SSC and ——— 0.8 spectral signatures, we calculated three mathematic 0.7 correlation coefficients commonly including logarithm, Index and linear relationships between 0.6 SSC and scalar reflectance which was directly 0.5 obtained by the radiometer in Sept. 2004 (see -1 400 500 600 700 800 900 Fig.7.(a)), between SSC and Rrs (sr ) which was estimated with measurable data respectively from Feb.2006 (see Fig.7.(b)) and July 2006 (see Fig.7.(c)). The purpose of Fig.7 is to find out which spectral wavelength between 500nm-900nm is 0.8 more sensitive to SSC variation. Fig.7 showed that correlation coefficient between SSC and Rrs ao corresponding to any wavelength, whatever is of logarithm, index or linear correlations, is relative 0.2 high by comparisons. However, it is difficult to establish a formula of relationship between SSC and Rrs corresponding to certain single 350 450 550 650 750 850 wavelength. It was implied that the SSC inversion vavelengrd (mu) can not only be determined by single spectral wavelength. The relationship of multi-wavelength composite should be considerable in the SSC Fig.7. Three mathematic correlations between SSC and corresponding spectral reflectance or Rrs. (a) data from inversion of high turbid water in Yangtze River Sept. 2004 (b) data from Feb.2006 (c) data from the estuary.

W]tli (!O

W]tli (!O



. 0. 4i.-•

experiment of July 2006.

Proc. of SPIE Vol. 6406 640608-5

SSC (g/l)

0. 5495 0. 505833333 0. 483666667 0. 389833333 0. 302 0. 2615 0. 260666667 0. 150833333 0. 105833333 0. 0915 0. 086833333 0. 077333333 0. 0665 0. 064 0. 053166667 0. 044666667 0. 043833333 0. 031333333 0. 017166667

0. 08 0. 07

Rr s ( 1/ sr )

0. 06 0. 05 0. 04 0. 03 0. 02 0. 01 0 1 2 3 No. of wavelength corresponding three spectral peaks

SRP =

Pi − Pj

∑ P +P

i =1, 2 j = 2,3

i

.

j

Two scatter-dot charts of the SRP and the SSC were listed in Fig.9, where data sets respectively from Sept. 2004 as well as July 2006. Meanwhile, two logarithmic relationships of SRP and SSC were given in Fig.9, and two high correlation coefficients were obtained from different data sets. Fig.9 showed that two logarithmic formulas were similar. However, it was not coincident we thought because original data sets among the formulas were

(a)

0.5 0.4

y = -0.1825Ln(x) - 0.1023 R2 = 0.9152

0.3 0.2 0.1 0 -0.1 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Suspended sediment concentration (g/l) (b)

0.6 Sediment response parameter

Fig.4, Fig.5 and Fig.6 showed that the Rrs corresponding to three spectral peaks were greatly sensitive to the SSC variations, which were reflected in Fig.8. Meanwhile, it was found that the ratio of P1 (represents the Rrs corresponding to 590nm) and P2 (represents the Rrs corresponding to 690nm), as well as the ratio of P2 and P3 (represents the Rrs corresponding to 810nm) gradually decreased with the SSC increasing. That situation demonstrates that the two ratios (P1 / P2 and P2 / P3) are accessible to the SSC variation. Thus, the amount of P1 / P2 plus P2 / P3 correlated with the SSC is thinkable. We respectively normalized the two ratios in order to balance their effect (or for scale identity), i.e. the P1 / P2 was rewritten into (P1-P2)/ (P1+P2), and the P2 / P3 was likewise done into (P2-P3)/ (P2+P3). In order to find an empirical relationship among them for the SSC inversion, here a parameter SRP (Sediment Response Parameter) was defined as

Sediment response parameter

Fig.8. Remote sensing reflectance of three spectral characteristic peaks under the SSC variations No.1: 590nm; No.2: 690nm; No.3: 810nm.

0.5 y = -0.1818Ln(x) - 0.1262 R2 = 0.9627

0.4 0.3 0.2 0.1 0 -0.1 0

0.1

0.2

0.3

0.4

0.5

Suspended sediment concentration (g/l) Fig.9. Scatter-dot charts and correlations between SRP and SSC (a) data sets from Sept. 2004; (b) data sets from July 2006

Proc. of SPIE Vol. 6406 640608-6

0.6

respectively from two measurements, i.e. different measuring date (one was in 2004, the other in 2006), different measuring technique (one was by directly measuring scalar reflectance, the other by estimating remote sensing reflectance) and different observing platform (one was based on measurement in situ, the other in an experimental barrel). It is concluded that the SRP mostly depends on the SSC variation, and is less related to measuring date, conditions and even technique.

5.

CONCLUSION AND FUTURE WORK

Although numerous in situ measurements were expected to be carried out in Yangtze River estuary during high and low tidal conditions, tidal cycle, flooded and dry seasons, some data obtained were not satisfied due to the complicated survey situations such as ship swaying, ship shadow, sky condition, hydrodynamic state, wind speed and etc. Therefore, an experiment in barrel as a complementary measurement was conducted on land, and spectral signature characters observed from which were compared with that from in situ measurement then. We found that spectral bands from 500 to 900nm could effectively reflect suspended sediment concentration of the high turbid water in Yangtze River estuary. All of the spectral signatures, whatever measured in situ or in laboratory, mostly showed three spectrum-peak characters in the bands area, and the signatures of three peaks are more sensitive to the SSC variation. The Rrs corresponding to the first peak is bigger than that corresponding to the second peak when SSC is about 1.0g/l. These characters will be useful to the selection of remotely sensor channel applied to SSC retrieval from remotely sensed data in high turbid water. A parameter SRP described a changeable rate of ratios among three spectral peaks was defined. It was indicated that SRP presented a logarithmic decreasing tendency with SSC increasing in this study. Here, two logarithmic relationships between SRP and SSC were established respectively for two measuring data. It was found that the two relationships were similar, although which were from two different measuring data. It is demonstrated that SRP is mostly determined by SSC amount itself, that is to say, the SRP as a parameter is able to reflect the SSC. This implies that SRP is capable of the SSC retrieval of satellite data in the estuary. Moreover, the SRP as a ratio is accessible to the application of “ocean color” satellite data, since satellite spectral bands ratio can minimize atmospheric influences. The future researches include the ideal design of experimental barrel or pool with a churn-dasher capable of uniformly mixing sediments, so that it can satisfy the need of turbid water with the high SSC (e.g. more than 1.0g/l). It is expected that numerous optical measurements will go on being conducted in order to validate the SRP. Moreover, the retrieval accuracy of the SSC from satellite data in the estuary will be improved, with future developments of high-spectral and –spatial resolution techniques. Acknowledgments: this work was supported by the foundations of China Education Ministry key research project (105076) and Shanghai science and technology project (04DZ12049). The authors are extremely grateful to Dingbo KUANG (Shanghai Institute of Technique Physics, Chinese Academy) for help with experiments in barrel, and my colleagues Jiufa LI, Yunxuan ZHOU, Heqing CHEN and Yi MENG for helps with the field work. In situ measurement in 2004 was conducted with the support of SKLEC ‘973’ oceanographic survey.

REFERENCE 1. 2. 3. 4. 5. 6.

Chen Shenliang, “Seasonal neap-spring variation of sediment concentration in the joint area between Yangtze Estuary and Hangzhou Bay”, Science in China (Series B), 44(supp.), 57-62, 2001. Shen Huanting, Pan Ding'an, Turbidity Maximum in the Changjiang Estuary, Ocean Press, Beijing, 2001. Mertes et al. “Estimating Suspended Sediment Concentration in Surface Waters of the Amazon River Wetlands from Landsat Images”, Remote Sensing of Environment, 43, 281-301, 1993 S.Tassn, “Local algorithms using SeaWiFS data for the retrieval of phytoplankton, pigments, suspended sediment and yellow substance in coastal waters”, Applied Optics, 33(12), 2369-2378,1994 S.Ouillon, P.Forget, J.M.Froidefond, J.J.Naudin, “Estimating suspended matter concentrations from SPOT data and from field measurements in the Rhone River plume”, Oceangraphic Literature Review, 45(4),712, 1998 David Doxarana, Jean-Marie Froidefonda, Samantha Lavenderb, Patrice Castainga, “Spectral signature of highly turbid waters. Application with SPOT data to quantify suspended particulate matter concentrations”, Remote Sensing of Environment, 81, 149– 161, 2002.

Proc. of SPIE Vol. 6406 640608-7

7. 8.

A.K. Mishra, “Retrieval of suspended sediment concentration in the estuarine waters using IRS-1C WiFS data”, International Journal of Applied Earth Observation and Geoinformation, 6, 83–95, 2004. Curtis D. Mobley, “Estimation of the remote-sensing reflectance from above-surface measurements”, APPLIED OPTICS, 38( 36), 7442-7455, 1999.

Proc. of SPIE Vol. 6406 640608-8