0213—1177/84 $0.00 + .50 Copyright © COSPAR
9l-96, 1984 Adv. Space No.11, Printed in Res. Great Vol.4, Britain. All pp. rights reserved.
CAPABILITY OF BHASKARA-II SATELLITE MICROWAVE RADIOMETER BRIGHTNESS TEMPERATURE DATA TO DISCRIMINATE SOIL MOISTURE CONDITIONS OF INDIAN LANDMASS K. S. Rao,* P. Venkatachalam,* A. Sowmya,~ A. K. Kandya** and T. J. Majumdar** *
Centre of Studies in Resources Engineering, lIT, Bombay,
India **lndian Space Research Organisation, SAC, Ahmedabad, India
ABSTRACT iiith the objective of developing !i~icrowaveRemote Sensing technology in the country, India has launched a series of Satellites Bhaskara—I and II with the microwave radiometer capability. In this paper, an attempt is made to demonstrate the capability of the brightness temperature data acquired by these radiometers to discriminate various soil moisture conditions of Indian land mass. The analysis show that large areas assessment of soil moisture is possible to a limited extent. INTRODUCT I cN
Large scale assessment and continuous monitoring of soil moisture is of great importance from the point of view of agriculture and hydrology. It is also important in predicting drought conditions which are widely prevalent in India. Microwave Remote Sensing has demonstrated its capability to estimate quantitatively the soil moisture content even upto one meter depth of the soil /1,2,3/. India has launched a series of satellites namely Bhaskara—I and II with microwave radiometers operating at 19.1, 22.235 G~z (for Bhaskara—I) and also 19.35, 22.235, 31 GHz (for Bhaskara—II). Though these frequencies are ideally suitable to study atmospheric parameters (water vapour and cloud), the frequencies around 19 GIiz can provide information on surface soil moisture 14/. This has been demonstrated through this paper by studying the brightness temperature data acquired during Feb. 1983. SELEcTION OF BRIGFT~NE5S TEMPERATURE DATA The brighthess temperature (T ) data acquired by Bhaskara—II Satellite Microwave Radiometers (SAMIR)Boperating at 19.35, 22.235 and 31.4 GHz /5/ has been used in the present study. This data comprises 23 passes during Feb. 1983, the details of which are provided in the paper by K.S.Rao et al /8/. The ground resolution of the data is a 100 km. diameter circular resolution cell with a temperature resolution of l°K.The broad specifications of Bhaskara II 5A!.~IR system are given by Calla et al. /6/. The ground traces of the data are shown in Fig.l which indicates that there is a good coverage of data over the Indian subcontinent. The data extends from noi~thern Himalayan snow regions to southern sea regions. SC.thENING CF 3r~IGi-IT?’~SS T~.PEi-iAT~REDATA
To have a quick look at the quality of the data, a computer program has been developed to plot brightness temperature versus spin number on a line printer, an example of which is shown in Fig.2. It has been found that the data for 31.4 GHz is noisy for many of the passes. However, in general, the data is very accurate from the point of view of ground location which can be studied on the basis of land—sea crossings for all the three radiometers. The data which is free from all such noise is considered for statistical analysis.
91
K.S. Rao at ci.
92
ATM~PHERICCcFiRECTIONS TO 19.35 GHz DATA The T data acquired by SAMIR is influenced by atmospheric water vapour and 1~quidwater (cloud). Therefore, the variations observed in ID data may be partially due to these effects. It has been found from A.P~T.cloud pictures that most of the data is free from clouds. K.S. Rao et al /7/ developed an operational model to account for atmospheric water vapour influence on 19.1 GHz. T data With the help of another measurement at 22.235 GHz. In the preseAt context, there is a slight change in the frequency (19.35 GHz) and so the coefficients of the model were recalculated suitable to 19.35 GHz and are given below. TB 19.35
=
Ao
+
A1 T~19.35
+
A2 T~22.235
...(1)
where T~19.35 is the TB free from water vapour influence; 19.35 and T~22.235 are the satellite measurements of TB at 19.35 and 22.235 GHz; Ao = 5.022; A1 = 1.399; and A2 = —0.422. The TB data acquired at 19.35 GHz has been corrected using the above model. SEPARATICt’! OF DATA ACQUIRED WER HIMALAYAN SNOJi REGIC~S
From the statistical analysis of Bhaskara—II SAMIR T data, K.S. Rao et al /8/ pointed out that there is a considerable spread ~n the T0 data corresponding to high to low soil moisture conditions. However, th~sspread in T cannot be attributed to soil moisture alone as wet snow/ice also has a s~mi1arrange of TB values. In this paper, an attempt is made to eliminate the data acquired over snow! ice regions with the help of geographical threshold (Refer Fig.l). This is possible over the Indian subcontinent as snow/ice occurs only over Himalayan regions /9/. The data over sea will not interfere with the data over land as their respective brightness temperature ranges are quite different. Hence, by putting a threshold on ~ the sea data can be easily separated from the land data. RESULTS AND DLSCWSION
The T for all the three spin number to visualise llite passes from sea to example of computer plot
radiometers have been plotted as a function of the general trends of T~variations as the sateland and then over Hima~ayansnow regions. An is given in Fig.2.
SPECIFIC 0B~E~VATICNS
vs. Spin number for all the passes, the following specific observati~ns are made a) Orbit No: 6764 date : 10.2.83 From the computer plots of T
A sudden rise in T~has been noticed at spin no: 30 spreading over 250 km. length. The change in ID is 16°K, l3°Kand 6°Kfor 31.4, 19.35 and 22.235 GHz respectively. From ~he ground traces, this location corresponds to Andaman Islands. b) Orbit No: 6767
date : 10.2.83
The data of this pass acquired over Himalayan regions also shows considerable variation in TB for all the three radiometers corresponding to different conditions of snow/ice and for partially ice covered regions. c) Orbit No.: 6824
date : 13.2.83 and Orbit No: 6869
date : 16.2.83
These two passes are over Sri Lanka island and it has been reflected in the TB data as shown below. Radiometer Orbit 6824 Orbit 6869 a1 (31.45 GHz) 174—217 (43 rise) a2 (19.35 GHz) 145—177 (32) 147—197 (so) a3 (22.235 GHz) 184—208 (24) 206—230 (24) —
Bhaskara—lI Satellite Microwave Radiometer Brightness Temperature Data
93
RESULTS The I data comprising 23 orbits has been pooled out radiometer—wise irres— pecti~eof the look angl~(The hook angle at which SAMIR data has been acquired are —5.60, —2.8 , +2.8 , +5.6 which are very close to NADIR/Sf. It was shown theoretically /10/ that the emissivity variations at low look angles are negligible) and histograms have been plotted as a function of T~.The histograms have been generated for the following sets of data for e~ch radiometer. (a) Complete data (b) Data acquired region (refer Cc) Data acquired
irrespective of geographical position. over land and s~aregions excluding Himalayan snow/ice Fig.l). only over snow/ice regions.
The histograms of these three categories have been normalised to the maximum occurrence of a particular radiometer and plotted superimposed one over the other to enable comparison. The sets of histograms for all the three radiometers and also the I data at 19.35 GHz corrected for atmospheric water vapour have been preseAted in Fig.3. Certain prominent features such as sea and land peak positions with RMS value, spread in T3 data over land corresponding to different conditions of soil and different conditions of snow/ice have been extracted from the histograms and are given in Table 1. ANALYSIS AND DISCU3SICN It is difficult to analyse each I data due to various reasons such as coarse resolution; nonavai1abi1it~of ground truth; limitations on the data etc. However, certain inferences can be drawn on the basis of statistical analysis. 1) All the histograms show two prominent peaks, one corresponding to sea and another correspondin8 to land regions. Though the T resolution of the radiometer system is 0.5 K /6/, the above peaks indicat~effective RMS error of the order of ±6°K. This may be due to varying conditions of sea and land as the total data used in the analysis is spread over about 12 days. 2) The influence of atmospheric water vapour on TB is considerable over sea compared to land. This can be noticed from the shift of sea peak position in the corrected data by about 8°Kcompared to the negligible shift of land peak position. 3) The spread in the snow/ice data is of the order of 45°Kwhich probably included certain portions of land due to the approximate way of choosing the geographical threshold for separating the data and also due to partial snow/ice cover portions of Himalayan regions. 4) Even after removing the snow covered data from the bulk of T0 data, considerable spread in T~data can be noticed for all the three radiometers and also for the data corrected for atmos8heric water vapour effect. In general, the spread is of the order of 45 K. 5) The above indicated spread in TB data acquired over land regions can be attributed to the following a)
b)
Since I is the product of emissivity and physical temperature, it is likely %hat the spread in T~ is due to the changes in physical temperature as the pooled data h~s been acquired over different timings of the day as well as different days of Feb.1983.
Varying ernissivity conditions of land have direct influence on the soil moisture conditions. Schmugg /11/ pointed out that for a one percent change in soil moisture, there is about 3°Kchange in T9. TB data as re~lected by peaks over and effective sea (i.e. resolution with effective 6) the Considering theland worst of theRMS value of ±6K),the spread in TD is much higher than the RMS error which definitely has a sh&e in soil moi~ture variations.
94
K.S. Rao at ai.
7) During the month of February, the surface soil moisture is very low and also 19 GHz frequency cannot penetrate more than one cm. Hence, the effect of soil moisture variations indicated in the histograms are quite reasonable. However, the choice of lower frequencies will give better results. CONCLUSI ONS In spite of various limitations of the data, it has been successfully demonstrated adopting the statistical approach for the analysis that the SA1.~IR data can provide some information on soil moisture conditions of Indian land mass, thus showing its capability to discriminate various soil moisture conditions of land mass in addition to its capability to study atmospheric phenomena. ACKNGVJLEDGEMENTS The authors are grateful to Prof.R.K. Katti, Head, CSRE for providing all the necessary facilities and encouragement all through the work. The authors also wish to thank Shri D.S. Kamat, Head, IPAD, RSA, SAC (iSao) for his encouragement. The useful discussions with Prof.M.A. Alasingra— char, Civil Engg. Dept., Pmof.G. Thyagaraja, Physics Dept., and Prof.G. Venkatachalam, Civil Eng~. ~ept., lIT—Bombay are thankfully acknowledged. The authors wish to express their special thanks to Utilization Cell, SAC (isao) for givino the Bhaskara—S/~LiR data without which the work could not have been done. Jo also acknowledge with gratitude all help extended by our colleagues. Table 1:
.~ad~.ometersignatures (Bri;htness Temperatures are in °K~ Atmospheric Ri R2 Corrected R2 R3 (31.4 3Hz) (19.35 3Hz) (19.35 3Hz) (22.235 GH~
1) Complete data Sea Peak
165 ±7
142 ±7
134
+
8
Land Peak 2) Excluding snow/ice Sea Peak
265
246 ±6
245
+
6
142
8
134 ~ 8
Land Peak 3) Snow/ice regions
265 228
245 ±7 215
229
4) Moist Soil range 5) Snow/ice range
2l~276 217—249
246 + 6 213 227 237 211—256 201—245
217—261 198—245
225—270 215—260
+
6
165 ±7 +
7
+
183 ±6 196 + 3 260 ~ 4 183 ~ 6 197 ~ 6
260 ±4
REFERENCES 1. E.G. Njoku and J.A. Kong; Theory of Passive microwave Remote Sensing of near—surface soil moisture. J.Geophy. Res. Vol.82, No.20, pp 3108—3118 (1977). 2. 1. Schmugg; Soil moisture sensing with microwave techniques, in: Fourth International Symposium on Remote Sensing of Environment, (1980) pp 487—505. 3. J.R. Wang and B.J. Choudhury; Remote Sensing of soil moisture content over bare field at 1.4 GHz frequency. J.Geophy. Res. Vol.86, No.C6, pp 5277—5282, June 1981. 4. W.J. Burke, I. Schmugge and J.F. Paris, ‘Comparison of 2.8 cm and 21 cm microwave Radiometer observations over soils with emission model calculations’, J.Geophy. Res. Vol.84, No.Cl (1979), pp 287—294. 5. Bhaskara—II SAMIR Data User’s Guide, SAC—RSA—V—IN—02, July 28, 1982.
Bhaskara—l.l
Si:~i1ite
~1icrjwave
Radiometer
BrLgh~ness Temperature
Data
93
6. O.P.N. Calla, S.S. Rana, 0. Raju and S. Balasubramanyarn, ‘SAMIR payload system chracteristics and performance’, in Apolications of Bhaskara—II TV and SA~IR data) Space_~pp1ications Centre, Ahmedabad, 1982. 7. K.S. Rao, Y.V.S. Murthy, A.K.S. Gopalan, R. Ramakrishnan and T.J. Majumdar, ‘Model for atmospheric corrections to Microwave Radiometer brightness temperature data’, Remote Sensing of Environment, Vol.13, pp 209—233, (1933). 8. K.S. Rao, A. Sowrnya, P.Venkatachalam, A.K. Kanidya and T.J. Majumdar, ‘An attempt to study soil moisture over Indian subcontinent using Bhaskara—II SAMI~ data’, in: Proc. S~1AR, Ahmedabad, 1983. 9. Lenduse maps of India, Prepared by Tatioial Atlas Czgar~isation (1975). 1O.L.T. Sang, ~. Njoku and J..-~. Kong; ‘Microv:avs Thermal emission from a stratified medium with nonuniform temperature distribution’, J.Appl.Phy. Vol.46, No.12, dec.1975. ll.T.Schmugge et al; ‘Remote Sensing of Soil moisture with Microwave Kadiometers’, J.Geophy. Res. Vol.79, pp 317—323 (1974).
4
LEGEND 1—Ri (314GHz) 2—R2 (19 35GHz) 3—R3 (22 235GHz) FIG 2 BRIGHTNESS TEMPERATURE VERSUS SPIN NUMBER
-
‘c~
/
~
—:13;
—
*-~---~-~
~LL 4’
~.
~.
-;~
•
.
~.
...
~.m F!QI — ~
-
--
_______________
t~s— sao~ ~ 1..as O~ U~m4RA—fl S~ ~Zfl 10 Zl.Z.U
O~T44C#~~
96
K.S. Rao at al.
130
N.— No. of occurrences TB—Brightness Temp. .
N
300 TB
100
140 180 220 260 HISTOGRAMS OF RADIOMETER R1(31.4GHz) DATA
100
160 180 220 260 300 HISTOGRAMS OF RADIOMETER R2 (19.35 0Hz) DATA
N 130
100
140 180 220 260 300 TB HISTOGRAMS OF CORRECTED BRIGHTNESS TEMP. DATA (19.35 0Hz)
NmPI~:d~~~ 130 Legend
100
140 180 220 260 300 TB HISTOGRAMS OF RADIOMETER R3 (22.2350Hz) DATA _F 1G. 3