JAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India, 2002 
phenomenon, however, could not be addressed. Spectral 
measurements in thermal and microwave region of the spectrum 
have shown some promise (Chase, 1969; Myers and Moore, 1972; 
Huntley, 1978; Heilman and Moore, 1982) in the detection of 
waterlogging due to rising ground water table. 
A study was, therefore, taken up to evaluate the potential of 
thermal sensor data for assessment of waterlogging due to rising 
ground water table in part of Mahanadi stage-I command area, 
Orissa state using day-and-night time Landsat-TM data acquired 
on November 2, 2000. 
The approach involves conversion of DN values of Landsat-TM 
thermal band (Band-6) data into radiance after geo-referencing 
and ultimately into at-satellite brightness temperature (Tb) using 
pre-defined calibration constants provided along with the satellite 
data, atmospheric correction, concurrent ground truth collection 
and correlation of at-satellite brightness temperatures (Tb) with 
the occurrence of waterlogging due to rising ground water table. 
The Ground water table data was collected during field 
verification campaign. Some observations were used for 
establishing relationship between thermal response and the 
incidence of waterlogging whereas the rest for validation of the 
relationship thus established. The brightness temperatures values 
were thresholded as per correlation results and satellite image was 
colour sliced accordingly. 
The standard false colour composite (FCC) image of the area 
generated from green, red and near-infrared bands of day-time 
Landsat-TM data along with corresponding day-time and night- 
time thermal data are appended as Fig-l. As evident from the 
Figure, areas with surface ponding, wetness or a thin film of water 
appear in different shades of blue/cyan colour. By virtue of high 
specific heat and thermal inertia, water warms up and cools down 
gradually. Consequently, in the day-time thermal image (single 
band image), areas with ground water closer to the surface, those 
with thin film of water on the surface and wet surface appear 
darker indicating thereby lower temperature as compared to 
terrain background which is manifested as light to very light gray 
tone (indicating higher temperature). During the night when 
ambient temperature drops down due to gradual cooling, water 
appear warmer (dull white to white colour) in the thermal image 
as compared to terrain background, which exhibits various shades 
of colour ranging from medium to dark gray. A comparison of 
day-time optical and thermal with night-time thermal data reveals 
an improved detectability of waterlogged areas in the latter than 
the former. 
Addtionally, an attempt has been made to correlate the areas with 
high ground water (less than 2m below. the surface) as observed 
during ground truth campaign and the thermal response as 
measured by the Landsat-TM. The at-satellite brightness 
temperature has been found to range from 18°C to 27°C at around 
9.30 PM on November 2, 2000. The areas with ground water table 
less than one meter have been found to be in the brightness 
temperature (Ty) range of 23.1?C to 24.1?C, whereas areas with 1- 
2 m ground water table had T, values between 22.3°C to 23.1°C. 
The areas with standing water had T, values ranging between 
25.1 to 26.1? C. And those with a thin film of water have Tb 
values ranging from 24.1 to 25.1 °C. The overall accuracy of the 
segmentation was 86.8%. From the foregoing it is quite evident 
that night-time thermal image offers a great potential for detecting 
potentially waterlogged areas. 
732 
2.2. The Impact of Mining on Agricultural Land 
Optical sensor data have been operationally used to study the 
dynamics of land disruption due to mining (Wier et al, 1973, 
Irons et al, 1980; Parks et al, 1987), water pollution assessment 
(Repic et al, 1991) and monitoring (Lathrop and Lillesand; 1986; 
Lillesand et al, 1987), and detection of land subsidence 
(Mechaffie and Seargent, 1985; Volk et al., 1990). 
Realizing the potential of spaceborne multispectral data in 
addressing various issues related to mining activities, a study was 
taken up to assess the impact of opencast iron ore mining on 
agricultural land in part of Goa, India using an erosion-deposition 
model proposed by Mitasova et al (1996). The database used in 
the study consists of IRS-1C LISS-III and PAN data with path - 
row nos. 96 - 62, acquired on February 11, 1997, and Survey of 
India topographic maps at 1:25,000 scale. 
The approach involves database preparation, ground truth 
collection, digital analysis as well as systematic visual 
interpretation of satellite data. Initially, the IRS-1C LISS-III and 
PAN data were georeferenced to 1:25,000 scale topographic maps 
followed by data fusion using Principle Component data fusion 
technique. Simultaneously, the contour information in Survey of 
India (SOI) topographic maps at 1:25,000 scale were digitized 
along with drainage line and a DEM was generated using 
ARC/INFO software. The satellite data along with DEM was used 
for visual interpretation of soil properties. For deriving 
information on land use / cover through digital analysis, 
resampled LISS-III (6 m resolution) and PAN individual data sets 
were used. Various land cover categories that were identified 
during ground truth campaign were used as training sets for 
classifying entire image using Gaussian maximum likelihood per- 
pixel classifier. 
To study the impact of mining on agricultural lands, initially, the 
spatial distribution of erosion and deposition rates need to be 
estimated. The UPSED model developed by Mitasova et al. 
(1996) was used in the present study to predict the erosion- 
deposition zones. The model is simple and predicts the spatial 
distribution of erosion / deposition as the divergence of sediment 
flow under steady state conditions. The outline of model is as 
follows: 
D = div q = Kt [grad h ] * S Sin B - h[ kp + kt] (1) 
where — D = Net rate of erosion / deposition 
K = Transportability of sediment 
h = Water depth estimated from upslope area 
S = Unit vector in steepest slope direction 
ß = Slope in degrees 
kp = terrain curvature in the direction of steepest 
slope 
kt = curvature tangential to a contour line projected to 
normal plane. 
The land cover information was used for predicting the runoff, 
cover and management factors. The information on soils was used 
to generate erodibility factor grid, and the DEM for generation of 
slope and aspect grids. For implementing the model in 
ARC/INFO GRID environment, source code provided by 
Mitasova and Mitas (1999) was used after incorporating