was estimated to be 3,833 ha and 3,907 from
LISS
-
IV
data and
RI-
SAT
-1 hybrid polarimetric data, respectively. The spatial agree-
ment between the rice class derived from
LISS
-
IV
and
RISAT
-1
FRS
was found to be 85 percent. The statistics of rice crop in
the selected administrative unit (Achanta Mandal) was found
to be very close to the Bureau of Economic and Statistics esti-
mate of 3,830 ha. The accuracies of
LISS
-
IV
and
RISAT
-1 show a
close match of statistics using the conventional method. These
were all indicators of the potential of single date hybrid polari-
metric
SAR
data for rice mapping and monitoring.
Conclusions
The single date hybrid polarimetric
RISAT
-1,
FRS
1 data was
used for mapping
rabi
(post-rainy) rice crop in West Goda-
vari, Andhra Pradesh, India. The data was subjected to Raney
hybrid polarimetric decomposition to understand the con-
tribution of various backscattering mechanisms (odd bounce
(single), double bounce (even), volume (multiple) scattering)
in the rice crop. The closest acquired
LISS
-IV imagery over the
same area was used for accuracy assessment. The overall ac-
curacy from
LISS
-IV and
RISAT
-1
FRS
data is 91 and 90 percent,
respectively, while the rice class kappa statistics were found
to be 0.94 and 0.92, respectively. The spatial agreement of the
rice cropped area derived from
RISAT
1 and
LISS
-IV data was
found to be 85 percent.
The rice cropped area was estimated to be 3,833, 3,907and
3,830 ha from
LISS
-
IV
,
RISAT
-1 hybrid polarimetric data and
Bureau of Economics and Statistics records, respectively. The
classification results clearly showed that the rice crop can be
clearly separated from all other land-use/land-cover classes,
but the periphery of the settlement class exhibits same magni-
tude of double bounce as that of rice. Also, this method of us-
ing single date
SAR
data is unique in the sense that, it uses dual
polarized hybrid polarimetric data to overcome cloud cover in
T
able
2. T
he
P
roducers
and
U
sers
A
ccuracies
of
D
ecomposed
risat
-1
and
liss
iv
C
lassifications
LISS IV
RISAT-1
Class Name
Producers Accuracy User Accuracy Kappa statistics Producers Accuracy User Accuracy Kappa statistics
Rice
98%
98%
0.9466
100%
96%
0.9228
Plantation
83%
88%
0.8664
89%
76%
0.7297
Fallow
79%
61%
0.5714
79%
79%
0.7638
Settlement
71%
88%
0.8633
57%
100%
1
Water
100%
94%
0.9342
100%
89%
0.8757
Overall Accuracy
91%
0.858
90%
0.8564
Figure 7. Classified images of
liss
-
iv
and Decomposed
risat
-1 image.
562
July 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
