Content uploaded by Kruti Upadhyay
Author content
All content in this area was uploaded by Kruti Upadhyay on Jul 02, 2022
Content may be subject to copyright.
RESEARCH ARTICLE
Deriving Sea Ice Images from Super Resolution SCATSAT-1 Data
over the Antarctic: Operational Method and Accuracy Assessment
Kruti Upadhyay
1
•Naveen Tripathi
2
•Bhasha Vachharajani
1
•D. Ram Rajak
2
•I. M. Bahuguna
2
Received: 8 June 2019 / Accepted: 4 July 2021 / Published online: 18 July 2021
ÓIndian Society of Remote Sensing 2021
Abstract
Sea ice has an intense impact on the polar environment, ocean circulation, weather and regional climate. Unexpected
melting of sea ice, which is considered as one of the climate change effects, has become a potential threat to the Earth’s
climate. The regular monitoring of sea ice and its extent has become very important towards understanding of sea ice
temporal dynamics. In this study, we present an operational technique of generation of sea ice images and sea ice area
(derived from the images) using level-4 data from Indian Scatterometer SCATSAT-1. Using hierarchical classification
rules, the threshold-based technique has been developed and applied to generate super-resolution (2.25 km) daily sea ice
images over the Antarctic for the years 2017 and 2018. The technique uses four SCATSAT-1 data products, i.e. Gamma0
[Horizontal (H) and Vertical (V)] and Brightness Temperature (Hand V) to classify sea ice, open water and other classes.
Classification accuracy has been assessed by comparing SCATSAT-1 sea ice images with those obtained from AMSR2 sea
ice concentration data. The comparison shows that there is around 96.1% matching of sea ice classification between
SCATSAT-1 and AMSR-2 SIC derived sea ice images. Hence, it indicates that the super-resolution data of SCATSAT-1 is
well capable of distinguishing sea ice from water.
Keywords Sea ice Sea ice concentration SCATSAT-1 Scatterometer AMSR-2
Introduction
Sea ice is a predominant surface feature of polar environ-
ment. It affects global ocean circulation and the world’s
habitat directly or indirectly. Changes in the dynamics of
sea ice affect the global climate system, that is why it is
very crucial to understand the components of sea-ice and
forces governing the sea ice dynamics. Many researchers
have studied the spatial distribution of sea ice trends and
inter-annual variability of sea ice and its influence to global
teleconnections patterns and oceanic processes, hence its
relation to global climate change (Rai and Pandey 2006;
Oza et al., 2011,2017). Though Arctic and Antarctic, both
are situated at poles, there have been observed significantly
different changes in their climate (Overland et al., 2008).
Some of the factors affecting the contrasting behaviour of
two polar regions include their dissimilar topographies,
greenhouse gases, CO2 sink, anthropogenic influences,
land/sea distribution, etc. (Turner & Overland, 2009).
For the polar studies, passive-microwave, synthetic
aperture radar (SAR), a scatterometer, altimeter and ther-
mal infrared radiometer (TIR) sensors, etc., have been used
by the researchers (Bhandari et al., 2002; Teleti & Luis,
2013). Scatterometer has an invaluable contribution in the
field of polar studies and has been extensively used in
mapping sea ice extent (Anderson & Long, 2005), quan-
tification of snow accumulation (Raleigh, 2013), melt
&Kruti Upadhyay
krutiben.uphd16@sot.pdpu.ac.in
Naveen Tripathi
naveent@sac.isro.gov.in
Bhasha Vachharajani
bhasha.vachharajani@sot.pdpu.ac.in
D. Ram Rajak
rajakdr@sac.isro.gov.in
I. M. Bahuguna
imbahuguna@sac.isro.gov.in
1
Department of Mathematics, Pandit Deendayal Petroleum
University, Gandhinagar, Gujarat, India
2
Space Applications Centre, Indian Space Research
Organization (ISRO), Ahmedabad, Gujarat, India
123
Journal of the Indian Society of Remote Sensing (October 2021) 49(10):2575–2581
https://doi.org/10.1007/s12524-021-01412-8(0123456789().,-volV)(0123456789().,-volV)
detection of sea ice-ice sheet (Howell et al., 2006), clas-
sifying sea ice types (Ulaby et al., 2014), sea ice motion
(Haarpaintner & Spreen, 2007), etc. It is also useful in
tracking large icebergs and measuring wind over polar
regions, etc. (Long, 2017).
Scatterometer Image Reconstruction with Filter (SIRF)
algorithm was developed for enhancement of scatterometer
image (Long et al., 1993). Various algorithms have been
developed for sea ice identification. Most of the algorithms
use the backscatter r0and polarization ratio for primary
classification. Estimation of sea ice extent had been carried
out using Mahalanobis distance classification using
NSCAT data (Remund & Long, 1999). Another algorithm
uses Active Polarization Ratio (APR) with a season-de-
pendent threshold for discrimination of sea ice and water.
Brigham Young University (referred to as BYU) algorithm
has been developed using QuikScat/SeaWinds products
(Haarpaintner et al., 2004). Various algorithms based on
maximum-likelihood classifier, Bayesian classifier, Fish-
er’s linear discriminant method and Geophysical model
function have been developed with QuikSCAT, ASCAT
and HY-2A scatterometer data (Belmonte Rivas et al.,
2012; Kern et al., 2007; Li et al., 2016; Remund & Long,
2014).
In this paper, we propose an advancement in sea ice
classification method using gamma naught (c0Þand
brightness temperature Tb
ðÞdata. The microwave bright-
ness temperature is a physical parameter of the object and
it is proportional to the emissivity of object. For sea ice, as
a mixture of different states of water, ice and snow, the
brightness temperature depends upon the emissivity of all
the objects. Hence, this can be another parameter to dis-
criminate sea ice and water. We have emphasised the fact
that the polarization ratios of sea ice and water are dif-
ferent. This can be due to the differences in their
backscattering values because of varied surface properties.
This ratio is lower in sea ice than in open ocean. We have
used mean values of different backscattered variables in
vertical and horizontal polarizations along with the stan-
dard deviation values for separation. Validation and accu-
racy assessment of the products have been provided using
Advanced Microwave Scanning Radiometer 2 (AMSR2)
data.
Study Area and Data Used
Our study focuses on the sea ice in the Antarctic or
Southern Ocean surrounding Antarctica Continent.
Antarctica is the coldest, driest and windiest continent
situated in the southernmost part of the Earth, surrounded
by oceans. Due to extreme cold weather conditions during
winter, ocean water freezes and floats on the surface of the
ocean. This floating ice affects the polar condition and
ocean circulation (NSIDC, 2018a).
We have used two data products from SCATSAT-1, HH
and VV polarizations of gamma naught (c0Þand the
brightness temperature (Tb) (SCATSAT-1, 2017). SCAT-
SAT-1, launched on 26th Sep 2016, is a Ku-band pencil
beam scatterometer operating at 13.515 GHz providing a
ground resolution cell of size 25 925 km. The level-1B
data suggest that all the static parameters are stable and
well within the specified range (Kumar et al., 2019). The
high resolution (2.25 km) Level 4 products are generated
from Level 1B products using Scatterometer Image
Reconstruction (or SIR) technique. Level 4 South Polar
products are available in polar stereographic projection at
2.25 km grid size with datum WGS84. The brightness
temperature (TB) data are derived from SCATSAT-1 noise
measurements, and it exhibits close match with collocated
microwave radiometer (AMSR-2) derived 10.65 GHz TB
data. HH and VV backscattering coefficients are used for
extracting different parameters for discrimination of sea ice
and water (SCATSAT-1) (Fig. 1).
SCATSAT-1 Sea Ice Discrimination
Parameters
Scattering coefficients are conventional measures of the
strength of radar signals. It is often expressed in dB, which
is a dimensionless number. These signals are reflected back
into the space by a distributed surface. Sigma naught (r0)is
defined with respect to the nominally horizontal plane and
has a significant variation with wavelength, incidence angle
and properties of the scattering surface as well as with
polarisation (Small, 2011). Gamma naught (c0) is described
as per the prevailing concepts of the backscattering nor-
malization. It is obtained using the local-incident angle
metric through the slope normalization (Small, 2011).
For clear discrimination of sea ice, we have taken four
parameters, namely gamma naught (c0) and brightness
temperature (Tb)ofH-pol and V-pol values. While study-
ing Antarctic region, using c0is more advantageous than r0
because it is less sensitive to the impact of radar incidence
angle (Small, 2011).
H-pol values have either similar signatures with their V-
pol values (H&V) or are quite higher compared with V-pol
values (H[V) in case of sea ice of SCATSAT-1. Figure 2
shows the raw images of c0and Tbof H-pol and V-pol data,
respectively.
2576 Journal of the Indian Society of Remote Sensing (October 2021) 49(10):2575–2581
123
Sea Ice Discrimination Methodology
We have taken four parameters (c0
H;c0
V;TbH ;Tby) to dis-
criminate sea ice and open ocean using hierarchical deci-
sion rules. The data are converted to parameter values from
an unsigned short integer value using respective conversion
coefficients provided. For the differentiation in backscatter
values, we have taken the polarization ratio of H-pol and V-
pol data which are Normalized Difference Gamma Naught
Index (NDGI) and Normalized Difference Brightness
Temperature Index (NDBI). Note that
NDGI ¼c0
Hc0
V
c0
Hþc0
V;NDBI
¼TbH TbV=
TbH þTbV
Having computed these indices, we form stacked layers,
comprising of NDGI, c0
H;c0
V, NDBI, TbH ;TbV :Since we are
interested only in sea ice, not in land ice, we have applied
the Antarctic land mask(Oza et al., 2012). This way, we
have been able to obtain images of masked stack layers of
six bands, out of which four are parameters and two are
normalized indices. We applied two-stage classification to
each of the images. In stage-1 classification, threshold
values are applied. These values are derived using ISO-
DATA (Iterative Self-Organizing Data Analysis Technique
Algorithm) clustering technique. Each of the image of
stacked layer is subject to k-means classification, which
results in 50 classes. With prior knowledge, using mean
and standard deviation, pure classes are separated. Further,
the total number of classes are increased, so as to have
more number of pure classes. For each of the pure classes,
we again find an average value of the means (l) and that of
the standard deviations (r). The threshold for a particular
class is obtained using the value falling in the interval (l-
2r,l?2r), as there is 95% chance of an image,
belonging to that pure class. After applying these threshold
values, the stacked image gets converted into just one, with
nine different classes.
Fig. 1 Daily global coverage of SCATSAT-1 scatterometer r0for 1st September 2018
Fig. 2 (a, b) Raw images of (H-pol, V-pol) c0from SCATSAT-1
scatterometer of 5th July 2017; (c, d) Raw images of (H-pol, V-pol) T
b
from SCATSAT-1 scatterometer SCATSAT-1 scatterometer of 5th
July 2017; the white hollow dot represents the area of ‘no
observation’
Journal of the Indian Society of Remote Sensing (October 2021) 49(10):2575–2581 2577
123
These classes are categorized into different classes of
sea ice and water. For further refinement, second stage
classification rules are applied. In this stage, sea ice
occurrence probability of 36 years (SIOP36) has been
applied. SIOP36 is sea ice occurrence probability of
36 years prepared using passive microwave sea ice con-
centration daily averaged data from the years 1978 to 2014.
It has been used to remove vague sea ice signatures at the
locations, where we never expect the existence of sea ice.
Therefore, we have given the clear boundary, out of which
detection of sea ice must not be carried out. After applying
SIOP36, class names and colour coding have been given to
get the classified image. Sea ice area has also been com-
puted using the classified sea ice map. Figure 3shows the
steps followed in the methodology, and Fig. 4shows the
intermediate steps while following the procedure. Fig-
ure 4c shows the final classified image in which white,
black and blue colours depict sea ice, water and Antarctic
mask, respectively.
The algorithm described above has been used to produce
the daily operational SCATSAT-1 polar sea ice maps for
Antarctic region.
Figure 5depicts the variations in sea ice over the year.
Antarctic sea ice has been clearly observed which reveals
the dynamicity of sea ice behaviour. Minimum sea ice
extent has been observed during the months of February–
March. It gains the maximum extent during the growth
phase during the months of September–October. After
achieving the peak value, it again experiences its decaying
phase.
Validation and Accuracy assessment
AMSR2 Sea Ice Concentration and NSIDC Sea Ice
Extent
Advanced Microwave Scanning Radiometer 2 (AMSR2)
sea ice concentration data have been used with a 15%
threshold (Kaleschke & Tian-Kunze, 2016). AMSR2 sen-
sor onboard satellite Global Change Observation Mission-
Water (GCOM-W) of Japan Aerospace Exploration
Agency (JAXA) provides daily data over sea ice region of
the Earth. The resolution of the data is 3.125 km. ASI
(ARTIST Sea Ice) algorithm has been used to generate the
data (Beitsch et al., 2014).
The continuous series of data sets of daily sea ice con-
centration disseminated by National Snow and Data Ice
Centre (NSIDC) have been used (NSIDC, 2018b). The
brightness temperature data have been derived from the
Special Sensor Microwave Imager/Sounder (SSMIS) sen-
sor from the Defense Meteorological Satellite Programme
F17 (DMSP F17) Satellite developed using the NASA
Team algorithm. We have used daily sea ice extent data
which are provided in the polar stereographic projection
with a 25 km resolution (Fetterer et al., 2017).
We have compared SCATSAT-1 SIA with AMSR2 SIC
data (with 15% threshold) (Fig. 6). This result demon-
strates a sensibly good agreement during both the seasons,
but it slightly varies in the peak duration. Additionally, SIA
derived from SCATSAT follows a trend similar to the one
found from NSIDC SIE (Fig. 6).
Fig. 3 Steps of stage one and stage two classification
2578 Journal of the Indian Society of Remote Sensing (October 2021) 49(10):2575–2581
123
Accuracy Assessment
For validation purpose, we have used SIC available from
AMSR2 radiometer. SIC values exceeding 15% are
retained and the corresponding area (SIA) has been com-
puted. Also, we have applied land mask, so as to take sea
ice and not land ice into account. The SIA thus computed
has been compared to the one obtained from SCATSAT-1.
To address minute details of SIE, we assess accuracies
month wise and season wise. For seasonal accuracy, we
consider winter season (July, August and September
months of the year 2017) and the subsequent summer
season (January, February and March of the year 2018) in
the same cycle. Since we acquire the same parameters from
different sensors (scatterometer v/s radiometer), this par-
ticular section may be regarded as ‘‘inter-sensor compar-
ison’’. It is remarkable to note that, this compares an active
sensor to a passive one. The data are in the form of daily
images, say S1 (SCATSAT-1 image) and R1 (AMSR2
image) for a given day. We compute the monthly means
(say Sand R) and then compare the features. This com-
parison brings some features common to both SCATSAT-1
and AMSR2 and some uncommon features. We categorize
Fig. 4 (a, b, c) Images obtained
at each classification stage
Fig. 5 (i, ii, iii, iv): Variations in Sea ice images on 15th day for the
months of: (i) February 2017, (ii) June 2017, (iii) September 2017,
(iv) December 2017
Fig. 6 Comparison of
SCATSAT-1 sea ice area with
AMSR2 and NSIDC data
Journal of the Indian Society of Remote Sensing (October 2021) 49(10):2575–2581 2579
123
the images as belonging to any of the three classes, as
defined in Table 1.
By Rand S, we mean the features not found in AMSR2
and SCATSAT-1, respectively.
These classes may be interpreted as matching (Class-I),
overestimation (Class-II) and underestimation (Class-III)
by SCATSAT-1 with reference to AMSR-2.
Table 2shows the monthly extent of matching (and
contrast) between SCATSAT-1 and AMSR2. Matching is
found to be the best during winter months (September and
October), when SIE is maximum and undergoes melting
phase. Additionally, during these months, SCATSAT-1
overestimates AMSR2 by 1.178% and underestimates by
0.18%, we may say there is a very small room for
SCATSAT-1, not being able to capture the features found
in AMSR2. The minimum matching is found during the
month of January (when the melting is at peak); in fact, a
large number of features shown by AMSR2 are lacking in
SCATSAT-1. We carry out a similar analysis for both the
seasons, in order to study the impact of seasons on these
measurements.
Tables 3and 4demonstrate the seasonal accuracy for
the dataset SCATSAT-1 and AMSR2. We find that during
the winter season, the matching is quite high (98%), which
suggests a very good accuracy of SCATSAT-1 with respect
to AMSR2. However, during summer, this accuracy redu-
ces to an average of 89%. Wet snow and water lying above
the sea ice layer (as a result of melting) could be the
probable reasons for low matching.
Conclusion
In this paper, we have presented an operational technique
for discriminating sea ice and open ocean and generating
sea ice images of the Antarctic region using SCATSAT-1
2.25 km data. Results show that SCATSAT-1 super-reso-
lution data are capable to distinguish sea ice from open
ocean using this operational method. Four Scatsat-1 data
parameters [Gamma0 (H&V) and Brightness Temperature
(H&V)] have been used to develop this technique.
Brightness temperature products have proved to be valu-
able for discrimination, when used with Gamma0. The
accuracy assessment and validation of the sea ice images
produced has been carried out by comparing sea ice images
from NSIDC sea ice extent and AMSR2 sea ice concen-
tration data. The validation result shows the overall
matching accuracy of around 96%. The technique dis-
criminates sea ice from water better in winter/refreeze
season than in summer/melt season. The presence of wet
snow or melt water on sea ice during melt season may be a
cause of reduced differentiation between the two classes.
These sea ice maps are publicly available at Visualization
of Earth Observation Data and Archival System (VEDAS)
Table 1 Class definition for comparison of AMSR2 and SCATSAT-1
Features in RR
SClass-I Class-II
SClass-III Void
Table 2 Month wise accuracy assessment (in %) of SCATSAT-1 sea
ice area with AMSR2 (June 2017 to May 2018)
Month Class-I Class-II Class-III
June, 2017 98.149 0.843 1.008
July, 2017 98.344 0.950 0.706
August, 2017 98.073 0.969 0.959
September, 2017 98.642 1.178 0.180
October, 2017 98.626 1.199 0.175
November, 2017 93.693 0.841 5.465
December, 2017 87.181 0.766 12.053
January, 2018 82.606 0.584 16.810
February, 2018 91.900 0.736 7.364
March, 2018 94.251 0.823 4.926
April, 2018 96.056 0.777 3.167
May, 2018 95.774 0.492 3.734
Average 94.441 0.847 4.712
Table 3 Winter season accuracy of SCATSAT-1 for the months July
to September for the year 2017
Winter season
Month 2017 Class1 Class2 Class3
July 98.344 0.950 0.706
August 98.073 0.969 0.959
September 98.642 1.178 0.180
Average 98.353 1.032 0.615
Table 4 Summer season accuracy of SCATSAT-1 for the months
January to March for the year 2018
Summer season
Month 2018 Class1 Class2 Class3
January 82.606 0.584 16.810
February 91.900 0.736 7.364
March 94.251 0.823 4.926
Average 89.586 0.714 9.700
2580 Journal of the Indian Society of Remote Sensing (October 2021) 49(10):2575–2581
123
web-portal (https://vedas.sac.gov.in/vedas_new/view/
south_pole.jsp).
Acknowledgements We would like to thank Director, PDPU and
Director, Space Applications Centre (SAC), ISRO for giving the
opportunity and providing the data sets. We would also like to thank
Deputy Director, EPSA for his inspiration and guidance.
References
Anderson, H. S., & Long, D. G. (2005). Sea ice mapping method for
SeaWinds. IEEE Transactions on Geoscience and Remote
Sensing, 43, 647–657. https://doi.org/10.1109/TGRS.2004.
842017
Beitsch, A., Kaleschke, L., & Kern, S. (2014). Investigating high-
resolution AMSR2 sea ice concentrations during the February
2013 fracture event in the beaufort sea. Remote Sensing, 6,
3841–3856. https://doi.org/10.3390/rs6053841
Belmonte Rivas, M., Verspeek, J., Verhoef, A., & Stoffelen, A.
(2012). Bayesian sea ice detection with the advanced scatterom-
eter ASCAT. IEEE Transactions on Geoscience and Remote
Sensing, 50, 2649–2657. https://doi.org/10.1109/TGRS.2011.
2182356
Bhandari, S. M., Dash, M. K., Vyas, N. K., et al. (2002). Microwave
Remote Sensing of Ice in the Antarctic Region from OCEAN-
SAT-1. In: Advances in marine and Antarctic science (p. 443).
A.P.H. Pub. Corp.
Fetterer, F., Knowles, K., Meier, W. N., et al. (2017). Sea Ice Index,
Version 3.0, Boulder, Colorado USA. NSIDC. In: National Snow
and Ice Data Center.
Haarpaintner, J., & Spreen, G. (2007). Use of enhanced-resolution
QuikSCAT/SeaWinds data for operational ice services and
climate research: Sea ice edge, type, concentration, and drift.
IEEE Transactions on Geoscience and Remote Sensing, 45,
3131–3137. https://doi.org/10.1109/TGRS.2007.895419
Haarpaintner, J., Tonboe, R. T., Long, D. G., & Van Woert, M. L.
(2004). Automatic detection and validity of the sea-ice edge: An
application of enhanced-resolution QuikScat/SeaWinds data.
IEEE Transactions on Geoscience and Remote Sensing, 42,
1433–1443. https://doi.org/10.1109/TGRS.2004.828195
Howell, S. E. L., Tivy, A., Yackel, J. J., & Scharien, R. K. (2006).
Application of a SeaWinds/QuikSCAT sea ice melt algorithm
for assessing melt dynamics in the Canadian Arctic Archipelago.
Journal of Geophysical Research, 111, C07025. https://doi.org/
10.1029/2005JC003193
Kaleschke, L., Tian-Kunze, X. (2016). AMSR2 ASI 3.125 km Sea Ice
Concentration Data. In: Institute of Oceanography: Universita
¨t
Hamburg, Germany Digital Media. http://ftp-projects.zmaw.de/
seaice/
Kern, S., Spreen, G., Kaleschke, L., et al. (2007). Polynya signature
simulation method polynya area in comparison to AMSR-E 89
GHz sea-ice concentrations in the Ross Sea and off the Adelie
Coast, Antarctica, for 2002–05: First results. Annals of Glaciol-
ogy, 46, 409–418. https://doi.org/10.3189/172756407782871585
Kumar, R., Bhowmick, S. A., Chakraborty, A., Sharma, A., Sharma,
S., Seemanth, M., Gupta, M., Chakraborty, P., Modi, J., Misra,
T., et al. (2019). Post-launch calibration–validation and data
quality evaluation of SCATSAT-1. Current Science,117(6),
973–982.
Li, M., Zhao, C., Zhao, Y., et al. (2016). Polar sea ice monitoring
using HY-2A scatterometer measurements. Remote Sens, 8, 688.
https://doi.org/10.3390/rs8080688
Long, D. G. (2016). Polar applications of spaceborne scatterometers.
IEEE Journal of Selected Topics in Applied Earth Observations
and Remote Sensing, 10(5), 2307–2320. https://doi.org/10.1109/
JSTARS.2016.2629418
Long, D. G., Hardin, P. J., & Whiting, P. T. (1993). Resolution
enhancement of spaceborne scatterometer data. IEEE Transac-
tions on Geoscience and Remote Sensing, 31, 700–715. https://
doi.org/10.1109/36.225536
NSIDC (2018a). All about sea ice. In: National Snow and Ice Data
Center. http://nsidc.org/cryosphere/seaice/index.html. Accessed
20 Jun 2018.
NSIDC (2018b). http://nsidc.org/data/seaice_index/archives.html..
Accessed on June 2018.
Overland, J. E., Turner, J., Francis, J., et al. (2008). The arctic and
antarctic: Two faces of climate change. Eos (washington DC),
89, 177–178. https://doi.org/10.1029/2008EO190001
Oza, S. R., RAJAK DR, K DASH M, , et al. (2017). Advances in
ANTARCTIC SEA ICE STUDIES IN India. Proceedings of the
Indian National Science Academy, 83, 427–435.
Oza, S. R., Singh, R. K. K., Srivastava, A., et al. (2011). Inter-annual
variations observed in spring and summer Antarctic sea ice
extent in recent decade. Mausam, 62, 633–640.
Oza, S. R., Singh, R. K. K., Vyas, N. K., & Sarkar, A. (2012). AN
ATLAS OF THE ARCTIC AND THE ANTARCTIC SEA ICE
TRENDS (1999-2009)-DERIVED FROM QUIKSCAT SCAT-
TEROMETER DATA.
Rai, S., & Pandey, A. C. (2006). Antarctic sea ice variability in recent
years and its relationship with Indian Ocean Sea Surface
Temperature. Journal of Indian Geophysics Union, 10(3),
219–229.
Raleigh, M. S. (2013). Quantification of uncertainties in snow
accumulation, snowmelt, and snow disappearance dates.
Remund, Q. P., & Long, D. G. (2014). A decade of QuikSCAT
scatterometer sea ice extent data. IEEE Transactions on
Geoscience and Remote Sensing, 52, 4281–4290. https://doi.
org/10.1109/TGRS.2013.2281056
Remund, Q. P., & Long, D. G. (1999). Sea ice extent mapping using
Ku band scatterometer data. Journal Geophysics Research
Ocean, 104, 11515–11527. https://doi.org/10.1029/98JC02373
SCATSAT-1 DPT SCATSAT-1 Level 4 Data Products Format
Document. Scatsat-1 Data Product Team. Scientific Report No.
SC1/DP/L4FORMAT-DOC/V1.1/JUL2017. Ahmedabad.
Small, D. (2011). Flattening gamma: Radiometric terrain correction
for SAR imagery. IEEE Transactions on Geoscience and Remote
Sensing, 49, 3081–3093.
Teleti, P. R., & Luis, A. J. (2013). Sea ice observations in polar
regions: Evolution of technologies in remote sensing. Interna-
tional Journal of Geosciences, 04, 1031–1050. https://doi.org/10.
4236/ijg.2013.47097
Turner, J., & Overland, J. (2009). Contrasting climate change in the
two polar regions. Polar Research, 28, 146–164. https://doi.org/
10.1111/j.1751-8369.2009.00128.x
Ulaby, F. T., Long, D., Blackwell, W., et al. (2014). Microwave radar
and radiometric remote sensing. University of Michigan Press.
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Journal of the Indian Society of Remote Sensing (October 2021) 49(10):2575–2581 2581
123