Detection of cryospheric parameters with artificial neural network over antarctic region using Ku-Band based ISRO's SCATSAT-1 data

Abstract

The antarctic region is an essential component of Earth's climate system and heat balance. Therefore, continuous monitoring of various cryospheric parameters (Sea-Ice Extent - SIE, and Snow Water Equivalent - SWE, among others) over Antarctica is essential to understand the state of global climate processes. This study presents a framework to detect and analyse the SIE and SWE using enhanced resolution data products from ISRO's (Indian Space Research Organisation) Scatterometer Satellite (SCATSAT-1) level-4 (operating at Ku-band, 13.5 GHz) data products. This framework is based on Artificial Neural Network (ANN) and Post-Classification Comparison (PCC) to detect the multitemporal variations using SCATSAT-1 images (sigmanaught, gamma-naught, and brightness temperature). For validation purposes, the classified maps and change maps are compared with Advanced Microwave Scanning Radiometer (AMSR2) derived SIE and SWE polar gridded datasets. Statistical analysis has confirmed the effectiveness (more than 88.75% accuracy) of the proposed framework with SCATSAT-1. We conclude that the ANN-PCC is a powerful technique to estimate the spatiotemporal dynamics of surface melting and freezing over Antarctica.