Publications

Patel, A; Mark, BG; Haritashya, UK; Bawa, A (2025). Twenty first century snow cover prediction using deep learning and climate model data in the Teesta basin, eastern Himalaya. CLIMATE DYNAMICS, 63(3), 156.

Abstract
The snow cover in the Teesta River basin (TRB), located in eastern Himalaya, plays crucial role in regional hydrology by influencing water availability, ecological processes, and socio-economic activities. This study assesses the spatio-temporal distribution of snow cover in the TRB for both the present (2000-2021) and future periods (2021-2040: early-century; 2041-2060: mid-century; 2081-2100: late-century). The analysis of spatio-temporal snow cover distribution was conducted using daily moderate resolution imaging spectroradiometer (MODIS) snow cover products (Terra and Aqua), which revealed a decreasing mean annual SCA trend at a rate of - 0.03%/yr (insignificant; p > 0.05) across the TRB during 2000-2021. We also assessed SCA distribution in relation to topographical factors, finding a declining trend across the basin, particularly above 3000 m elevation. On the other hand, future snow cover predictions were made using a long short-term memory (LSTM) deep learning model under two climate scenarios (SSP245 and SSP585). The LSTM model was trained on three predictors (precipitation, minimum and maximum temperatures) and validated against independent dataset. The model's performance was evaluated using root mean square error (RMSE), correlation coefficient (CC), and Nash-Sutcliffe efficiency (NSE), showing robust predictive capability (RMSE = 3.8, NSE = 0.86, CC = 0.94) during testing. Under both future scenarios, annual precipitation, and minimum and maximum temperatures are projected to increase, while snow cover is expected to follow a decreasing trend during the early, mid, and for the late centuries. Overall, this research enhances understanding of snow cover changes in the TRB and provides valuable insights for policymakers, water resource managers, and local communities. Additionally, the predictive model developed in this study offers a useful tool for proactive decision-making and adaptive strategies in response to evolving snow cover dynamics.

DOI:
10.1007/s00382-025-07643-6

ISSN:
1432-0894