Modelling Potential Zones of Gangotri Glacier Using GIS and ML in the Wake of Physico-Climatic Factors

The modern episode of climate change has rendered glaciers highly susceptible to meltdown. Polar areas are losing ice mass faster than ever in the Holocene history. The largest ice deposits outside poles are found in the ranges of the Himalayas. These Himalayan glaciers are crucial for the human community for sustaining perennial rivers of Indian sub-continent. Therefore, proper study of these glaciers is a must. However, the dynamics of these alpine glaciers are intricate, unlike polar ice sheets overlain on simpler terrain. Therefore, studies of alpine glaciers should give due acknowledgement to complex in situ physiology along with ex situ factors like episodes of global warming. This study is an attempt to model the potential zones of glacial mass balance, i.e. potential zones of ablation, equilibrium, and accumulation using the fuzzy product method of GIS. Elevation, slope, insolation, land surface temperature and Normalized Difference Snow Index are utilized as causative member raster for potential zone modelling. Principle component analysis eigenvalues are supplied as weights of member rasters. The study also seeks to explain the mutual interrelationship of physico-climatic factors using random forest regression for which a random sample of pixels of causative factor rasters is extracted and used to explain the existing mutual association. A comparative analysis of two time periods (1995–2015) is taken into account for the quantification of areal change in the potential zones of Gangotri. It is found that the potential zone of ablation has increased tremendously at the expense of the potential zones of equilibrium and ablation. The result of random forest regression is safely acceptable at a mean square residual <= 0.020.