ANALYSING THE STATUS AND FUTURE CHANGES OF THE CRYOSPHERE AND ITS RELATION WITH CLIMATE CHANGE FOR THE HIMALAYAN REGION

Abstract

The Karakoram and Himalayan (KH) cryosphere (in terms of snow and glaciers) plays a significant role in managing the ecosystem and supporting livelihood and economic development. The KH region experiences large variations in snow cover and glaciers in a warming environment. The potential adverse effect of changing cryosphere causes a cascading implication on water availability and generates a condition of water stress in the future. Other than this, the change in water storage with increasing temperature can create drought-like conditions and affect people living in the region. Therefore, monitoring cryospheric changes and their interaction with climate are essential to understanding present climate sensitivity and future water availability. The melting of the cryosphere provides water for the region where the livelihood of millions of people depends upon meltwater of snow and glacier during the summer season. However, the continuous monitoring of snow cover at a large spatial extent is challenging due to harsh climatic conditions and rugged topography. For this, the use of remote sensing presented a great advantage in snow cover monitoring. This thesis uses satellite observations for spatio-temporal snow cover monitoring at basinal and regional scales. The cloud blocks are the major limitation of optical remote sensing data that hinders the original snow cover information in the high-mountain terrains. To overcome this limitation, a spatially distributed cloud removal methodology is developed to ensure that all the necessary physical-based considerations and topographical variations are correctly incorporated in the method. This non-spectral sequential methodology includes a combination of multi-sensor data, temporal filter, nearest neighborhood filter, zonal snowline filter, and multiday-backward replacement filter. The cloud-gap-filled snow cover outcomes are validated with a direct and indirect approach to assess the accuracy of the methodology over the Chenab River basin. The results suggested that the spatially distributed cloud removal methodology can bridge the gap between regional observation-based snow cover and cloud blocks. After analyzing the methodology performance, the snow cover distribution is assessed at spatio-temporal scale along with topographical parameters. Further, we have established a relationship between snow cover and essential climatic drivers. After developing a cloud removal methodology, the snow cover distribution and trends have been carried out over the KH region. The methodology-related uncertainties are quantified for understanding the exact error inherent in the cloud gap-filling approach. The cloud-gap-filled Snow Cover Area (SCA) is compared with Landsat-8, and the relationship with in situ observations (snowfall and temperature) is also examined. Then, the Snow Cover Day (SCD) and nine snow cover timing indices have been assessed to explore the snow cover characteristics The interconnection between SCA and meteorological variables is evaluated, suggesting a higher correlation with temperature as well as shortwave radiation. Other than this, a sensitivity analysis is performed, indicating a higher sensitivity with radiations towards SCA than other selected variables. On the other hand, we have explored the variation of energy balance components and then measured the point-scale Surface Energy Balance (SEB) for the Phuche glacier, upper Ganglass catchment, Ladakh range. The meteorological variables are recorded at 5600 m a.s.l., altitude, which is mounted at the ridge of the glacier. And the point-scale SEB ablation is validated with the stake measured total melt. Despite this, we have also quantified the point-based and glacier-wide Mass Balance (MB) calculated using the stake measurement for Phuche and Khardung glaciers, Ladakh range during 2014–2017. The Equilibrium Line Altitude (ELA) and Accumulation Area Ratio (AAR) are calculated to assess the year-wise glacier changes for the hydrological year. The result suggests that the Khardung glacier experienced 3.7 times more mass loss relative to the Phuche glacier. In the glacier mass variation, the cold-arid region is considered because this region comes under climatic zone (western Himalayas) where mass variation is large and the majority of the glaciers (~79%) of this region have a smaller surface area (< 0.75 km2), and only 4% of glaciers are > 2 km2. Due to the smaller glacier area and scarce precipitation, we have selected these two glaciers to understand the glacier's direct response to climate fluctuation. It was noted that the small size glaciers are a good indicator of climate change. Another objective of this thesis is to present the water storage change over the major river basin of India in order to assess the water availability and water stress in the future. We have used the twin satellite gravimetric data to analyze the total water storage change, groundwater recharge and also quantify the condition which causes drought. The total water storage change and groundwater recharge are measured over the Ganga River basin from 2003 to 2016. In this study, various groundwater recharge estimation methods are applied and validated with the in situ observational well. And the best-fitted recharge method with higher accuracy is used to establish the link between the recharge change and hydrometeorological variables. The relationship of ground recharge with other factors (total withdrawal, irrigation-based groundwater abstraction, population density (domestic factor), and overall water stress) are established to conclude the exact picture of groundwater reduction. Further, a new drought index is developed to map the drought occurrence and severity that incorporate meteorological and hydrological conditions in drought identification. The generated index is applied over the Indus, Ganga, and Brahmaputra river basins. Results are compared with the past drought occurrence and other well-established drought indices. The model output suggested that the index has the capability to map the agricultural, meteorological, and hydrological droughts to a broader area. The last component of the study in this thesis presents a comprehensive approach for predicting the discharge over the Sutlej River basin using different Long Short-Term Memory (LSTM) deep learning models. The combination of best-suited variables (climatic and SCA) is selected based on their correlation and recursive feature elimination techniques. After finalizing the dataset, the hyperparameter tuning was done and set the best parameters to enhance the model performance. We have compared five different LSTM model architectures over the selected dataset. The bidirectional LSTM (BLSTM) outperformed other LSTM architectures during the training and testing stages. Further, we have also compared the normal BLSTM with Principle Component Analysis (PCA)-based BLSTM models over the study area. The PCA-based BLSTM performed well during the training stage, and this model was further used for forecasting the discharge over the other selected gauging sites. To derive the snow cover and glaciers changes, we have used a methodology to connect the cryosphere changes with the spatial extent and climatic interaction. We have utilized remote sensing data; however, the in situ observation is used to calibrate and validate the obtained results. The cloud removal methodology development and their variation with terrain parameters are presented in objective-1. The snow cover distribution and their trend are discussed in objective 2.1. In comparison, the glacier energy balance and mass balance variation are quantified in objective 2.2. And water storage change, recharge modeling, and identification of drought occurrence in objective 2.3. Finally, the discharge prediction model is developed in objective 3. Overall, the outcomes and model development in this thesis is likely to be important for the research community to understand the snow cover and glacier variation in the present and their implication on discharge in the future. This thesis can also be a benchmark for modelers working in the high-mountain region and facing challenges in terms of cloud blocks. The present work also produces a solution for the drought modeling (hydrological and meteorological) which can be mapped using the remote sensing-based index. In a broader context, the results of this thesis can be used for predicting the snow cover changes and their interaction with climate change. It also helps to manage the water availability and even reduce the condition of water stress in the future by designing laws by the decision-maker for balancing the ecosystem. Keywords: Basin; Drought; Climate change; Energy balance; Glacier; Himalayas; Mass balance; Recharge; Remote Sensing; Snow cover variability; Water resources.