Akademik Veri Yönetim Sistemi
ARABIAN JOURNAL OF GEOSCIENCES, cilt.15, ss.1-18, 2022 (Hakemli Dergi)
In situ glacier discharge and sediment observations are uncommon in the Himalayan region because of the complex terrain and bad weather conditions. This research is the frst study of the glaciers investigated to collect and forecast in situ glacier melt SSC (suspended sediment concentration) data from streams associated with the Pindari and Kafni glaciers in the central Himalayan region valley (Pindar basin) during three consecutive years (2017–2019). Stream discharge and sedimentation play a crucial role in hydroelectric power projects located in the Himalayan mountain regions. The problem is severe in the food season due to excessive sediment concentration. In the Pindari and Kafni glacier stream dynamics, discharge, precipitation, and temperature were identifed as major regulating components of variations in sediment concentration. Multiple linear regression (MLR) and artifcial neural network (ANN) models were used. A bivariate correlation test was carried out, with a signifcant p-value of less than 0.05. The analytical measurement used daily values calculated between 2017 and 2018. MLR analysis revealed that the precipitation and SSC are not proportional since precipitation has a negative beta coefcient. The normalized importance of precipitation concerning discharge was determined to range between 11.54 and 76.1%. Statistical indices evaluated the performance of the used models, specifcally residual sum of squares error (RSS), relative error (RE), and mean squared error (MSE). When predicting future SSCs for Pindari and Kafni streams, the ANN model outperforms the MLR model. The results clearly show that extreme events such as foodings and landslides cannot be predictable considering the research area based on the collected in situ hydro-meteorological data. In light of the results, it is thought that there are other factors, such as solar radiation, that afect discharge values and thus sediment transport. Sustained multi-year observations using machine learning applications could improve regional water resources assessment and management and regulate the policy to develop multi-purpose hydroelectric projects in the region.