Local and global determinants of hydroclimatic variability in Central Asia

Abstract

Mountains are vital "water towers" for semi-arid and arid regions, with snowpack accumulated during the cold season being the primary source of river streamflow during the warm season. Observational and knowledge gaps on local snowpack dynamics complicate the prediction of river discharge, particularly in Central Asia, where high interannual climate variability causes significant fluctuations in river discharge. This thesis aims to improve the understanding of hydroclimatic variability in the Central Asian mountains to advance seasonal streamflow prediction. Using data-driven and ensemble methods, it explores empirical relationships between climate, topography, and snowmass, as well as the climate teleconnections of seasonal precipitation and streamflow in the mountainous basins. The thesis also assesses the suitability of global reanalyses and satellite data for hydrological forecasting. I present a novel, parsimonious snow model with robust generalizability and transferability performance for accurate snow mass estimation in data-limited regions. On seasonal to annual scales, hydroclimatic variability in Central Asia is driven by multiple global ocean-atmospheric oscillations, primarily the El Niño–Southern Oscillation, the Pacific Decadal Oscillation, and the North Atlantic Oscillation. Most teleconnections have longer temporal legacies with varying spatial effects on seasonal precipitation, making them valuable long-lead predictors. Finally, I present a seasonal streamflow forecasting framework for Central Asia that integrates catchment snowpack as initial hydrological conditions and climate oscillation indices as precursors of future climate variability. The findings and methods contribute to hydrological modelling and operational forecasting in data-scarce, snowmelt dominated catchments, enhancing understanding of climate teleconnections and emphasizing the value of large-scale oscillation indices for snow-based streamflow forecasts.