How Hydrologic Processes Differ Spatially in a Large Basin: Multisite and Multiobjective Modeling in the Tarim River Basin

Abstract

Water resources are essential to ecosystems and social economies worldwide, especially in the deserts and oases of the Tarim River Basin, whose water originates largely from alpine mountains characterized by complicated hydrological processes and scarce hydrometeorological observations. This paper presents multisite and multiobjective modeling of hydrological processes in the whole Tarim River Basin, covering 32 catchments in total. The study uses the Soil and Water Assessment Tool, extended by incorporating a degree‐day glacier melt module to enable modeling of glacier melt in the alpine mountains. The multiobjective calibration approach of ε‐Nondominated Sorting Genetic Algorithm‐II was implemented with the two objective functions of the Nash‐Sutcliffe value of daily streamflow and the bias of simulated glacier melt contribution to streamflow. Based on the combined use of the Morris sensitivity technique and hierarchy cluster, the 32 catchments in the study area are divided into six groups according to their dominating hydrological processes, for example, glacier melt, snowmelt, groundwater, and routing. The multiobjective calibration was satisfactory, with 22 of the 32 catchments showing Nash‐Sutcliffe values of daily streamflow larger than 0.6 and the bias of simulated glacier melt contribution to streamflow values smaller than 0.05. Model performance was highly dependent on meteorological data availability, in that low data availability led to poor model performance, while factors such as catchment area and mean annual snowfall had little influence on model performance. The results indicate that multisite and multiobjective calibration enables consistent and comprehensive examination of the spatially different hydrological processes in a large basin and provides information for further assessment of the impact of climate change on water availability.

Glacio‐hydrological model calibration and evaluation

“…This approach of grouped response units is also called a semi-distributed approach. In large catchments, a first division is often done by creating sub-catchments, before HRU or elevation zones are defined (e.g., Adnan et al, 2019;Fang et al, 2018). In both discretization cases, the spatial resolution of the model is determined by the grid cell size (e.g., 25 m in Huss et al (2008) and $8 km in Zhang et al (2013)), or the HRU size or other model units.…”

“…However, Frans et al (2016) argue that manual calibration was used because of spatial and temporal inconsistencies in the observed data. Regarding automatic calibration techniques, in the reviewed studies around 15 studies used a Monte Carlo approach, and around 30 used a more sophisticated algorithm such as the SCE-UA algorithm (Chen et al, 2017), ε-NSGAII (e.g., Duethmann et al, 2015;Fang et al, 2018), a combination of evolutionary and steepest gradient descent algorithm (Jost et al, 2012), or GAP (Meyer et al, 2019).…”

“…Around 35 different glacio‐hydrological model codes have been used to model glacierized catchments in the 145 reviewed papers (Table A1). Most used models are versions of the HBV model (e.g, Finger et al, 2015; Konz & Seibert, 2010; Stahl et al, 2008), the SWAT model (e.g., Fang et al, 2018; Garee et al, 2017), and the TOPKAPI(‐ETH) model (Pellicciotti et al, 2012; Ragettli et al, 2013) and represent altogether ca. 35% of the studies (Table 2).…”

Recent Changes in Water Discharge in Snow and Glacier Melt-Dominated Rivers in the Tienshan Mountains, Central Asia

“…The contribution of snow-glacier meltwater to runoff is up to 45%, with precipitation accounting for about 33.1% [80]. For the Toxkan River, glacier meltwater accounts for only about 21.3% of the total annual runoff [81], with the rest coming from snow meltwater (26.7%) [82] and rainfall [26], while for the Kumalak River, glacier meltwater makes up around 54%~58.65% of the runoff [81,83,84]. runoff [81], with the rest coming from snow meltwater (26.7%) [82] and rainfall [26], while for the Kumalak River, glacier meltwater makes up around 54%~58.65% of the runoff [81,83,84].…”

Historic and Simulated Desert-Oasis Ecotone Changes in the Arid Tarim River Basin, China

“…The drainage network in the basin consists of the main stream of the Tarim River and 144 drainage systems associated with nine major tributary basins: the Yarkand River, the Aksu River, the Kaidu-Kongque River, the Hotan River, the Kaxgar River, the Weigan River, the Dina River, the Keriya River, and the Qarqan River. The tributaries to the main stream of the Tarim River form a centripetal shape around the Tarim Basin [25]. The Tarim River is a dissipative inland river whose runoff is mainly supplied by meltwater from glaciers and snow.…”