Effect of soil spatial aggregation caused by the calculation unit division on runoff and sediment load simulation in the SWAT model
Juan Xiao,Yao Wang,Jinxing Sun,Shuishi Xie,Yan Huang,Zhiwei Wan,Lihong Meng,Xiujuan Li,Keyuan Zhong
DOI: https://doi.org/10.1016/j.jhydrol.2023.130345
IF: 6.4
2023-10-19
Journal of Hydrology
Abstract:Distributed hydrological model divides watersheds into distinct calculation units to simulate hydrological processes. Within the model, different calculation unit division schemes induce input elements, such as soil and land use, to certain degrees of generalization, thus affecting the simulation result. Analyzing the effect of the calculation unit division on the simulation result is an important methodology to reduce uncertainty and improve the simulation accuracy of the model. Previous researchers have focused on the comprehensive effect of underlying surface generalization (such as soil, land use, and terrain generalization) caused by the calculation unit division on the simulation results, while it did not separate the effect of the generalization of the individual input factor on the simulation result. In this paper, the Dongjiang headwater region in southern China was selected as the study area. The single type of land use and slope were input into the Soil and Water Assessment Tool (SWAT) model to fix the effects of land use and terrain caused by calculation unit division on the simulation results. Eleven calculation unit division schemes were used to analyze the impact of soil spatial aggregation on runoff and sediment load simulations. The calculation unit division schemes included 11 sub-watershed division schemes and their hydrological response unit (HRU) thresholds were set at 8 %. The drainage area thresholds of the 11 sub-watershed division schemes were 30 ha, 60 ha, 120 ha, 240 ha, 480 ha, 1000 ha, 1600 ha, 2200 ha, 2800 ha, 3400 ha and 4000 ha. The results demonstrated that the soil spatial distribution showed an obvious aggregation with an increase in the drainage area threshold in the SWAT model. Of these, the Krasnozem soil type which covers the largest area in the watershed, increased significantly with an increase in the drainage area threshold. However, the areas of all the other soil types decreased significantly. The Lioth soil type covers the smallest area in the watershed decreased and even disappeared when the threshold in the SWAT model was greater than 1000 ha. Furthermore, soil spatial aggregation caused a significant increase in the annual runoff and sediment load (P < 0.05), with the sediment load (Re = 29.78 %) increasing at a greater rate compared to runoff (Re = 1.74 %). Increased soil aggregation significantly increased the sediment loads in the maximum 1-day, the continuous maximum 5-day, the continuous maximum 7-day, and the sediment load in flood processes (P < 0.001). Refining the calculation unit division scheme will cause a significant increase in the number of calculation units. Due to the computer processing power limitations, we cannot further refine the calculation unit division scheme. Based on our current operational calculation unit division scheme, we did not identify the optimal division scheme in the study region. These results can provide reference information for future research into hydrological model uncertainty and can improve the SWAT model simulation accuracy.
geosciences, multidisciplinary,water resources,engineering, civil