A modified SWAT model for mechanistic simulation of soil water‐salt transport and the interactions with shallow groundwater

Danning Mu,Yihao Xun,Ya Gao,Dongyang Ren,Chen Sun,Xin Ma,Guanhua Huang,Xu Xu

DOI: https://doi.org/10.1002/hyp.14980

IF: 3.784

2023-09-05

Hydrological Processes

Abstract:This article is to develop a new mechanistic vadose zone (VZ) module coupling with an improved groundwater‐balance module for the well‐known SWAT model to enhance the model capabilities of existing hydrological models. The modified SWAT model (M‐SWAT) is an effective tool for analysing complex hydrology processes and suitable for various application scenarios with good accuracy, acceptable data demand and satisfactory computing time, especially for watersheds with shallow groundwater environments. Current hydrological models still face difficulties in accurately simulating vadose zone (VZ) hydrological processes in arid watersheds with shallow groundwater environments. Therefore, this article introduces a new mechanistic VZ module coupled with an improved groundwater‐balance (GWB) module to enhance the capabilities of the well‐known Soil Water Assessment Tools (SWAT) model. The main improvements and features of the modified SWAT model (M‐SWAT) include: a numerical VZ module to accurately simulate soil water‐salt transport under variably saturated conditions by solving the head‐formed Richards' equation and advection‐dispersion equation; a GWB module (on a sub‐basin scale for shallow aquifer) coupled with VZ module by proposing an effective way of boundary information exchanges; and a more reasonable water‐salt stress functions (i.e., based on matric head and osmotic head) to calculate root water uptake and plant growth. The M‐SWAT was then tested and evaluated with two‐year observation data from the Yangchang canal command area (YCA, an experimental site) and the Jiyuan Irrigation System (Jiyuan, region‐scale), located in the arid upper Yellow River basin of northwest China. The model produced a good agreement between the simulated and observed data in both calibration and validation with sufficiently small root mean square error (RMSE). The determination coefficient (R2) was greater than 0.51, 0.36, and 0.87 in simulating the soil water‐salt dynamics, groundwater depth fluctuations, and leaf area index development, respectively. Case testing also proved that the M‐SWAT had sufficient computational efficiency and stability for regional application. Further comparisons with previous studies indicated that the M‐SWAT could more accurately and rationally simulate the soil water‐salt dynamics and their interactions in arid watersheds with shallow groundwater tables.

water resources