Using Sobol’ sensitivity analysis to identify important model layers of a prediction model for simulating soil nitrogen transport in surface runoff and subsurface water

• Published in: Stochastic environmental research and risk assessment Vol. 39; no. 2; pp. 747 - 763
• Main Authors: Zhang, Shiwei, Tong, Juxiu, Ye, Ming
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.02.2025; Springer Nature B.V
• Subjects: Agricultural land; Agricultural runoff; Ammonia; Aquatic Pollution; Chemistry and Earth Sciences; Computational Intelligence; Computer Science; Earth and Environmental Science; Earth Sciences; Environment; Math. Appl. in Environmental Science; Nitrates; Nitrogen; Nonpoint source pollution; Original Paper; Parameter identification; Parameter sensitivity; Parameter uncertainty; Physics; Pollution dispersion; Pollution sources; Ponding; Prediction models; Probability Theory and Stochastic Processes; Runoff; Sensitivity analysis; Soil analysis; Soil conditions; Soil improvement; Soil layers; Soil mixing; Soil pollution; Soil surfaces; Soils; Statistics for Engineering; Subsurface water; Surface runoff; Surface water; Uncertainty analysis; Urea; Waste Water Technology; Water Management; Water Pollution Control; Sobol GSA; HYDRUS-1D; Model parameter; The prediction model of soil nitrogen transport
• ISSN: 1436-3240; 1436-3259
• DOI: 10.1007/s00477-024-02894-x

Due to complex soil conditions of farmlands, a model simulating soil nitrogen transport in surface runoff and subsurface water is subject to uncertainty not only in model parameters but also in representation of soil layers in model layers. This study presents a model of nitrogen transport with the following three layers: a ponding-runoff layer, a soil mixing layer, and a underlying soil layer, which have a total of nine parameters. The model simulates concentrations of urea, ammonia nitrogen, and nitrate nitrogen in surface water and at the depths of 2 cm and 32 cm below soil surface. The main focus of this study is to identify the most important model layer that is crucial for simulating soil nitrogen transport. The global sensitivity analysis method of Sobol’ is used to examine importance of the three model layers, because the Sobol’ method can consider impacts of individual parameters and parameters interactions on model simulations. Results show the following: (1) the soil mixing layer is the most important one for simulating urea transport in surface water, and the underlying soil layer is the most important one for simulating NH 4 + -N and NO 3 − -N transport in surface water, (2) the underlying soil layer is the most important one for simulating nitrogen transport at the depths of 2 cm and 32 cm below soil surface, and (3) parameters interactions have a negligible impact on the simulations. These results can be used for improvement of the model and reduction of model prediction uncertainty, and provide a reference for subsequent studies on soil nitrogen transport to reduce agricultural non-point sources pollution.