Strong sensitivity of watershed-scale, ecohydrologic model predictions to soil moisture

Authors: PIGNOTTI, GARETT - Purdue University; RATHJENS, HENDRIK - Stone Environmental Consulting; CHAUBEY, INDRAJEET - University Of Connecticut; Williams, Mark; CRAWFORD, MELBA - Purdue University

Submitted to: Journal of Environmental Modeling and Software
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/29/2021
Publication Date: 8/4/2021
Citation: Pignotti, G., Rathjens, H., Chaubey, I., Williams, M.R., Crawford, M. 2021. Strong sensitivity of watershed-scale, ecohydrologic model predictions to soil moisture. Journal of Environmental Modeling and Software. 144. Article 105162. https://doi.org/10.1016/j.envsoft.2021.105162.
DOI: https://doi.org/10.1016/j.envsoft.2021.105162

Interpretive Summary: Watershed models including the Soil and Water Assessment Tool (SWAT) are used globally to evaluate scenarios that inform land management and policy. It is important that we understand how well models represent known and predicted processes and that we quantify the uncertainty in model predictions. In this study, we assessed the the sensitivity of the SWAT model to soil moisture and its impact on discharge and nutrient loading in two watersheds (Cedar Creek watershed, northeastern Indiana; Little River watershed, southwestern Georgia). When soil moisture was varied by 0.5 to 2 standard deviations, predictions of surface runoff (42-107%), evapotranspiration (29-76%) and nitrate loss (19-369%) varied considerably from baseline conditions. Findings highlight the need to collect soil moisture data and improve soil moisture simulations.

Technical Abstract: It is currently unclear how changes to modeled soil water simulations influence key predictions in complex, ecohydrologic models, particularly beyond hydrology. We assessed sensitivity to soil moisture in the Soil and Water Assessment Tool by artificially perturbing simulated soil moisture and analyzing targeted prediction change. Results demonstrated a strong, process-dependent sensitivity of model predictions to perturbations in soil moisture. At the basin scale, the absolute values of differences from a baseline simulation for surface runoff, evapotranspiration, and nitrate loss ranged from 42% to 107%, 29% to 76%, and 19% to 369%, respectively, for soil moisture changes between 0.5 to 2 standard deviations. Landscape-scale results illustrated the impact of modeled land management actions, notably fertilizer applications, on determining the magnitude of the change. The findings highlight and motivate the need to improve soil moisture simulations, suggesting the possible accuracy improvement for key model predictions by more precisely simulating soil moisture.