The Transferability of Terrestrial Water Balance Components under Uncertainty and Nonstationarity: A Case Study of the Coastal Plain Watershed in the Southeastern USA

The challenges posed by nonstationarity in predicting catchment water balance components motivated this study to test the stationary versus nonstationarity hypothesis and detect changes in the watershed response to land use land cover (LULC) alterations, and climate variability and change. The focus is on a two‐step procedure that includes model calibration of Soil and Water Assessment Tool using a sequential Bayesian uncertainty algorithm (i.e. sequential uncertainty fitting), followed by nonstationary assessment of water balance component using extreme value analysis over an Atlantic coastal plain watershed in the southeastern USA. Analysis suggests that the uncertainty of Soil and Water Assessment Tool model is statistically aligned with LULC alterations that increased the sensitivity of Manning's roughness coefficient, transmission loss and the resistance of the soil matrix to water flow. Changes in LULC along with variability in the magnitude, timing and frequency of precipitation diminished surface runoff and groundwater contribution to the river system whereas it increased evapotranspiration with a substantial decline in water storage capacity. Nonstationary assessment of water balance using extreme value analysis model further revealed a functional form of stationary behaviour (no trends) prior to LULC alteration while large amplification was detected during post‐changes. The results and findings presented in this paper confirm our hypothesis about a combined effect of climate and LULC changes on hydrological functions and that variation of these fingerprints elucidates the presence of nonstationarity in the watershed system. Copyright © 2017 John Wiley & Sons, Ltd.