Global Sensitivity Analysis for Multiple Scenarios and Models of Nitrogen Processes

Abstract:

Modeling nitrogen process in soil is a long-lasting challenge partly because of the uncertainties from parameters, models and scenarios. It may be difficult to identify a suitable model and its corresponding parameters.This study assesses the global sensitivity indices for parameters of multiple models and scenarios on nitrogen processes. The majority of existing nitrogen dynamics models consider nitrification and denitrification as a first-order decay process or a Michaelis-Menten model, while various reduction functions are used to reflect the impact of environmental soil conditions. To determine the model uncertainty, 9 alternative models were designed based on NP2D model in this study. These models have the similar descriptions of nitrogen process but are different in the cal reduction functions of soil water and temperature. A global sensitivity analysis of each models under various scenarios was evaluated. Results show that in our synthetic cases of nitrogen transport and transformation, the global sensitivity indices vary between each models and scenarios. Larger indices for parameters of nitrification are obtained than the ones of denitrification in 6 models, while an inverse relationship is revealed in the rest 3 models. Parameters of soil temperature reduction functions are more sensitive than those of soil water reduction functions. When the soil water and temperature increase separately or together, parameters of denitrification gain their sensitivity, but the indices for parameters of soil temperature reduction functions decrease simultaneously. Our results indicate that identifying important parameters may be biased if ignoring the model and scenario uncertainties. This problem can be resolved by using the global sensitivity indices for multiple models and multiple scenarios. The new indices is useful to determine the relative contributions from different models and scenarios.