@article{gledocs_11858_10387,
author = {Wu, Songjun and Tetzlaff, Doerthe and Yang, Xiaoqiang and Soulsby, Chris},
title = {Identifying Dominant Processes in Time and Space: Time‐Varying Spatial Sensitivity Analysis for a Grid‐Based Nitrate Model},
year = {2022-08-04},
volume = {58},
number = {8},
publisher = {},
publisher = {},

abstract = {Distributed models have been increasingly applied at finer spatiotemporal resolution. However, most diagnostic analyses aggregate performance measures in space or time, which might bias subsequent inferences. Accordingly, this study explores an approach for quantifying the parameter sensitivity in a spatiotemporally explicit way. We applied the Morris method to screen key parameters within four different sampling spaces in a grid‐based model (mHM‐Nitrate) for NO3‐N simulation in a mixed landuse catchment using a 1‐year moving window for each grid. The results showed that an overly wide range of aquatic denitrification rates could mask the sensitivity of the other parameters, leading to their spatial patterns only related to the proximity to outlet. With adjusted parameter space, spatial sensitivity patterns were determined by NO3‐N inputs and hydrological transport capacity, while temporal dynamics were regulated by annual wetness conditions. The relative proportion of parameter sensitivity further indicated the shifts in dominant hydrological/NO3‐N processes between wet and dry years. By identifying not only which parameter(s) is(are) influential, but where and when such influences occur, spatial sensitivity analysis can help evaluate current model parameterization. Given the marked sensitivity in agricultural areas, we suggest that the current NO3‐N parameterization scheme (land use‐dependent) could be further disentangled in these regions (e.g., into croplands with different rotation strategies) but aggregated in non‐agricultural areas; while hydrological parameterization could be resolved into a finer level (from spatially constant to land use‐dependent especially in nutrient‐rich regions). The spatiotemporal sensitivity pattern also highlights NO3‐N transport within soil layers as a focus for future model development.},
note = { \url {http://resolver.sub.uni-goettingen.de/purl?gldocs-11858/10387}},
}