Comparative Assessment of Two Global Sensitivity Approaches Considering Model and Parameter Uncertainty

Abstract

Global Sensitivity Analysis (GSA) is key to assisting appraisal of the behavior of hydrological systems through model diagnosis considering multiple sources of uncertainty. Uncertainty sources typically comprise incomplete knowledge in (a) conceptual and mathematical formulation of models and (b) parameters embedded in the models. In this context, there is the need for detailed investigations aimed at a robust quantification of the importance of model and parameter uncertainties in a rigorous multi‐model context. This study aims at evaluating and comparing two modern multi‐model GSA methodologies. These are the first GSA approaches embedding both model and parameter uncertainty sources and encompass the variance‐based framework based on Sobol indices (as derived by Dai & Ye, 2015, https://doi.org/10.1016/j.jhydrol.2015.06.034) and the moment‐based approach upon which the formulation of the multi‐model AMA indices (as derived by Dell'Oca et al., 2020, https://doi.org/10.1029/2019wr025754) is based. We provide an assessment of various aspects of sensitivity upon considering a joint analysis of these two approaches in a multi‐model context. Our work relies on well‐established scenarios that comprise (a) a synthetic setting related to reactive transport across a groundwater system and (b) an experimentally‐based study considering heavy metal sorption onto a soil. Our study documents that the joint use of these GSA approaches can provide different while complementary information to assess mutual consistency of approaches and to enrich the information content provided by GSA under model and parameter uncertainty. While being related to groundwater settings, our results can be considered as reference for future GSA studies coping with model and parameter uncertainty.

Novel Ensemble Models Based on the Split‐Point Sampling and Node Attribute Subsampling Classifier for Groundwater Potential Mapping

“…We obtained the required data for these factors from various sources and used them to generate thematic maps. For this purpose, we utilized an SRTM Digital Elevation Model (DEM) from USGS with a resolution of 90m for the generation of the altitude, altitude difference, slope, curvature aspect, TWI, and river density maps (Dai, Liu, et al, 2024). We extracted the LULC map of the Sesan River basin from the global LULC maps produced by Karra et al (2021).…”

Recent Advances in Dielectric Properties-Based Soil Water Content Measurements

“…To evaluate the uncertainty limit of a dielectric probe to detect the liquid water content in variably frozen soils, the dielectric mixing model has been extended to frozen circumstances [102]. To describe the characteristic curves of soil freezing, this tiny uncertainty limit is helpful, but due to the regional diversity of soil parameters, the field calibration of dielectric devices presents challenges [71,110]. Both gravimetric measurements and those performed with the same instrument installed are undertaken on many samples to calibrate the probe.…”

Integration of the Non-linear Time Series GARCH Model with Fuzzy Model Optimized with Water Cycle Algorithm for River Streamflow Forecasting

“…Uncertainty in these models stems from three distinct areas: the input data used the desired model's structure and the models' parameters. The uncertainty and risk associated with employing rainfall-runoff models have been the focus of significant recent work [3,4].…”