Towards simplification of hydrologic modeling: identification of dominant processes

In this work, we investigate methods for gaining greater insight from hydrological model runs conducted for uncertainty quantification and model differentiation. We frame the sensitivity analysis questions in terms of the main purposes of sensitivity analysis: parameter prioritization, trend identification, and interaction quantification. For parameter prioritization, we consider variance‐based sensitivity measures, sensitivity indices based on the L1‐norm, the Kuiper metric, and the sensitivity indices of the DELSA methods. For trend identification, we investigate insights derived from graphing the one‐way ANOVA sensitivity functions, the recently introduced CUSUNORO plots, and derivative scatterplots. For interaction quantification, we consider information delivered by variance‐based sensitivity indices. We rely on the so‐called given‐data principle, in which results from a set of model runs are used to perform a defined set of analyses. One avoids using specific designs for each insight, thus controlling the computational burden. The methodology is applied to a hydrological model of a river in Belgium simulated using the well‐established Framework for Understanding Structural Errors (FUSE) on five alternative configurations. The findings show that the integration of the chosen methods provides insights unavailable in most other analyses.

Section: Introductionmentioning

confidence: 98%

Making the most out of a hydrological model data set: Sensitivity analyses to open the model black‐box

Borgonovo

Plischke

et al. 2017

“…For example, previous work on sensitivity tend to focus solely on the sensitivity to forcing parameters such as temperature and precipitation (Christensen et al, ; Das et al, ; Harding et al, ; Vano et al, ) but largely do not consider the range of responses anticipated from changing internal parameters. There are notable exceptions, including Painter et al () and Deems et al's () adjustment of albedo parameters and Markstrom et al (). Quantifying parameter sensitivity to projected climate change answers key questions, including the followings.…”

Section: Introductionmentioning

confidence: 99%

Global Sensitivity of Simulated Water Balance Indicators Under Future Climate Change in the Colorado Basin

Bennett

Blanco

Jonko

et al. 2018

The Colorado River Basin is a fundamentally important river for society, ecology, and energy in the United States. Streamflow estimates are often provided using modeling tools which rely on uncertain parameters; sensitivity analysis can help determine which parameters impact model results. Despite the fact that simulated flows respond to changing climate and vegetation in the basin, parameter sensitivity of the simulations under climate change has rarely been considered. In this study, we conduct a global sensitivity analysis to relate changes in runoff, evapotranspiration, snow water equivalent, and soil moisture to model parameters in the Variable Infiltration Capacity (VIC) hydrologic model. We combine global sensitivity analysis with a space‐filling Latin Hypercube Sampling of the model parameter space and statistical emulation of the VIC model to examine sensitivities to uncertainties in 46 model parameters following a variance‐based approach. We find that snow‐dominated regions are much more sensitive to uncertainties in VIC parameters. Although baseflow and runoff changes respond to parameters used in previous sensitivity studies, we discover new key parameter sensitivities. For instance, changes in runoff and evapotranspiration are sensitive to albedo, while changes in snow water equivalent are sensitive to canopy fraction and Leaf Area Index (LAI) in the VIC model. It is critical for improved modeling to narrow uncertainty in these parameters through improved observations and field studies. This is important because LAI and albedo are anticipated to change under future climate and narrowing uncertainty is paramount to advance our application of models such as VIC for water resource management.

“…In the conventional procedure of system model development, a system process is represented by a single process model, and the process models are integrated to form a single system model. Based on this conceptualization, to identify the important processes, sensitivity analysis is conducted for the parameters of the process models; a process is considered to be important if the parameters of the process model are important to system model outputs of interest [e.g., Saltelli et al ., ; Hill et al ., ; Markstrom et al ., ]. However, using a single process model is questionable, when available data and knowledge support use of multiple process models [ Beven , ; Bredehoeft , ; Neuman , ; Gupta and Nearing , ; Clark et al ., ].…”

Section: Introductionmentioning

confidence: 99%

A new process sensitivity index to identify important system processes under process model and parametric uncertainty

Dai

Walker

et al. 2017

A hydrological model consists of multiple process level submodels, and each submodel represents a process key to the operation of the simulated system. Global sensitivity analysis methods have been widely used to identify important processes for system model development and improvement. The existing methods of global sensitivity analysis only consider parametric uncertainty, and are not capable of handling model uncertainty caused by multiple process models that arise from competing hypotheses about one or more processes. To address this problem, this study develops a new method to probe model output sensitivity to competing process models by integrating model averaging methods with variance‐based global sensitivity analysis. A process sensitivity index is derived as a single summary measure of relative process importance, and the index includes variance in model outputs caused by uncertainty in both process models and their parameters. For demonstration, the new index is used to assign importance to the processes of recharge and geology in a synthetic study of groundwater reactive transport modeling. The recharge process is simulated by two models that convert precipitation to recharge, and the geology process is simulated by two models of hydraulic conductivity. Each process model has its own random parameters. The new process sensitivity index is mathematically general, and can be applied to a wide range of problems in hydrology and beyond.