From maps to movies: high-resolution time-varying sensitivity analysis for spatially distributed watershed models

Abstract

Abstract. Distributed watershed models are now widely used in practice to simulate runoff responses at high spatial and temporal resolutions. Counter to this purpose, diagnostic analyses of distributed models currently aggregate performance measures in space and/or time and are thus disconnected from the models' operational and scientific goals. To address this disconnect, this study contributes a novel approach for computing and visualizing time-varying global sensitivity indices for spatially distributed model parameters. The high-resolution model diagnostics employ the method of Morris to identify evolving patterns in dominant model processes at sub-daily timescales over a six-month period. The method is demonstrated on the United States National Weather Service's Hydrology Laboratory Research Distributed Hydrologic Model (HL-RDHM) in the Blue River watershed, Oklahoma, USA. Three hydrologic events are selected from within the six-month period to investigate the patterns in spatiotemporal sensitivities that emerge as a function of forcing patterns as well as wet-to-dry transitions. Events with similar magnitudes and durations exhibit significantly different performance controls in space and time, indicating that the diagnostic inferences drawn from representative events will be heavily biased by the a priori selection of those events. By contrast, this study demonstrates highresolution time-varying sensitivity analysis, requiring no assumptions regarding representative events and allowing modelers to identify transitions between sets of dominant parameters or processes a posteriori. The proposed approach details the dynamics of parameter sensitivity in nearly continuous time, providing critical diagnostic insights into the underlying model processes driving predictions. Furthermore, the approach offers the potential to identify transition points between dominant parameters and processes in the absence of observations, such as under nonstationarity.

Flood and drought hydrologic monitoring: the role of model parameter uncertainty

“…Instead, there tend to be multiple parameter sets that satisfy the observations; in hydrology this is known as model parameter equifinality (Beven, 2006). Although the performance might be similar for a given calibration metric, the results can vary significantly when comparing other metrics, timescales, or variables (Gupta et al, 2008;Herman et al, 2013;Wagener and Gupta, 2005;Reusser and Zehe, 2011;Reusser et al, 2009;Clark and Vrugt, 2006).…”

Moment-based metrics for global sensitivity analysis of hydrological systems

“…is widely used in the literature (e.g., Homma and Saltelli, 1996;Chun et al, 2000;Borgonovo, 2007;Sudret, 2008;A. Dell'Oca et al: Global sensitivity analysis of hydrological systems …”

A meta-modeling approach for spatio-temporal uncertainty and sensitivity analysis: an application for a cellular automata-based Urban growth and land-use change model

“…In light of the above mentioned studies, U-SA is essential to increase the credibility of the UG-LUC models, as well as to test the model robustness in producing realistic outputs for future implementations (Jantz et al 2003, Jantz and Goetz 2005, Kim 2013). Additionally, a spatially-explicit approach is also crucial for spatial models that do not entirely depend on scalar inputs but that are also affected by spatial relationships (Herman et al 2013, Abily et al 2016, Variance, which is a central measure in error propagation and SA, allows partitioning the effects of uncertainty in the input factors on model output (Kyriakidis and Goodchild 2006). There is a range of different theoretical and methodological approaches to variance-based SA in the literature.…”

“…In particular, an uncertainty and SA (U-SA) approach that is both independent from model structure and capable of handling interaction effects is important for complex, nonlinear models (Halls 2002, Chen et al 2010, Roura-Pascual et al 2010, Moreau et al 2013, Xu and Zhang 2013, Saint-Geours et al 2014. Moreover, for spatio-temporal models, where the simulation results are spatially distributed, it is additionally important to identify not only the source(s) of uncertainty but also its location(s) at a specific time (Herman et al 2013, Abily et al 2016. The resulting uncertainty maps can assist in locating uncertainty hot spots, and the sensitivity maps help further in identifying the spatial pattern of influential input factors behind the areas of high uncertainty (Ligmann-Zielinska 2013, Şalap-Ayça and Jankowski 2016).…”