Global Sensitivity Analysis (GSA) is increasingly used in the development and assessment of environmental models. Here we present a Matlab/Octave toolbox for the application of GSA, called SAFE (Sensitivity Analysis For Everybody). It implements several established GSA methods and allows for easily integrating others. All methods implemented in SAFE support the assessment of the robustness and convergence of sensitivity indices. Furthermore, SAFE includes numerous visualisation tools for the effective investigation and communication of GSA results. The toolbox is designed to make GSA accessible to non-specialist users, and to provide a fully commented code for more experienced users to complement their own tools. The documentation includes a set of workflow scripts with practical guidelines on how to apply GSA and how to use the toolbox. SAFE is open source and freely available for academic and non-commercial purpose. Ultimately, SAFE aims at contributing towards improving the diffusion and quality of GSA practice in the environmental modelling community
We address two critical choices in Global Sensitivity Analysis (GSA): the choice of the sample size and of the threshold for the identification of insensitive input factors. Guidance to assist users with those two choices is still insufficient. We aim at filling this gap. Firstly, we define criteria to quantify the convergence of sensitivity indices, of ranking and of screening, based on a bootstrap approach. Secondly, we investigate the screening threshold with a quantitative validation procedure for screening results. We apply the proposed methodologies to three hydrological models with varying complexity utilizing three widely-used GSA methods (RSA, Morris, Sobol’). We demonstrate that convergence of screening and ranking can be reached before sensitivity estimates stabilize. Convergence dynamics appear to be case-dependent, which suggests that “fit-for-all” rules for sample sizes should not be used. Other modellers can easily adopt our criteria and procedures for a wide range of GSA methods and cases
Global Sensitivity Analysis (GSA) is an essential technique to support the calibration of environmental models by identifying the influential parameters (screening) and ranking them.In this paper, the widely-used variance-based method (Sobol') and the recently proposed moment-independent PAWN method for GSA are applied to the Soil and Water Assessment Tool (SWAT), and compared in terms of ranking and screening results of 26 SWAT parameters. In order to set a threshold for parameter screening, we propose the use of a dummy parameter, which has no influence on the model output. The sensitivity index of the dummy parameter is calculated from sampled data, without changing the model equations. We find that Sobol' and PAWN identify the same 12 influential parameters but rank them differently, and discuss how this result may be related to the limitations of the Sobol' method when the output distribution is asymmetric. We compare the results of Sobol' and PAWN methods for a complex environmental model.We introduce a dummy parameter to support parameter screening.Convergence rates and screening results are similar for Sobol' and PAWN.Differences among sensitivity indices of influential parameters are larger for PAWN.
16 August 2022 in Environmental Modelling & Software
Karst aquifers are an important source of drinking water in many regions of the world. Karst areas are highly permeable and produce large amounts of groundwater recharge, while surface runoff is often negligible. As a result, recharge in these systems may have a different sensitivity to climate and land cover changes than in other less permeable systems. However, little is known about the combined impact of climate and land cover changes in karst areas at large scales. In particular, the representation of land cover, and its controls on evapotranspiration, has been very limited in previous karst hydrological models. In this study, we address this gap (1) by introducing the first large-scale hydrological model including an explicit representation of both karst and land cover properties, and (2) by providing an in-depth analysis of the model's recharge production behaviour. To achieve these aims, we replace the empirical approach to evapotranspiration estimation of a previous large-scale karst recharge model (VarKarst) with an explicit, mechanistic and parsimonious approach in the new model (V2Karst V1.1). We demonstrate the plausibility of V2Karst simulations at four carbonate rock FLUXNET sites by assessing the model's ability to reproduce observed evapotranspiration and soil moisture patterns and by showing that the controlling modelled processes are in line with expectations. Additional virtual experiments with synthetic input data systematically explore the sensitivities of recharge to precipitation characteristics (overall amount and temporal distribution) and land cover properties. This approach confirms that these sensitivities agree with expectations and provides first insights into the potential impacts of future change. V2Karst is the first model that enables the study of the joint impacts of large-scale land cover and climate changes on groundwater recharge in karst regions.
Worldwide surface waters suffer from the presence of nitrogen (N) compounds causing eutrophication and deterioration of the water quality. Despite many Europe-wide legislation’s, we still observe high N levels across many water bodies in Europe. Information on long-term annual soil N surplus is needed to better understand these N levels and inform future management strategies. Here, we reconstructed and analysed the annual long-term N surplus for both agricultural and non-agricultural soils across Europe at a 5 arcmin (≈10 km at the equator) spatial resolution for more than a century (1850–2019). The dataset consists of 16 N surplus estimates that account for the uncertainties resulting from input data sources and methodological choices in major components of the N surplus. We documented the consistency and plausibility of our estimates by comparing them with previous studies and discussed about possible avenues for further improvements. Importantly, our dataset offers the flexibility of aggregating the N surplus at any spatial scale of relevance to support water and land management strategies.