HESS Opinions &amp;quot;Crash tests for a standardized evaluation of hydrological models&amp;quot;

Abstract. Many fields, such as drought-risk assessment or reservoir management, can benefit from long-range streamflow forecasts. Climatology has long been used in longrange streamflow forecasting. Conditioning methods have been proposed to select or weight relevant historical time series from climatology. They are often based on general circulation model (GCM) outputs that are specific to the forecast date due to the initialisation of GCMs on current conditions. This study investigates the impact of conditioning methods on the performance of seasonal streamflow forecasts. Four conditioning statistics based on seasonal forecasts of cumulative precipitation and the standardised precipitation index were used to select relevant traces within historical streamflows and precipitation respectively. This resulted in eight conditioned streamflow forecast scenarios. These scenarios were compared to the climatology of historical streamflows, the ensemble streamflow prediction approach and the streamflow forecasts obtained from ECMWF System 4 precipitation forecasts. The impact of conditioning was assessed in terms of forecast sharpness (spread), reliability, overall performance and low-flow event detection. Results showed that conditioning past observations on seasonal precipitation indices generally improves forecast sharpness, but may reduce reliability, with respect to climatology. Conversely, conditioned ensembles were more reliable but less sharp than streamflow forecasts derived from System 4 precipitation. Forecast attributes from conditioned and unconditioned ensembles are illustrated for a case of drought-risk forecasting: the 2003 drought in France. In the case of low-flow forecasting, conditioning results in ensembles that can better assess weekly deficit volumes and durations over a wider range of lead times.

Section: Catchments and Hydrological Modelmentioning

confidence: 98%

Seasonal streamflow forecasting by conditioning climatology with precipitation indices

Crochemore¹,

Ramos²,

Pappenberger³

et al. 2017

“…This may improve the reliability of its internal processes representation (e.g. unsaturated flow and groundwater flow) and subsequently its applicability for ungauged situations (based on Andréassian et al, 2007Andréassian et al, , 2009Andréassian et al, , 2010Bronstert, 2004). In this respect, we have been monitoring groundwater level close to the outlet of the studied reach of the Middle Jaguaribe River, whose first results were presented in Costa et al (2012).…”

Section: A C Costa Et Al: a Channel Transmission Losses Model For mentioning

confidence: 99%

A channel transmission losses model for different dryland rivers

Costa

Bronstert

Araújo

2012

Abstract. Channel transmission losses in drylands take place normally in extensive alluvial channels or streambeds underlain by fractured rocks. They can play an important role in streamflow rates, groundwater recharge, freshwater supply and channel-associated ecosystems. We aim to develop a process-oriented, semi-distributed channel transmission losses model, using process formulations which are suitable for data-scarce dryland environments and applicable to both hydraulically disconnected losing streams and hydraulically connected losing(/gaining) streams. This approach should be able to cover a large variation in climate and hydro-geologic controls, which are typically found in dryland regions of the Earth. Our model was first evaluated for a losing/gaining, hydraulically connected 30 km reach of the Middle Jaguaribe River (MJR), Ceará, Brazil, which drains a catchment area of 20 000 km 2 . Secondly, we applied it to a small losing, hydraulically disconnected 1.5 km channel reach in the Walnut Gulch Experimental Watershed (WGEW), Arizona, USA. The model was able to predict reliably the streamflow volume and peak for both case studies without using any parameter calibration procedure. We have shown that the evaluation of the hypotheses on the dominant hydrological processes was fundamental for reducing structural model uncertainties and improving the streamflow prediction. For instance, in the case of the large river reach (MJR), it was shown that both lateral stream-aquifer water fluxes and groundwater flow in the underlying alluvium parallel to the river course are necessary to predict streamflow volume and channel transmission losses, the former process being more relevant than the latter. Regarding model uncertainty, it was shown that the approaches, which were applied for the unsaturated zone processes (highly nonlinear with elaborate numerical solutions), are much more sensitive to parameter variability than those approaches which were used for the saturated zone (mathematically simple water budgeting in aquifer columns, including backwater effects). In case of the MJR-application, we have seen that structural uncertainties due to the limited knowledge of the subsurface saturated system interactions (i.e. groundwater coupling with channel water; possible groundwater flow parallel to the river) were more relevant than those related to the subsurface parameter variability. In case of the WEGW application we have seen that the non-linearity involved in the unsaturated flow processes in disconnected dryland river systems (controlled by the unsaturated zone) generally contain far more model uncertainties than do connected systems controlled by the saturated flow. Therefore, the degree of aridity of a dryland river may be an indicator of potential model uncertainty and subsequent attainable predictability of the system.

“…When this is the case, it may be said that the hypothesis that a model structure "suits a catchment" cannot be rejected. To test if dominant processes are represented by a given model structure, it is important to have a sound method to evaluate which model structure suits better for a certain catchment and to understand the reasons behind it (Kirchner, 2006;Andréassian et al, 2009).…”

Section: Introductionmentioning

confidence: 99%

A framework to assess the realism of model structures using hydrological signatures

Boer-Euser

Winsemius

Hrachowitz

et al. 2013

231

230

Abstract. The use of flexible hydrological model structures for hypothesis testing requires an objective and diagnostic method to identify whether a rainfall-runoff model structure is suitable for a certain catchment. To determine if a model structure is realistic, i.e. if it captures the relevant runoff processes, both performance and consistency are important. We define performance as the ability of a model structure to mimic a specific part of the hydrological behaviour in a specific catchment. This can be assessed based on evaluation criteria, such as the goodness of fit of specific hydrological signatures obtained from hydrological data. Consistency is defined as the ability of a model structure to adequately reproduce several hydrological signatures simultaneously while using the same set of parameter values. In this paper we describe and demonstrate a new evaluation Framework for Assessing the Realism of Model structures (FARM). The evaluation framework tests for both performance and consistency using a principal component analysis on a range of evaluation criteria, all emphasizing different hydrological behaviour. The utility of this evaluation framework is demonstrated in a case study of two small headwater catchments (Maimai, New Zealand, and Wollefsbach, Luxembourg). Eight different hydrological signatures and eleven model structures have been used for this study. The results suggest that some model structures may reveal the same degree of performance for selected evaluation criteria while showing differences in consistency. The results also show that some model structures have a higher performance and consistency than others. The principal component analysis in combination with several hydrological signatures is shown to be useful to visualise the performance and consistency of a model structure for the study catchments. With this framework performance and consistency are evaluated to identify which model structure suits a catchment better compared to other model structures. Until now the framework has only been based on a qualitative analysis and not yet on a quantitative analysis.