Performance Of Hydrological Models Research Articles

Use of statistically and dynamically downscaled atmospheric model output for hydrologic simulations in three mountainous basins in the western United States

This paper examines the hydrologic model performance in three snowmelt-dominated basins in the western United States to dynamically- and statistically downscaled output from the National Centers for Environmental Prediction/National Center for Atmospheric Research Reanalysis (NCEP). Runoff produced using a distributed hydrologic model is compared using daily precipitation and maximum and minimum temperature timeseries derived from the following sources: (1) NCEP output (horizontal grid spacing of approximately 210 km); (2) dynamically downscaled (DDS) NCEP output using a Regional Climate Model (RegCM2, horizontal grid spacing of approximately 52 km); (3) statistically downscaled (SDS) NCEP output; (4) spatially averaged measured data used to calibrate the hydrologic model (Best-Sta) and (5) spatially averaged measured data derived from stations located within the area of the RegCM2 model output used for each basin, but excluding Best-Sta set (All-Sta). In all three basins the SDS-based simulations of daily runoff were as good as runoff produced using the Best-Sta timeseries. The NCEP, DDS, and All-Sta timeseries were able to capture the gross aspects of the seasonal cycles of precipitation and temperature. However, in all three basins, the NCEP-, DDS-, and All-Sta-based simulations of runoff showed little skill on a daily basis. When the precipitation and temperature biases were corrected in the NCEP, DDS, and All-Sta timeseries, the accuracy of the daily runoff simulations improved dramatically, but, with the exception of the bias-corrected All-Sta data set, these simulations were never as accurate as the SDS-based simulations. This need for a bias correction may be somewhat troubling, but in the case of the large station-timeseries (All-Sta), the bias correction did indeed ‘correct’ for the change in scale. It is unknown if bias corrections to model output will be valid in a future climate. Future work is warranted to identify the causes for (and removal of) systematic biases in DDS simulations, and improve DDS simulations of daily variability in local climate. Until then, SDS based simulations of runoff appear to be the safer downscaling choice.

Niels Henrik Abel and his Times: Called Too Soon by Flames Afar

It has been repeatedly shown that conceptual hydrologic model performance degrades under conditions that deviate from those of the calibration period. In this study, we describe three experiments that aim to understand and address this problem using the conceptual model GR4J over 164 Australian catchments. The first is an investigation of model transferability, where parameters calibrated under certain conditions are applied to simulate both similar and contrasting conditions. We find that model performance degradation is more dependent on the conditions under which the model is tested than the conditions under which it is calibrated. Because dry periods are typically more difficult to simulate than wet periods, this means that transferring dry-calibrated parameters to wet periods is more successful than transferring wet-calibrated parameters to dry periods. For both wet and dry periods the best results were obtained when climatically similar calibration periods were used, suggesting that targeted use of climatically similar calibration data could improve predictive capacity. To this end, a second experiment was designed that preferentially weights modeled series calibrated under different conditions, with series associated with more climatically similar calibration periods weighted more heavily. While this improved model performance in most cases, the success was variable across the different catchments. Given that the model weighting scheme could not easily be generalized to all catchments in the sample, a third experiment was conducted where each model parameter was defined dynamically as a function of climate. The dynamic parameters were calibrated separately for each model, so the individual sensitivities of each catchment to climate conditions could be captured. While this also gave performance improvements, especially under drier testing conditions, the results continued to vary between catchments and there was no clear pattern in the parameter variation. This suggests that nonstationarity can be captured in different parameters for models of different catchments. While both model weighting and dynamic parameters can benefit overall conceptual model performance, it seems that reliable improvements across large samples of catchments may be difficult to achieve without more physically realistic model structures.

Discharge analysis and runoff mapping applied to the evaluation of model performance

Performance of hydrological models is usually assessed using the Nash criteria, based on differences between daily discharge time series. Two ways are investigated to extend the comparison tests. The first one is based on other statistical analyses performed at the gauging stations. The second is based on interpolation procedures that are specific to discharge data. These approaches are used within Gewex-Rhône, a research project aiming to achieve a coupled modelling of continental and atmospheric water cycle components at mesoscale. Two versions of a distributed hydrological model (MODCOU and CIRSE) were applied to the French part of the Rhône basin (86200 km2) within this project. They provided a considerable amount of simulated daily discharges to be objectively compared to observations.

Analysis of aggregation and disaggregation effects for grid-based hydrological models and the development of improved precipitation disaggregation procedures for GCMs

Abstract. Appropriate representation of hydrological processes within atmospheric General Circulation Models (GCMs) is important with respect to internal model dynamics (e.g. surface feedback effects on atmospheric fluxes, continental runoff production) and to simulation of terrestrial impacts of climate change. However, at the scale of a GCM grid-square, several methodological problems arise. Spatial disaggregation of grid-square average climatological parameters is required in particular to produce appropriate point intensities from average precipitation. Conversely, aggregation of land surface heterogeneity is necessary for grid-scale or catchment scale application. The performance of grid-based hydrological models is evaluated for two large (104km2) UK catchments. Simple schemes, using sub-grid average of individual land use at 40 km scale and with no calibration, perform well at the annual time-scale and, with the addition of a (calibrated) routing component, at the daily and monthly time-scale. Decoupling of hillslope and channel routing does not necessarily improve performance or identifiability. Scale dependence is investigated through application of distribution functions for rainfall and soil moisture at 100 km scale. The results depend on climate, but show interdependence of the representation of sub-grid rainfall and soil moisture distribution. Rainfall distribution is analysed directly using radar rainfall data from the UK and the Arkansas Red River, USA. Among other properties, the scale dependence of spatial coverage upon radar pixel resolution and GCM grid-scale, as well as the serial correlation of coverages are investigated. This leads to a revised methodology for GCM application, as a simple extension of current procedures. A new location-based approach using an image processing technique is then presented, to allow for the preservation of the spatial memory of the process.

Use of objective criteria for the assessment of biogeochemical ecosystem models

Ecosystem modeling is confronted with complex biological systems and changing environmental conditions. A model which describes ecosystem behavior under all conditions has not been found yet and there does not exist one `true' model for a specific ecosystem. Often ecosystem models describe the measured data more or less well, but most judging criteria for model performance are rather subjective. Furthermore, from a mathematical view point the calibrations of ecosystem models are hardly ever unique. The aim of this study was to develop and use criteria which permit an objective comparison of different models to the observed field data and to each other. A given model which describes a specific system significantly better will be declared the `valid' model while the other will be rejected. The term `valid' is used here in a sense that any model that could not be proven invalid would be a valid model for the system. We used the biogeochemical soil models MAGIC (Cosby, B.J., Hornberger, G.M., Wright, R.F., 1985. Modelling the effects of acid deposition: assessment of a lumped-parameter model of soil water and stream water chemistry. Water Resour. Res. 21, 51–63) and the SO-Model (derived from the Batch Equilibrium Model (BEM; Prenzel, J., 1991. Introduction to BEM (Batch Equilibrium Model), vol 28. Berichte des Forschungszentrums Waldökosysteme/Waldsterben, Göttingen, 51 pp.). The data set used was the soil solution chemistry in a forest ecosystem of the Solling area (North-West Germany). To test the performance of the models four criteria were used: the efficiency (Martinec, J., Rango, A., 1989. Merits of statistical criteria for the performance of hydrologic models. Water Resour. Bull. 25 (2), 421–432; Hinzman, L.D., Kane, D.L., 1991. Snow hydrology of a headwater artic basin; 2. Conceptual analysis and computer modelling. Water Resour. Res. 27, 1111–1121), the Normalized Mean Absolute Error ( NMAE, given by Janssen, P.H.M., Heuberger, P.S.C., 1995. Calibration of process orientated models. In: van Grinvsen, J.J.M. (Ed.), Modelling Water, Carbon and Nutrient Cycles in Forests: Application of 16 Simulation Models to a Spruce Stand at Solling, Germany. Ecological Modelling, vol. 83, pp. 55–66), the confidence interval test ( CIT, developed in this study) and the model rejection criteria (Sun, N.Z., 1994. Inverse Problems in Groundwater Modelling. Dordrecht, 337 pp.). Whereas the efficiency and NMAE are related to the averaged data, the CIT and the model rejection criteria include the spatial heterogeneity at every time step. When evaluated visually, both model results might be accepted. From the application of the model performance criteria we selected the MAGIC model as the `valid' model for our system.

Validation and Intercomparison of Different Updating Procedures for Real-Time Forecasting

The paper presents a classification and a review of the updating procedures currently used in real-time flood forecasting modelling. On the basis of results from the WMO project ‘Simulated Real-Time Intercomparion of Hydrological Models’, comprising more than 10 commonly used hydrological models and a variety of different updating procedures, an analysis of the relative importance of updating procedures and hydrological simulation models is provided. In particular, an intercomparison is made beteween two models (NAMS11/MIKE11 and NAMKAL) consisting of the same hydrological model (NAM conceptual rainfall-runoff) but containing different routing modules (linear reservoirs versus hydraulic routing) and different updating procedures (error prediction versus state variable updating based on an extended Kalman filter). A main conclusion is that updating procedures significantly improve the performances of hydrological models for short-range forecasting. Furthermore, there are no clear conclusions regarding which type of updating procedure performs the better. However. intercomparison of the NAMS11 and NAMKAL models indicates that the extended Kalman filter is marginally better than an error prediction model in cases where the basic hydrological model simulation is good. Finally, it is concluded that the basic simulation is very essential for accurate forecasts, and that the better the basic simulations are the better the updating routines in general function. This puts emphasis on the importance of thoroughly calibrating and validating the hydrological simulation models before applying them together with updating routines in operational real-time forecasting.

Assessing the Performance of the NELUP Hydrological Models for River Basin Planning

The hydrological modelling capability within the NERC/ESRC Land Use Programme is built around two simulation codes, SHETRAN and ARNO. Together these provide a comprehensive tool for analysing the potential impacts of land use change on the hydrology of a catchment. Both models have been validated for five-year flow simulations on the prototype Tyne catchment. Results from the SHETRAN nitrate simulations are also presented with the limited available validation data. Finally, the value of the hydrology models is discussed in the context of river basin planning.

Approximate confidence intervals for verification criteria of the WMO intercomparison of snowmelt runoff models

Abstract The comparison of the performances of hydrological models is usually based on the values of criteria which are functions of the differences between estimated and observed flows. In comparing models we have to consider whether the differences between performance criterion values may be due to sampling variation. To deal with this problem, the comparison of approximate confidence intervals of criterion values using the jackknife statistic is proposed. The method is applied to the results of the recent WMO project for the intercomparison of conceptual models of snowmelt runoff.

Entropy as a measure of hydrologic data uncertainty and model performance

This paper extends the original use of entropy by Amorocho and Espildora as a measure of uncertainty in hydrologic data and the reduction in that uncertainty due to application of a model. In many situations calculations of entropy are better based on a proportional rather than a fixed class interval. General equations are given for the proportional class interval, are solved for assumed log-normal and gamma distributions, and extended to data series with zero values. The solutions are shown to be independent of the units of measurement of the original data. The results are applied to comparative evaluation of hydrologic models using the same and different data sets, and a new criterion of model performance is suggested for the latter case.