Michael Butts

IMPROVING STREAMFLOW SIMULATIONS AND FLOOD FORECASTS WITH MULTIMODEL ENSEMBLES

Methods for improving the hydrological simulation through the use of multi-model ensembles (MME) are demonstrated. In recent years, the meteorological community has exploited several MME combination techniques as a means for improving short-term weather and seasonal climate forecasts. Within the hydrological community, little work has been carried out to explore the benefits of MMEs for streamflow simulations. This study examines the use of MMEs for improving streamflow simulation, including their potential benefits to flood simulation. Multimodel ensembles are generated using ten distinct model structures derived using a new hydrological modelling tool. These model structures were applied to a US NWS test catchment, the Blue River and evaluated using a split-sample procedure. The resulting ensemble can be used to make probabilistic simulations that characterise model structure uncertainty. Furthermore it is shown that the ensemble average of all 10 models performs better than any single model in split sample test. Using regression methods, improvements in ensemble simulations using linear combinations of the ensemble members were explored. The performance of the resulting weighted ensemble is similar to the simple ensemble average but uses a smaller ensemble. This may provide a means to identify which model structures provide significant contributions to accurate hydrological simulation.

Flood forecasting in the anglian region

The Anglian Region of the Environment Agency of England and Wales has replaced a fragmented and in most places technically limited flood forecasting capability with a state of the art system (the Anglian Flow Forecast and Modelling System – AFFMS). This has created a platform from which to tackle issues of accuracy, reliability and timeliness and has expanded access to forecasts to a wide customer base. Perhaps most importantly, it will help flood event managers move from a relatively passive (monitoring) mindset, to a forecasting approach that considers “what will, or what might happen?” However, the demands upon flood forecasting are increasing beyond even this newly acquired capability This paper reviews the technical and organisational improvements achieved in the Anglian Region within the context of the traditional structure of forecast delivery. It then considers future challenges, proposing a development of this structure to a probabilistic or risk-based approach. This approach explicitly incorporates the limitations inherent in predicting flood events due to the natural variability and inherent uncertainty of hydrological systems. It develops the proactive forecasting mindset further to consider “how likely is a particular outcome?” and “what are the likely consequences?” It also provides a framework within which technical developments can be prioritised and expectations managed Although the specific circumstances of Anglian Region are unlikely to recur elsewhere, it seems likely that the need for risk forecasting will emerge elsewhere, at least within service-orientated societies. The paper therefore concludes that, if forecast delivery is to keep pace with rising demands, highest priority should be given to coupling probabilistic forecasting with forecasting the depth and velocity of inundation. This will require advances in forecasting science

DATA ASSIMILATION TO IMPROVE FORECAST QUALITY OF RIVER BASIN MODELS

Ideally, real-time flood management decisions must be based on an understanding of nthe uncertainties and associated risks. It is therefore central for effective flood nmanagement tools to provide reliable estimates of the forecast uncertainty. Only by nquantifying the inherent uncertainties involved in flood forecasting can effective nreal-time flood management and warning be carried out. Forecast uncertainty requires nthe estimation of the uncertainties associated with both the hydrological model ninputs (e.g. precipitation observations and forecasts), model structure, nparameterisation and calibration, and methodologies that predict how the nuncertainties from different sources propagate through the hydrological and nhydraulic system.nnWithin the EU 5th framework project FLOODRELIEF, an ensemble-based approach has been ndeveloped to address the issue of handling and quantifying forecasting and modelling nuncertainties. A general stochastic framework for flood forecast modelling is npresented based on the Ensemble Kalman Filter (Evensen, 1994). The Kalman filter nprovides a natural framework for determining how the different sources of nuncertainty propagate through the hydrological and hydraulic models and to reduce nforecast uncertainty via data assimilation of real-time observations. An evaluation nof this framework is presented for several case studies including the US NWS study ncatchment, the Blue river basin and the Welland and Glen River Basin in the UK.nnTwo methods for introducing uncertainties into the model are compared:n1. Stochastic errors are added to the runoff calculated by the catchment model. nOnly states in the river channel model are updatedn2. Stochastic errors are added to the input to the catchment model (e.g. nprecipitation and evaporation). States in both the catchment model and in the river nchannel model are updated.nnIn particular, an investigation of the value of these two approaches for rapidly nresponding river basins versus more slowly responding systems is presented. As nexpected it is observed that updating in both the catchment model and the river nchannel model has a longer lasting effect on the forecast than updating in the river nchannel alone. Finally the results of this evaluation highlight the fact that one of nthe major outstanding problems in estimating the forecast uncertainty is the ncharacterisation of the sources of uncertainty.nnReferences:nEvensen, G. (1994), Sequential data assimilation with a nonlinear quasi-geostrophic nmodel using Monte Carlo methods to forecast error statistics, J. Geophysical nResearch, vol. 99, no. C5, pp. 10143-10162.

DYNAMIC COUPLING OF HYDROLOGICAL AND ATMOSPHERIC MODELS TO EXAMINE FEEDBACK EFFECTS

Understanding the interaction between terrestrial microclimate, hydrology and necology is a key to determining the effect of land-use and climate change on nhydrological systems. Traditionally, the hydrological impacts of climate change have nbeen based on driving hydrological models with the output of regional climate nmodels. These climate models often operate at spatial and temporal scales that are nmuch larger than the scales required to analyse the effects on the hydrological nsystem. This is in part because of computational limitations and in part because of nthe physics of the regional models do not justify much higher resolution. nFurthermore there is an inherent contradiction in this approach since these climate nmodels include their own hydrological model component. Similarly in analysing the nhydrological effects of land-use change the feedback to the meteorological system is noften neglected.nnTo address these issues a dynamically coupled hydrological and meteorological model nsystem has been developed for evaluating interactions at hydrological (catchment) nscales. Such a coupled system provides a unique framework for investigation of land nsurface - atmosphere interactions at hydrological scales. A comprehensive nhydrological modelling system describing the terrestrial component of the nhydrological cycle MIKE SHE has been modified to allow coupling to a local scale nmeteorological model. The coupling exploits the European Open Modelling Interface nand Environment (OpenMI) Open MI to link the two models systems. As simulations can nbe run with and without coupling to the meteorological model, it is possible to nevaluate the impact of feedbacks between the two systems on hydrological npredictions. The uncoupled system has been validated against remote sensing and neddy correlation measurements at the field and landscape scale describing the nhydrological and energy fluxes on the land-surface. The coupled system is then nvalidated against field data describing both the atmosphere and hydrological system. nFinally, a sensitivity analysis is carried out to examine the sensitivity of nhydrological predictions to atmospheric feedbacks to identify where feedback will nsignificantly affect the simulated impact.

Effects of future climate change on groundwater in Denmark

The water supply in Denmark is entirely based on groundwater and it is therefore of nconcern how climate change will affect the groundwater reserves in the future. We nhave analyzed this problem by retrieving climatological output data from a regional nclimate model and using this data in a distributed hydrological model, focusing on a nselected watershed in the country. nGlobal climate models simulate the global climate system with greenhouse gas nconcentrations representing observed and possible future conditions, but their nlimited regional detail makes them less suitable for hydrological impact studies. nTherefore, regional climate models with a limited model domain and higher resolution nare utilized. In this study the regional climate model HIRHAM from the Danish nMeteorological Institute is used with boundary conditions generated by the global nclimate model HadAM3H from Hadley Centre. The concentrations of greenhouse gas and naerosols, have been applied based on the IPCC SRES A2 emission scenario. nThe climate model output consists of data for two time slices, one for a period nrepresenting recent climate (1961-1990; control run) and one for the future climate n(2071-2100; scenario run). The climate output used is daily precipitation, ntemperature and potential evapotranspiration, all at a 12x12 km resolution. nClimate models are subject to systematic biases, so climate model data cannot be nused directly in hydrological models. Different transfer methods exist to transfer nthe signal of climate change, of which two will be examined in this study. One is nthe so-called delta change approach, that alters the original hydrological model ninput data with a factor deduced from the climate model output data to generate the ninput data for the hydrological simulation of the future scenario. The other method nis the direct method, that uses the climate model output directly as input for the nhydrological model after correcting the climate output with factors based on the nbias between the climate model control scenario data and the original hydrological nmodel input data.nWe use the so-called DK model as the model for the hydrological analysis. The DK-nmodel divides Denmark into 10 regions and it is based on the distributed nhydrological model code MIKE SHE with a horizontal spatial discretization of 1 x 1 nkm2. One of these regions has been selected as the study area for the impact nanalysis.