Hydrological modelling using convective scale rainfall modelling – phase 2

Environment Agency. 2010 Hydrological modelling using convective scale rainfall modelling – phase 2. Bristol, UK, Environment Agency, 162pp. (SC060087/R2)

Before downloading, please read NORA policies.
Phase_2_Project_Report_SCHO0210BRYQ-e-e.pdf - Published Version

Download (10MB)


Hydrological models have the capability to provide useful river flow predictions and flood warnings. The aim of this project, ‘Hydrological Modelling using Convective Scale Rainfall Modelling’, is to investigate which models and associated computational methods would allow best use of the latest Met Office developments in numerical weather prediction (NWP). Two recent enhancements in particular offer interesting opportunities and open the door to the use of probabilistic flood forecasting. These two developments are: • operation of the nowcasting system STEPS (Short Term Ensemble Prediction System) at 2 km resolution; • a new system for longer term numerical weather prediction called MOGREPS (Met Office Global and Regional Ensemble Prediction System). The three-phase project is concerned primarily with: • how to use high resolution (convective scale) rainfall forecasts effectively for flood forecasting; • how to make operational the use of ensembles of numerical weather prediction (MOGREPS) in flood forecasting and warning within the Environment Agency’s National Flood Forecasting System (NFFS). This report presents the results of Phase 2 (pilot case study). During Phase 1 (inventory and data collection), the storm event at Boscastle on 16 August 2004 was selected as the case study to test various hydrological modelling concepts for the transformation of high resolution rainfall predictions into accurate flood forecasts. Three hydrological models (one lumped and two distributed) were applied to the north Cornish catchments affected by the Boscastle event: • Probability Distributed Moisture (PDM) model; • physical–conceptual Grid-to-Grid (G2G) model; • physics-based Representative Elementary Watershed (REW) model. The three models were configured and calibrated for the three gauged catchments (Ottery, Tamar and Camel) selected as the focus of the case study. Raingauge-adjusted radar rainfall data produced using HyradK were used as model input. This phase also coupled the latest Met Office high resolution NWP products with the distributed hydrological model developed by the Centre for Hydrology & Ecology at Wallingford and considered the future potential of ensemble convective scale rainfall predictions. The term ‘distributed forecasting’ in this sense means the use of a spatially distributed (grid-based) hydrological model to forecast ‘everywhere’. This contrasts with current hydrological model networks that comprise a connected set of (normally) lumped rainfall-runoff (‘catchment’) models feeding into hydrological and hydrodynamic river models which provide forecasts only at specific locations. Ensemble forecasting was configured in a test NFFS system for two Environment Agency Regions (Thames and North East) set up at Deltares in the Netherlands as part of the Delft Flood Early Warning System (Delft-FEWS) and receiving MOGREPS forecasts from the Met Office. Particular attention was given to the effect on system performance as it is necessary, when running the models in ensemble mode, to repeat the forecast workflow 24 times. Conclusions Phase 2 demonstrated that a distributed hydrological model (set up using a digital terrain model) can be operated on the National Flood Forecasting System platform with short enough run times for use in real-time forecasting. The PDM model gave excellent performance across catchments but was insensitive to the storm pattern. The G2G model gave good performance across catchments; ungauged performance was also good. The REW model gave good performance for winter periods but overall performance was only reasonable; ungauged performance was on a par with gauged performance. Distributed models were considered to have a number of advantages including: • sensitivity to spatio–temporal structure of storms; • helpful in understanding storm and catchment shaping of flash floods; • can identify locations vulnerable to flooding; • help forecast floods shaped by ‘unusual’ storm and catchment conditions absent from the historical record; • provide a complete spatial picture of flood hazard across a region; • respond sensibly to ensemble rainfall forecasts that vary in position. The performance of the G2G model was considered particularly promising with a number of its attributes being relevant to convective scale probabilistic flood forecasting. Recommendations • Hydrological modelling concepts to be carried through to Phase 3. - The distributed G2G model shows promise for Area-wide flood forecasting at gauged and ungauged locations. - An extended G2G formulation incorporating soil/geology datasets should be considered. • Case study selection for Phase 3. A more focused ‘regional assessment’ should be undertaken based on an area within Midlands Region affected by the summer 2007 floods and utilising raingauge data in combination with radar data for improved rainfall estimation as model input. More detailed analyses (including use of high resolution NWP pseudo ensembles) should be performed for the selected case study area in the Midlands Region. • Continuation of the MOGREPS trial into Phase 3. This will allow: - more events to be captured; - additional forecast stability tables to be added to the test configuration; - a wider range of Environment Agency staff to assess the results. A more radical recommendation is to trial the G2G model countrywide across England and Wales in Phase 3. At the end of Phase 3 (analysis and verification), overall conclusions will be drawn on the general benefits of using high resolution NWP as input into a hydrological model for flood forecasting. In addition, a possible approach using the hydrological models – and calibration and computation methods – will be formulated.

Item Type: Publication - Report (UNSPECIFIED)
Programmes: CEH Topics & Objectives 2009 - 2012 > Water > WA Topic 3 - Science for Water Management > WA - 3.1 - Develop next generation methods for river flow frequency estimation and forecasting
CEH Sections: Boorman (to September 2014)
ISBN: 9781-849111829
Funders/Sponsors: Environment Agency, Centre for Ecology & Hydrology
Additional Keywords: flood forecasting, hydrological modelling, numerical weather prediction, probabilistic forecasting, uncertainty
NORA Subject Terms: Meteorology and Climatology
Atmospheric Sciences
Date made live: 30 Mar 2011 09:43 +0 (UTC)

Actions (login required)

View Item View Item

Document Downloads

Downloads for past 30 days

Downloads per month over past year

More statistics for this item...