Methodology/work description :
While an improving of the integration of radar and raingauge data is a key objective of the existing MUSIC (EVK1-CT-2000-00058) research this proposed project will concentrate on integrating these with NWP models for forecasting precipitation; thus an overall improvement of flood prediction capability is expected, especially for the longer warning lead times required by end-users. The focus of this wp is the assessment of this crucial aspect of the proposal.
The assessment of bias and spatial anomalies carried out in WP9 will guide the development of the best methodology. In particular, ways of assimilating data from rain gauges and radar into the precipitation forecasts will be investigated. The expected improvement in the accuracy of the forecasted precipitation from the use of other data sources will be assessed. In all cases the most important criterion will be the performance in estimating flood discharges in accordance with the end-user requirements determined in WP12. Thus all candidate methodologies will be tested by using the resulting combined estimates as inputs to the hydrological catchment models and comparing the full hydrograph of the forecasted flood with the measured hydrograph.
This work will be done at both small and medium catchment scales. The small (120 km2) Irish catchment of partner 9 is capable of producing flash floods and will require special 5 km grid size HIRLAM runs for the periods before and during severe rainstorms. In the Irish catchment, the discharge from 10 sub-catchments (average area 10 km2) will be recorded during the project. This provides a uniquely fine spatial density of discharge measurements to be able to detect spatial anomalies in the input precipitation fields. It will allow us to model and check how the spatial distributions of rainfall are converted into runoff in each of the sub-basins. It will also allow us to determine the degree of complexity required in a catchment model for flood forecasting in different environments (rural, urban, large/small). The catchment of partner 4 is considerably larger, and the operational 20 km grid resolution is expected to be sufficiently accurate.
The relationship between forecast lead time and accuracy and reliability will be assessed for the recommended methodology. Of particular concern will be the forecast lead time at which NWP becomes useful. This determines cross-over between the shorter lead times for which the methods originally developed during the TELFLOOD (ENV6-CT96-0257) project are best and the longer lead times for which the method developed here should be better. The Basic minimum requirements for an operational system will be determined at an early stage by the end-user group in WP12. Subsequently, guidance on the most useful combinations of forecast lead-time and accuracy will be determined in consultation with the end-user group.