Abstract/Summary

Intense rainfall can lead to flash flooding and may cause disruption, damage and loss of life. Since flooding from intense rainfall (FFIR) events are of a short duration and occur within a limited area, they are generally poorly predicted by Numerical Weather Prediction (NWP) models. This is because of the high spatio-temporal resolution required and because of the way the convective rainfall is described in the model. Moreover, the hydrological process descriptions of Land Surface Models (LSMs) are not necessarily suitable to deal with cases of intense rainfall. In this study different representations of infiltration into the soil were developed in the JULES land surface scheme with the aim of improving prediction of the amount of surface runoff, and thus ultimately river flow. Infiltration and surface runoff are explored in a test case of intense rainfall with a variable maximum infiltration. The modelled hydraulic conductivity profile is modified with depth to reduce the rate of outgoing fluxes. The new infiltration scheme is then applied to different UK catchments. The resulting river flow is evaluated against a benchmark river flow calculated using default infiltration in JULES and also observations. The results demonstrate improved representation of the highest flows with this new variable maxiumum infiltration scheme in some catchments but limited improvement elsewhere. This scheme shows best improvement in the wettest areas of the UK where the annual mean precipitation is above 1200 mm. This work highlights the requirement for substantial further work on the hydrological process representation in JULES.