Improved distributed runoff modelling of urbanised catchments by integration of multi-resolution remote sensing

The runoff amount and intensity on catchment scale is strongly related to the spatial distribution of impervious area cover, which is the predominant cover type in urbanized area. This can only be taken effectively into account when a fully- distributed hydrological model is used. In this paper we investigate the assessment of imperviousness by a multi- resolution remote sensing technique. The remote sensing approach uses a classified high resolution (HR) Ikonos image that covers part of the research area to train a neural network based sub-pixel classification model that estimates impervious surface cover proportions within the pixels of a medium-resolution (MR) Landsat ETM+ image that covers the entire area. The GIS based distributed WetSpa model was used for studying the influence of different imperviousness scenarios on runoff generation with an hourly time step. It shows that estimates of imperviousness derived from satellite data may strongly improve those made by experts, as well as the necessity of application of fully-distributed grid-based hydrological models for urban runoff simulation.