Detecting weather radar clutter using satellite-based nowcasting products |
| | Abstract | This contribution presents the initial results from experiments with detection of weather radar clutter by information fusion with satellite based nowcasting products.
Previous studies using information fusion of weather radar data and first generation Meteosat imagery have shown promising results for the detecting and removal of clutter. Naturally, the improved spatio-temporal resolution of the Meteosat Second Generation sensors, coupled with its increased number of spectral bands, is expected to yield even better detection accuracies.
Weather radar data from three C-band Doppler weather radars of the Danish Meteorological Institute has been extracted for cases of severe to moderate cases of land and sea clutter. For comparison, cases of clutter free data has also been analyzed.
The satellite-based dataset used is an operational meteorological product developed within the 'Nowcasting Satellite Application Facility' of EUMETSAT and is based on multispectral images from the SEVIRI sensor of the Meteosat-8 platform. Of special interest is the 'Precipitating Clouds' product, which uses the spectral information coupled with surface temperatures from Numerical Weather Predictions to assign probabilities for three rain intensity classes (heavy precipitation, light to moderate
precipitation, and no precipitation) to every pixel in the Meteosat-8 images.
Image fusion of the satellite image product and the radar composite images was done by resampling to a common grid of a spatial resolution dictated by the resolution of the radar data. Subsequently, a supervised classifier was developed based on training data selected by a weather radar expert.
Results of classification of data from several different meteorological events are shown. Cases of widespread sea clutter caused by anomalous propagation are especially seen to be detected with high accuracy using the proposed method, however, the satellite based dataset overestimates precipitation which results in weather radar clutter being misclassified as precipitation. Other sources of errors were identified, such as geometrical and temporal misalignments arising from the use of multi-source datasets. | | Type | Conference paper [With referee] | | Conference | Proceedings of the Fourth European Conference on Radar in Meteorology and Hydrology (ERAD) 2006, 18-22 September, 2006 | | Year | 2006 Month September pp. 153-156 | | Address | Barcelona, Spain | | Electronic version(s) | [pdf] | | BibTeX data | [bibtex] | | IMM Group(s) | Image Analysis & Computer Graphics, Geoinformatics |
|