Rainfall spatial modeling from different data sources

Rainfall is a phenomenon difficult to model and predict, for the strong spatial differences in data and the presence of many zero collected values. Observed data come from rain gauges, sparsely distributed on ground. These observations can be accompanied by other measurements, like reflectivity radar data. Rainfall prediction is a fundamental issue: all available data ought to be treated together for obtaining more precise results. In this work, we investigate whether radar data can contribute to improve spatial statistical prediction, comparing kriging estimates based on rain gauges with kriging on rain gauges and radar rainfall data acting as an external drift. Results are encouraging about enriching prediction with radar information.