Thursday, 17 January 2002: 1:42 PM
Distribution of daily rain gauge observations over hourly intervals with the aid of satellite precipitation estimates
In most nations of the world, daily observations of rainfall greatly outnumber hourly observations. Yet for hydrological monitoring, the hourly or 6-hourly rainfall data are preferred as input into river forecasting models. Geostationary satellite-based rainfall estimates are available at least every hour for continental scale regions, but their accuracy is inferior to the point measurements at gauges. We describe a technique, developed for Mexico, that partitions the daily rainfall amount into hourly intervals by using the hourly co-located satellite precipitation estimate. The technique preserves the accuracy of the daily observation at that location while making good use of the frequency of the satellite product. Furthermore, the local bias of the satellite estimate is computed every day. By using the bias to re-calculate new satellite estimates, one can create pseudo-hourly precipitation estimates in real time, even before new daily gauge observations are available at 12 UTC; and one can interpolate the bias correction factor between the gauge locations, which improves the satellite estimate.