Area-intensity probability distributions of rainfall based on a large sample of radar data

The history of weather radar is, among other applications, a manifestation of quantitative precipitation estimation (QPE), shown e.g. by Collier and Hardaker (2003). The main objective in such activities is to develop quantitatively accurate hydrological applications, especially nowcasting and warning systems of flash floods. The advantage of weather radars compared to conventional gauge networks is the availability of measurements in real time, which exhibit typically 100 times better time resolution and 10 000 times better spatial resolution than national gauge networks. High resolution facilitates the derivation of rainfall intensities from small areas and short time periods. On the other hand, the absolute accuracy of radar based rain rate in a randomly selected bin is not very good. Two dBs is considered a good achievement as commonly at long ranges the bias is of the order of 5-10 dBs figures which were generally not accepted prior to the work of Joss and Waldvogel (1990). Much work has been devoted in removing systematic biases and random errors from the radar estimates of rainfall rate (R) and accumulated precipitation e.g. due to clutter and nonmeteorological targets (Peura (2002)), calibration errors (Huuskonen (2001)) and sampling differences between gauges and radar (Germann and Joss (2003), Koistinen et al. (2003a)). As sufficient absolute accuracy can’t be guaranteed radar data has not yet been widely used for deriving climatologically reliable statistics of precipitation. Those would be valuable as national hydrological planning standards are typically based on, let’s say, 50 years old gauge data. If the challenge of inaccuracy can be solved to an acceptable level, radar measurements