Radar measurement of rainfall by differential propagation phase: A pilot experiment

Abstract

A pilot project concerning the measurement of rainfall by polarization diversity radar, conducted jointly by the Alberta Research Council and the University of Essex in the summer of 1989, is described. The objective of the experiment was to test the theory that differential propagation phase shift can give a better estimate of rainfall with high rain rates (about 30 mm h^?1 and greater) than reflectivity measurements.

The project comprised a field experiment that was conducted in central Alberta during the period 20 July to 2 August 1989. The field experiment involved observing storms within about a 70‐km radius from Red Deer with the ARC S‐band polarization diversity radar and measuring rainfall at the ground through a network of fixed, volunteer observers and a mobile storm‐chase operation.

Theory describing how differential propagation phase may be extracted from the data recorded by the radar system is given.

Data collected on three days during the experiment (27 July, 29 July and 2 August) have been analysed and the results are presented. A total of 31 samples of total rainfall amount were collected on these days. All but three of the 31 radar rainfall amount predictions obtained from the differential propagation phase are within a factor of 2 of the rainfall observed at the ground. In fact, the average ratio of the total rainfall amount predicted from the differential propagation phase to the total rainfall amount measured at the ground is 1.16 for the 31 samples. This suggests that, on average, the total rainfall amount predicted from the differential propagation phase is only 16% higher than that measured at the ground. Of the 31 events, over a third involved some contamination of the differential propagation phase measurement through hail. Furthermore, because the K_DP technique does not rely on parameters dependent on precipitation characteristics or adjustment factors, the technique can be validated in a convenient location and then applied anywhere.

On the other hand, the radar rainfall amount estimates derived from Z‐R relations represent, in general, a large overestimation of the rainfall amount observed at the ground, especially when hail is present. No attempt was made to adjust or calibrate the radar rainfall estimates with raingauge data.