Monday, 6 August 2007
Halls C & D (Cairns Convention Center)
More than 50 years have passed since scientists first attempted to make quantitative precipitation estimates from weather radars. Despite the significant progress that has been made in this area, the widespread operational use of radar data for quantitative hydrology is still regarded by many as unreliable, partly because of the poor quality of radar data that has been available for this purpose historically. Even when well-quality-controlled data have been used, the belief is that until there is complete agreement between the readings at the rain gauge and the rainfall estimates from the radar at the corresponding location, the radar data cannot be relied upon for quantitative precipitation estimates. It is believed (incorrectly in our view) that any disparities are due entirely to incorrect Z-R relationships We argue that the main problem is that the sampling characteristics of raingages and radars are quite different. Standard raingages measure rainfall at the ground surface, in a funnel a few inches in diameter, with a time resolution that depends on the rainfall rate. In contrast, radars estimate the amount of liquid water in drop form in a large volume (of several cubic kilometers), at some distance above the ground, at a relatively low temporal sampling interval of often 5 minutes or longer), which then falls to the ground with a variable fall speed and variable horizontal drift due to the wind patterns. These sampling mismatches mean that the radar and raingages are not actually observing the same raindrops, so it is not reasonable to expect then to agree on a minute-by-minute or even on an hour-by-hour basis. The attribution of the discrepancies between raingage and radar exclusively to errors in the Z-R relationship is not credible in light of these large sampling errors. In fact, continuous correction schemes are likely to introduce significant extra noise into the data, particularly in summer convective conditions. On the other hand, raingages and radar observations should agree on a longer time scale accumulation (on a monthly or even an annual basis) as radar miscalibration can obviously lead to systematic errors. The question that remains to be addressed to the satisfaction of the hydrological community is how raingage and radar data can best be combined to yield the optimum input data set for hydrological modeling.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner