Systematic variations of raindrop size spectra with altitude derived from wind profiler: Measurements for TRMM PR evaluation

The precipitation radar (PR) onboard the Tropical Rainfall Measuring Mission (TRMM) provides the 3-dimentional rain structures over the wide area of ocean and land. The derived rain rate from the PR is overall in good agreement with the surface measurement. However, there are still some discrepancies between the PR estimate and the rain rate derived from the microwave radiometer on the TRMM satellite. One of the key issues to validate the TRMM PR is a raindrop size distribution (DSD). Measurements of DSD are also crucial for better understanding of the rain formation processes and for the improvement of rain rate estimate from ground-based radar and the precipitation radar onboard the Tropical Rainfall Measuring Mission (TRMM). Many studies have reported the nature of DSD at the ground by using a disdrometer. Numerical models of the evolution of raindrop size spectra, however show significant changes of DSD during fall of raindrops, thus the measurements of the vertical profile of DSD are needed.

We have measured vertical profiles of DSD using 443 MHz wind profilers at Tsukuba and Okinawa, Japan. In the estimation of DSD, we have applied an iterative retrieval technique involving repeated convolutions. This technique assumes no particular shape of DSD and can be used to derive detailed raindrop size distributions. It, therefore enables small changes of DSD with altitude to be observed. Results show that significant vertical variations of DSD appear during fall of raindrops. We have calculated various integral parameters from the derived DSD, such as reflectivity factor Z, total number of raindrops N, second moment, and median volume diameter D0. Reflectivity factor correlates well with rain rate R, which is in agreement with previous studies. The second moment depends strongly on R and is also linearly related to NT. Physically, the second moment is related to the total surface area of raindrops. The probability of drop collision is proportional to the square of the drop size. This may lead to the strong relation between R and S. Statistically, no apparent correlation is shown between R and D0, as in previous studies. Vertical profiles from one event, however, show a clear inverse relationship between N to D0. Many profiles of N and D0, interestingly show the similar inverse relations. These systematic features are useful for the improvement of our understanding the rain formation process, the evaluation of the TRMM PR and the development of rainfall-estimate algorithms for the dual-frequency radar planned on board the GPM.