Monday, 28 August 2023: 5:00 PM
Great Lakes A (Hyatt Regency Minneapolis)
The Dual-frequency Precipitation Radar (DPR) onboard the Global Precipitation Measurement (GPM) core mission, operating at Ka and Ku-bands, brought new opportunities to estimate the characteristics of cloud microphysics, such as amount, intensity, and type of precipitation, as well as microphysical characteristics of rainfall globally. The operational GPM precipitation algorithm uses a three-parameter —shape (µ), mass-weighted mean diameter (Dm), and normalized intercept parameter (Nw)— Normalized Gamma Size Distribution (NGSD) to describe cloud particle size distributions (PSD). The shape (µ) parameter has been traditionally fixed, and set to be equal to three, to solve the precipitation retrieval system of equations. In this study, we test the hypothesis that the lack of degrees of freedom in a three-parameter NGSD estimation is a significant source of uncertainty using microphysical measurements from The NASA Clouds, Aerosol, and Monsoon Processes Philippines Experiment (CAMP2Ex) field campaign. The CAMP2Ex field campaign conducted 19 research flights near the Philippines using two NASA aircraft platforms, NASA P-3 and SPEC Learjet aircraft, with remote sensing and in situ microphysical probes to characterize cloud and precipitation structures under varying tropical rainfall conditions.
In our study, the Two-Dimensional Spectrometer (2DS) and the high-volume precipitation spectrometer (HVPS) in situ measurements were used to compute cloud bulk properties, NGSD size distribution parameters, and rainfall rates at 1-Hz resolution and compared with those retrieved using a GPM-like precipitation algorithm over tropical cumulus (Cu) and cumulus congestus (Cu-con) clouds. Results show that both NGSD models, with and without a fixed µ parameter, do not accurately represent in-situ PSD measurements on Cu and Cu-con clouds, leading to significant errors in retrieving cloud microphysical parameters. The mean absolute error (MAE) and root mean square error (RMSE) values for Dm, rainfall rate (R), and the Nw parameter, which is analyzed on a logarithmic scale, were 0.47 mm and 0.68 mm, 22.3 mm hr-1 and 632.6 mm hr-1, and 9.2 log10(mm-1 m-3) and 12.4 log10(mm-1 m-3), respectively. Potential improvements in developing an analytical retrieval algorithm from a dual wavelength-NGSD framework are also examined. A novel neural network algorithm is applied to retrieve the three parameters of the PSD. Retrieved microphysical parameters and rainfall rates from the neural network framework are compared with the GPM baseline algorithms, and the strengths and weaknesses of these approaches are discussed.
In our study, the Two-Dimensional Spectrometer (2DS) and the high-volume precipitation spectrometer (HVPS) in situ measurements were used to compute cloud bulk properties, NGSD size distribution parameters, and rainfall rates at 1-Hz resolution and compared with those retrieved using a GPM-like precipitation algorithm over tropical cumulus (Cu) and cumulus congestus (Cu-con) clouds. Results show that both NGSD models, with and without a fixed µ parameter, do not accurately represent in-situ PSD measurements on Cu and Cu-con clouds, leading to significant errors in retrieving cloud microphysical parameters. The mean absolute error (MAE) and root mean square error (RMSE) values for Dm, rainfall rate (R), and the Nw parameter, which is analyzed on a logarithmic scale, were 0.47 mm and 0.68 mm, 22.3 mm hr-1 and 632.6 mm hr-1, and 9.2 log10(mm-1 m-3) and 12.4 log10(mm-1 m-3), respectively. Potential improvements in developing an analytical retrieval algorithm from a dual wavelength-NGSD framework are also examined. A novel neural network algorithm is applied to retrieve the three parameters of the PSD. Retrieved microphysical parameters and rainfall rates from the neural network framework are compared with the GPM baseline algorithms, and the strengths and weaknesses of these approaches are discussed.

