Drop Size Distribution Temporal Sample Sensitivity Analysis on Radar-Rain Relations and Gamma Model Parameters

Rainfall has spatial and temporal variability that is related to drop size distribution (DSD) temporal sampling. Radar-rainfall algorithms are typically based on an assumed radar scattering and raindrop physical models which include fitting a mathematical model such as a gamma function with three parameters—intercept (N0), shape (µ), and slope (Λ) or median drop diameter (D0)—to the DSD. The objective of this study is to characterize the sensitivity of inferred gamma parameter and radar-rain relations to the assumed sample time of rain DSD from disdrometer data. Selecting sample times to optimize these relations can ultimately help improve satellite rainfall algorithms and associated ground validation efforts.

Previous studies have often used a 1-minute DSD sample based largely on trial-and-error empirical experience; however, there has been little documentation as to why a 1 minute sample time is optimal. If the radar and disdrometer observations are to be inter-compared, the appropriate spatial and temporal scales and disdrometer sampling time should be considered to a first order. For example, the ground-based radar gate spacing is on the order of 100 meters while the Global Precipitation Measurement (GPM) Mission Dual-Precipitation Radar (DPR), launched in February 2014, has a footprint of 5 km. These are two highly different spatial scales; however, the DSD sampling time studies continue to use 1 minute samples for comparisons. Furthermore, the following characteristics of hydrometeors and radars need to be taken into account when optimizing sample times: convective versus stratiform, distribution of drop velocities, storm motion, and radar beam width.

Based on simple principles, it can be shown that 1, 2, 3, 4-minute (ground-based radars), and 8-minute (GPM-DPR) DSD temporal sampling are all consistent with typical radar spatial scales and physical processes. Therefore, in this study various sample time integrations are analyzed from 2-dimensional video disdrometer (2DVD) DSD data from the NASA's GPM Iowa Flood Studies (IFloodS) ground validation (GV) experiment. The truncated method of moments is employed to estimate gamma fit parameters from binned 2DVD data. The gamma fit parameters are then used to calculate DSD moments as well as input into T-matrix and Muller matrix scattering models for the calculation of radar observables (e.g., radar reflectivity (Zh), differential reflectivity (Zdr), specific differential phase (Kdp), and others). Various radar-rain and gamma parameter relations will be tested to document any relative biases from the different temporal sampling. Preliminary results show that D0 (Zdr) 4th order fit compared to truth show less than 10% bias while the gamma parameters µ-Λ relationship (quadratic) show normalized bias up to 10%. Other relations that are planned to be tested are attenuation (Ah and Adp) and Dual Frequency Ratio (DFR) as well as, DFR-D0, radar-rainrate (R) estimation from R(Zh), R(Zh,Zdr), and R(Kdp, Zdr) which are all related to GPM-GV science goals as well as GPM radar relations.