Using Observed Drop Size Distributions in Light Rainfall: Application to the Combined DPR-GMI Algorithm

Abstract:

The drop size distributions in light rainfall (defined here as rates < 1 mm/h) arguably show the highest variability for a given rain rate (R) due to large variations in the “normalized” intercept parameter (N_w) which is negatively correlated with the mass-weighted mean diameter (D_m) whereas the scaled-normalized shape is remarkably stable. Recent observations of the DSD using collocated high resolution optical array probe for drizzle and small drops (D<1 mm) and 2D-video disdrometer for larger sizes (>0.75 mm) have shown that normalizing N(D) by N_w and scaling D by D_m [or in compact notation N(D)=N_wh(x) where x=D/D_m], leads to the generalized gamma model as a good “climatological” fit for h(x) describable by two shape parameters (μ, c). The GPM DPR and combined algorithms use the legacy standard gamma model for h(x) with one shape parameter μ fixed at 2 or 3 which gives a very different convex down shape for small x (<0.5) relative to the concave up shape for the generalized gamma. This is due to small drop truncation when only one disdrometer is used (the legacy Joss disdrometer from TRMM era). However, this affects the lower order moments (0-2) quite strongly and less so for moments 3 and higher. In particular, the accurate retrieval of N_w with low bias and low standard error in light rainfall has proven to be difficult for the combined algorithm since in such cases the apriori state assumes more importance. Hence, we propose to use DSDs from C-130 aircraft data in Drizzle (0ceanic Shallow Warm Rain Stratocumulus Clouds SE Pacific west of Chile) using “fast” 2D-cloud probe (25 microns resolution). Additionally, we have DSD data from collocated optical array probe (50 microns resolution) and 2D-video disdrometer from Greeley, CO and Huntsville, AL. We also have oceanic DSDs from the ODM 470 optical disdrometer on the R/V Investigator during its voyage south of Tasmania in 2018 along with the C-band dual pol radar during several GPM overpasses. These diverse DSD datasets will be used to demonstrate the stability of the scaled-normalized h(x) and the N_w-D_m co-variability for light rain. Specific scattering tables using these DSDs and the derived N_w-D_m relationships will be constructed and used in the combined algorithm in a research mode for evaluation relative to the current approach.

The figure shows the scatter plot log₁₀(N_w) [N_w in mm^-1 m^-3] vs D_m based on 4,650 1-s DSDs from “fast” 2D-cloud probe on C-130 in shallow stratocumulus (gold points with D_m< 0.5 mm. The other gold points are from 2,928 3-min DSDs from Greeley, CO and Huntsville, AL (D_m>0.4 mm). The blue circles are mean values from a large database of 1-min DSDs from Oklahoma. The black solid line is the smoothed spline fit based on the gold points only. It is remarkable that the Oklahoma mean values fit so well to the smoothed spline curve for D_m> 0.5 mm. Note also the “slower fall off" of log₁₀(N_w) between 1<D_m<2 mm is consistent between the Greeley/Huntsville and Oklahoma DSDs.