Evaluation of the GPM-DPR snowfall detection capability: Comparison with CloudSat-CPR

The primary goal of this study is to assess the DPR's ability to identify snowfall using near-coincident CloudSat Cloud Profiling Radar (CPR) observations and products as an independent reference dataset. CloudSat near global coverage and high sensitivity of the W-band CPR make it very suitable for snowfall-related research. While DPR/CPR radar sensitivity disparities contribute substantially to snowfall detection differences, this study also analyzes other factors such as precipitation phase discriminators that produce snowfall identification discrepancies. Results show that even if the occurrence of snowfall events correctly detected by DPR products is quite small compared to CPR (around 5-7%), the fraction of snowfall mass is not negligible (29-34%). A direct comparison of CPR and DPR reflectivities illustrates that DPR misdetection is worsened by a noise-reducing DPR algorithm component that corrects the measured reflectivity. This procedure eliminates the receiver noise and side lobe clutter effects, but also removes radar signals related to snowfall events that are associated with relatively low reflectivity values. In an effort to increase DPR signal fidelity associated with snowfall, this paper proposes a simple algorithm using matched DPR Ku/Ka radar reflectivities producing an increase of the fraction of snowfall mass detected by DPR up to 59%.