This study uses Global Precipitation Measurement (GPM) Microwave Imager (GMI) and Ka-precipitation radar observations to quantify the snowfall detection performance for different channel (frequency) combinations. Results showed that the low-frequency-channel set contains limited snow detection information with a 0.34 probability of detection (POD). Much better performance is evident using the high-frequency channels (i.e., POD = 0.74). In addition, if only one high-frequency channel is allowed to be added to the low-frequency-channel set, adding the 183 ± 3 GHz channel presents the largest POD improvement (from 0.34 to 0.50). However, this does not imply that the water vapor is the key information for snowfall detection. Only using the high-frequency water vapor channels showed poor snowfall detection with POD at 0.13. Further analysis of all 8191 possible GMI channel combinations showed that the 166-GHz channels are indispensable for any channel combination with POD greater than 0.70. This suggests that the scattering signature, not the water vapor effect, is essential for snowfall detection. Data analysis and model simulation support this explanation. Finally, the GPM constellation radiometers are grouped into six categories based on the channel availability and their snowfall detection capability is estimated, using channels available on GMI. It is found that type-4 radiometer (all channels) has the best snowfall detection performance with a POD of 0.77. The POD values are only slightly smaller for the type-3 radiometer (high-frequency channels) and type-5 radiometer (all channels except 183 channels).

This study first quantifies the virga precipitation occurrence percentage using three spaceborne radar observations, including Tropical Rainfall Measuring Mission Precipitation Radar (PR), Global Precipitation Measurement dual‐frequency PR (Ku‐band PR/Ka‐band PR [KuPR/KaPR]), and CloudSat Cloud Profiling Radar (CPR). PR and KuPR/KaPR show that virga occurrence percentage is over 30% in arid regions (e.g., Sahara desert and deserts of Australia). CPR reveals similar virga geospatial distribution. However, the virga percentage based on CPR is about twice as large as that based on PR and KuPR/KaPR due to better detection sensitivity. Results also show that the majority of the virga clouds are altostratus and cirrus. Second, we investigate the virga precipitation contribution to the passive microwave radiometer false precipitation detection. Virga precipitation accounts for as much as 50% (30%) of Tropical Rainfall Measuring Mission Microwave Imager (Global Precipitation Measurement Microwave Imager) false detection results in arid regions. The underlying reason is because precipitation detection over land primarily relies on the ice scattering signature, while virga and light precipitation (e.g., 1 mm/hr) have similar amounts of ice water path in arid regions.

A prototype precipitation retrieval algorithm over land has been developed by utilizing 4 year National Mosaic and Multi‐Sensor Quantitative Precipitation Estimation and Special Sensor Microwave Imager/Sounder coincident data sets. One of the unique features of this algorithm is using the ancillary parameters (i.e., surface type, surface temperature, land elevation, and ice layer thickness) to stratify the single database into many smaller but more homogeneous databases, in which both the surface condition and precipitation vertical structure are similar. It is found that the probability of detection (POD) increases about 8% and 12% by using stratified databases for rainfall and snowfall detection, respectively. In addition, by considering the relative humidity at lower troposphere and the vertical velocity at 700 hPa in the precipitation detection process, the POD for snowfall detection is further increased by 20.4% from 56.0% to 76.4%. The better result is evident in both ends of the retrieved rain rate when the stratified databases are used, especially when the rain rate is greater than 30 mm/h. Similarly, the retrieved snowfall rate using stratified databases also outperforms that using single database. The correlation between retrieved and observed rain rates from stratified databases is 0.63, while it is 0.42 using the single database. The root‐mean‐square error is reduced by 50.3% from 2.07 to 0.98 by using stratified databases. The retrieved snow rates from stratified database are also better correlated with observations and possess smaller root‐mean‐square error. Additionally, the precipitation overestimation from the single database over the western United States is largely mitigated when the stratified databases are utilized. It is further demonstrated that over the majority of the stratified databases, the relationship between precipitation rate and brightness temperature is much closer to that from the corresponding category in the validation databases, rather than that from the single database. Therefore, overall superior performance using the stratified databases for both the precipitation detection and retrieval is achieved.

This study assesses the level-2 precipitation estimates from 10 radiometers relative to Global Precipitation Measurement (GPM) Ku-band precipitation radar (KuPR) in two parts. First, nine sensors—four imagers [Advanced Microwave Scanning Radiometer 2 (AMSR2) and three Special Sensor Microwave Imager/Sounders (SSMISs)] and five sounders [Advanced Technology Microwave Sounder (ATMS) and four Microwave Humidity Sounders (MHSs)]—are evaluated over the 65°S–65°N region. Over ocean, imagers outperform sounders, primarily due to the usage of low-frequency channels. Furthermore, AMSR2 is clearly superior to SSMISs, likely due to the finer footprint size. Over land all sensors perform similarly except the noticeably worse performance from ATMS and SSMIS-F17. Second, we include the Sondeur Atmospherique du Profil d’Humidite Intertropicale par Radiometrie (SAPHIR) into the evaluation process, contrasting it against other sensors in the SAPHIR latitudes (30°S–30°N). SAPHIR has a slightly worse detection capability than other sounders over ocean but comparable detection performance to MHSs over land. The intensity estimates from SAPHIR show a larger normalized root-mean-square-error over both land and ocean, likely because only 183.3-GHz channels are available. Currently, imagers are preferred to sounders when level-2 estimates are incorporated into level-3 products. Our results suggest a sensor-specific priority order. Over ocean, this study indicates a priority order of AMSR2, SSMISs, MHSs and ATMS, and SAPHIR. Over land, SSMIS-F17, ATMS and SAPHIR should be given a lower priority than the other sensors.

The first step to accurately measure global snowfall is to separate rainfall from snowfall correctly (i.e., precipitation phase discrimination). This study first evaluates the phase discrimination performance in four remote sensing datasets, including observations from ground radar, spaceborne radars, and spaceborne radiometer, relative to ground observations. Results show that the snowfall discrimination accuracy varies greatly among these datasets ranging from 42% to 96%, dependent on whether and how the temperature information are considered. For example, over half of the snowfall from the Global Precipitation Measurement Mission (GPM) spaceborne radar is actually rainfall at the surface since it detects snowfall in the air without considering the temperature information close to the surface. Second, we evaluate the discrimination performance using the temperature information from four reanalysis datasets. It is found that MERRA2 temperature close to the surface is colder than the other three datasets, leading to more rainfall being misclassified as snowfall.

This study investigates the correlation between the upwelling microwave brightness temperature measured by satellite radiometer and surface precipitation rate from ground radar observations at different time lags. Results show that brightness temperatures correlate more strongly with the lagged surface precipitation rate than the simultaneous surface precipitation rate. The lag time for snowfall ranges from 30 to 60 min. This time lag effect has an important influence when evaluating the satellite retrieval results relative to ground observations. For example, the falsely identified pixels can decrease by as much as 23.88% when considering a 30‐min lag time. Furthermore, the satellite‐retrieved snowfall rate shows much stronger correlation with the time‐lagged surface snowfall rate than the simultaneous snowfall rate in cold environments and for tall storms. This work implies that the time of the level‐2 swath‐retrieved snowfall rate needs to shift forward when incorporated into the level‐3 gridded products.