Satellite Remote Sensing of Snowfall

There has been so far no global climatology of snowfall produced on the basis of observational data, although snowfall is important for monitoring hazardous weather and understanding hydrological cycle. Surface-based gauge measurements are spotty and only exist over land areas. To measure snowfall globally, observations from satellite-borne instruments such as CloudSat radar and high frequency passive microwave radiometers become inevitable. In this presentation, I will report our research results on satellite snowfall retrievals and point out some of the obstacles. It consists of the following topics. (1) Separation of rain and snow. Before performing retrieval, we need to know whether it is a raining or a snowing condition. Since rain and snow have very different radar/radiometric signatures, a mis-identification can lead to an order of magnitude difference in precipitation rate. Satellite observations themselves are often unable to make the distinction; ancillary data are commonly used. (2) Active sensing. Cloud radar from CloudSat (and future EarthCare) provides the best opportunity so far for measuring global snowfall because of its excellent minimum detection. Preliminary global snowfall maps have been developed based on CloudSat data. Remaining issues include the radar reflectivity's high dependence on particles' size distribution, shape and orientation, attenuation, and intrinsic saturation to high snowfall rates. (3) Passive sensing. High-frequency microwave radiometers provide the most abundant satellite observations so far for measuring snowfall. Radar-trained radiometer algorithms have had some success in detecting and retrieving snowfall. Since radiometric signature is the integration over an entire vertical column, issues such as surface contamination and masking by supercooled liquid water are obstacles in snowfall retrievals. A good strategy may be to perform separated retrievals for different weather types and/or surface conditions.