Winter Precipitation Classification from Two Dimensional Video Disdrometer Data: Observations from Gcpex and Lpvex Experiments

The liquid-equivalent snow rate (SR) quantitative precipitation estimation (QPE) using radar (both ground-base and satellite) is very important component of GPM research. Winter storms can have with different type of precipitations such as, pristine ice crystal, sleet, graupel-like snow, rimed and aggregated snow, and each type of precipitations have very different micro-physical properties (e.g. size, density, fall speed, dielectric constant), to estimate SR accurately using radar becomes a challenge. The first step to get a proper estimator from radar observations is the precipitation classification. The form of different winter precipitation relates to the environment especially the temperature and humidity. In addition, the different precipitation type also correspond to different shape and density resulting in different fall speeds. Recent research shows that 2DVD is able to observe winter precipitation successfully. The 2DVD can measure the fall speed of each particle falling into its observing area, and provide the contours from two orthogonal views. It is straight forward to estimate the diameter of snowflake from two views of 2DVD and compute the particle size distribution (PSD). In order to classify the different type of precipitation, we must not only estimate the diameter but also define several geometric features. In this paper we explore a technique to estimate features from the contours of 2D-Video images. In this paper, we analyze observations from two GPM winter campaigns namely, GCPEX and LPVEX. From these data sets , we derive the to get the statistics of fall speed and geometric features developed in this paper. Subsequently these are used to classify winter precipitation into four catalogs, namely rain, sleet, graupel-like snow and fluffy snow (may form by aggregation or aggregation with riming). We apply this procedure to two GCPEx cases which mixed with different type of precipitation in different time interval. Since we are using the contouring data and fall speed, this procedure may also apply to other image instruments which are capable to measure the fall speed such as the hydrometeor velocity and shape detector (HVSD) or Multi-Angle Snowflake Camera (MASC). Subsequently these classifications are used in the assumptions to simulate radar observations.