2B.5 Comparison of Ground-Based Observations for Detecting Falling Snow

Monday, 8 January 2018: 11:30 AM
Room 18B (ACC) (Austin, Texas)
Bailey Swick, St. Cloud State Univ., Thief River Falls, MN; and A. Tokay, D. A. Marks, J. L. Pippitt, D. B. Wolff, and W. Petersen

One of the four level one requirements of the National Aeronautics and Space Administration Global Precipitation Measurement (GPM) program is to detect falling snow at an effective resolution of five km of the dual frequency precipitation radar (DPR) and 15 km of the GPM microwave imager. Two operational weather radar products, the Hydrometeor Identification (HID) and Multi Radar Multi Sensor (MRMS), are employed as a validation tool for GPM DPR precipitation type algorithms over the continental United States. The former product classifies the precipitation type at ten different classes for each radar pixel at a given range and azimuth. The latter product outputs percent snow at 12.5% increments at 0.01x0.01 degree pixel. MRMS also contains the radar quality index (RQI) ranging between 0 and 1. RQI has two components: beam blockage and beam spreading and ascent. For the purpose of this study, an RQI of 0.9 or higher was considered as good data. The temporal resolution of the HID was dependent on the volume coverage pattern the radar was operating in, but typically ranged between five and ten minutes. MRMS had a temporal resolution of 30 minutes.

This study uses the Automated Surface Observing System (ASOS) present weather sensor to evaluate the performance of HID and level-III MRMS derived products. ASOS reports present weather, surface temperature, and several other observables at one-minute resolution. The database consisted of 11 ASOS stations covered by ten radar sites from the winter of 2014-2015. A total of 143 days were identified as snow, cold rain, or mixed precipitation event. Qualitative and quantitative assessments were performed taking into account the freezing level and the radar sampling height. For the qualitative assessment, event based diagrams were made to compare each product at its native resolution. For the quantitative assessment, comparisons were made at two resolutions. First, the ASOS observations were averaged to match the irregular resolution of HID and compared. Second, both ASOS observations and HID were averaged to 30-minute resolution and compared with MRMS data. To determine the precipitation type and HID classification for multiple observations over time, a weighted sorting method is used: The original classification was maintained for ASOS. The HID was generalized with category of snow, rain, mixed, other, or none. A classification of mixed occurred when HID was mostly wet snow, or there was an equal number of classifications of rain and snow.

For the 30-minute comparisons, the sample size was 6,128. Of these observations, 3,871 (63%) had an RQI above 0.9. Above the threshold, HID classified as snow 88% and mixed 5% of the time when snow was observed at the ground. HID classified as rain 33.6% and mixed 35% of the time when rain was observed at the ground. MRMS, on the other hand, reported 100% of snow when only 41% of the time snow was observed at the ground. When MRMS reported no snow, the snow was actually at the ground 50% of the time. This is an ongoing study. The study will also expand including the sensitivity of the HID and MRMS precipitation algorithms to the station location, radar site, beam height, event type, and freezing level.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner