12.3 A Multi-algorithm Reanalysis-based Freezing Precipitation Dataset for Climate Studies in the South Central U.S

Thursday, 14 January 2016: 4:00 PM
Room 245 ( New Orleans Ernest N. Morial Convention Center)
Esther Mullens, South Central Climate Science Center, Norman, OK; and R. A. McPherson

Freezing precipitation, including freezing rain, drizzle, and ice pellets, presents substantial hazards to transportation, energy, and infrastructure. The southern United States in particular is a region where winter storms produce multiple phases of precipitation, and in some cases, damaging ice storms. Most quantitative climatological length (> 30 years) datasets for freezing precipitation are obtained from in-situ instrumentation, such as National Weather Service (NWS) automated observing system stations (ĎASOS'), which can be spatially inhomogeneous and of coarse resolution. This work investigates whether the 32 km North American Regional Reanalysis (NARR) is viable for developing a more comprehensive and spatially gridded dataset for freezing precipitation, and its associated meteorological environment, with a focus on the south central United States. NARR already includes categorical precipitation type as a variable. Nonetheless, to permit a translatable method across other gridded multi-dimensional reanalyzes, and eventually downscaled climate model projections, and to overcome biases from individual algorithm assumptions, a multi-algorithm approach (Bourgouin, Baldwin, and Partial Thickness) is used to extract environmental conditions, counts, and liquid water equivalent (LWE) for freezing precipitation. Due to the microphysical complexity, we do not attempt to individually resolve for ice pellets and freezing rain. The resulting dataset is evaluated spatially and temporally against hourly and daily counts and LWE from 13 first-order NWS stations over regions of Texas, Oklahoma, Missouri, Arkansas, and Louisiana. Additional validation datasets include National Climatic Data Center Storm Data, and MPING observations. The derived dataset shows very good statistical agreement with many of the station sites, and is able to reproduce years with heavy ice events, however it is typically unable to resolve freezing drizzle conditions. Peak magnitudes of LWE tend to be underestimated for portions of the western and central domain, and overestimated to the south. As an additional check, we develop a comparable dataset from the NARR Eta model categorical freezing and rain and ice pellet data, and contrast its performance at each station site against our multi-algorithm approach. We show that this value-added product could be a useful tool for climatological research, vulnerability and hazard analyses, e.g., in the transportation and energy sectors. The benefits and caveats of these datasets will be discussed, and we will also briefly demonstrate the individual performance of each algorithm.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner