Comparison of Classification Methods for Winter Surface Precipitation Phase over the Korean Peninsula Using a KMA Operational NWP Model

초록

The performance of classification methods for surface precipitation phase was compared across different winter precipitation types over the Korean Peninsula using the reanalysis and forecast data of the Korean operational numerical weather prediction model, KIM (Korean Integrated Model), operated by the Korea Meteorological Administration. Four classification methods—based on wet-bulb temperature, 1,000-500 hPa thickness, 1,000-700 hPa thickness, and the revised Matsuo scheme—were applied. Winter precipitation cases over the Korean Peninsula from 2020 to 2022 were selected based on three types: Bohai Bay Trough (BBT), Warm ADvection (WAD), and Mid-latitude Low-Pressure (MLP) system. The comparison of observed precipitation phases and classification using the KIM analysis field showed that the wet-bulb temperature-based method exhibited higher hit rates across all precipitation types. The 1,000- 700 hPa thickness and revised Matsuo methods tended to classify several precipitation phases as mixed ones, which resulted in not only deteriorating hit rates for both snow and rain but also a decrease in false alarm rates. The 1000-500 hPa thickness-based method overestimated snow by misclassifying rain as snow. Meanwhile, the wet-bulb temperature-based method consistently exhibited the highest performance in both 12-hour and 24-hour forecast fields for BBT and WAD, compared to other methods. However, compared to the analysis field, the forecast field showed lower wet-bulb temperatures as the forecast time increased, which led to an increased tendency to misclassify rain as snow using the wet-bulb temperature-based method. Other methods showed similar classification performance in forecast fields compared to the analysis fields.

키워드