전지구예측모델 개발을 위한 검증분석 도구: Korean Integrated Model Analysis/Evaluation Tool (KAT)

Abstract

Forecast verification is a critical component for advancing weather research and model development. At the Korea Institute of Atmospheric Prediction Systems (KIAPS), a verification and evaluation package has been developed to support the enhancement of global numerical weather prediction models. The Korean Integrated Model (KIM) Analysis/Evaluation Tool (KAT), is designed to conduct statistical verification for medium-range forecasts and provide graphical displays that help identify forecast errors, thus offering valuable feedback to model developers. To meet the specific needs of forecasters, KAT includes evaluation tools tailored for high-impact weather events such as heavy rains, typhoons, and heatwaves, allowing users to assess the strengths and weaknesses of forecasting systems. Since its initial launch, KAT has undergone various refinements, prioritizing optimization of computational efficiency through parallel processing for major statistical analyses. Furthermore, KAT evaluates model stability and climatological representation in long-term simulations, which are critical for improving numerical prediction models. This technical note aims to outline the core structure of the KAT verification methodology, elucidate the scientific principles and perspectives that inform its design, and provide practical details on verification metrics, testbed setups, observations, and relevant references for users interested in similar objectives.