Multiscale feature analysis of forecast errors of 500 <scp>hPa</scp> geopotential height for the <scp>CMA‐GFS</scp> model

Using ERA5 reanalysis data from March 2021 to February 2022 and the China Meteorological Administration Global Forecasting System (CMA-GFS) operational forecast dataset of 500 hPa geopotential height in Northern Hemisphere same period, multiscale features errors are analyzed. The results indicate that anomaly correlation coefficient (ACC) its components keep decreasing with extension lead time, there no seasonal differences evolution ACC. effective skills by season for CMA-GFS model above 6 days at multiscale, highest winter planetary-scale components. In space, significant observed locations extreme values height, spatial distribution reflects inadequate prediction intensity large-scale trough ridge systems middle high latitudes phase-shift small troughs ridges latitudes. Generally, original field component show wavelike or banded distribution, synoptic-scale always distributed latitudinal patterns alternating between positive negative, without land, sea, terrain. first empirical orthogonal function modes almost retain their respective feature. temporal, spring, summer, autumn time series all have quasi-biweekly negative phase transitions within monthly scale, transition only occurs around January 1st. These deepen understanding possible causes provide ideas improvement revision model.