资源环境科技发展态势分析平台

Subseasonal-to-Seasonal (S2S) forecasts of weather and climate extremes are being increasingly demanded by water resource managers, operational forecasters, and other users in the applications community. This study uses hindcast data from the European Centre for Medium-Range Weather Forecasts (ECMWF) S2S forecast system to evaluate global subseasonal prediction skill of atmospheric rivers (ARs), which are intense lower tropospheric plumes of moisture transport that often project strongly onto extreme precipitation. An aggregate quantity is introduced to assess AR subseasonal prediction skill, defined as the number of AR days occurring over a week-long period (AR1wk occurrence). The observed pattern of seasonal mean AR1wk occurrence strongly resembles the general pattern of daily AR frequency. The ECMWF S2S forecast system generally shows positive (negative) biases relative to reanalysis in the mid-latitude regions in summer (winter) of up to 0.5-1.0 AR days in AR1wk occurrence in regions of highest AR activity. ECMWF AR1wk occurrence forecast skill outperforms a reference forecast based on monthly climatology of AR1wk occurrence at week-3 (14-20days) lead over a number of subtropical to midlatitude regions, with slightly better skill evident in wintertime. The magnitude and subseasonal forecast skill of AR1wk occurrence are shown to vary interannually, and both quantities are modulated during certain phases of the El Nino-Southern Oscillation, Arctic Oscillation, Pacific-North America teleconnection pattern, and Madden-Julian Oscillation.