92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Thursday, 26 January 2012
Verification of the T213L31 Ensemble Prediction System of CMA Over the East Asia Area
Hall E (New Orleans Convention Center )
Huiling Yuan, Nanjing University, and NOAA/ESRL, Nanjing, Jiangsu, China; and J. Song, Y. Wang, S. Chen, and J. Ming

This paper investigates the performance of the ensemble prediction system (T213L31 EPS) of China Meteorological Administration (CMA) in the year of 2008 over the East Asia area, using the analysis and 1- to 10-day forecast data from the Observing System Research and Predictability Experiment (THORPEX) Interactive Grand Global Ensemble (TIGGE). Two synoptic variables including 500 hPa geopotential height (Z500) and 850 hPa temperature (T850) have been evaluated. The ensemble-mean forecast of the T213L31 EPS is more accurate and skillful than its corresponding control forecast, which proves the value of the TIGGE forecasts. As the lead time increases, the superiority of ensemble-mean forecast is more significant. The probabilistic prediction of Z500 is more skillful than T850. Most importantly, the ensemble spread of Z500 is overestimated (overdispersion) compared to the error of the ensemble-mean forecast at the lead time of 1-2 day, and then becomes underdispersive for longer lead times. However, underdispersion is prevalent in the T850 ensemble forecast at all lead times. Overdispersion and underdispersion in the T213L31 EPS indicate the large and small perturbations adding to the initial analysis, respectively. The spread-error relationship between the ensemble-mean forecast error and ensemble spread provides meaningful guidance, which may help improve the quality of the T213L31 EPS in the future. In addition, selected high-impact weather events in East Asia are analyzed in detail, including the synoptic weather conditions, rainfall distribution, moisture fields, forecast skill, and so on.

Key words: TIGGE, the T213L31 ensemble prediction system, probabilistic prediction, ensemble spread

Supplementary URL: