Potential Predictability of Summer Mean Precipitation in a Dynamical Seasonal Prediction System with Systematic Error Correction.

Potential predictability of summer mean precipitation over the globe is investigated using data obtained from seasonal prediction experiments for 21 yr from 1979 to 1999 using the Korea Meteorological Administration–Seoul National University (KMA–SNU) seasonal prediction system. This experiment is a part of the Climate Variability and Predictability Program (CLIVAR) Seasonal Model Intercomparison Project II (SMIP II). The observed SSTs are used for the external boundary condition of the model integration; thus, the present study assesses the upper limit of predictability of the seasonal prediction system. The analysis shows that the tropical precipitation is largely controlled by the given SST condition and is thus predictable, particularly in the ENSO region. But the extratropical precipitation is less predictable due to the large contribution of the internal atmospheric processes to the seasonal mean. The systematic error of the ensemble mean prediction is particularly large in the subtropical western Pacific, where the air–sea interaction is active and thus the two-tier approach of the present prediction experiment is not appropriate for correct predictions in the region. The statistical postprocessing method based on singular value decomposition corrects a large part of the systematic errors over the globe. In particular, about two-thirds of the total errors in the western Pacific are corrected by the postprocessing method. As a result, the potential predictability of the summer-mean precipitation is greatly enhanced over most of the globe by the statistical correction method; the 21-yr-averaged pattern-correlation value between the predictions and their observed counterparts is changed from 0.31 before the correction to 0.48 after the correction for the global domain and from 0.04 before the correction to 0.26 after the correction for the Asian monsoon and the western Pacific region. [ABSTRACT FROM AUTHOR]