Abstract
This study was conducted to evaluate the prediction accuracies of THe Observing system Research and Predictability EXperiment (THORPEX) Interactive Grand Global Ensemble (TIGGE) data at six operational forecast centers using the root-mean square difference (RMSD) and Brier score (BS) from April to July 2012. And it was performed to test the precipitation predictability of ensemble prediction systems (EPS) on the onset of the summer rainy season, the day of withdrawal in spring drought over South Korea on 29 June 2012 with use of the ensemble mean precipitation, ensemble probability precipitation, 10-day lag ensemble forecasts (ensemble mean and probability precipitation), and effective drought index (EDI). The RMSD analysis of atmospheric variables (geopotential-height at 500 hPa, temperature at 850 hPa, sea-level pressure and specific humidity at 850 hPa) showed that the prediction accuracies of the EPS at the Meteorological Service of Canada (CMC) and China Meteorological Administration (CMA) were poor and those at the European Center for Medium-Range Weather Forecasts (ECMWF) and Korea Meteorological Administration (KMA) were good. Also, ECMWF and KMA showed better results than other EPSs for predicting precipitation in the BS distributions. It is also evaluated that the onset of the summer rainy season could be predicted using ensemble-mean precipitation from 4-day leading time at all forecast centers. In addition, the spatial distributions of predicted precipitation of the EPS at KMA and the Met Office of the United Kingdom (UKMO) were similar to those of observed precipitation; thus, the predictability showed good performance. The precipitation probability forecasts of EPS at CMA, the National Centers for Environmental Prediction (NCEP), and UKMO (ECMWF and KMA) at 1-day lead time produced over-forecasting (under-forecasting) in the reliability diagram. And all the ones at 2∼4-day lead time showed under-forecasting. Also, the precipitation on onset day of the summer rainy season could be predicted from a 4-day lead time to initial time by using the 10-day lag ensemble mean and probability forecasts. Additionally, the predictability for withdrawal day of spring drought to be ended due to precipitation on onset day of summer rainy season was evaluated using Effective Drought Index (EDI) to be calculated by ensemble mean precipitation forecasts and spreads at five EPSs.
Highlights
Uncertainty exists in weather predictions generated using numerical models
This study compared the ability of ensemble prediction system (EPS) from six operational forecasting centers (CMA, CMC, European Center for Medium-Range Weather Forecasts (ECMWF), National Centers for Environmental Prediction (NCEP), Korea Meteorological Administration (KMA), and UKMO) to predict the onset of the summer rainy season and withdrawal of spring drought over South Korea through use of the ensemble mean and probability precipitation, which was quantitatively observed on 29 June 2012
Except for the KMA, which was not available to use at Q850, a pattern of semidiurnal variation was observed in the root-mean square difference (RMSD) distribution for specific humidity, and the NCEP showed a large improvement in forecast accuracy as the forecast period was reduced (Fig. 4)
Summary
Uncertainty exists in weather predictions generated using numerical models. Some uncertainty is due to the sensitivity and dependence of the data on the initial conditions. After running a medium-range weather prediction using EPSs in the ECMWF and the NCEP, other operational forecast centers constructed EPSs using their own numerical models and different data assimilation methods; medium-range ensemble predictions are often implemented autonomously. Because the performance methods (e.g., initial condition field and method of initial perturbation) of EPSs used by the various Numerical Weather Prediction (NWP) centers are different from each other, many studies have focused on making comparisons and verifying data from the various EPSs, single-, and multi-models using deterministic and probabilistic verification indices (Matsueda and Tanaka 2008; Park et al 2008; Johnson and Swinbank 2009). This study compared the ability of EPSs from six operational forecasting centers (CMA, CMC, ECMWF, NCEP, KMA, and UKMO) to predict the onset of the summer rainy season and withdrawal of spring drought over South Korea through use of the ensemble mean and probability precipitation, which was quantitatively observed on 29 June 2012. The ensemble mean and spread of the Effective Drought Index at all centers were compared
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
