Abstract:Real-time seasonal climate prediction was performed in China during the extremely strong El Niño event of 2015/2016,through a combination of dynamical and statistical climate prediction.Generally,real-time summer(winter) climate prediction in China starts in February(October) in every year.The results showed that,although the NCEP-CFSv2 coupled model predicted the evolution of the extremely strong El Niño event in 2015/2016 well,its performance in predicting the summer rainfall anomaly of 2015 and the winter temperature of 2016 at 160 stations in China was limited.Compared to observation,CFSv2 predicted a stronger East Asian summer monsoon and weaker East Asian winter monsoon.One of the reasons for this is that CFSv2 is poor at predicting the extratropical climate system.Thus,based on the climate prediction direct outputs of the NCEP-CFSv2 model,we created a hybrid dynamical and statistical prediction model for forecasting the precipitation anomaly and temperature anomaly at 160 stations in China in 2015/2016.The skill of the hybrid of statistical and dynamical prediction model was higher than that of the direct prediction results of the NCEP-CFSv2 model.The spatial anomaly correlation coefficient(ACC) of summer rainfall at 160 stations in China in 2015 increased from 0.21 to 0.31(exceeding the 99% significance level),along with the percentage of the same sign of the rainfall anomaly improving to 60% from 50%.The model reproducedthe observed flood pattern in southern China,as well as the drought pattern in summer 2015.Meanwhile,the prediction ACC of winter temperature in China in 2016 increased to 0.32 from 0.19,and the percentage of the same sign of the temperature anomaly increased to 75% from 62%.Moreover,the year-to-year increment prediction method proposed by Fan et al.(2007) was applied successfully to predict summer rainfall over the Yangtze River valley in 2015,and winter temperature over North China in 2016.The year-to-year increment method predicts the year-to-year increment of the climate variable instead of the climate anomaly,in which the year-to-year increment of the climate is defined as the climate variable of the current year minus that of the previous year.The year-to-year increment of the climate variable was firstly predicted by the statistical or dynamical model,and then the predicted climate anomaly or climate variable of the current year could be obtained by adding the predicted year-to-year increment to the observed one of the previous year.The advantage of the year-to-year increment is that it can amplify the prediction signal,especially the extra tropical climate signal.Furthermore,as the observed climate in the previous year is an accurate value containing the interannual and interdecadal signals,it further promotes the level of accuracy in predicting the interannual and interdecadal climate variable.The results showed that the summer rainfall anomaly over the middle and lower reaches of the Yangtze River valley in 2015 could be predicted successfully by the year-to-year increment;the predicted(observed) value in 2015 was 38.6%(31%).Meanwhile,the upward trend of the summer rainfall anomaly over the middle and lower reaches of the Yangtze River valley since the 1980s,and the downward trend since 2000,were also reproduced.The model reproduced the warming trend since the 1970s,and the slowly cooling trend since 1998,with the predicted(observed) winter temperature anomaly over North China being 1.20℃(0.51℃).However,there is still a long way to go in terms of improving the prediction skill level in China to a sufficiently high level.The extratropical climate prediction skill should be improved by improvement to thedynamical model.It is necessary to explore how to combine the dynamical climate model with the statistical climate model more effectively.Importantly,climate theory,methods and techniques,models,as well as climate dynamics suited for climate variability in China,should be further developed.