Evaluation Of Sub-seasonal Predictive Ability For The Extreme Heavy Snow Event In Early 2008 Based On S2S Models

Sub-seasonal prediction is an important part of seamless forecasting systems and a bridge between weather forecast and seasonal climate prediction.However,for the restrictions of data,models and methods,it is difficult to accurately predict the weather processes over sub-seasonal time scale,especially for extreme events.Therefore,taking the extreme heavy snow event in early 2008 as an example,8 models participating in the S2 S project are selected to evaluate the sub-seasonal predictive ability from the aspects of regional cumulative precipitation and anomalous atmospheric circulation leading to this extreme event.The effects of persistent anomalous atmospheric circulation and low-frequency signals such as the arctic oscillation(AO),tropical interseasonal oscillation(MJO),and the quasi-biweekly oscillation(QBWO)over the South China Sea-western North Pacific(SCS-WNP)on the ability to predict extreme events are analyzed.The persistent anomalous atmospheric circulation and low frequency signals are discussed as the predictable sources of the extreme event.In addition,we can understand the prediction level of CMA model among the other seven models by comparing the prediction skills for the extreme event.The main results are as follows:1.Evaluation of the models' ability to predict the extreme event on the sub-seasonal time scaleBased on the handcast results of S2 S models,the sub-seasonal prediction ability of 8 models for the extreme heavy snow event in early 2008 is compared and analyzed by using correlation coefficient and root-mean-square error.Most models can well predict the spatial distribution of cumulative precipitation over southern China 2 weeks in advance.ECMWF model has the best prediction ability,and it can predict the location and direction of rainbelts at the lead time of 3 weeks.The prediction skill of CMA model for regional cumulative precipitation is at the lower medium among the eight models,and precipitation intensity predicted is obviously weaker than observations.All the models predict the sudden increase of precipitation well,the prediction ability of the models declines rapidly with the longer lead time.The prediction ability of the models for atmospheric circulation is consistent with that for precipitation.The models can reasonably predict the atmospheric circulation within 2 weeks in advance,but the bias of atmospheric circulation predicted begins to increase at the lead time of 3 weeks.In general,the models have high prediction skill for 500 h Pa height field in the extreme snowstorm event,however,the prediction skill for the southwest airflow transports to southern China at 700 h Pa is significantly low.The difference of prediction ability for southwest airflow is the main reason for the difference of prediction ability of this event among the 8 models.2.The influence of persistent anomalous atmospheric circulation on the predictability of extreme heavy snow eventPersistent anomalous atmospheric circulation,such as 500 h Pa blocking high,low trough over central Asia and southwest airflow at low latitude,were the direct cause for the extreme event,which lasted for more than 20 days.The 8 models can reasonably predict the blocking high and low troughat at mid-high latitudes 3-4 weeks in advance,but for the southwest airflow,the prediction ability of the models is mostly less than two weeks in advance,among which ECMWF and UKMO models have better prediction ability for southwest airflow,and CMA and BOM models have weak prediction ability.In this event,there is an obvious concurrent variation between the subtropical jet and the polar jet.Most of the models have strong intensity prediction on the polar jet and weak intensity prediction on the subtropical jet.The two jets are closely related to the cold and warm air activities of this extreme event,especially the subtropical jet stream is closely related to the warm air transported to southern China.The difference of the model's prediction ability to the upper jets affects the prediction ability to the extreme heavy snow event.The strength and movement of this jet stream during the snowstorms can affect the eastward movement and intensity of the ridge and though over west Asia,thus affecting the atmospheric circulation over East Asia.UKMO and ECMWF models are more accurate than other models in predicting the intensity evolution of the Middle East jet stream during the extreme event.It can be seen that the persistent anomalous atmospheric circulation is one of the sources of the predictability of the successful prediction of this extreme event.3.The influence of low frequency signals on the predictability of extreme heavy snow eventAO,MJO and QBWO over SCS-WNP are three important low frequency signals affecting this extreme heavy snow event.The study has found that the AO is very active at the beginning of 2008.AO index changes from negative to positive on January 15 and remains strong positive phase until early February.The positive AO is conducive to the maintenance of blocking situation over the mid-high latitude,which provides a weak cold air condition for the snowstorms over south China.The prediction ability of the 8 models for AO is different.The prediction of ECCC,ECMWF,NCEP and UKMO are basically consistent with the observation,while those of other models are quite different.In the first two periods of the extreme event,the convective activity of MJO is mainly in phase 7,and transfers to the phase 2-3 in last two periods.When the MJO is in phase 2-3,the convective activity in the Indian Ocean is active and there is obvious eastward propagation signal in late January.Most models can predict this east propagation of MJO signals except for BOM and CMA.In addition,the QBWO over SCS-WNP begins to propagate northward from the equator,and arrives over south China in late January,which corresponds to the enhancement of the southwest airflow and the sudden increase of precipitation.ECMWF and UKMO models can predict the northward propagating signal.The difference in the prediction ability to the low frequency signals leads to the difference in the prediction ability to the atmospheric circulation,and finally affects the prediction ability of the models to the extreme event.Therefore,low frequency signals are also a source of predictability in accurately predicting this extreme heavy snow event.