Large-scale Circulation Features Of The Cluster Extreme High Temperature Events And Model Predictive Ability At Subseasonal Time Scale

Under global warming,the high temperature extremes have significantly increased.Previous studies have found that heat wave events are usually associated with largescale circulation anomalies.However,whether the circulation characteristics during the cluster high temperature events in different regions are the same remains to be studied.Our paper aims at revealing the large-scale circulation features during the cluster high temperature events in South China and North China.Moreover,in order to learn about the sub-seasonal predictive skills of BCC_CSM for the cluster high temperature events and its specific level in the worldwide,the difference of forecast skills of BCC_CSM for the cluster high temperature events in different regions and the forecast skills of other ten models in the S2 S have been investigated in this study.The main results are as follows:Based on the surface air temperature data from 1981 to 2017 at 2479 stations and the identification method of cluster high temperature events,a total of 38 cluster high temperature events are identified,with 18 events in South China and 20 events in North China.During the cluster high temperature events in South China and North China,we can learn that there are both positive geopotential height anomalies in the upper troposphere and anomalous anticyclones in the lower troposphere.The whole layer of the troposphere is equipped with equivalent barotropic structure.However,during the cluster high temperature events in different regions,the circulation characteristics are of great difference.The extreme high temperature in North China is mainly affected by the ridge in the mid-latitudes accompanied with the center of the South Asian high(SAH)in the Iranian plateau.While the cluster high temperature events in South China are jointly affected by the westward shift of the west Pacific subtropical high(WPSH)and the eastward shift of the SAH.Besides,the location and intensity of the subtropical jet are also different.During the cluster high temperature events in North China,the center of subtropical jet is to the west of the climatological mean and its intensity is relatively weak.While the location of the subtropical jet is close to the climatological mean during the cluster high temperature events in the South China.Moreover,the lowfrequency wave trains are also different.In the lower troposphere,there are inverse phase wave trains over East China during the cluster high temperature events in different regions.In the upper troposphere,the extreme high temperature over North China is also affected by two wave trains emanating downstream from the Atlantic region,one propagating eastwards along high latitude and the other along the subtropical jet stream.Both of them converge over North China,leading to the enhancement of the geopotential height anomalies.But the low-frequency wave train during the cluster high temperature events over South China is not so significant according to the wave activity flux.The predictive skills of BCC_CSM for the cluster high temperature events in different regions are investigated using the observed station data,reanalysis data and the hindcast data of BCC_CSM.For the cluster high temperature events in North China,the model is able to reproduce the spatial distribution characteristics of temperature anomalies and corresponding circulation patterns ahead of 10 days when the performance of the model is good.But the intensity of positive temperature anomalies and geopotential height anomalies predicted by the model weakens and the center migrates eastwards compared to the observation.The forecast skill is limited at 1 week when the performance is relatively poor.When leading 1 week,the model can reproduce the positive temperature anomalies and geopotential height anomalies during the cluster high temperature events.However,for the cluster high temperature events in South China,the forecast skills for air temperature are limited at 1 week.The location and intensity of the temperature positive anomalies are greatly deviated from the observation after leading 1 week.In order to explore the reasons for the decrease of skillful prediction at different lead times,the vorticity advection and the temperature advection variation with height are analyzed using the geopotential height tendency equation.We have studied that the relative vorticity advection contributes more to the circulation features of cluster high temperature events in North China.In the cases with higher forecast skills in North China,the relative vorticity advection ahead of 10 days predicted by BCC_CSM is consistent with that using the reanalysis data,which is able to predict the positive geopotential height anomalies.But in the cases with lower skills,the relative vorticity advection is contrary to that using the reanalysis data,which cannot reasonably reflect the circulation features.The absolute vorticity advection plays a major role in predicting the location of WPSH during the cluster high temperature events in South China.When leading 5 days,the negative absolute vorticity advection is much weaker than that using the reanalysis data,leading to the limited skills in predicting the westward shift of the WPSH.The forecast skills of other ten models in the S2 S are specifically investigated in the following part.When leading 2 weeks,the ECMWF shows higher predictive skills for the cluster high temperature events in North China than other models.And for the cluster high temperature events in South China,the forecast skills of ECMWF and NCEP are higher than that of other models.The predictive skills of the Bo M model for the cluster high temperature events in South and North China are much lower than other models.For the BCC_CSM model,the forecast skills for the cluster high temperature events in North China are above average among the other ten models.However,the forecast skills of BCC_CSM for the cluster high temperature events in South China are relatively weak,with the anomaly correlation coefficient(ACC)less than 0.2 when leading 2 weeks.The causes of the difference in forecast skills of ECMWF and BCC_CSM model are further investigated.Compared to the BCC_CSM model,the ECMWF model can reasonably reproduce the spatial characteristics of the positive temperature anomalies and the geopotential height anomalies ahead of 10 days when the performance is good.Particularly,the ECMWF can well reproduce the westward shift of the WPSH ahead of 10 days,which shows much higher skills than that of the BCC_CSM.However,similarly to the BCC_CSM model,the ECMWF model cannot reproduce the spatial distribution features of the positive temperature anomalies and geopotential height anomalies after leading 1 week when the performance is relatively poor.The geopotential height tendency equation is analyzed to explore the reasons for the decrease of skillful prediction at different lead times.In the cases with higher forecast skills,the vorticity advection predicted by the ECMWF model during the cluster high temperature events in South and North China is consistent with that using the reanalysis data,which is able to reproduce the circulation features.In the cases with lower forecast skills,the vorticity advection predicted by the ECMWF is contrary to that using the reanalysis data,which cannot reproduce the circulation patterns during the cluster high temperature events.