Attribution And Projection Study Of Changes In Summer Precipitation In The Yangtze-huai River Valley Based On Associated Large-scale Circulation Pattern

The Yangtze-Huai River Valley(YHRV)is one of the regions where heavy precipitation frequently occurs,and the heavy precipitation is prone to result in more disastrous consequence for human life and economy.For precipitation change in this area,attribution study on the responses of dynamic and thermodynamic factors and projection study under stabilized warming scenarios are both all hot issues in the current climate change research field.After a concise analysis of summer precipitation change in YHRV,the large-scale circulation patterns(CP)influencing summer precipitation in the YHRV are classified into 20 classes using a 5×4 competition layer through a new-emerging data mining tool-self-organizing map.The relationship between changes in the characteristics of these CPs and summer precipitation are analyzed and the contributions of dynamic and thermodynamic factors to precipitation changes are quantitatively assessed.After evaluating the performance of models in simulating the characteristics of the CPs,two sets of experiment driven with different forcings from Met Office Hadley Centre system-HadGEM3 are used to explore anthropogenic influence on the way that above-mentioned large-scale CPs response.Additionally,by employing the simulations under stabilized and transient 1.5? and 2? warming scenarios from CESM,responses of dynamic and thermodynamic factors and total precipitation changes under different warming scenarios are comparatively analyzed.The main conclusions are summarized as follows:The CPs successfully capture the activities of key synoptic systems such as west pacific subtropical high(WPSH),blocking high and mid-latitude trough in summer.Typical CPs mainly include EAP teleconnection type,single blocking high pattern,single low trough pattern,tropical cyclone influencing pattern,northernly-located or weak WPSH influencing pattern.The distribution and intensity of precipitation in the YHRV varies between CPs under which key synoptic systems show different location and strength.Overall,the intensity of precipitation under EAP pattern and the single blocking high pattern is relatively strong,while the precipitation under northernly-located or weak WPSH and the tropical cyclone influencing pattern are much less than the climatology.According to the intensity of precipitation under each CP relative to climatology,CPs are classified into dry,wet and neutral types.There is a significant increase in the frequency of wet type and neutral type,while the dry CPs show a significant decrease.The correlation coefficient between the annual total frequency of wet and neutral types and the area-averaged precipitation series in summer over YHRV reaches 0.49.Moreover,changes in the frequency of most CPs and summer precipitation in YHRV are of the same sign,which indicates a positive contribution of dynamic factor.The annual mean and maximum duration of the wet type show a significant increasing trend and the wet type tend to appear in the form of longer durations,whereas it is the opposite for dry type.The more frequent lengthened wet type and less shortened dry type provide more favorable conditions for the occurrence and maintenance of precipitation,which is the main reason for the increasing summer precipitation in the YHRV.Compared to ERA reanalysis,HadGEM3-A and CESM exhibit much larger magnitude in the differences between the frequency of the CPs.For any of the dry,wet or neutral type,the models show both an underestimation and overestimation for frequency of CPs.Both models tend to provide a longer annual mean and maximum duration,especially for CESM.The correlation coefficients of the transition probability distributions between the two models and the ERA reanalysis both reach higher than 0.9,which indicate the models show high skills for the possible forms and probabilities of the transition between CPs.For the distributions of the characteristics of the CPs over 1961-2013,the ensemble means of the models show an underestimation of the range without outliers and the interquartile range,but ensemble members show a much better skill.In addition,except a few CPs,the frequencies obtained from the two models and the ERA reanalysis all belong to the same distribution,and the spread of the probability density functions(PDF)of the ensemble members can encompass the PDFs from ERA reanalysis.Although CESM show poor performance in the interannual variability of the frequency of CPs and ensemble spread,CESM and HadGEM3-A both show very high skill for the direction and amplitude of the change between the two periods before and after 1980 and the trend and PDFs of the total frequency of the selected dry and wet CPs.Therefore,for the attribution and projection which emphasize the trends and long-term changes in direction and magnitude,HadGEM3-A and CESM are of high reliability.For the two periods before and after 1980,signs of precipitation change caused by dynamic factors are not only more consistent with the changes in total precipitation,but also show significantly larger magnitude than precipitation change attributable to thermodynamic factor and the combined dynamic and thermodynamic factor.In total,the contribution of dynamic change reaches 50.26%.No matter for 1980-2013 or 1994-2013,the PDFs of the mean absolute error for the characteristics of the CPs under ALL forcing show statistically significant(p<0.05)shifts toward lower MAE scores than that under NAT forcing,especially in 1994-2013.Furthermore,for the trend of the frequency of the selected dry and wet CPs and the shift of the PDFs between 1994-2013 and 1961-1980,the ALL forcing shows consistent sign with the ERA reanalysis,while NAT forcing is not.And in 1994-2013,the PDFs of the frequencies of the selected dry and wet CPs under the ALL forcing significantly shift toward the less and more directions respectively than that under NAT forcing.In 1994-2013,if the mean frequency in ERA reanalysis is used as the threshold,the return period of the selected dry and wet CPs at this threshold would be respectively prolonged and shortened,with a risk ratio of 0.34([0.29 0.39])and 1.29([1.19 I.37]).The different responses of the frequency change of CPs to the two forcings prove that human influences have changed the way that CPs occur which is associated with summer precipitation in YHRV,resulting in longer and more wet CPs but shorter and less dry CPs.This lead to the increasing trend of summer precipitation in the YHRV.In the stabilized warming scenario,the responses of dynamic and thermodynamic factors to warming are both stepwise.In the transient warming scenario,the dynamic factor was significantly adjusted when global warming reaches 1.5?,but it shows no obvious response to the additional 0.5? rise.The average precipitation changes influenced by the thermal factors show both increase and decrease at 1.5? warming,but they all increase significantly at 2? warming.Under all scenarios,the changes of dynamic factors exhibit decreasing trend of dry type but increasing trend of the other two types,which facilitates the increase of summer precipitation in the YHRV.Under the influence of the thermodynamic factors,the mean precipitation under the neutral and wet type is obviously more under the stabilized warming scenario than that under the transient scenario,while it is opposite for the dry type.Under both stabilized and transient warming scenarios,the summer mean precipitation in the YHRV will increase stepwisely,and the magnitude of the increase under transient scenario is slightly smaller.Additionally,the increase of precipitation from the current climate to a warming of 1.5? in the stabilized scenario is greater,while in the transient scenario the change from 1.5? to 2? warming is greater.For regional characteristics of precipitation changes,the increase of the mean summer precipitation and the maximum one-day precipitation in the YHRV under both warming scenarios is larger along the Yangtze River and in the south of the Yangtze River.The increase in the northwestern part is smaller,where precipitation in the transient 1.5? scenario is even reduced.The transient scenario tends to underestimate the increase in intensity and risk of heavy precipitation in the YHRV relative to the stabilized 1.5? and 2? scenarios but tends to overestimate the increase in intensity and risk resulting from an additional 0.5?warming.In addition,the increase in the mean summer precipitation and maximum one-day precipitation caused by the additional 0.5? warming is mainly concentrated in the south of the Yangtze River in the stabilized scenario,whereas in the transient scenario it is mainly concentrated in the southeast coasts and northwest.