Evaluation Of CMIP5 Models’ Capacity In Simulating The Eurasian Blocking Highs And Projection Of Its Change Trend

Since the middle of the 20th century, the intensity and frequency of extreme weather and climate events have been obviously changed. Related research shown that the occurrence of disastrous climate events are often closely associated with sustainable blocking events, therefore, how to accurately estimate the change of blocking is worthy of studying. Firstly, based on the NCEP/NCAR reanalysis data, the variation characteristics of different intensity and duration blocking over Northern Hemisphere are analyzed in detail. The frequency of blocking events from 1956 to 2005 over three key blocking regions over Eurasia have been analyzed, the characteristics of height and circulation field during the occurrence of long duration blocking events have been also analyzed. Based on this, the capability of CMIP5 (phase 5 of the Coupled Model Inter-comparison Project) models in simulating the spatial and temporal distribution characteristics of blocking, the frequency of different blocking events and the variation trend of blocking events over Eurasia have been assessed. Furthermore, based on the different emission scenarios, the possible change of Eurasian blocking has been estimated from 2016 to 2065. The main results are as follows:(1) The results present that the blocking frequency of the Atlantic region is the highest in three frequent region over Northern Hemisphere, while that of Eurasia region is the lowest. The frequency of strong intensity blocking all show significant decreasing trend in the three region, especially for the Atlantic region. It is found that the frequency of the long duration blocking obviously increased since the 21st century. The Pacific short duration blocking showed a hint of increasing frequency, but significant decreasing in Atlantic region.(2) In the three key blocking region over Eurasia, the blocking events in the Ural region is highest, and wintertime is slightly more than summertime; the blocking events frequency of the Okhotsk region is less than that in the Ural region, and the frequency is more distinct in wintertime compared with that in summertime. the blocking events in the Baikal region is lowest over the three region, while the frequency in summertime is more than that in wintertime. From the perspective of the long duration blocking frequency, the Okhotsk region is obviously greater than that in the Ural and Baikal region. When the occurrence of long duration blocking in the typical month of wintertime(February), the reanalysis data showed that the distribution characteristics of 500hPa height anomaly field are "three positive-three negative" over the three regions, while the location and intensity of anomaly center present different characteristics. It is also found that the distribution characteristics of 500hPa height anomaly field over Eastern and Northern Hemisphere is "positive-negative-positive" in the Ural and Okhotsk region during the occurrence of the long duration blocking in the typical month of summertime(July).(3) Comparison with NCEP/NCAR reanalysis data reveal that blocking days are on average underestimated by the majority of CMIP5 models. Most of models generally underestimate blocking days from June to August in the summertime, it also shows underestimation in the wintertime, but for the 120°E-160°E region, nine models significantly overestimate the blocking days, in contrast, the other four models (BCC-CSM1.1, CanESM2, CCSM4, CNRM-CM5 models) underestimate the results. The analysis documented that the simulation ability to blocking days in wintertime are superior to those in summertime. Overall, the multi-models ensemble underestimate the blocking days in summertime, and underestimation of the blocking days is evident for Ural region in winter, while the Okhotsk region present significant overestimation in winter.(4) Most of CMIP5 models could largely reproduce the zonal distribution of blocking days in summertime, wintertime and the whole year. It is found that the zonal distribution show "two peak one valley pattern" in wintertime and the whole year, but it show "three peak two valley pattern" in summertime. Furthermore, the majority of models can simulate the characteristics that the blocking days in wintertime is significantly greater than that in summertime, especially for the Okhotsk region.A quantitative analysis of the Taylor figure of blocking days revealed that the correlation coefficients between the modeled and reanalysis in wintertime are superior to those in summertime, the capacity of simulating the zonal distribution generally for five models (BCC-CSMl.1,CanESM2,CCSM4, CNRM-CM5, GFDL-ESM2G) are better than other models.(5) Most of CMIP5 models can simulate the distribution characteristics that the number of blocking events generally exhibits an exponential decrease with duration. Comparison CMIP5 models to NCEP/NCAR reanalysis data, reveals that, the frequency of blocking events are partially underestimated, especially for short duration blocking, and the bias is much more robust in summertime than in wintertime; the frequency of blocking events in summertime is consistent with NCEP/NCAR data over Okhotsk, but it shows overestimation in wintertime. Overall, three models (BCC-CSM1.1, CanESM2, CCSM4) can better represent the blocking events over three regions. It is found that the long duration blocking events present significant increasing trends in summertime over the three region, whereas slight decreasing trend is confined in wintertime; in most models, the variation trend of blocking events can be simulated in recent years, and the ability of the simulation in short duration is slightly superior to the long duration blocking; the simulation results by most models also shows that the blocking events trend in winter is better than that in summertime over Ural region, while the trend coefficient in summertime is largely accordance with NCEP/NCAR reanalysis data over Okhotsk region.(6) Model projections of blocking days for the 21st century (2016-2065) are examined under the RCP 4.5 and RCP 8.5 pathways over Asia. The multi-model ensemble results indicate that blocking days significantly increase from January to May under RCP4.5 emission scenario, it shows obvious decrease from June to August, it present a hint of increasing blocking days during the period September to December. The blocking days differ minimally between RCP4.5 and RCP8.5, but the increasing amplitude under RCP8.5 is less than RCP4.5, while the decreasing amplitude under RCP8.5 is more than RCP4.5.(7) In comparison to NCEP/NCAR reanalysis results, RCP4.5 and RCP8.5 emission scenarios all show significant decreases in the blocking events frequency over both the Ural and Baikal regions, especially for short duration blocking, with a hint of increasing blocking events frequency over Okhotsk region. For the long duration blocking, the frequency is obviously decreased by about 40% under two emission scenarios over Ural region, it shows slightly increasing trend over Okhotsk region (RCP4.5:5%; RCP8.5:2%). The variation trend of medium and long duration blocking from 2016 to 2065 over Ural region is largely consistent with projecting trend over Okhotsk region, furthermore, the trend is predicted to increase in wintertime under RCP4.5 scenario, while under RCP8.5 scenario, it exhibits decreasing trend, the projection in summertime is opposite to that in wintertime, which shows a decreasing trend under RCP4.5 and increasing trend under RCP8.5.In conclusion, the above results could provide important reference to predict the trend of extreme weather and climate over Asia region.