ERA-Interim和ERA-5再分析资料中台风模拟精度评估

Assessment of typhoons in ERA-Interim and ERA-5 reanalysis datasets

针对海洋水动力模拟中常用的10 m高度风速和海平面气压参数,通过对欧洲中期天气预报中心(ECMWF)的ERA-Interim(ERA-I)和其最新开发的ERA-5两套再分析风资料中的台风中心、最大风速和中心气压等要素进行识别,分析了2013—2015年两套资料对西北太平洋台风的模拟精度。结果表明:ECMWF再分析风资料对台风中心位置的反映存在一定偏差,台风强度越低,台风中心位置模拟的偏差越大;再分析资料中的台风最大风速较实测偏低,中心低压较实测偏高;ERA-5资料对台风的模拟精度显著优于ERA-I资料,但在西北太平洋仍低估了台风的强度,且台风强度越高,低估越明显。基于ERA-5资料建立了台风风场再分析最大风速与实测最大风速的相关关系,提出了在不具备实测台风数据的情况下直接修正ERA-5资料的方法。通过与实测资料比较,修正后风场显著优于修正前。

Based on two of the most commonly used parameters in ocean hydrodynamic simulation,i.e.10 m wind speed and sea level pressure,we obtained the typhoon centres,maximum wind speeds and central air pressures from the ECMWF ERA-Interim(ERA-I)reanalysis dataset and the most recently released ERA-5 reanalysis dataset,and used them to evaluate their representation of typhoons in the northwest Pacific region from 2013 to 2015.The results indicate that there are differences in the locations of the typhoon centres provided by the ECMWF reanalysis wind data and the measurement data.When the typhoon intensity decreases,the bias in the location of the typhoon centre in the ECMWF data increases.The maximum wind speed of typhoons in the reanalysis data is lower than that in the measurement data,whereas the central air pressure is higher.Although the precision of ERA-5 wind data is greatly improved for the simulation of typhoons from that of the ERA-I data,it is still not sufficiently accurate in the description of typhoon intensity in the northwest Pacific Ocean,which is underestimated.When the typhoon intensity increases,this underestimation is more obvious.In this study,we established an empirical correlation between the maximum wind speed of typhoons in ERA-5 and best-track data.On this basis,the ERA-5 data can be improved using a direct modification method without the requirement for observed typhoon track data.After validation based on the measured wind speed,the modified ERA-5 reanalysis data during the typhoon period was confirmed to have greatly improved accuracy compared to the original data.