多模式预报产品对长江中下游地区早稻高温热害识别能力的比较

Base the Multi-model Forecasting Products Compared Simulation Capability of Heat Damage on Early Rice in the Middle and Lower Reaches of Yangtze River

利用CLDAS实况融合产品,比较分析欧洲中心高分辨率模式(ECMWF_HR)、GRAPES全球模式(GRAPES_GFS)以及中央气象台指导预报(SCMOC)3种模式预报产品对2021-2023年长江中下游地区早稻孕穗-成熟期提前1~3d、1~5d、1~7d条件下高温热害发生与强度的识别能力。结果表明:(1)针对高温热害的发生预报,SCMOC产品识别效果最优,各预报时段命中率(Probability of detection,POD)均大于0.6,TS(Threat score)评分在0.49~0.59;GRAPES_GFS产品的POD同样大于0.6,但误报率(False alarm ratio,FAR)大于0.3,TS评分为0.45~0.52;ECWMF_HR产品的识别效果最差,POD小于0.4,TS评分小于0.3。(2)对于高温热害强度识别,SCMOC产品与CLDAS产品的危害热积温数值最为接近,相关系数(Cor)大于0.6,均方根误差(RMSE)随预报时段增加分别为1.57℃·d、2.57℃·d、3.43℃·d;GRAPES_GFS产品对湖北大部和江西北部的预报偏强,对湖南南部预报明显偏弱,且与CLDAS产品的Cor较SCMOC产品偏低约0.06,RMSE偏高约0.4℃·d;ECMWF_HR产品与CLDAS产品的Cor在各预报时段均小于0.4,RMSE较SCMOC产品偏高0.5~1.0℃·d。(3)针对不同年份的高温热害,SCMOC与GRAPES_GFS产品对研究区域早稻高温热害影响较重的2022年和2023年识别效果较好,但前者对高温热害影响较弱的2021年识别效果更优;ECMWF_HR产品对2021-2023年研究区域高温热害的发生与强度均呈显著偏弱预报。综上所述,SCMOC预报产品对长江中下游早稻高温热害的识别效果较优,能为早稻高温热害防灾减灾工作提供参考。

Using CLDAS products,the European Centre for Medium-Range Weather Forecasts high resolution model(ECMWF_HR),Global/Regional assimilation and prediction system-Global Forecast System(GRAPES_GFS)and the data of the national meteorological center forecast(SCMOC)were analyzed for identifying the occurrence and intensity of heat damage under the condition of 1-3d,1-5d and 1-7d in advance during the early rice booting maturity period in the middle and lower reaches of the Yangtze river from 2021 to 2023.The results showed that:(1)for the identification effect on the occurrence of heat damage,SCMOC products had better identification effect,with the probability of detection(POD)greater than 0.6 for each prediction period.The TS score between 0.49 to 0.59.The POD of GRAPES_GFS product was also greater than 0.6,but the false alarm rate(FAR)was greater than 0.3,with the TS score of 0.45 to 0.52.The ECWMF_HR product had the worst discriminative effect,with POD less than 0.4 and TS score less than 0.3.(2)For the identification effect of heat damage intensity,the value of accumulated heat damage of SCMOC products and CLDAS products were the closest,with the correlation coefficient(Cor)greater than 0.6.The root mean square error(RMSE)increased the forecast period,with values of 1.57℃·d,2.57℃·d,and 3.43℃·d,respectively.GRAPES_GFS product had a strong forecast for most of Hubei and northern Jiangxi,while the forecast for central and southern Hunan was significantly weak.The Cor of GRAPES_GFS product with the CLDAS product was about 0.06 lower than that of the SCMOC product,and the RMSE was about 0.4℃·d higher.The Cor of ECMWF_HR products and CLDAS products was less than 0.4 in each forecast period,and the RMSE was 0.5℃·d to 1.0℃·d higher than that of the SCMOC products.(3)For the heat damage in different years,SCMOC and GRAPES_GFS products had good identification effect in 2022 and 2023,but the former had better identification effect in 2021.The ECMWF_HR product showed a significant weak forecast for the occurrence and intensity of heat damage in the study area from 2021 to 2023.In summary,the SCMOC prediction product has better identification effect on the heat damage of early rice in the middle and lower reaches of the Yangtze River,which can provide reference to carry out disaster prevention and reduction work on the heat damage of early rice.