ERA5與ERA_interim降水資料於中印半島地區的表現評估與表現差異原因探究

Abstract

ERA5(ECMWF Reanalysis v5)是歐洲中期天氣預報中心(European Centre for Medium-Range Weather Forecasts；ECMWF)的第五代全球大氣再分析資料，近年ERA5逐漸取代ERA_interim(ECMWF Reanalysis - Interim)作為ECMWF主要的再分析資料，因此瞭解ERA5與ERA_interim資料的優缺點差異，有其必要性與重要性。本研究採用亞洲降水高解析度陸地觀測網格資料(Asian Precipitation Highly Resolved Observational Data Integration Towards Evaluation；APHRODITE v1901)作為降水的參考基準，以評估2000至2015年期間中印半島地區(100°E-110°E，8°N-24°N)ERA5與ERA_interim兩組再分析資料描繪四季降水的能力。此外，本研究亦透過環境場與降水類型的分析，進一步探討造成兩組再分析資料模擬降水差異的可能原因。在降水時間分析方面，本研究針對年週期變化、四季之年際變化進行分析。研究結果顯示，在年週期變化方面，與ERA_interim相比，ERA5在年週期時間序列的相位變化與量值變化都與APHRODITE較為接近，因此估算年週期時間序列的表現優於ERA_interim，但ERA5和ERA_interim兩者皆存在高估降水的情況，ERA5高估降水的情況主要集中在6月至8月期間，ERA_interim高估降水的情況主要集中在1月至5月期間。在四季之年際變化方面，大多數的統計分析結果顯示，與ERA_interim相比，ERA5在四季中皆具有較高的相關係數，因此，ERA5四季降水的年際相位變化與APHRODITE較為接近。在降水空間分析方面，中印半島的四季降水在空間分布依季節變化而有所不同，在華南地區以及泰國、寮國、柬埔寨三國交界處等地勢平緩的區域，ERA5與ERA_interim描繪降水的能力較佳。而隨著降水強度的增加，ERA5與ERA_interim對降水掌握的能力降低。另外在環境場空間分析方面，發現降水的分布與風場的輻合、輻散分布相關性高，降水偏高的位置多為風場輻合處。而ERA_interim降水較ERA5偏高的位置也可觀察到相對的風場輻合較強，顯示兩組再分析資料描繪降水的能力差異可能與其風場的表現差異有關。最後針對ERA5與ERA_interim的總降水、對流降水與大尺度降水的空間分析，發現兩組再分析資料的總降水差異在夏、秋兩季以對流降水為主導，在春、冬兩季則是對流降水與大尺度降水貢獻相當。ERA5 (ECMWF Reanalysis v5) is the 5th generation of global atmospheric data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF). In recent years, ERA5 is replacing the ERA_interim (ECMWF Reanalysis - Interim) as the ECMWF’s major reanalysis data. Thus, it is necessary and important to know the differences between ERA5 and ERA_interim. This study compares the capacities of ERA5 and ERA_interim in depicting seasonal precipitation variation in the Indochina Peninsula (100°E-110°E，8°N-24°N) from 2000 to 2015. The evaluation use the Asian Precipitation Highly Resolved Observational Data Integration Towards Evaluation (APHRODITE v1901) and the Integrated Multi-satellite Retrievals for GPM (IMERG V06) as the observational base. Additionally, this study also discusses the possible causes of differences between the ability of ERA5 and ERA_interim in depicting the precipitation variations, in a view of atmospheric circulation changes and precipitation type changes.For the temporal precipitation changes, this study examines the annual cycle, the interannual rainfall variation for the four season. The analysis suggests that ERA5 performs more closer than ERA_interim to APHRODITE in terms of changes in phases and magnitudes of annual rainfall variation. However, both ERA5 and ERA_interim tend to overestimate precipitation, and the ERA5 mainly overestimates the June to August rainfall, while the ERA_interim mainly overestimates the January to May rainfall. As for the interannual variation of four season rainfall, most statistics suggest that ERA5 outperforms ERA_interim with a higher correlation coefficient. Thus, the interannual phase evolution of rainfall depicted by ERA5 is more closer than ERA_interim to APHRODITE. In terms of precipitation spatial analysis, the spatial distribution of seasonal precipitation in the Indo-China Peninsula varies according to the seasons. In southern China, the junction of Thailand and Laos, and other areas with gentle terrain, the ability of ERA5 and ERA_interim in depicting the precipitation is better than that over the high mountains. With the increase of precipitation intensity, the ability of ERA5 and ERA_interim to depict precipitation change decreases.Regarding spatial analysis of the environmental field, it appears that the distribution of precipitation has a high correlation with the convergence of the wind field. Overall, locations with high precipitation are mostly at the convergence of the wind field. We found that where the precipitation of ERA_interim is being higher than that ofERA5, the relative wind field convergence tends to be stronger too. This implies that the differences between the precipitation of ERA5 and ERA_interim might be partly related to the differences between their related wind field. Finally, according to the spatial analysis of total precipitation, convective precipitation and large-scale precipitation of ERA5 and ERA_interim, it is found that the difference in total precipitation between ERA5 and ERA_interim is dominated by the difference in convective precipitation for summer and autumn. In contrast, for spring and winter, both the difference in convective precipitation and the difference in large-scale precipitation are important to the difference in total precipitation between ERA5 and ERA_interim.