地理学报 ›› 2019, Vol. 74 ›› Issue (7): 1305-1318.doi: 10.11821/dlxb201907003

• 气候变化与地表过程 • 上一篇    下一篇

全球降水计划IMERG和GSMaP反演降水在四川地区的精度评估

曾岁康1,2,雍斌1,2()   

  1. 1.河海大学水文水资源与水利工程科学国家重点实验室,南京 210098
    2.河海大学地球科学与工程学院,南京 211100
  • 收稿日期:2018-12-03 修回日期:2019-04-23 出版日期:2019-07-25 发布日期:2019-07-23
  • 通讯作者: 雍斌 E-mail:yongbin_hhu@126.com
  • 作者简介:曾岁康(1995-), 男, 四川泸州人, 硕士生, 主要从事GIS在水文气象中的应用研究。E-mail: zengsk678@163.com
  • 基金资助:
    国家重点研发计划(2018YFA0605402)

Evaluation of the GPM-based IMERG and GSMaP precipitation estimates over the Sichuan region

ZENG Suikang1,2,YONG Bin1,2()   

  1. 1.State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering, Hohai University, Nanjing 210098, China
    2.School of Earth Sciences and Engineering, Hohai University, Nanjing 211100, China
  • Received:2018-12-03 Revised:2019-04-23 Online:2019-07-25 Published:2019-07-23
  • Contact: YONG Bin E-mail:yongbin_hhu@126.com
  • Supported by:
    National Key R&D Program of China(2018YFA0605402)

摘要:

IMERG和GSMaP是全球降水计划(GPM)时代最主要的高分辨率降水产品。为研究其在中国四川地区的适用性,以中国气象局提供的自动气象站融合降水数据为参考基准,采用6种统计指数分析了IMERG(IMERG_Uncal, IMERG_Cal)和GSMaP(GSMaP_MVK, GSMaP_Gauge)系列产品在四川的误差特征。结果表明:① 在日和小时尺度上,GSMaP系列产品均高估地面降水观测,GSMaP_MVK高估最显著,校正产品GSMaP_Gauge的相关系数(CC)、相对偏差(BIAS)和均方根误差(RMSE)较GSMaP_MVK均有较大提高,尤其对川西高原降水的高估现象改善明显,而IMERG_Uncal存在低估川西高原降水、轻微高估四川盆地降水的问题,校正产品IMERG_Cal一定程度上降低了对川西高原降水的低估现象,但整体精度(CC, RMSE)提高不明显。② IMERG系列产品对降水事件的探测准确性更好,GSMaP_Gauge虽然在四川表现出较高的命中率(POD),但存在较多的误报降水,在盆地和四川南部各产品均表现出较高的POD和关键成功指数(CSI)以及低误报率(FAR),而四川西北部表现最差,尤其是在无自动站分布地区。③ 4套降水产品中,IMERG_Cal表现出最好的探测强降水和弱降水的能力,具有一定的监测极端降水的潜力。总体上,IMERG和GSMaP在盆地的反演精度优于高原山区,校正产品精度优于纯卫星产品,不同地形地区精度差异明显,表明对卫星降水产品进行不同地形误差订正仍是未来降水反演工作的重点和难点。

关键词: IMERG, GSMaP, 卫星降水, 四川, 误差特性, 地形

Abstract:

The Integrated Multi-satellitE Retrievals for GPM (IMERG) and Global Satellite Mapping of Precipitation (GSMaP) are two high precisely multi-satellite precipitation estimates in the GPM era. In order to evaluate the applicability of both IMERG and GSMaP series products (IMERG_Uncal and IMERG_Cal, GSMaP_MVK and GSMaP_Gauge) over the Sichuan region in China, six statistical indices are used to systematically analyze the error characteristics of these products, benchmarked by a set of ground-based dataset from China Meterological Administration (CMA). Results show that: (1) All products show the dramatic regional difference over Sichuan at both daily and hourly scales. The GSMaP series products overestimate precipitation and the most overestimations occur over the high altitude areas located in the Western Sichuan. GSMaP_Gauge shows relatively higher correlation coefficient and lower relative bias and root mean square error due to the employment of gauge-based adjustments. On the contrary, IMERG_Uncal shows underestimation over the mountainous areas, while the relatively slight overestimation appears in the basin area with lower elevation at both daily and hourly time scales, suggesting that gauge-calibrated dataset IMERG_Cal has effectively improved the relative bias in the mountainous areas but not in the flat basin area. (2) By synthesizing the three classified statistical indices, IMERG series products exhibit better potentials in detecting precipitation events. Although GSMaP_Gauge shows a higher hit rate of precipitation, it has more false alarm ratios of precipitation. All products show better hit rate and lower false alarm rate over basin area and southern Sichuan. Furthermore, it is found that the ground-based dataset has some errors in those areas without meteorological stations, which leads to the apparent uncertainty in assessing the accuracy of satellite precipitation products over the Northwest Sichuan Plateau. (3) IMERG_Cal performs better in capturing the rainfall amounts and events compared with other products, especially for the lowest and highest rainfall intensity ranges, demonstrating its application potential for monitoring the extreme weather events. Overall, both IMERG and GSMaP estimates have relatively high uncertainties over the mountainous areas than ones over the flat basin areas. Additionally, the gauge-calibrated products obviously outperform the uncalibrated datasets. On the basis of the findings, future efforts focus on reducing and correcting the errors and biases of satellite precipitation estimates by considering both spatio-temporal characteristics and the topographical information.

Key words: IMERG, GSMaP, satellite precipitation products, Sichuan, error characteristics, terrain