地球信息科学学报 ›› 2015, Vol. 17 ›› Issue (1): 108-117.doi: 10.3724/SP.J.1047.2015.00108

• • 上一篇    下一篇

多源遥感数据的降水空间降尺度研究——以川渝地区为例

嵇涛1(), 刘睿1,2,*(), 杨华1, 何太蓉1, 吴建峰1   

  1. 1. 重庆师范大学 重庆市GIS应用研究重点实验室,重庆 400047
    2. 华东师范大学 上海市城市化生态过程与生态恢复重点实验室,上海 200062
  • 收稿日期:2014-04-11 修回日期:2014-05-16 出版日期:2015-01-10 发布日期:2015-01-05
  • 通讯作者: 刘睿 E-mail:taoji_scholar@126.com;liur@lreis.ac.cn
  • 作者简介:

    作者简介:嵇涛(1988-),男,江苏扬州人,硕士生,主要从事资源环境遥感与GIS研究。E-mail:taoji_scholar@126.com

  • 基金资助:
    国家自然科学基金面上项目(41271411、40771135);重庆市科委软科学计划项目(CSTC2011CX-rkxA0280);科技基础性工作专项项目“格网化资源环境综合科学调查规范”(2011FY110400)

Spatial Downscaling of Precipitation Using Multi-source Remote Sensing Data: A Case Study of Sichuan-Chongqing Region

JI Tao1(), LIU Rui1,2,*(), YANG Hua1, HE Tairong1, WU Jianfeng1   

  1. 1. Key Laboratory of GIS Application, Chongqing Municipal Education Commission, Chongqing Normal University, Chongqing 400047, China
    2. Shanghai Key Laboratory for Urban Ecology and Sustainability, East China Normal University, Shanghai 200062, China
  • Received:2014-04-11 Revised:2014-05-16 Online:2015-01-10 Published:2015-01-05
  • Contact: LIU Rui E-mail:taoji_scholar@126.com;liur@lreis.ac.cn
  • About author:

    *The author: CHEN Nan, E-mail:fjcn99@163.com

摘要:

大量研究表明,通过传统地面气象站点实测的单点数据,不能有效地反映降水的空间变化特征。目前,以遥感数据获取的降水产品已得到了广泛的应用,但在地形地势复杂区域,遥感降水产品的空间分辨率与数据精度等方面仍然存在着极大的不足。因此,本文以四川重庆(川渝)地区为例,通过建立降水产品降尺度算法,以实现降水产品的降尺度估算,提高降水数据的空间分辨率。依据在不同尺度下(0.25°、0.50°、0.75°和1.00°),TRMM 3B43、地理因子,以及MOD13A3(NDVI)之间存在的相关关系,构建了多元回归模型。通过对比这4种尺度下的回归模型,选择其中精度最高的作为最终的降尺度算法,然后再把这种降尺度算法应用到1 km分辨率下进行降水估算。进一步,以区域差异分析(GDA)和区域比率分析法(GRA)对降尺度估算的降水数据进行校正,并结合部分地面气象站点实测的降水数据进行验证。验证结果表明:降尺度算法是可靠的,能有效提升降水产品的空间分辨率,同时GDA和GRA校正方法能减小误差,进一步提升降水估算的精度,满足区域地表过程应用的需求。

关键词: 空间降尺度, TRMM降水, NDVI, 地理因子, 川渝地区

Abstract:

Precipitation data with high spatial resolution is deemed necessary for hydrology, meteorology, ecology and others. Currently there are mainly two sources of precipitation estimation: meteorological stations and remote sensing technology. However, a large number of studies demonstrated that the measurements acquired from conventional meteorological stations are single points of data, and they can not reflect the spatial variation of precipitation effectively, especially in studying the more complex areas. While the technology of remote sensing can not only improve the quality of the actual observations, but also be able to produce reasonably high resolution gridded precipitation fields. These products obtained by satellites have been widely used in previous studies. However, when applied to complex topography region, the spatial resolution of these products is too coarse and data accuracy is not high. Therefore, we present a statistical downscaling algorithm based on the relationships between precipitation and other environmental associated factors such as topography and vegetation in the Sichuan-Chongqing region, which was developed for downscaling the spatial precipitation fields with these remote sensing products. This algorithm is demonstrated with the Tropical Rainfall Measuring Mission (TRMM) 3B43 dataset, the Digital Elevation Model (DEM) from ASTER Global Digital Elevation Model (ASTER GDEM) and Moderate resolution Imaging Spectroradiometer (MODIS) 13A3 dataset. The statistical relationship among precipitation, geographical factors and Normalized Difference Vegetation Index (NDVI), which is a representation for vegetation, is variable at different scales; therefore, a multiple non-linear regression model was established under four different scales (0.25°, 0.50°, 0.75° and 1.00°, respectively). By applying a downscaling methodology, TRMM 3B43 0.25°×0.25° precipitation fields were downscaled to 1 km×1 km pixel resolution for each year from 2000 to 2011. By comparing these four regression models, we first select the regression model with the highest accuracy as the final downscaling algorithm, and then apply this downscaling algorithm (0.25°) to 1 km resolution for the estimation of high precision in this study. Second, the calibration of downscaling precipitation was conducted based on Geographical Difference Analysis (GDA) and Geographical Ratio Analysis (GRA). The final downscaling estimation results were validated by applying part of meteorological stations measured precipitation data for a period of 12 years in Sichuan-Chongqing region. As a whole, these results indicated downscaling algorithm is reliable, and can effectively improve spatial resolution of precipitation products. The resultant best value of the 1 km annual precipitation data are achieved through downscaling followed by GDA and GRA calibration for most cases. And the downscaling 1 km annual precipitation has not only been significantly improved in the spatial resolution, but also corresponded well with TRMM 3B43 precipitation data and meteorological stations data achieved for Sichuan-Chongqing region.

Key words: spatial downscaling, TRMM precipitation, NDVI, geographical factors, Sichuan-Chongqing region