浙江农业学报 ›› 2019, Vol. 31 ›› Issue (5): 846-854.DOI: 10.3969/j.issn.1004-1524.2019.05.22

• 综述 • 上一篇    

被动微波土壤水分产品真实性检验研究进展

王春梅1,2, 顾行发1,2, 余涛1,2, 周翔1,2, 占玉林1,2, 韩乐然1,3, 谢秋霞1,3   

  1. 1.中国科学院 遥感与数字地球研究所,北京 100094;
    2.国家高分专项应用技术中心,北京 100094;
    3.中国科学院大学,北京 100049
  • 收稿日期:2018-12-05 出版日期:2019-05-25 发布日期:2019-05-23
  • 作者简介:王春梅(1978—),女,山东日照人,博士,副研究员,主要从事水资源遥感与真实性检验研究。E-mail: wangcm@radi.ac.cn
  • 基金资助:
    国家自然科学基金(41501400);国家发改委民用空间基础设施陆地观测卫星共性应用支撑平台项目(17QFGW02KJ);国家重点研发计划(2018YFB0504800,2018YFB0504804);自主部署基金(Y5SJ0600CX)

Recent advances of validation methods in passive microwave remote sensing of soil moisture products

WANG Chunmei1,2, GU Xingfa1,2, YU Tao1,2, ZHOU Xiang1,2, ZHAN Yulin1,2, HAN Leran1,3, XIE Qiuxia1,3   

  1. 1. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100094, China;
    2. High Resolution Application Technology Center, Beijing 100094, China;
    3. University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2018-12-05 Online:2019-05-25 Published:2019-05-23

摘要: 随着全球气候变化和陆面数据同化研究对土壤水分反演精度要求的不断提高,大尺度被动微波土壤水分产品的真实性检验变得极为重要。如何获取可以代表卫星观测尺度“真值”、并能表征空间异质性的土壤水分观测场,成为被动微波土壤水分产品真实性检验的关键问题。土壤水分观测场的构建涉及地面观测、同步观测、尺度转换等关键环节,可通过“点代面”和“多源信息融合”这2个技术途径进行构建。简要总结了应用于大尺度土壤水分的5种典型的真实性检验方法,包括实测样本数据检验、影像数据交叉检验、模型模拟检验、影响因素检验和传统地统计检验。这5种方法或缺乏对先验知识的有效利用,或忽略地面实测的重要性,或在综合利用多源数据类型的先验知识信息方面不足。随着贝叶斯最大熵理论的发展,基于贝叶斯最大熵理论和先验知识的大尺度土壤水分产品真实性检验有望发展成为一种可靠的方法。贝叶斯最大熵理论的优势在于,能够提供灵活的数据利用方式,使多种来源、多种类型的数据集有机会同时被用于卫星观测尺度的时空分析,生成高分辨率土壤水分数字地图,从而为大尺度土壤水分产品的真实性检验研究提供一个新的途径。

关键词: 土壤水分, 真实性检验, 贝叶斯最大熵

Abstract: The validation of large-scale soil moisture products in passive microwave remote sensing takes a key role for global climate change and data assimilation studies. However, it is difficult to obtain the real soil moisture values that can represent the soil moisture values and reflect spatial heterogeneity of large-scale passive microwave remote sensing products. It has become a key issue for the validation of passive microwave soil moisture products. At present, two technical approaches were used to build soil moisture observation network: the scale-up methods and the multi-source information fusion methods. Both two approaches involved the ground observation, simultaneous observation, and scale conversion, etc. The progress made in validation methods of large-scale soil moisture products was reviewed in this paper, and five typical methods were summarized, including the measured sample test, image data cross-test, model simulation test, influencing factor test and traditional geostatistical test. The first two methods belongs to the “spot generation” scale conversion test, and they attach importance to the spatial representation of the ground sample. Theoretical basis of the model simulation test and influencing factor test is the relationship between soil moisture and prior knowledge. However, these two methods neglect the importance of sample data. Traditional geostatistical test takes into account both sample points and prior knowledge, but it is insufficient in the comprehensive utilization of prior knowledge information of multi-source data types. Currently, the Bayesian maximum entropy method is widely used in spatial prediction. In some studies, this method has been applied to multi-source data fusion. The advantage of Bayesian maximum entropy test provides a flexible way of data utilization. This method allows multiple sources and types of data sets to be used for spatial analysis of satellite observation scale at the same time, generates high resolution reference map of soil moisture observation field, completes the validation of soil moisture products, and provides a basis for the authenticity test of large-scale soil moisture products. Thus, it would provide a new way for the validation of large-scale soil moisture products.

Key words: soil moisture, validation, Bias maximum entropy

中图分类号: