论文部分内容阅读
尺度问题是遥感科学研究的一个关键科学问题,但其理论和方法的发展严重受限于稀缺的多尺度观测数据。黑河生态水文遥感试验(Hi WATER)的核心目标之一是开展多尺度观测以支持尺度转换研究。本文综述了Hi WATER中定点观测的尺度上推研究进展,内容包括:(1)尝试严格定义了空间平均、空间尺度上推、观测足迹、代表性误差、观测真值等概念;(2)介绍了Hi WATER获取的多尺度(单点—像元—区域—流域)生态水文观测数据;(3)发展了基于地统计理论的多尺度采样方法,改进了基于时间稳定性的采样方法;(4)定量评估了辐射、碳通量、土壤水分、地表温度单点观测的代表性误差,实证了异质性地表遥感产品真实性检验的不确定性主要来源于观测的时空代表性;(5)发展了定点观测的尺度上推方法,将克里格方法推广至回归克里格、面到面、不等精度观测等情形,发展了贝叶斯框架下的非线性尺度上推方法,实证了引入遥感观测作为协同信息可显著提高尺度上推的精度。总之,Hi WATER初步形成了从采样设计、多尺度观测、代表性误差的度量、尺度上推新方法到真实性检验的研究框架。
The scale problem is a key scientific issue in remote sensing science, but the development of its theory and method is severely limited by the scarcity of multi-scale observation data. One of the core goals of Hi WATER is to conduct multi-scale observations to support scale conversion studies. This paper summarizes the research progress of scale up-push in fixed point observations in Hi WATER, including: (1) Attempting to define strictly the concepts of spatial averaging, spatial scale up-push, observation footprint, representative error and observed true value; The multi-scale (single point-pixel-area-catchment) eco-hydrological observation data obtained by Hi WATER; (3) The multi-scale sampling method based on geostatistical theory was developed and the sampling method based on time stability was improved; (4) ) Quantitatively evaluated the representative errors of single point observations of radiation, carbon flux, soil moisture and surface temperature, and demonstrated the uncertainty of the authenticity test of heterogeneous surface remote sensing products mainly from the spatial and temporal representation of observations. (5) This paper develops the standard push-up method of fixed-point observation, extends the Kriging method to regression kriging, surface-to-surface, unequal precision observation and so on, develops the non-linear scale push method under the Bayesian framework, The introduction of remote sensing observations as a collaborative information can significantly improve the accuracy of scale up. In a word, Hi WATER has initially formed the research frame from the sampling design, multi-scale observation, representative error measurement, and scale-up method to the authenticity test.