论文部分内容阅读
定量遥感反演由于观测信息量不足,往往是“病态”反演,而且在区域研究中格外突出.本文以MODIS250m估算区域LAI作为典型案例,提出了基于空间知识的多尺度多阶段目标决策反演方法,研究先验知识的引入及其合理使用,试图为解决区域遥感“病态”反演提供一个合理的解决方案.首先,利用不同分辨率的MODIS影像(1km,500m和250m)提取地表的多尺度信息,并将其融入到MODIS较低分辨率数据的多阶段目标决策反演上,通过降低空间异质性影响,提高了粗尺度数据反演参数的准确性;然后,粗尺度反演结果作为细尺度反演的先验知识,再次参与反演,通过多次反演,先验知识实现多次更新.从MODIS_1km到MODIS_250m,在每一个尺度的反演中,用最敏感的数据反演最不确定的参数,实现了有限的数据在模型空间中的合理分配.基于空间知识的多尺度多阶段目标决策反演方法,融合了地面实测数据、空间知识、多尺度遥感观测数据,反演是一个逐步细化的过程,待反演参数的初始期望更加合理、不确定性范围有效缩小,反演目标更加明确.最后利用MODIS数据反演黑河中游农作物区LAI对该方法进行了验证,结果表明这种反演方法较以往传统的区域性参数获取方法更为准确、可靠.
Due to the lack of observational information, quantitative remote sensing inversion is often “sick” inversion and especially prominent in regional studies.In this paper, a typical case of MODIS250m estimation region LAI is proposed, and a multi-scale and multi-stage target decision based on spatial knowledge (1km, 500m and 250m) using MODIS images with different resolutions, this paper proposes a new method to solve the regional remote sensing “sickness” inversion based on the inversion method and the introduction of prior knowledge and its rational use. The multi-scale surface information is extracted and integrated into multi-stage target decision retrieval of lower resolution MODIS data. By reducing the influence of spatial heterogeneity, the accuracy of inversion parameters of coarse-scale data is improved. Then, the coarse As a priori knowledge of microscale inversion, scale inversion results are again involved in inversion, and multiple updates are made through multiple inversions and prior knowledge. From MODIS_1km to MODIS_250m, in each scale inversion, the most sensitive The most uncertain parameters are retrieved and the reasonable distribution of finite data in the model space is achieved.A multi-scale and multi-stage target decision retrieval method based on spatial knowledge is combined with the measured data from the ground, Space knowledge and multi-scale remote sensing data, the inversion is a process of gradual refinement, the initial expectation of inversion parameters is more reasonable, the scope of uncertainty is effectively reduced, and the inversion objective is more clear.Finally, MODIS data is used to inverse the middle reaches of the Heihe River This method is validated by LAI in crop area. The results show that this inversion method is more accurate and reliable than the traditional method of obtaining regional parameters.