Quantitative Fusion of Multi-sensor Observations with Combination of Physical and Empirical Models

Fusion of multi-sensor observations of satellites would improve land surface monitoring. Physical-based model is the most popular method for retrieval of atmospheric and land surface parameters from remote sensing data for its applicability for large area, while empirical model is more efficient and requires fewer constraints. In this paper, a pixel-based quantitative fusion algorithm of multi-sensor observations with combination of physical and empirical model is presented. Firstly, the parameter was retrieved from one sensor data based on physical model, and then used to establish the pixel-based empirical relationships with measurements of another sensor. Thus, the two retrieval methods could be combined and the observations of two sensors could also be fused. The algorithm was applied to fuse MISR with multi-angular measurements and MODIS with high temporal resolution for retrieval of Leaf Area Index (LAI). The results were evaluated using field measurements in Changbaishan.