Inversion Of Vegetation Parameters In Chengde Using Multi-angle Satellite Data

Vegetation is the total name we called the plants growing on the earth. It plays an important role in the ecosystem, as well as a renewable resource of this planet. Many ecological processes such as photosynthesis, evaporation, evapotranspire, decomposition, and so on, correlated with vegetation. Remote sensing is an effective application to monitor the global plants. Satellite can observe the earth from out space, without the restriction of the nature and society, as well as get the data of large area immediately. According to the remote sensing data, people can quantify the affection on the global vegetation from the global changing. The reason of investigating the vegetation is: it can affect the balance of the soil-gas system, as well as play an important role in the cycle of the climate, hydrological and biochemical. Vegetation also is the sensitive index about the environment. Those are the reason why vegetation always concerned by scientists and every countries. Whether or not can retrieve every kinds of important parameters is deeper study of the analysis and technology about the remote sensing data quantifying. In this field the inversion of vegetation canopy parameters is one of the most outstanding problems about the vegetation remote sensing. It will be a long term influence if we can solve the issue successfully. Because of the restrictions of the models and the inversion applications, the study is depended on the improvement of the physical models, that is, we need an effective and accurate BRDF (Bidirectional Reflectance Distribution Function) model. But it still not established, as well as not all model could be used to retrieve the vegetation parameters. We can both not be able to get the fitting results if the models'sensitivities are hyper sensibility or hypoallergenic. The high sensitivity would cause the inversion error is big, in another hand, low sensitivity would cause the inversion result is incredible.The purpose of the study in the paper is to retrieve the vegetation parameters of the Chengde area, using the multi-angle remote sensing data and the DART(Discrete Anisotropic Radiative Transfer) model look up table, in order to get 9 vegetation parameters'changing situation from November 2005 to October 2006. Then to analysis and evaluate the inversion result, and the DART model look up table.To describe the BRDF character, DART model is one of the most outstanding three dimensions radiative transfer models. DART model can divide the vertical direction and horizontal direction of the study scene independently, that can save the compute resource. DART model divided vertical direction of the three dimensions into smaller pieces than horizontal direction, that is, because the accurate needs of the vertical direction radiative transfer. Individual cells are identified with the x, y, and z coordinates of their centers. The total cell number is I = (?X·?Y·?Z)/(?x·?y·?z), where ?X, ?Y, and ?Z are the Cartesian dimensions of the scene. Cells are used for simulating different types of scene elements. Two approaches can be used to specify the optical properties of each individual cell: first is to investigate the scene in field, second is to survey the radiative of the whole scene, as well as to calculate or simulate the radiative transfer of the scene using high resolution data. We use the two applications to establish the vegetation inversion look up table.Because of the observation modes of the simulated and the POLDER sensor are not always the same, we need to delete the unmatched POLDER data in order to make the observation modes to keep the same. We cannot go on until the POLDER data's observation mode matched with the look up table's. After finishing the screening we need to convert the POLDER BRDF value into BRF, for matching the look up table BRF. The bands matching also are necessary, and we find that there are four bands are the same between the POLDER data and look up table: 490nm, 670nm, 865nm, 1020nm. The matching algorithm is based on the least square method. That is, to calculate the square of the differences of each band and sum of them. Then we pick out the minimum sum, and find out using which BRF of the look up table we calculated. At last we pick out the parameters linked with upper BRF of the look up table.After comparing the inversion parameter with the MODIS LAI product, we find that the result of the inversion is obviously larger than MODIS data. The reason is the parameter result of the inversion reflected elder trees, the leaves are very dense, and the growth condition are very good, so the LAI value varied slightly.