An Algorithm Study On Normalization Of Vegetation Index Based On Multi-source Remote Sensing Data

Vegetation index is an important parameter to describe the characteristics of vegetation. Long time series of vegetation index is a simple, effective and experience measure parameter to obtain vegetation cover status and biophysical parameters of vegetation, which is useful for the global vegetation detection and meaningful for the carbon circle study. Vegetation index products of single remote sensing data have a big flaw in the accuracy and continuity of space and time, while the cooperative inversion of multi-source remote sensing data has attracted more and more attention. But some factors formed obstacles in cooperative inversion of multi-source remote sensing data, such as the performance of remote sensor and atmospheric condition. So the study on normalization of vegetation index based on multi-source remote sensing data had an important significance on quantitative remote sensing and cooperative inversion of multi-source remote sensing data.The paper takes EOS/MODIS, FY-3A/MERSI, FY-3A/VIRR, FY-3B/MERSI and FY-3B/VIRR for example to study how to better use multi-source remote sensing data.All the research data is moderate and low resolution remote sensing satellites data.Among them, the radiation performance of EOS/MODIS is stable and it is the basis to analysis the reasons caused the difference of vegetation index from these multi-source remote sensing data. An algorithm of normalization of vegetation index based on multi-source remote sensing data is proposed. The main works of this thesis are listed as follows:Firstly, the paper selected the data of FY-3A/MERSI, FY-3A/VIRR,FY-3B/MERSI, FY-3B/VIRR and EOS/MODIS which is covering the midstream of Heihe river basin and less cloudiness. The land use type of the study area is arable land.The vegetation index of the selected data is obtained after the data is processed. Secondly,atmospheric radiative transfer model MODTRAN, the vegetation index before and after cross-calibration, equivalent surface reflectance and observation geometry of satellite sensor were used to analysis the reasons caused the difference of vegetation index from these multi-source remote sensing data. The reasons include atmospheric water vapor,radiometric calibration, spectral response function and viewing angle. In the end,atmospheric radiative transfer model MODTRAN was used to realize the normalizationof vegetation index. The change of arable land had been taken into account to enhance the accuracy of the produced coefficients of BRDF model and spectral matching factor. Two sets of coefficients of BRDF model and corresponding spectral matching factors were produced for bare soil and vegetation period. After verification, the accuracy and stability of the normalized vegetation index has increased compared with a prior-oriented. The algorithm can be used in cooperative inversion of multi-source remote sensing data.