Study On Vegetation Coverage Retrieval Based On Multi - Source Remote Sensing

Vegetation coverage is the important data that describes the ecosystem, and plays an important role in the global ecosystem.. This paper in Qinghai Lake Basin as the study area, using modern satellite remote sensing technology, GPS technology with “fisheye” image technology to get the ground measurements of vegetation coverage data, using multi temporal remote sensing platform, multi resolution and different imaging mechanism of active and passive satellite remote sensing data. On the basis of calculation and analysis in different vegetation index, the establishment of the statistical regression model, Dimidiate Pixel Model and model of high resolution image and low resolution image cooperative inversion of vegetation coverage degree, elected best suited to the local areas, the vegetation cover degree of inversion model and vegetation index. This paper draws the following conclusions through analysis and comparison:1.Three the statistical regression model of the vegetation coverage is established by 1 kinds of optical remote sensing data, and the results are well correlated with the observed values.. Among them, the best vegetation index of high-resolution GF-1 and the medium resolution image Landsat8 is GEMI, and the best vegetation index of the low resolution image MODIS is EVI.2. Based on the active remote sensing data Radarsat2, the statistical regression model of the vegetation coverage of Radarsat2 was carried out, and the correlation R of the log regression model was the highest, and the correlation was 0.4900. Compared to the statistical regression model of the vegetation coverage of passive remote sensing, the correlation coefficient of active remote sensing data Radarsat2 is slightly lower.. So, the single radar data is not suitable for the inversion of vegetation cover in the region..3. Different resolution passive remote sensing data set up statistical regression model, have better inversion results. But in general, the inversion accuracy of GF-1 data is the highest, and its correlation coefficient R is 0.8455.4. Based on the principle of the two pixel model of the normalized difference vegetation index, the vegetation coverage of the Qinghai Lake basin is retrieved.. The estimation of Qinghai Lake watershed vegetation coverage based on MODIS Dimidiate Pixel Model, the correlation coefficient was 0.610 4. In order to improve the accuracy of the model, the choice of extracting Dimidiate Pixel Model NDVIsoil and ndviveg parameters in higher resolution landsat8 data, estimation of substituting the Dimidiate Pixel Model formula of vegetation coverage degree, the correlation coefficient for 7054.5. Consider monitoring the actual cost and timeliness issues, choose to use the higher temporal resolution MODIS data and the higher spatial resolution GF-1 synergy inversion, the income of vegetation coverage degree of inversion model has the best effect, the correlation coefficient is 0.7948.