Research On The Estimation Of Fractional Vegetation Cover And The Validation Of Fractional Vegetation Cover Product

As a quantitative parameter, the fractional vegetation cover(FVC) is an important element in vegetation monitoring, numerical weather prediction, regional and global climate modeling and global change monitoring. Accurate estimation of FVC is important for modeling earth system processes. To properly use FVC products in further studies, the assessments of the accuracies and uncertainties in them is crucial.This dissertation is supported by the National High Technology Research and Development Program of China and dependent on Jingyuetan Remote Sensing Experiment Station. In this dissertation, we carried out long term experiments of fractional vegetation cover over croplands and developed a method of ground sampling for the validation of remote sensing products. Combined with ground sampling measuremts, we focus on the estimation of fractional vegetation and its validation. The results show that:(1) An automatic extraction system of fractional vegetation cover is developed, which is a viable ground-based method of FVC measurements. This system is a basis of the dissertation.(2) The FVC estimation models based on vegetation indice are developed by using ground FVC and Landsat 8 OLI images. The accuracies of SAVI and MSAVI on the estimation of fractional vegetation cover are higher than other vegetation index. The accuracies of Gutman, Carlson, Baret and SDVI models are in the order of Baret>Carlson >SDVI >Gutman. The accuracy of Baret model is improved by modifying the paremeter of K_p/K_VI, which is sensitive to the types and growth of vegetation.(3) Four time serial EO-1 Hyperion imagery are used to investigate the uncertainty introduced by spectral response functions of OLI, MODIS and SPOT-VGT sensors in the estimation of fractional vegetation cover. The uncertaintity of HJ-1B CCD sensor on the estimation of fractional vegetation cover is higher than OLI. The VEGETAION simulated NDVI is lower than MODIS at the beginning growth of vegetation. The VEGETAION NDVI is greater than MODIS NDVI at the mature period of vegetation. The FVC of VEGETAION is lower than those of MODIS and OLI for both forest and crop. The highest average relative difference between FVCVGT and FVCOLI is about-17.8%.(4) A concept of mean length variability is proposed to compare the difference in spatial heterogeneity detected by variables with different magnitudes. The important temporal changes in spatial heterogeneity observed by NDVI, NIR and red bands over cropland are a result of changes in the fraction of vegetation cover. The red reflectance is sensitive to soil properties while the NIR reflectance responds to vegetation. The spatial heterogeneity of red reflectance decreases and that of the NIR reflectance increases with the growth of vegetation. The NDVI value shows the greatest heterogeneity in the early stage of crop growth. With an increase in the image pixel size, the spatial heterogeneity quantified by the mean length variability of the NDVI, NIR and red variables tends to be the same.(5) For its effective stratification strategies and its criteria of minimum mean square estimation error, MSN demonstrates the lowest mean squared estimation error for estimating the means of stratified nonhomogeneous surface. MSN is an effective method for estimating the spatial means for heterogeneous surfaces. For a given variance of estimation, the MSN program can export the optimal sampling strategy. A sampling strategy of heterogeneous pixel in different scales is developed by variance of estimation.(6) Global CYCLOPES, GEOV1 and regional Australian MODIS FVC products were compared over six evaluating units with different biome types across the Australian continent during the 2000-2012 period. GEOV1 and Australian MODIS FVC products were validated by comparison with 443 FVC in-situ measurements over the period 2011-2012. GEOV1 and Australian MODIS products describe consistent temporal profiles over most biome types while withn differences in magnitude. The Australian MODIS FVC shows the highest values while CYCLOPES the lowest. GEOV1 and Australian MODIS provide similar agreement with in-situ Australian FVC values over Australian continent with a RMSE around 0.08-014.