In this thesis the temperate typical steppe of North China is taken as the research object. Within the research area 49 sample fields are chosen respectively from the areas with high, medium and low vegetation fractional coverage. Sample plots are taken from each sample field through the process of nesting. Namely, a 30m nested sample plot is chosen randomly from each 60m sample plot, and then a 15m nested sample plot is further chosen from each 30m sample plot. Within these sample plots of different sizes, three 1m sample plots are chosen from each 15m sample plot, five 1m sample plots are chosen from each 30m sample plot, and ten 1m sample plots are chosen from each 60m sample plot. For each 1m sample plot, a digital camera is positioned at a 2-meter height from the herbage with its lens directed vertically downwards towards the ground and taking pictures of the herbage. The vegetation fractional coverage is estimated from the patterns of the images thus taken by the camera. Based on the data thus obtained from the above-described field measurement and the data obtained through ASTER remote sensing, the experiential model of vegetation fractional coverage is established.
The procedures for the establishment of the model are as follows:
(1) the calculation of the vegetation fractional coverage indicated by each digital photo taken at the field measurement and the vegetation fractional coverage of each sample field based on its particular size, using the model of extracting vegetation fractional coverage;
(2) the calculation of the various vegetation indices (RVI,NDVI,NDGI) indicated by the ASTER;
(3) the match of the sample fields and their corresponding areas of ASTER image pixels according to the sample fields′ respective sizes;
(4) the calculation of the medians of the vegetation indices of a particular sample field′s corresponding area of ASTER image pixels , based on the size of that particular sample field;
(5) the establishment of the regression model (including the lineal model and the nonlinear model) between the vegetation fractional coverage values derived from the field measurement and those derived from the ASTER data, based on the size of that particular sample field;
(6) the administration of a significance test on the regression model thus established.
In the establishment of the experiential model, the following two points are investigated: the correlation between the vegetation indices (NDVI,NDGI,RVI) and the vegetation fractional coverage and the effects of the sizes of the sample plots on the experiential model. In addition, a comparison is made between the experiential model and the sub-pixel model which is currently in wide use to test the applicability of the sub-pixel model for the temperate typical steppe of North China. The results of the comparison show:
(1) For the monitoring of the vegetation fractional coverage of grassland of the temperate typical steppe of North China, it is appropriate to use vegetation index NDVI, which is more effective and relevant than RVI and NDGI;
(2) The establishment of experiential model of the vegetation fractional coverage is much affected by the choice of different sizes for the sample plots on the earth. With regard to the temperate typical steppe of North China, to choose large sample plots from the research area and select their medians can eliminate to some extent the negative effects brought about by a mismatch of the data of the field measurement and the data of the remote sensing measurement so as to enhance the significance level of the correlation coefficient between the vegetation indices and the vegetation fractional coverage.
(3) Compared with the experiential model of vegetation fractional coverage based on large-size sample plots, the accuracy of sub-pixel model is relatively weaker for the estimation of the vegetation fractional coverage of the temperate typical steppe of North China.