The Vegetation Of The Scale Of The Physical And Chemical Parameters Inversion Effect And Sensitivity Analysis

As one of the most important components of the ecosystem, vegetation provides matter and energy for almost all organism including humans. Many ecological processes about the exchange in matter and energy, like photosynthesis, evapotranspiration, respiratory, primary productivity, decomposition and so on, are closely related to vegetation biophysical and biochemical variables, so estimating these vegetation variables has a vital significance. Remote sensing data plays a crucial role in monitoring earth biosphere and dynamic change of vegetation. With the development of hyperspectral imaging techniques, devising vegetation biophysical and biochemical variables quantitatively becomes a research hotspot.Remote sensing technique provides a series of remote sensing data with different spatial and spectral resolution, and the scale issues in remote sensing application follow. The unique scale characteristics make the scale issues in vegetation physicochemical parameters inversion more prominent. Based on the field experiment, airborne AISA hyperspectral data as well as PROSAIL model, we researched systemically the scale issues in vegetation physicochemical parameter inversion and the sensitivity change of vegetation reflectance to vegetation physicochemical parameter, which were wished to provide adequate theoretical basis for accurately inversing vegetation physicochemical parameters.Based on the field experiment, the relationships between chlorophyll content and vegetation reflectance, derivative of the reflectance and vegetation indices in leaf and canopy scale were analyzed quantitatively. In addition, the factor analysis method and the fractal dimension were used in this paper to retrieve vegetation physicochemical parameters. The results show that there are great differences of relationships between leaf scale and canopy scale, which illustrate the scale problems in the physicochemical parameters inversion to a certain extent. The results also prove that the factor analysis method reducing dimension and the fractal dimension reflecting synthetic variations of vegetation reflectance have potential applications on vegetation physicochemical parameters inversion.In this paper, leaf scale, canopy scale and pixel scale were proposed as three consecutive scales in vegetation physicochemical parameters inversion, and the scale effect was analyzed from two steps of obtaining landmark reflectance or spectral index and inversing parameters. Based on the AISA hyperspectral data and the simulated data from PROSPECT model and PROSAIL model, we analyzed the spatial scale effects of vegetation reflectance and vegetation indices at leaf, canopy and pixel scale, and the effect of different combination type in the pixel, different vegetation coverage and different spatial resolution on the physicochemical parameters inversionat pixel scale. The results show that from leaf scale to canopy scale, the scale variation of vegetation reflectance is most strongly impacted by LAI, while by vegetation coverage from canopy scale to pixel scale. At pixel scale, the linear vegetation index DVI shows a scale-invariant feature, while the nonlinear index NDV1shows significant scale variation, which is more obvious in the continuous woodland. The relationship between the inversed chlorophyll content and the true content has close related to the element type and the vegetation area in the pixel. With the reduced spatial resolution, the difference between the inversed chlorophyll content and the true content gradually decreases.Based on the simulated data from PROSPECT and PROSAIL model, the sensitivity of vegetation reflectance at leaf, canopy and pixel scale to vegetation physiochemical parameters (chlorophyll content, carotenoid content, water content, dry matter content,leaf structure parameter, LAI, leaf angle and soil background) were analyzed adopting the EFAST global sensitivity analysis method. The results show that at leaf scale, chlorophyll, water content and leaf structure parameter are the most sensitive variables to leaf reflectance. At canopy scale, LAI is the most important variable to canopy reflectance in lower foliage cover (LAI<3), and as LAI increased, chlorophyll a+b, dry matter content, and water content control the variation of canopy reflectance in VIS, NIR and SWIR regions respectively. At pixel scale, the vegetation area in the pixel is the most important parameter to the reflectance variety, and the other variables like LAI, chlorophyll, water content, dry matter content, soil background and son on have no significant contribution to the variation of pixel reflectance. In addition, we analyzed the sensitivity of physicochemical parameters to spectral indices with scale changing, and the best hyperspectral indices and multispectral indices used to retrieve chlorophyll content, carotenoid content, water content, dry matter content and LAI are chose at different scale, by calculating and comparing the correlation coefficient between vegetation indices and vegetation physiochemical variables.Based on the study mentioned above, we analyzed the problems existed in vegetation physicochemical parameters diversion at present, and discussed the overview of its trends.