The knowledge of foliar biochemical concentration provides us a deep understanding of many esosystem functions

such as photosynthesis

nutrient cycling

and so on. Generally speaking

two ways of biochemical concentration estimation by remote sensing

namely statistical regression and physical model inversion

have been studied in parallel. There are many limitations in statistical regression method. So

direct inversion of physical model is desired. From the view of inverting physical models to retrieve biochemical concentration

inversion methods were analyzed at leaf and canopy level respectively. At leaf level

using laboratory measured spectra and PROSPECT model

water and chlorophyll concentration were inverted quite accurately. Through comparing true spectra with reconstructed spectra using true inputs

results were got as that accurate inversion of some parametes are based on the model which should describe the effect of this parameter accurately. At simulated canopy level

preliminary inversion of biochemical content directly from canopy spectra showed poor accuracy especially for chlorophyll. Based on the idea of multi stage inversion

following inversion was decomposed into 2 parts: first from canopy spectra to leaf spectra

then from leaf spectra to biochemical content. By this step by step inversion strategy

biochemical concentration was inverted accurately ultimately.It should be noted that in this paper the canopy spectra was simulated by model

further testifying with real spectra is needed which we will take into consideation in near future. Also other inversion algorithms than iterative optimization will be tested.