Leaf growth will change along with time, so species spectral characteristics will be affected. The researches on spectral changing rule of the same tree leaf under the conditions of different time and spectral characteristics of different tree leaves under the condition of the same time can not only provide theoretical basis for vegetation leaf spectral change rule along with time, but also are the keys of vegetation information accurate identification with hyperspectral remote sensing. Ten kinds of common tree species in Beijing were selected and leaf spectrum of each tree species in different time was observed by using spectrometer. At the same time, the observed spectrum was dealt with first order differential and typical bands analysis. Then spectrum difference of different tree leaves at the same time were contrasted and spectrum change laws of the same species in different time were analyzed, and the effective bands of species identification by hyperspectral remote sensing under the condition of different time were explored. The results showed that the leaf spectrum of different tree species had significant changes along with time but the changing rules were different, and there were significant differences among different tree species leaves spectrum in the same time, so it proved a theoretical basis for high precision tree species identification. This study aims at providing basic data and theoretical support for tree species identification of hyperspectral remote sensing and the building of leaf spectrum base database.