Retrieval chlorophyll concentration models could be classified into three categories

namely empirical approach

semi-empirical/analytical approach and analytical approach.Among those algorithms

the semi-empirical/analytical algorithm is applied widely.But as chlorophyll concentration changes

retrieval models as well as parameter in models and retrieval precision are changed accordingly.Different season has different chlorophyll concentration in water body

character of water reflect-spectral and sensitivity band which are better related to chlorophyll concentration also have difference

resulting in the difference in retrieval models and parameters in models for each season.This paper first analyzes the seasonal difference of chlorophyll a（Chla） concentration data which were collected from January to October in 2005

and then analyzes reflect-spectral data which were collected synchronous with water sample from April to October in 2005.We divided the time span into spring

summer and autumn

built different models for the three seasons and compared them

so as to build an optimal model for each season.The research result indicates that: in spring and autumn

the band-ration algorithm has better retrieval precision

among different models

logarithm model

liner model and quadratic equation model has higher relativity;in summer

differential algorithm has better retrieval result

three models built dependent on the algorithm has high relativity.