Research On The Retrieval Of Fraction Of Absorbedphotosphotosynthetically Active Radiation (FPAR) Based On FY-3C MERSI Data

FAPAR / FPAR(Fraction of Absorbed Photosynthetically Active Radiation)is the ratio of Photosynthetically active radiation(PAR)absorbed by plant leaves and canopy that can be used for the accumulation and conversion of biomass to the total PAR,and it is a basic variable parameter that describes the energy exchange process of internal structure of the plant and the canopy-atmosphere material,so it is very important to obtain FPAR accurately.In this thesis,the model and algorithm of FPAR inversion are used as research topics,and focusing on the FPAR inversion strategy suitable for FY-3C Medium Resolution Imaging Spectrometer(MERIS)data,then provide reference and basis for the generation of large-area and long-term sequence FPAR products for FengYun-3C(FY-3C)data.The main work and achievements of the thesis are as follows:(1)Selection of models and algorithms for the inversion of FPAR.Analyze common vegetation canopy physical models,including Radiation Transfer models,such as PROSAIL models,ACRM models,Geometric Optical models and computer simulation models;and analyze common inversion algorithms,including optimization algorithms,neural network algorithms,Look-up table algorithm to study the relatively more practicable models and algorithms suitable for the inversion of FPAR from FengYun data.According to the applicability and advantages and disadvantages of the model and algorithm,and referring to the universality of the inversion method,this thesis selects the PROSAIL model and the look-up table(LUT)algorithm.(2)The strategy of FPAR inversion from FengYun data.Using the Heihe experimental area to analyze the applicability of PROSAIL model and LUT algorithm for the inversion of FPAR.The sensitivity of the input parameters of the PROSAIL model needs to be analyzed before the inversion.Based on the same model and algorithm,the inversion of FPAR from FY-3C MERSI data and MOD09 surface reflectance data have been generated,and combining MODIS FPAR products the three FPAR results compare with the Heihe FPAR products published by the Scientific Data Center for Cold and Arid Region.The correlation coefficients(R)were 0.5746,0.8922,and 0.7991;and the Root Mean Square Error(RMSE)were 0.2029,0.0848,0.0929.The results show that the correlation of FPAR from MOD09 surface reflectance data based on the research method is higher then MODIS FPAR products,so the feasibility of the model and algorithm is proved.At the same time,the correlation coefficients(R)of FPAR from the FY-3C MERSI data is low,so the inversion strategy of FPAR needs to be optimized for the characteristics of the FengYun data.It mainly includes the Pop convolution from the spectral response function of FengYun data,the classification of vegetation,and the optimization of the look-up table.The solution is to replace the minimum RMSE value with the average of the first 10% of the RMSE in the cost function as the final FPAR inversion value.(3)Application and verification analysis based on the strategy of FPAR inversion from FY-3C MERSI data.The typical research area of 5 types of grassland,cropland,shrub,broadleaf forest and needleleaf forest was used for batch inversion using the strategy of FPAR inversion to analyze the seasonal variation characteristics of FPAR of different vegetation types.Using the method of cross-validation,the FPAR product data of MODIS,GEOV1 and GLASS were selected for accuracy verification.The results show that the FPAR retrieved from FY-3C MERSI data has good spatial and temporal consistency and correlation with the three FPAR products.The difference between FY-FPAR and other three FPAR products is concentrated around 0(+0.4 interval)in the spatial distribution;the time series curves of four FPAR products are close to each other and consistent with the seasonal growth of vegetation;and the FY-FPAR is relatively more suitable for the inversion of cropland and grassland during the lush period in the correlation degree.Therefore,the inversion strategy provides the application value for quantitative inversion of vegetation parameters from FY-3C data.