Evaluating And Optimizing Modis Gross Primary Production Of Maize

There is a great uncertainty in vegetation carbon cycle,which is easily affected by environment,climate,land use change and so on.Gross primary productivity(GPP)is the amount of carbon fixed by vegetation through photosynthesis at unit time and unit area.Therefore,it is of great significance to quantify the vegetation GPP for the simulation of the carbon cycle.The widely used MOD17 model estimates GPP base on the light energy utilization.But there is a great uncertainty in the estimation because original MODIS GPP product often underestimates GPP values.Aiming at this phenomenon,the model input parameters and measured data of 9 maize stations are used for GPP algorithm uncertainty analysis to improve and optimize the GPP model of maize.The main methods and conclusions are as follows:(1)The 8-day MODIS surface reflectance products(MOD09A1)are used to make vegetation index curves through band operation.The time series vegetation index curve is reconstructed by Harmonic Analysis of Time Series,which could remove the outliers caused by cloud and sensor noise.The maize key phenological phase is extracted with the dynamic threshold method based on EVI time series curves and phenological change in different regions of maize is analysed.(2)Analyzing the effects of input parameters derived from three sources(meteorological,biome-specific,and fraction of absorbed photosynthetically active radiation(FPAR)parameters)on the MOD17 model behavior.The results show that all the parameters have different effects on the simulation results.(3)Applying the improved MOD17 model to estimate the 9 maize site GPP in 20 years.Simulation result shows that the seasonal dynamics accord with measured GPP during the growth period.But the estimated GPP is higher than the flux GPP in regreening stage and mature stage,lower in heading stage.(4)The REG-PEM model is used to estimate the GPP of maize then compared with MOD17 model GPP and flux GPP.The result shows that REG-PEM model is more effective and more accurate in GPP estimation.