Study On Retrieving Leaf Area Index Of Paddy Land Based On Time Series GF-1 Images And PROSAIL Model

The tasseled cap triangle(TCT)–leaf area index(LAI)isoline model interprets the distribution of LAI isoline in the spectral space constituted by reflectance of red and near-infrared(NIR)band,Comparing to the statistical relationship models,it is potential to retrieve LAI with higher accuracy based on the TCT-LAI isoline model.The distribution of LAI contours in Red-NIR spectral space(slope and intercept)is affected by vegetation coverage,while the vegetation coverage of rice at different growth stages is different.Study on the trajectory characteristics of reflectance in the red and near-infrared wavebands of paddy land during the whole growth period.Establishing the TCT-LAI contour model suitable for the whole growing period of paddy land is of great significance for large-scale inversion of LAI,rice field growth monitoring and agricultural yield estimation.Based on PROSAIL model and time series measured data,the applicability of PROSAIL model to paddy land at different growing period was verified by introducing vegetation coverage information.Combined with the Sobol global sensitivity analysis,model suitability analysis and measured data ranges,a combination model of PROSAIL model parameters suitable for remote sensing inversion of paddy land was proposed,and the numerical model of triangular tasseled cap of red-near infrared reflectance and leaf area index were established during the entire growth period of paddy fields.Based on this,the rice leaf area index of GF-1 remote sensing image of time series was retrieved,and the accuracy of model and LAI inversion accuracy were verified by measured LAI.The main conclusions are as follows:(1)During the entire growth period of rice fields,the leaf area index first increased and then decreased.The content of chlorophyll fluctuates continuously during the whole growth period,and reaches the peak at the filling stage and heading stage.The average leaf inclination angle first increased and then decreased and then increased during the entire growth period.The height of the plant increased first and then decreased or tended to be stable.The time series NDVI of rice paddy field based on the measured spectrum and the time sequence NDVI trajectories based on image pixel are both increasing and then decreasing,and the maximum value is reached at the jointing heading stage.(2)The reflectance of Landsat 8 OLI paddy field in the same time phase has linear correlation with the reflectance of the red near infrared band of GF-1 paddy field,and the correlation coefficient is 0.65 and 0.83 respectively.(3)When the vegetation coverage is large,the matching results of the measured canopy spectral curve(taking into account the vegetation coverage)and the simulated spectrum curve are good.The RMSE of the matching quantification value is less than0.04,and the two curves are also nearly the same in shape.When the vegetation coverage is small,the growth background(water or wet soil)of rice field has a greater impact on the canopy spectrum,and the matching results between the simulated spectrum curve and the measured spectrum curve are poor,and the corresponding RMSE value is also large.(4)Chlorophyll content is the main factor affecting canopy reflectance in the visible light range.The total sensitivity to canopy reflectance is about 85%.Leaf area index,average leaf inclination and leaf dry matter content are the three main parameters affecting canopy reflectance in the near-infrared region.The total sensitivity of leaf area index and average leaf inclination to canopy reflectivity exceeds 80%.(5)The inverted leaf area index of rice field with GF-1 FWV image first increased and then decreased during the whole growth period of the paddy field,reaching the maximum value at the shooting-Booting stage,and the LAI variation characteristics were in line with the growth and development process of rice.(6)There is a good linear correlation between LAI measured in Changle Town and LAI based on measured spectrum inversion.The correlation coefficient R~2 is 0.92;the measured LAI in Dongyuan Township is less related to the LAI inversion,and R~2 is0.61.As a whole,the retrieval accuracy of the look-up table for the tasseled cap triangle(TCT)–leaf area index(LAI)isoline distribution pattern of the paddy field,which is suitable for the whole growth period,is high.The LAI values of individual pixels in the rice field in the observation point corresponding to the GF-1 WFV image are consistent with the measured LAI values in the whole time series,and they both show the tendency of increasing first and then decreasing,and reach the maximum at jointing heading stage.