Chlorophy Remote Sensing Inversion Of Winter Wheat Based On Hyper-spectral And GF-1 Satellite Imagery

Wheat is one of the world's most widely distributed food crops and one of the main crops in China.The chlorophyll content is closely related to the photosynthesis ability and growth state of plants.It is a good indicator of photosynthetic capacity,nutritional physiology and aging process of vegetation.The determination of its content is of great significance to crop growth monitoring,fertilization regulation and yield assessment.By using hyperspectral data and satellite image data to invert parameters closely related to crop growth and yield,such as chlorophyll,the monitoring of crop growth can be realized,and provide guidance and reference value for crop production.In this study,winter wheat in different regions of different growth stages were taken as the research object.Through field experiments,winter wheat hyperspectral data,GF1 satellite image data and chlorophyll content?Chl?were obtained.Through calculation and mathematical statistical analysis,the optimal estimation model of winter wheat chlorophyll content was established,and the spatial inversion and accuracy verification of chlorophyll content in the canopy of winter wheat at jointing stage were conducted with the help of GF-1 image.It provides theoretical basis and technical support for winter wheat growth monitoring and accurate field management.The main conclusions are as follows:?1?The chlorophyll content?Chl?of winter wheat leaves was significantly different at different growth stages,and it showed a gradual increase with the progress of growth process.The chlorophyll content?Chl?of winter wheat canopy increased first and then decreased with the progress of growth.Over all,the Chl values at the leaf scale are greater than the Chl values at the canopy scale.?2?With the increase of chlorophyll content,the reflectance of the original spectrum decreases in the visible region and the reflectance increases in the near-infrared region.The spectral reflectance of the leaves with different chlorophyll levels is higher than that of the canopy,and is more pronounced in the visible light range.At different Chl levels,the red edge features of the leaves are different,and the red edge position is constantly"red shifted"with the increase of Chl,and there is a phenomenon of"double peak"or"multiple peak".With the advancement of the growth period,the spectral reflectance of the leaf is lower and lower in the visible light range,and the reflectance in the near-infrared band is getting higher and higher.Canopy spectral reflectance first decreases and then increases in the visible band,but inversely in the near infrared band.Under different growth stages and different chlorophyll contents,the red edge position of winter wheat canopy was distributed near 735nm,while the red edge position of winter wheat leaves was distributed around 710 nm.?3?In the four growth stages,the correlation between the first derivative and the Chl value are stronger than the correlation between the original spectrum and the Chl value,and the correlation between the leaf spectrum and Chl are stronger than that of the canopy.The sensitive band is selected as the independent variable to invert the leaf Chl,and the accuracy of the model fitting is poor except for the filling period.Choosing a high-correlation"three-sided"parameter to establish a Chl estimation model,at the canopy scale,except for the flowering period optimal model is based on the yellow-edge position?_yellowellow model,the other growth period optimal models are based on?SD_r-SD_b?/?SD_r+SD_b?model;at the leaf scale,the optimal models for jointing,heading,flowering and filling periods are based on?SD_r-SD_y?/?SD_r+SD_y?,SD_r/SD_b,?red,?SD_r-SD_y?/?SD_r+SD_y?.In addition to the filling period,the accuracy of winter wheat Chl estimation model based on three-sided parameters in other growth stages was improved compared with the single factor model.?4?The relationship between the 15 vegetable indices and Chl was analyzed,and 8vegetable indices are selected to construct the Chl single factor estimation model.At the canopy scale,the four optimal models of corresponding growth period are based on OSAVI,PRI,PRI and VOG2.At the leaf scale,the four optimal models of corresponding growth period are based on FD_730/525,FD_730/525,VOG2 and FD_{?730-525?/?730+525?}.Except for the model of canopy scale in the filling stage,which is based on the sensitive band D₇₅₁,the accuracy of other models based on vegetation index are higher than that of the three-sided parameters.These are the best single factor estimation models for winter wheat Chl value.?5?The selected hyperspectral parameters with high accuracy were taken as independent variables,and the partial least squares regression?PLSR?and support vector regression?SVR?methods were used to construct the models,which showed good fitting and prediction accuracy in each growth period.The accuracy of the model constructed by PLSR is better than that of the single factor estimation model.The accuracy of the model constructed by SVR is better than that of the model constructed by PLSR,which is the best model for estimating the Chl content of winter wheat.?6?The spatial inversion and accuracy verification of the chlorophyll content of winter wheat at the jointing stage of the winter wheat with high score satellite image showed that the best model is the model based on GNDVI.The model R² is 0.713,the RMSE is 2.288,and the RE is 4.5%;validate R² is 0.714,RMSE is 2.228,and RE is 4.4%;remote sensing mapping results verify that R² is 0.729,RMSE is 2.446,and RE is 6.0%.It is feasible to use the spectral index to monitor the Chl value of winter wheat based on GF1 satellite data.