Study On Canopy Reflectance Modeling Of Row Crops

Canopy reflectance modeling is a basic study on vegetation remote sensing,which provides core algorithms for physics-based inversion of the biophysical parameter.In agricultural research,the crop canopy is divided into continuous canopy and row canopy.In the continuous canopy,its canopy structure is simple,hence its study is relatively mature.Unlike the continuous canopy of crops,the canopy of row crops has a row structure that reflects crop growth.Therefore,the development of canopy reflectance modeling of row crops and the establishment of accurate and reliable canopy reflectance models will improve the accuracy of seasonal biophysical parameters in remote sensing inversion.In the quantitative remote sensing of vegetation,the reflectance model of row crops has received more and more attention.According to the physical mechanism,canopy reflectance models can be divided into three approaches: radiative transfer(RT),geometric optics(GO),and computer simulation.RT and GO approaches have higher calculation accuracy and the fastest computational time.Therefore,RT and GO approaches are widely used in remote sensing inversion.The previous RT models for row crops ignored the horizontal radiative transfer,which had led to the imbalance of the vegetation radiation energy,causing the poor performance of the reflectance simulation at large viewing zenith angles.Moreover,in previous GO models for row crops,ignoring multiple scattering can lead to an underestimation of reflectance.Although previous studies had considered multiple scattering contribution in modeling forest radiative transfer,row crops have their unique geometric characteristics.Therefore,the modeling approach originally applied to forests cannot be directly applied to row crops.Based on the above limitation,this thesis attempts to study on canopy reflectance modeling of row crops,the specific content and results are as follows:(1)To address the poor performance of the reflectance simulation at large viewing zenith angles,this study developed a modified four-stream radiative transfer(MFS)equations describing the canopy closure(row)that considers horizonal radiative transfer based on the classic four-stream radiative transfer equations.Meanwhile,for the between-row area,an integral form of radiative transfer equation was also derived.After solving the established quation(equations),the MFS model was developed by considering both the single scattering and multiple scattering contributions.The MFS model was validated using both computer simulation results and in-situ measurements.The validation showed that the MFS model can simulate the multi-angle reflectance of the crop at different growth stages,and the accuracy of the simulated results from the MFS model was comparable to computer simulation(RMSE<0.019).Moreover,the MFS model can be successfully used to simulate the reflectance of continuous(RMSE=0.012)as well as row crop canopies(RMSEs <0.018).Based on the above results,it is shown that the MFS model addressed this limitation of poor simulation at the large view zenith angle in the previous model.(2)To address an underestimation of reflectance caused by ignoring multiple scattering contribution,this study first considered the overlapping relationship of the gap probabilities involved in single scattering contribution.Based on the accurate calculation of single scattering contribution,this study introduced the adding method and the solution of the integral form of radiative transfer equation into row modeling,and derived the equations of multiple scattering suitable for GO approach.Based on these modifications,this study developed a Row model that can accurately describe the single scattering and multiple scattering contributions in the row crops.The model was validated using both computer simulations and in-situ measurements.The validation showed that the Row model can be used to simulate the multi-angle reflectance of the crop at different growth stages,and their accuracy were comparable to computer simulation(RMSEs<0.0404).In addition,the Row model can accurately simulate single scattering near hotspot(RMSEs <0.0057)and multiple scattering(RMSEs <0.0062).Based on the above results,it is shown that on the basis of addressing calculation deviation of the single scattering near the hotspot,the Row model developed based on the GO approach accurately calculated the multiple scattering contribution.(3)In an application example,this study used OMIS-II(Operational Modular Imaging Spectrometer-II)and OLI(Operational Land Imager)images as data sources.The canopy reflectance model of row crops proposed in this study was used to invert fractional vegetation cover vegetation(FVC)of row crops.This study attempted to address the simulated suboptimal phenomenon caused by the continuous crop models in the high spatial resolution images with obvious row structure.Comparing to the common algorithms,the results showed that the mean absolute deviation of the MFS model is 3%,the mean absolute deviation of the Row model is 17.9%.The results reinforced the conclusion that the canopy reflectance of row models proposed in this study has good practical application in remote sensing inversion of vegetation parameters for row crops.