Influence Of Aggregation Effect On The Forest Canopy BRDF Characteristics

Forests are an essential component of terrestrial ecosystems,and the growth and dynamics of forests can reflect the stability of ecosystems and maintain ecological balance.Surface albedo is the main factor affecting the energy balance of an ecosystem.When studying surface albedo in remote sensing applications,people usually assume that the surface is a Lambertian;it is assumed that the surface reflects uniformly in all directions and that the surface albedo is independent of the incident direction and observation direction.However,the vast majority of features in nature are non-Lambertian,and the reflected radiation from features is anisotropic,meaning it is reflected in all directions and is not uniform in intensity.The reflectance of the forest canopy shows strong anisotropy at different wavelengths and solar-sensor geometries,and the variation of canopy reflectance with observation angle is usually described by the bidirectional reflectance distribution function(BRDF).The high-precision inversion study of vegetation structure parameters has long been of great scientific significance and application value in the field of quantitative remote sensing.Since forest canopy BRDF properties are closely related to vegetation physiological and biochemical parameters and canopy structural parameters,BRDF is widely used to invert key canopy structural parameters,such as leaf area index(LAI),tree height,and clumping index(CI).Among them,LAI and CI are the most important canopy structure parameters for global carbon and water cycle simulations by land surface ecological process models.Previous studies have mainly focused on the development of canopy structure parameter inversion algorithms,for example,how to accurately estimate global vegetation LAI and CI products from BRDF remote sensing data.However,orthorectification studies of canopy BRDF properties from the perspective of canopy structure parameters are still relatively few,and there is especially limited knowledge about how the BRDF properties of forest canopies with complex 3D structures vary with CI.It is challenging to estimate the effect of canopy structure on BRDF using traditional radiative transfer modeling methods,while LESS(Larg E-Scale remote sensing data and image Simulation framework),a3D radiative transfer model based on a ray-tracing algorithm,can effectively simulate canopy directional reflectance while fully considering the multi-component spectral and structural characteristics of vegetation.In view of this,based on the rich ground truth data and MODIS BRDF products provided by the WATER Platform,this paper applied the LESS model to construct a 3D forest scene with a single wood as the basic unit,taking Picea crassifolia as an example,and carried out three aspects of research work:(1)accuracy verification of the LESS model;(2)influence of the aggregation effect on canopy BRDF;(3)sensitivity analysis of canopy BRDF.The following summarizes the thesis’s primary research topic and findings:(1)A three-dimensional forest scene was constructed based on the canopy structure parameter-driven LESS model obtained from the ground sample survey,and the reflectance was calculated using the semi-empirical kernel-driven two-way reflectance model Ross Thick-Li Sparse-Reciprocal(RTLSR)and the MODIS BRDF parameter product(MCD43A1)was used for the LESS model simulated canopy BRDF.These results were compared and validated to test the validity of the canopy BRDF properties simulated by the LESS model.The results showed that the canopy BRDF simulated by the LESS model had high agreement with MODIS BRDF data in the red band(R~2=0.97,RMSE=0.03)and near-infrared band(R~2=0.94,RMSE=0.05),indicating that the LESS model is feasible for studying canopy BRDF properties.Compared with the canopy BRDF estimated based on MCD43A1 data,the forest canopy BRDF simulated by the LESS model also exhibited reasonable hot spot effects,roof effects,and bowl edge effects.(2)The canopy reflectance in the red band decreases continuously with increasing CI,and that in the NIR band increases continuously.The canopy reflectance in the red band is more sensitive to the change in CI than that in the NIR band.Although the strong atmospheric effects in the red band may lead to uncertainties in CI estimation,this study strongly recommends the use of clear-sky observations in the red band for CI inversion,considering the high sensitivity of the red band to CI changes.(3)The study constructed three forest scenes with typical aggregation effects using the LESS model,changed the scene LAI,and analyzed the sensitivity of canopy BRDF to LAI in different CI forest scenes.The results show that the canopy BRF(Bidirectional Reflectance Factor)in the red band decreases continuously with increasing LAI and shows a clear roof-like shape.On the contrary,the canopy BRF in the NIR band kept increasing with the increase in LAI and showed a clear bowl shape.(4)The sensitivity of BRDF to the canopy structure parameters differed among the different observation planes.The analysis results showed that in the principal plane,the canopy BRF in the red band showed an overall decreasing trend with increasing SZA,while the canopy BRF in the NIR band showed an overall increasing trend.In the cross principal plane,with the increase of SZA,the canopy BRF of both red and NIR bands showed a decreasing trend.This conclusion,that the canopy BRF in the red band decreases and the canopy BRF in the NIR band increases with increasing CI is consistent under both observation planes,but the canopy directional reflectance in the principal plane is more differentiated and more favorable for CI inversion.(5)The influence of sun position change on canopy BRDF qualities is crucial when using time-series multi-angle canopy reflectance to invert canopy structural parameters.The study investigates the effect of SZA on canopy BRDF by combining the aggregation effect of forest scenes in the LESS model.The results show that with the increase in SZA,the peak reflectance at the hotspot in the red band shows a decreasing trend,while the peak reflectance at the hotspot in the NIR band shows an increasing trend.The effect of the SZA change is more obvious in the NIR band than in the red band.The multi-angle observations obtained in summer were more suitable for inversion of the canopy aggregation index than those obtained in winter,and the canopy BRDF in the red band was more sensitive to the change of CI than that in the NIR band.The multi-angle reflectance data in the red band obtained under all SZA and VZA conditions contributed to the inversion of CI,while in the NIR band,the multi-angle reflectance data in the main plane obtained during April to September were sensitive to CI changes only for forests with high aggregation(CI<0.6)and insensitive to CI changes for forests with low aggregation(CI>0.6).Therefore,to make full use of satellite data in different seasons,the effect of SZA variation on multi-angle reflectance data needs to be considered.(6)The observation angle has a strong influence on the canopy BRDF characteristics.The relative deviations of canopy reflectance were calculated for different solar zenith angles and observed zenith angles for different aggregation index forest scenes.It was found that the canopy BRDF varied more drastically in the near-infrared band than in the red band with increasing VZA;however,the relative deviation of the canopy BRDF was larger in the red band than in the near-infrared band.The canopy BRDF was insensitive to the variation of CI between0.4 and 0.6 and VZA between 20°and 50°in the red band.On the contrary,canopy BRDF is sensitive to VZA variations greater than 30°in the near-infrared band.The study recommends selecting directional reflectance observations at approximately 40°VZA in the red band and directional reflectance observations near the zenith direction of VZA in the NIR band and using time-series multi-angle optical remote sensing data when performing vegetation canopy structure parameter inversion to reduce the uncertainty of CI estimation results due to seasonal variation in SZA.The results of this thesis provide a scientific basis for the inversion of canopy structure parameters to use BRDF remote sensing products.