Characteristic Of Primary Shapes Implied In Directional Reflectance And Model Construction Over Typical Vegetation Covers

As the inherent attribute of land cover,the heterogeneity of directional reflectance brings challenge to the normalization of remote sensing data,meanwhile enriches the methods of extracting land cover structure parameters.The continuous accumulation of multi-angle observations from laboratory,ground,near ground and spaceborne remote sensing provides effective priori knowledge for the construction of multi-angle forward model,backward inversion of land parameters and the correction of angle effect for data,especially the investigation for the reflection characteristic in principal plane.Currently,the spatial resolution of main satellite sensors with multiangle observation varies from 500 m to 6km and the standard operational BRDF data products are mainly limited by several datasets such as MODIS and POLDER.There is still a lack of multiangle observations and data products with medium or high resolution.Thus,how to use the existing priori knowledge with lower resolution effectively in the applications of quantitative remote sensing with medium or high resolution is an urgent problem to be solved.Several classical BRDF shape indicators contain the characteristics of directional reflectance for land covers to a certain degree,but there are still some defects in capturing hot spot and the detailed primary shape characteristics.This work studies the key shape characteristics of the directional reflectance for land covers by investigating the change characteristics of directional reflectance for land covers in multiple spatiotemporal scales,constructs BRDF shape vector models,compares the constructed models with other shape indicators to test the performance,verifies their potential abilities to capture land surface heterogeneity in application cases.The main research contents are as follows:(1)We selected six typical vegetation covers which are ENF,Cropland,Grassland,Savannas,DBF,EBF sites for their representativeness in structure.Based on standard operational MCD43 anisotropic data products,the characteristics of primary shapes implied in directional reflectance for various vegetation covers in three broad bands of Near infrared(0.7-5.0μm),Short wave(0.3-5.0μm),Visible(0.3-0.7μm)were investigated.We studied the change characteristics of directional reflectance in temporal scales from daily to 8 days,16 days,month and spatial scales from500 m to 5600 m and their impact on the estimation of albedo.(2)We selected multiple feature points as the basic for the modeling in the principal plane for its rich heterogeneity characteristics.Using the basic change law of segmented primary shape and the primary shape angles for feature points to construct the 6-component PAV(Partial Anisotropic Vector)and 3-component AEV(Angular Effect Vector)models,which could describe the overall characteristics of the reflection primary shape and the bluntness or sharpness of feature points.Compared with the classical shape indicators,the effectiveness of vector models was evaluated from both theoretical and experimental aspects.The cosine similarity and the average error obtained from the error transfer function were used to verify the representativeness of the shape vectors compared to the original primary shape.(3)The hot spot component of the vector model was used to investigate the growth cycle of land vegetation and the dynamic change characteristics of vegetation cover in the process of ecological restoration of Haizhou mining area in Fuxin.The potentiality of hot spot component to capture the characteristic of heterogeneity for land covers and the structure change in the restoration process of vegetation after reclamation in Haizhou mining area were verified,meanwhile the ability of NDVI index to adjust BRDF’s local shape was evaluated.The main conclusions of this study are as follows:(1)The heterogeneity characteristics of the vegetation covers vary with the spatiotemporal scales,both the magnitude and shape in principal plane have changed to some extent.With the scale increases from 8 days to 16 days and month,the average difference of magnitude assessed by overall reflectance for principal plane in the three broad bands is 0.15%,0.22%,0.36%,and the average difference of shape assessed by the dispersion of reflectance variation is 0.121,0.170,0.241.Meanwhile,the changes can induce the differences that can be captured on the subsequent estimation of albedo.(2)The results show that vector models are effective in describing the overall primary shape,reflecting scattering effect weight and the capture of local variations compared with classical shape indicators.The correlation between the vector model based on the slope of reflectance and the local variation(99.11%)is significantly higher than the classical indicators based on the ratio of reflectance(16.13%).In Near infrared,Short wave and Visible band,the correlation between the hot spot component of the vector model and the scattering effect weight reaches 0.936,0.945,0.863,the representativeness of PAV to the original primary shape is 0.980,0.979,0.969 and the representativeness of AEV is 0.987,0.991,0.994 on average.(3)The hot spot component of AEV has the potentiality to capture the growth cycle of land vegetation and monitor the change of land cover in the mining area.The change of hot spot angle in the restoration process of vegetation after reclamation in Haizhou mining area is related to the variation of LAI parameter for land cover.This verified that the change of primary shape angle for hot spot could provide a possible method for the extraction of the characteristics of vegetation structure,while the NDVI constructed by the reflectance in Red and Near infrared band is difficult to establish a strong correlation with the BRDF of land cover in the internal mechanism.There are 67 figures,11 tables,and 134 references.