Considering the simplicity of flight route planning, orthorectified images obtained from nadir observations are widely used in remote sensing. However, they are always insufficient to represent the anisotropic reflectance and three-dimensional (3D) structural information of objects. Therefore, multi-angle observation information can enhance target information and potentially improve the accuracy of target classification and recognition. In this study, we investigated the potential of anisotropic reflectance information in land cover classification. By employing the DJI P4M multispectral observation system, multi-angle multi-spectral reflectance images for five land cover types were captured at bare soil, concrete roads, grassland, apricot tree, and red broom cypress areas. Subsequently, the AFX-based BRDF archetypes model and the kernel-driven model were used to reconstruct the bidirectional reflectance distribution function (BRDF). Finally, land cover classification was performed using three types of machine learning algorithm considering different BRDF features and band combinations. The results indicate that, compared to nadir directional reflectance, multi-angle feature sets can improve the overall classification accuracy up to 24%. Compared to using single-band information, band combinations can also improve that up to 54%. The overall accuracy using the feature set of kernel-driven model parameters and nadir reflectance was also enhanced significantly, which can reach 86% using green-red-near infrared band combinations. This work demonstrates the contribution of multi-angle multi-spectral information to natural and artificial land cover classification.

The anisotropic reflectance characteristic (i.e., bidirectional reflectance distribution function (BRDF) effect) lays the foundation of quantitative remote sensing, while it is difficult to reconstruct at high spatial resolution due to near-nadir small-angle observations. Based on the traditional archetype inversion algorithm, this study explores and evaluates a two-step method considering vegetation structure clustering to retrieve fine-scale BRDF, using LESS simulated sufficient multiangle reflectances in forest scenes at multiple scales. First, full inversion of the kernel-driven model RTLSR_C was performed to investigate the BRDF feature at 1–50 m. Subsequently, the sensitivity of normalized difference vegetation index (NDVI) and anisotropic flat index (AFX) to forest structure was comprehensively analyzed. For reflectances at small view zenith angles (VZAs) of 0° and 15° close to Sentinel multispectral instrument (MSI) observations at 5, 10, and 25 m, forest structure cluster-based BRDF parameters were first retrieved based on accumulated intraclass reflectances, where prior BRDF information from 50-m LESS data and 500-m MODIS global representative sites were compared. Finally, cluster-based BRDF parameters were used as prior archetypes to retrieve pixel BRDF. Results show good fitting root mean square errors (RMSEs) for full inversion, with most average RMSEs below 0.05, and spectral and angular reflectance are sensitive to the fraction of vegetation cover (FVC), tree height, and crown length at 5 m and larger scales. In addition, the two-step method based on structure clustering yields a slightly higher BRDF inversion accuracy than that of the one-step method, and the traditional one-step archetype inversion algorithm demonstrates its simplicity and accuracy. This study provides new insights for fine-scale BRDF inversion.

Earth observation technology plays an important role in military reconnaissance, agriculture and forestry plant protection, emergency disaster relief, etc. In this paper, we mainly apply a prototype inversion algorithm: based on the multi-angle multi-spatial scale data of three scenes including trees, buildings and mixed objects simulated by the three-dimensional radiative transfer model LESS, we take the 500m/pixel coarse-resolution BRDF as an archetype priori information to reconstruct the high-resolution multi-angle information at 10m/pixel level. Then, sensitive feature indices are used to categorize the reconstructed 10m images, and the simulated 10m images are used as standard values to evaluate the classification accuracy. The results show that the average accuracies of BRDF inversion for the three scenes are 96.2%, 50.08% and 71.43%, respectively, and the plant class is more suitable for this inversion model. In terms of the classification accuracies, the three scenes are 60.52%, 96.84% and 76.60%, respectively, with building scene shows the highest accuracy.

The surface reflectance anisotropy is usually described by bidirectional reflectance distribution function (BRDF), and the BRDF archetypes extracted based on the Anisotropic Flat Index (AFX) have been used as the prior reflectance anisotropy knowledge of quantitative inversion. In this study, we introduce a new index, i.e., the Perpendicular Anisotropic Flat Index (PAFX), based on the AFX in a prior study, to refine the BRDF classification of the land surface. Based on the MODIS BRDF parameter products sampled at global intervals over multiple land cover types in 2015, we used AFX and PAFX as classification indicators to divide MODIS BRDF parameter space into some orthogonal clusters through use of the ISODATA (Iterative Self organizing data Analysis technique) Clustering Algorithm. As a case study, the MODIS BRDF parameters are divided into three classes by using the AFX and PAFX, respectively, in an orthogonal way, and therefore a total of nine cluster classes of BRDF parameters are generated, correspondingly generating nine BRDF archetypes. We use the average BRDF fitting error, algorithm complexity and classification accuracy to evaluate this classification method for classifying BRDF parameters, showing that the difference within the BRDF archetypes class is significantly reduced after the introduction of PAFX (RMSE = 0.0083) compared with only using AFX (RMSE = 0.0139), and the error for fitting all BRDF shapes with the average BRDF shape, i.e., the BRDF archetype presents a V-shaped trend as functions of AFX and PAFX, indicating an optimized minimum. In general, jointing the AFX and PAFX has improved the classification accuracy of the MODIS BRDF parameter space and is expected a further application in near future.

Clumping index (CI) field measurements based on the logarithmic gap fraction averaging (LX) method are widely used. However, some challenges regarding this method have been recognized; e.g., CI overestimation or underestimation occurs in the sampling units where there is no measurement gap, which creates major uncertainties in CI field measurements. To address this issue, we proposed an improved eight-connected LX method that replaces null gap units with the arithmetic mean of the gaps in the eight connected neighbouring units, considering the neighbouring connections of the natural foliage extension. To validate this method, we designed two controlled experimental schemes based on simulated digital hemispheric photography (DHP) images through the LargE-Scale Remote Sensing Data and Image Simulation Framework (LESS) model considering the leaf area index (LAI) and leaf angle distribution (LAD), respectively, together with collected field measurements. The results showed that our method could almost prevent overestimation and improve underestimated CIs by nearly 20%. In addition, CIs of our method had the smallest error compared to the “true” CIs (error<0.1). In conclusion, our method can significantly improve the data quality of the simulation results relative to the existing methods and present potentials in the CI measurements of upcoming field campaigns.

Foliage Clumping Index (CI) is an important structural parameter within vegetation canopy, and current satellite-borne CI products mainly retrieved by using the linear relationship between the CI and the normalized difference between hotspot and dark spot (NDHD), while there is no directly model to calculate the CI of mixed forest. The objective of this paper is to propose a new method to calculate the mixed forest CI (MFCI) and access the ability to response the spatial heterogeneity of mixed forest pixels. The results show that: (1) The accuracy of MFCI is significantly higher than that of existing MODIS CI products, the average error can be reduced by 6.3% (2) the total sensitivity of MFCI to spatial heterogeneity is high(>0.6), and with the highest sensitivity to bare soil.

The surface anisotropy, usually described as bidirectional reflectance distribution function (BRDF), plays a key role in the quantitative remote sensing. Numerous BRDF studies are focus on sensors onboard polar-orbiting satellites such as POLDER and MODIS based on multi-angle sensors or accumulative observations through multiple days. Notably, sensors onboard geostationary satellites can also obtain multi-angle reflectances benefited from their high revisit frequencies, while by which only a few BRDF studies have been completed. In this study, we aim to evaluate the BRDF information collected from time-series directional observations of the Advanced Himawari Imager (AHI) onboard geostationary satellite Himawari-8. The multi-angle reflectances of 6 × 7 km POLDER and 0.05° AHI at mixed forest area during a whole month were collected. Generally, AHI has a good weight of Determination (WoD) of 0.03, as well as small fit-RMSEs of 0.0055 and 0.0235 in the red and NIR bands based on the kernel-driven Ross-Li BRDF model, which shows promising potential to provide BRDF information with a good quality.

Bidirectional Reflectance Distribution Function (BRDF) is an important component in quantitative remote sensing. In this study, we explored the relationship of sampling distribution and reconstructed BRDF curve for snow surface. In order to get enough observations, we utilize the field-measured data to analyze the different BRDF under different sampling pattern. This work can be very meaningful because it is hard to acquire sufficient measurements in all directions especially for snow in the real life.

Clumping index (CI) characterizes the clumping degree of vegetation canopy foliage relative to the random distribution, which is an important vegetation structure parameter. Digital Hemispherical Photography (DHP) is widely used in ground CI measurement and one of the key steps using this method is to determine the sampling resolution. This study presents a set of sampling way including 30 sampling methods with 17 levels of sampling resolutions. They were applied to four vegetation types and different growth stages of crops to explore the variation of CI with the decrease of sampling resolution of 17 levels. The results show that with the decrease of sampling resolution at 17 levels, the average increase of CI in the four vegetation types was 26%, 29%, 14% and 35%, and in different growth stages of soybean crop, the increase of CI was different which could be up to 60% at the maximum increase.

Clumping Index (CI) is an important vegetation structure parameter, which describes the grouping of leaves relative to the random distribution. Multi-angle data of the POLarization and Directionality of Earth Reflectance (POLDER) sensor (about 6×7 km) and the MODerate resolution Imaging Spectradiometer (MODIS) (500 m) are two main sources for global CI products. To better understand the variability inherent in CIs of such different spatial resolutions and optimize the used of CI products, extensive POLDER CIs and corresponding MODIS CIs were retrieved and compared in this study. Field measurements were conduct in one selected POLDER pixel in Hebei, China. Our results showed that POLDER and MODIS CIs presented relative good consistency (R2=0.65, RMSE=0.07, bias=0.003), and the correlation coefficient can reach 0.98 at the class level. Both POLDER CI (0.67) and MODIS CI (0.62±0.06) showed good consistency with field CIs (0.64±0.10) and POLDER CI was more likely to overestimate than MODIS CI.

Surface albedo is closely related to the Earth’s energy budget and is usually estimated by integrating remotely sensed bidirectional reflectance distribution function (BRDF) data based on the widely used Ross–Li kernel-driven models. However, for large solar zenith angles (i.e., SZAs > 70°), albedo estimation using the operational algorithm of the Moderate Resolution Imaging Spectroradiometer (MODIS), i.e., RossThick-LiSparseReciprocal (RTLSR), is not recommended because it is reported to somewhat underestimate the black-sky albedo (BSA) at large SZAs based on ground albedo measurements. Recently, various combinations of the Ross–Li BRDF models with improved capabilities have been developed, and the assessments of these models based on worldwide satellite BRDF data with good spatial sampling, particularly at the large view and solar angles, will be important to improve an understanding of their performance in estimating intrinsic albedos. Following previous studies, the objective of this study is to further assess a series of hotspot-corrected Ross–Li models by demonstrating their ability to fit the POLarization and Directionality of the Earth’s Reflectances (POLDER) data sets and estimate albedo, especially at large SZAs, based on selected concurrent POLDER and MODIS data. The hotspot-corrected RTLSR model obtained by combining the RossThickChen and LiSparseReciprocalChen kernels (RTLSR_C) shows the best fitting ability, with a high cumulative frequency of small root-mean-square errors (RMSEs), thus confirming previous conclusions. Model differences mainly appear in albedo estimates, especially BSA estimates at large SZAs. The BSAs estimated by other models are significantly different from the RTLSR_C estimates in the near-infrared (NIR) and red bands as the SZA increases to approximately 60° and 70°, respectively. In this case, RossThinChen-LiSparseReciprocalChen (RTNLSR_C) yields higher BSA estimates than those of RTLSR_C. Comparisons of the MODIS and POLDER albedos estimated with Ross–Li models show that models with the RossThinChen kernel yield higher BSA estimates than those of the RTLSR_C model as the SZA increases. The results indicate that the retrieved albedo is likely to be more accurate with appropriately selected kernels for BRDF models at large SZAs, providing guidance for selecting suitable combinations of multiple kernels.

Anisotropy information from moderate-to-coarse-resolution sensors [e.g., 500-m Moderate Resolution Imaging Spectroradiometer (MODIS)] is widely applied to estimate high-resolution surface albedo. Simulated albedos using MODIS bidirectional reflectance distribution function (BRDF) parameters as prior knowledge based on the kernel-driven model are employed to build and assess the lookup table (LUT) of the direct-estimation method, which is then used to estimate high-resolution albedos directly from top-of-atmosphere (TOA) reflectance data (e.g., Landsat albedo). Previously, the errors in the simulated albedos were not considered in building and assessing the LUT. In this article, daytime time-series (30 min) of snow-free albedo measurements with sufficient solar zenith angles (SZAs) were introduced to build the LUT for snow-free Landsat TM surface shortwave broadband albedo (TM albedo) retrievals, together with TOA-simulated reflectance by concurrent daily MODIS BRDF parameters. The assessment utilizes an independent data set and shows larger discrepancies between the estimated and measured albedos [root-mean-square errors (RMSEs) of >0.03 at SZAs ≥ 60°] than those in previous articles. To reduce inconsistencies between the MODIS BRDF parameters and the observed albedos, as well as possible spatial resolution differences between the MODIS and Landsat data, we adopted a correction strategy that first linearly adjusts the MODIS BRDF parameters to match the albedo measurements by a magnitude method, and second, the TOA reflectance simulations were further corrected by concurrent TM reflectances. The developed algorithm shows a significant improvement after using such corrections as a priori (RMSE <; 0.02 at SZA ≤ 75°). The validation indicates improved accuracies in the TM albedo estimation. These improvements may provide potential albedo estimations for nadir-viewing high-resolution sensors using coarse-resolution anisotropy information.

Surface albedo is a key parameter controlling the earth energy budget, which can be estimated by integrating the Bidirectional Reflectance Distribution Function (BRDF). The semi-empirical kernel-driven BRDF models has been widely used in BRDF/Albedo products, MODIS products for instance (Schaaf et al., 2002). However, these albedo products are suspect under large SZA (Liu et al., 2009). With the hotspot improved kernel-driven models, it is necessary to assess the property of these models. In this study, two POLDER datasets are utilized to access these models by root-mean-square error (RMSE) and relative RMSE (RMSE_r). Then, cross-comparison between albedos estimated by improved models under several SZAs is analyzed by POLDER and the concurrence MODIS pixels. This study is aimed at choosing suitable models for albedo estimation under different angular situations to retrieve more accurate albedo.

Leaf area index (LAI) is an important structure parameter to illuminate the fractions of solar radiation absorbed, transmitted and reflected by the plant canopy, and also a useful reference for ecological and meteorological modeling. The GLAS full-waveform Lidar data of ICESat satellite are easily available and global coverage, which can also provide detailed forest canopy structure information in the GLAS footprint. In this study, we show a LAI estimation method from the GLAS waveform Lidar data at footprint level. Firstly, Gaussian decomposition method is used to process the raw GLAS waveform data to identify ground echo energy and canopy echo energy. In addition, the optical height threshold (HT) to separate the canopy and ground in the GLAS waveform has been discussed, and the result show that 3 m is the optical HT in our study area. Secondly, a reflectance correction method is used to calculate the laser penetration ratio (PC) of forest covered GLAS footprints based on the ground echo energy and canopy echo energy. Thirdly, the relationship between the between the field-measured LAIs and PCis constructed based on the Beer-Lambert law. The determination coefficient (R2) is 0.69 and the root mean square error (RMSE) is 0.64. The performance of the GLAS-derived LAIs is also evaluated using the 20 field-measured LAIs. The result indicates that the GLAS-derived LAIs have a high accordance with the field measurements (R2=0.67, RMSE=0.52). The result suggests that the GLAS waveform data can be used to retrieval LAI for various ecological applications.

Recently, Jiao et al. developed a snow kernel in the kernel-driven bidirectional reflectance distribution function (BRDF) model framework to better characterize the anisotropic reflectance of pure snow surface. In this study, we assess performances of this snow kernel in the kernel-driven model framework and original kernel-driven model (hereinafter named the RTS and RTR models) in retrieving snow albedo based on the bicontinuous photon tracking (bic-PT) model. Our results show that: (1) The spectral albedo retrieved by these two models has a high consistency with the bic-PT model. However, the results of the spectral albedo for RTR model has a slight underestimation, especially at SZA=0° in the visible bands, and the RTS model performs well compared with simulated data. (2) The albedo retrieved by these two models agrees reasonably well with the simulated data (R2=~0.9). Yet, the result of the RTR model underestimates 0.50% and 0.52% compared simulated albedo in the red and near-infrared bands, respectively, and the RTS model has a negligible bias for all bands. This assessment provide a priori knowledge of these two models performances, and thus, suggests the RTS model can be applied to future researches of estimating snow albedo.

The foliage Clumping Index (CI) is a key vegetation structure parameter for leaf area index (LAI) estimating and ecological modelling. Previously, several global CI products have been retrieved from the Collection V005 Moderate Resolution Imaging Spectroradiometer (MODIS) Bidirectional Reflectance Distribution Function (BRDF) products with a temporal resolution of one year, one month or 8 day. The Collection V006 MODIS BRDF products provide a chance to retrieve a global CI product with higher temporal resolution and data accuracy. However, the large size of the Collection V006 MODIS BRDF products (~150 terabyte from January 2001 to December 2017) and the retrieved CI products (~9 terabyte) increases the difficulty in retrieving and publishing the global CI product. In this study, we develop a software tool that enable users to produce CI product of their desired date, region and temporal resolution based on the Collection V006 MODIS land cover type and BRDF products. The software tool reduces the requirements of the processing and storage capacity for researchers and thus facilitate the publication and widespread application of the CI product.

Clumping index (CI) is an important vegetation structure parameter in the estimation of leaf area index (LAI) and the modeling of ecological and meteorological process. With the development of surface process modeling and remote sensing technology, high resolution CI product is urgently needed but no appropriate high resolution multi-angle reflectance satellite data is currently available to produce such product. In recent years, random forest algorithm has been widely used in the derivation of high resolution products from remote sensing data. In this study, the random forest algorithm was used to estimate Landsat CI basing on MODIS and field data. The developed predictive model was validated using 26 field measurements and the predicted CI shown a good consistency with the field CI (R2=0.63, bias=0.005, RMSE=0.10).

The surface anisotropy plays a key role in the quantitative remote sensing inversion, which is usually described as bidirectional reflectance distribution function (BRDF). Studies show that BRDF sampling has a significant effect on parameter inversion such as albedo. However, BRDF samplings are complex, and only specific samplings were considered in previous studies. In this study, we investigated the sensitivity of BRDF sampling to albedo and the normalized difference between hotspot and dark spot (NDHD) angular index based on the kernel-driven Ross-Li BRDF model. Albedo and NDHD calculated by a set of dense sampling airborne data were used as the reference data, and inversion results from many sparse samplings were compared to the reference results. The result shows the overall number, plane, range and symmetry in observing condition of BRDF sampling can affect albedo and NDHD estimation. Among typical sensors, POLDER shows best sampling while Landsat shows largest errors.

The hotspot effect is a typical angular reflectance signature of vegetation canopies and contains important information for the retrieval of vegetation structural parameters. To date, the hotspot effect of various analytical bidirectional reflectance distribution function (BRDF) models (e.g., the PROSAIL model) has rarely been assessed by multiangular measurements with sufficient hotspot observations due to the lack of accurate hotspot measurements (for field measurements) or appropriate methods (for airborne and spaceborne measurements). In this paper, we develop a method to further improve the hotspot effect of the kernel-driven model and design a framework to utilize the improved kernel-driven model as a bridge to assess the hotspot effect of the PROSAIL model with Polarization and Directionality of the Earth Reflectance (POLDER) hotspot observations. The results indicate that the proposed method further improves the fits between the models and the observations in the vicinity of the hotspot direction, particularly in the rare situations where the geometric-optical scattering component governs the target reflectance. In addition, the hotspot signature indicated by the PROSAIL multiangular data shows a larger variability than that of POLDER observations. C1and C2in the improved kerneldriven model can be used as benchmarked parameters to qualify the amplitude and width of the hotspot effect for the simulated multiangular data of physical BRDF models and thus present the potential for the assessment and analysis of the hotspot effect of physical models, which, in return, helps retrieve the structural parameters of vegetation canopies from hotspot signatures.

It has been a hot study field to extract forest structure parameter using Airborne LiDAR. Since footprints of Airborne LiDAR data are discontinuously distributed with small data coverage, therefore, it is impossible to obtain the forest structure information of continuous region using Airborne LiDAR data alone. The MODIS BRDF shape indicators contain the information regarding 3-D structure of forest and have the possibility to retrieve the structural parameters of forest. In this study, we select Howland Forest, Harvard Forest, La Selva Forest and Bartlett Forest as experimental areas, and aim to construct a canopy height estimation model from the airborne Laser Vegetation Imaging Sensor (LVIS) data and MODIS BRDF shape indicators. Firstly, H100 canopy height was extracted from the LVIS data and the MODIS BRDF shape indicators were calculated based on MODIS data. Secondly, using the Random Forest algorithm to develop a canopy height estimation model with H100 canopy height data and MODIS BRDF shape indicators. Finally, 10-fold cross-validation method is used to evaluate the accuracy of the model, and the validation results show that the MODIS BRDF shape indicators can be estimated forest canopy heights in high accuracy.

The foliage Clumping Index (CI) quantifies the level of foliage grouping within a distinct canopy structure relative to a random distribution. It is a key structure parameter for the ecological, hydrological, and land surface models. In this study, we investigate the influence of snow cover on the seasonal variation of global CI products derived from the Moderate-resolution Imaging Spectroradiometer (MODIS) Bidirectional Reflectance Distribution Function (BRDF) parameter products using the improved RTCLSR kernel-driven model. Results indicated that the cover of snow can lead to a much larger CI and thus considerably decrease the quality of the CI product. Statistics in 2006 indicates that more than 85% low quality pixels are covered by the snow. The average CI for evergreen needleleaf forests in winter will decrease about 0.1 after deducing the influence of snow covered pixels. The influence of snow cover should be carefully considered and corrected when analyzing the seasonal variation of the global CI product.

Semi-empirical kernel-driven Bidirectional Reflectance Distribution Function (BRDF) model has been developed to retrieve the BRDF shapes of the observed surface from multi-angle measurements. At present, hemispherical diffuse irradiance is usually blended in the multi-angle measurements. The blend of diffuse irradiance will smooth the intrinsic BRDF shapes of observed surface. Therefore, there is a need to correct the diffuse irradiance to get the ideal BRDF shapes when the multi-angle measurements is processed by the kernel-driven model. In this article, we develop a method for kernel-driven model to correct the blended hemispherical diffuse irradiance in the multi-angle measurements. Multi-angle data simulated by the PROSAIL model are used to validate the efficiency of the method. The result indicates that the simulated reflectance after correction agree well with the measurements without diffuse irradiance.

The surface anisotropy characteristics have important significance in the quantitative remote sensing inversion. The kernel-driven model can express the surface anisotropy well and widely used in remote sensing, and the PROSAIL model is a mature vegetation canopy model which can describe complex vegetation structure, therefore studying surface anisotropy variance of the two models is a key point to combine them for further research. We simulate surface reflectance data with complex vegetation structure through the PROSAIL model, with the RossThick-LiSparseR(RTLSR) model and its extended model of Chen(RTCLSR) considering hotspot effect, we analyze anisotropy variance. The result shows: (1) The overall fitting effect is good, the average fitting RMSE is about 0.0071 in red band and 0.0342 in near infrared band; (2) AFX is sensitive to some vegetation structure parameters; (3) C1 and C2 in Chen model is inversely proportional to each other in different Hspot, while proportional in different LAI.

Normalized Difference Vegetation Index (NDVI) is one of the most widely used vegetation indexes because the NDVI of typical type of land coverage has a clear distinction on a large-scale image, especially has a valid highlight for vegetation. Anisotropic reflectance characteristics of natural land surface affect the retrieval of NDVI, which means the variation of solar zenith angle and view zenith angle leads to different values of NDVI. This paper uses kernel-driven models to simulate the reflectance of different solar and view zenith angles to calculate the corresponding NDVI values and contrasts the influences of bidirectional reflectance on NDVI under different degrees of the coverage of surface.

The semi-empirical kernel-driven Bidirectional Reflectance Distribution Function (BRDF) models have been widely used in many remotely sensed BRDF/albedo products such as MODIS products for their simplicity and physical interpretation[1]. Based on anisotropic flat index (AFX) derived from the model[2], magnitude inversion algorithm which takes BRDF archetypes as prior knowledge have been proposed[3]. In order to validate the ability of AFX to indicate land surface anisotropy, the relations between AFX and BRDF shapes of several typical land cover types are described in this paper. Besides, we compare the values of prototype inversion algorithm based on AFX with the original full inversion algorithm. Finally, we study relations between AFX and LAI with the simulation data of the two-layer canopy reflectance model of KUUSK to apply AFX, which may guide future research of retrieving the vegetation parameters.