Search

Accurate land surface reflectance plays an important role in the accurate inversion of surface parameters, and atmospheric correction plays a decisive role in obtaining accurate reflectance. For GF-1 WFV and GF-6 WFV images, there are two major issues to be addressed, including the spectral differences between nadir with far off-nadir pixels and the spatial variability of atmospheric components for wide imaging. Therefore, this study focuses on these two issues using the Sensor Invariant Atmospheric Correction (SIAC) method. Our results reveal that the SIAC approach improves the correlation accuracy from 0.8868 to 0.9173 for GF-1 WFV image compared with Sentinel-2 reflectance, from 0.9530 to 0.9620 for GF-6 WFV image compared with the results using FLAASH model. For alleviating wide-swathed anisotropy, the directional imaging angle is calculated with the result ranging from 5.6450° to 33.7497°. Furthermore, the atmospheric components have been inversed pixel by pixel with obvious spatial variation. And the correlation of inversed aerosol optical thickness (AOT) and total column water vapor (TCWV) with a spatial resolution of 500 m TCWV with measured results of AERONET (AErosol RObotic NETwork) observation stations are 0.9175 and 0.4442, respectively. These results reveal that the atmospheric correction method works well, which is effective for the wide swath of Chinese GF-1 WFV and GF-6 WFV images on land.

Corn is the dominant crop planted in Northeast China, and its accurate and timely mapping is important for food security and agricultural management in China. However, the absence of enough labels is challenging for corn accurate mapping in a regional area using machine learning methods or deep learning methods. In this study, an efficient way of automatic labelling and mapping of corn planted areas by combining Global Ecosystem Dynamics Investigation (GEDI) data and Sentinel-2 images is proposed. We explore the height and vertical structure differences between corn and other crops derived from GEDI features and generate labels automatically by referencing crop type products and transferring models from historical years. The trained learning networks of automatic labelling from GEDI points and the trained decision trees of the Random Forest (RF) classifier can be transferred to corn mapping in arbitrary target years. The Sentinel-2 features are combined to perform wall-to-wall corn mapping using a random forest algorithm and GEDI-based labels. This approach is used to map corn planted areas in Northeast China from 2019 to 2022, and the classification results are validated using independent labels collected in field campaigns in 2023, published maps, and official statistics. Our classification results reveal that our proposed method achieves high accuracy and robustness with an average overall accuracy of 0.91 validated using testing labels from spatial-type stratified sampling. The correlation coefficient (R2) between our classified result with the official statistical data and published classification results reach 0.96 and 0.98, respectively. These results demonstrate the potential of GEDI data for automatic label collection for vegetation with height difference and provide a new approach for efficient crop mapping on a large-scale.

Disentangling drought impacts on plant photosynthesis is crucial for projecting future terrestrial carbon dynamics. We examined the separate responses of canopy structure and leaf physiology to an extreme summer drought that occurred in 2011 over Southwest China, where the weather was humid and radiation was the main growth-limiting factor. Canopy structure and leaf physiology were, respectively, represented by near-infrared reflectance of vegetation (NIRv) derived from MODIS data and leaf scale fluorescence yield (Φf) derived from both continuous SIF (CSIF) and global OCO-2 SIF (GOSIF). We detected contrasting responses of canopy structure and leaf physiology to drought with a 14.0% increase in NIRv, compared with 12.6 or 19.3% decreases in Φf from CSIF and GOSIF, respectively. The increase in structure resulted in a slight carbon change, due to water deficit-induced physiological constraints. The net ecosystem effect was a 7.5% (CSIF), 1.2% (GOSIF), and −2.96% (EC-LUE GPP) change in photosynthesis. Our study improves understanding of complex vegetation responses of plant photosynthesis to drought and may contribute to the reconciliation of contrasting observed directions in plant responses to drought in cloudy regions via remote sensing.

The bidirectional reflectance distribution function (BRDF) of the land surface contains information relating to its physical structure and composition. Accurate BRDF modeling for heterogeneous pixels is important for global ecosystem monitoring and radiation balance studies. However, the original kernel-driven models, which many operational BRDF/Albedo algorithms have adopted, do not explicitly consider the heterogeneity within heterogeneous pixels, which may result in large fitting residuals. In this paper, we attempted to improve the fitting ability of the kernel-driven models over heterogeneous pixels by changing the inversion approach and proposed a dynamic weighted least squares (DWLS) inversion approach. The performance of DWLS and the traditional ordinary least squares (OLS) inversion approach were compared using simulated data. We also evaluated its ability to reconstruct multiangle satellite observations and provide accurate BRDF using unmanned aerial vehicle observations. The results show that the developed DWLS approach improves the accuracy of modeled BRDF of heterogeneous pixels. The DWLS approach applied to satellite observations shows better performance than the OLS method in study regions and exhibits smaller mean fitting residuals both in the red and near-infrared bands. The DWLS approach also shows higher BRDF modeling accuracy than the OLS approach with unmanned aerial vehicle observations. These results indicate that the DWLS inversion approach can be a better choice when kernel-driven models are used for heterogeneous pixels.

It is imperative to rapidly and precisely acquire crop lodging area and severity for disaster prevention and yield prediction. However, estimation of crop lodging area at a large scale remains challenging due to the relatively low sensitivity of remote sensing signal to the lodging variation, limited availability of remote sensing images, and lodging statistical data. This study proposes a new method for lodging area estimation based on the optimal grid cell of Sentinel-2 and crop lodging percentage, overcoming the limitation of traditional pixel-based mapping approaches that fail to obtain quantitative lodging information. Basing the spatial aggregation method, we analyzed the optimal grid size of Sentinel-2 data for lodging percentage estimation. Then we investigated the spectral response for different lodging percentage levels and analyzed the potential of lodging percentage estimation for Sentinel-2 metrics (including selected spectral bands and their derived vegetation indexes (VIs)). A quantitative model was established between the training set and the Sentinel-2 metrics using the random forest (RF) algorithm. Finally, around 1462.62 ha fields from six counties or districts in Heilongjiang province in China were estimated for lodging percentage. Results indicate that the proposed method can estimate the crop lodging percentage on the testing set with an R2 and RMSE of 0.64 and 25.24, respectively, which can explain around 95 % spatial variation of lodging crop. Moreover, the overall magnitude of reflectance increased with the increase in lodging percentage. Among all Sentinel-2 optimal metrics, the Green, SWIR1, and Red edge 1 bands are the most crucial indicators for lodging percentage estimation. Our results on lodging percentage estimation in the study area indicate that there is more lodging maize in the Meilisidawoerzu district than in other areas. Although typhoons passed over Fuyu and Lindian counties, the lodging percentage in these areas is relatively low. The lodging percentage map has great value in agriculture management and insurance claim.

Spring phenology plays an essential role in climate change, terrestrial ecosystem, and public health. Field-based monitoring and understanding of changes in spring phenology for long periods and in large regions are challenging due to the limited in-site observations. Space-based remotely sensed observations offer great potentials for monitoring decadal spring phenology changes from regional to global scales. However, the coarse-scale remotely sensed observations are insufficient to capture fine-scale spring phenology dynamics, especially in urban areas, and this makes it challenging for understanding the combined effects of climate change and urbanization on spring phenology. We derived the start of phenology season (SOS) in New York State using 30 m Landsat observations from 1990 to 2015 to understand the impact of the environment and urbanization on SOS. The results show that SOS for different years reveals heterogeneous spatial distribution. Most regions of New York State have been experiencing significant spring phenology changes in form of earlier onset of vegetation greening, ranging from 0.2 to 0.6 day/year during 1990 to 2015, and this trend varies slightly with latitudes and urbanization levels. Further, spatial correlation analysis shows that the increase in temperature and urbanization could both promote the advancement of SOS. However, the effect of urbanization (partial correlation coefficient (R) ranges from −0.289 to −0.542) on SOS is greater than the effect of temperature (R ranges from 0.006 to −0.192). The study generates a high spatio-temporal resolution spring phenology dataset for ecological, environmental and public health studies, especially in urban areas, and reveals the importance of better accounting for the urbanization effects when quantifying the SOS dynamics in phenology models.

The management of crop residue covering is a vital part of conservation tillage, which protects black soil by reducing soil erosion and increasing soil organic carbon. Accurate and rapid classification of corn residue-covered types is significant for monitoring crop residue management. The remote sensing technology using high spatial resolution images is an effective means to classify the crop residue-covered areas quickly and objectively in the regional area. Unfortunately, the classification of crop residue-covered area is tricky because there is intra-object heterogeneity, as a two-edged sword of high resolution, and spectral confusion resulting from different straw mulching ways. Therefore, this study focuses on exploring the multi-scale feature fusion method and classification method to classify the corn residue-covered areas effectively and accurately using Chinese high-resolution GF-2 PMS images in the regional area. First, the multi-scale image features are built by compressing pixel domain details with the wavelet and principal component analysis (PCA), which has been verified to effectively alleviate intra-object heterogeneity of corn residue-covered areas on GF-2 PMS images. Second, the optimal image dataset (OID) is identified by comparing model accuracy based on the fusion of different features. Third, the 1D-CNN_CA method is proposed by combining one-dimensional convolutional neural networks (1D-CNN) and attention mechanisms, which are used to classify corn residue-covered areas based on the OID. Comparison of the naive Bayesian (NB), random forest (RF), support vector machine (SVM), and 1D-CNN methods indicate that the residue-covered areas can be classified effectively using the 1D-CNN-CA method with the highest accuracy (Kappa: 96.92% and overall accuracy (OA): 97.26%). Finally, the most appropriate machine learning model and the connected domain calibration method are combined to improve the visualization, which are further used to classify the corn residue-covered areas into three covering types. In addition, the study showed the superiority of multi-scale image features by comparing the contribution of the different image features in the classification of corn residue-covered areas.

Topography significantly complicates the radiative transfer process of vegetation and further causes variation in reflectance observed by remote sensors. Leaf area index (LAI) inversion based on reflectance data is subsequently influenced by topography. Neglecting the topographic effects may lead to large biases when estimating LAI over rugged terrain. How the topography influences the LAI inversion process has rarely been explored. In this article, the topographic effects on LAI inversion over sloped terrain are quantitatively investigated and analyzed based on a dataset generated from the discrete anisotropy radiative transfer model. An artificial neural network (ANN) model is established to represent the flat surface LAI inversion algorithms. Then, the reflectance of sloped terrain is input into the ANN model to obtain the biased LAI inversion values. The results reveal that topography effects on LAI inversion are related to canopy density and generally lead to an underestimation except for sparse canopies. The mean relative bias could reach 51% when the slope angle reaches 60°. The variation trends of inverted LAI are closely related to the local incident angle. The different levels of bias in reflectance at red and near-infrared bands lead to different patterns of inversion errors for different canopies densities. Finally, we compared the existing strategies (geometric correction and topographic correction strategies) designed for LAI inversion over sloped terrain. It is found that these strategies apply in different situations. The results are helpful in understanding the topographic effects and further finding a better strategy for LAI inversion over sloped terrain.

In July 2021, a flooding event, which attracted the attention of the whole country and even the world, broke out in Henan, resulting in dramatic losses across multiple fields (e.g., economic and agricultural). The basin at the junction of Hebi, Xinxiang, and Anyang was the most affected region, as the spread of water from the Wei river submerged surrounding agricultural land (e.g., corn-dominated). To comprehensively evaluate the flooding impacts, we proposed a framework to detect the flooding area and evaluated the degree of loss using satellite time series data. First, we proposed a double-Gaussian model to adaptively determine the threshold for flooding extraction using Synthetic Aperture Radar (SAR) data. Then, we evaluated the disaster levels of flooding with field survey samples and optical satellite images. Finally, given that crops vary in their resilience to flooding, we measured the vegetation index change before and after the flooding event using satellite time series data. We found the proposed double-Gaussian model could accurately extract the flooding area, showing great potential to support in-time flooding evaluation. We also showed that the multispectral satellite images could potentially support the classification of disaster levels (i.e., normal, slight, moderate, and severe), with an overall accuracy of 88%. Although these crops were temporarily affected by this flooding event, most recovered soon, especially for the slightly and moderately affected regions. Overall, the distribution of resilience of these affected crops was basically in line with the results of classified disaster levels. The proposed framework provides a comprehensive aspect to the retrospective study of the flooding process on crops with diverse disaster levels and resilience. It can provide rapid and timely flood damage assessment and support emergency management and disaster verification work.

Efficient, fast, and accurate crop lodging monitoring is urgent for farmers, agronomists, insurance loss adjusters, and policymakers. This study aims to explore the potential of Chinese GF-1 PMS high-spatial-resolution images for corn lodging monitoring and to find a robust and efficient way to identify corn lodging accurately and efficiently. Three groups of image features and five machine-learning approaches are used for classifying non-lodged, moderately lodged, and severely lodged areas. Our results reveal that (1) the combination of spectral bands, optimized vegetation indexes, and texture features classify corn lodging with an overall accuracy of 93.81% and a Kappa coefficient of 0.91. (2) The random forest is an efficient, robust, and easy classifier to identify corn lodging with the F1-score of 0.95, 0.92, and 0.95 for non-lodged, moderately lodged, and severely lodged areas, respectively. (3) The GF-1 PMS image has great potential for identifying corn lodging on a regional scale.

Sun-induced chlorophyll fluorescence (SIF) has shown potential in quantifying plant responses to environmental changes by which abiotic drivers are dominated. However, SIF is a mixed signal influenced by factors such as leaf physiology, canopy structure, and sun-sensor geometry. Whether the physiological information contained in SIF can better quantify crop disease stresses dominated by biological drivers, and clearly explain the physiological variability of stressed crops, has not yet been sufficiently explored. On this basis, we took winter wheat naturally infected with stripe rust as the research object and conducted a study on the responses of physiological signals and reflectivity spectrum signals to crop disease stress dominated by biological drivers, based on in situ canopy-scale and leaf-scale data. Physiological signals include SIF, SIFyield (normalized by absorbed photosynthetically active radiation), fluorescence yield (ΦF) retrieved by NIRvP (non-physiological components of canopy SIF) and relative fluorescence yield (ΦF-r) retrieved by near-infrared radiance of vegetation (NIRvR). Reflectance spectrum signals include normalized difference vegetation index (NDVI) and near-infrared reflectance of vegetation (NIRv). At the canopy scale, six signals reached extremely significant correlations (P < 0.001) with disease severity levels (SL) under comprehensive experimental conditions (SL without dividing the experimental samples) and light disease conditions (SL < 20%). The strongest correlation between NDVI and SL (R = 0.69) was observed under the comprehensive experimental conditions, followed by NIRv (R = 0.56), ΦF-r (R = 0.53) and SIF (R = 0.51), and the response of ΦF (R = 0.45) and SIFyield (R = 0.34) to SL was weak. Under lightly diseased conditions, ΦF-r (R = 0.62) showed the strongest response to disease, followed by SIFyield (R = 0.60), SIF (R = 0.56) and NIRv (R = 0.54). The weakest correlation was observed between ΦF and SL (R = 0.51), which also showed a result approximating NDVI (R = 0.52). In the case of a high level of crop disease severity, NDVI showed advantages in disease monitoring. In the early stage of crop diseases, which we pay more attention to, compared with SIF and reflectivity spectrum signals, ΦF-r estimated by the newly proposed ‘NIRvR approach’ (which uses SIF together with NIRvR (i.e., SIF/ NIRvR) as a substitute for ΦF) showed superior ability to monitor crop physiological stress, and was more sensitive to plant physiological variation. At the leaf scale, the response of SIF to SL was stronger than that of NDVI. These results validate the potential of ΦF-r estimated by the NIRvR approach to monitoring disease stress dominated by biological drivers, thus providing a new research avenue for quantifying crop responses to disease stress.

Vegetation indices are widely used to derive land surface phenology (LSP). However, due to inconsistent illumination geometries, reflectance varies with solar zenith angles (SZA), which in turn affects the vegetation indices, and thus the derived LSP. To examine the SZA effect on LSP, the MODIS bidirectional reflectance distribution function (BRDF) product and a BRDF model were employed to derive LSPs under several constant SZAs (i.e., 0°, 15°, 30°, 45°, and 60°) in the Harvard Forest, Massachusetts, USA. The LSPs derived under varying SZAs from the MODIS nadir BRDF-adjusted reflectance (NBAR) and MODIS vegetation index products were used as baselines. The results show that with increasing SZA, NDVI increases but EVI decreases. The magnitude of SZA-induced NDVI/EVI changes suggests that EVI is more sensitive to varying SZAs than NDVI. NDVI and EVI are comparable in deriving the start of season (SOS), but EVI is more accurate when deriving the end of season (EOS). Specifically, NDVI/EVI-derived SOSs are relatively close to those derived from ground measurements, with an absolute mean difference of 8.01 days for NDVI-derived SOSs and 9.07 days for EVI-derived SOSs over ten years. However, a considerable lag exists for EOSs derived from vegetation indices, especially from the NDVI time series, with an absolute mean difference of 14.67 days relative to that derived from ground measurements. The SOSs derived from NDVI time series are generally earlier, while those from EVI time series are delayed. In contrast, the EOSs derived from NDVI time series are delayed; those derived from the simulated EVI time series under a fixed illumination geometry are also delayed, but those derived from the products with varying illumination geometries (i.e., MODIS NBAR product and MODIS vegetation index product) are advanced. LSPs derived from varying illumination geometries could lead to a difference spanning from a few days to a month in this case study, which highlights the importance of normalizing the illumination geometry when deriving LSP from NDVI/EVI time series.

Satellite remote sensing of vegetation at regional to global scales is undertaken at considerable variations in solar zenith angle (SZA) across space and time, yet the extent to which these SZA variations matter for the retrieval of phenology remains largely unknown. Here we examined the effect of seasonal and spatial variations in SZA on retrieving vegetation phenology from time series of the Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) across a study area in southeastern Australia encompassing forest, woodland, and grassland sites. The vegetation indices (VI) data span two years and are from the Advanced Himawari Imager (AHI), which is onboard the Japanese Himawari-8 geostationary satellite. The semi-empirical RossThick-LiSparse-Reciprocal (RTLSR) bidirectional reflectance distribution function (BRDF) model was inverted for each spectral band on a daily basis using 10-minute reflectances acquired by H-8 AHI at different sun-view geometries for each site. The inverted RTLSR model was then used to forward calculate surface reflectance at three constant SZAs (20°, 40°, 60°) and one seasonally varying SZA (local solar noon), all normalised to nadir view. Time series of NDVI and EVI adjusted to different SZAs at nadir view were then computed, from which phenological metrics such as start and end of growing season were retrieved. Results showed that NDVI sensitivity to SZA was on average nearly five times greater than EVI sensitivity. VI sensitivity to SZA also varied among sites (biome types) and phenological stages, with NDVI sensitivity being higher during the minimum greenness period than during the peak greenness period. Seasonal SZA variations altered the temporal profiles of both NDVI and EVI, with more pronounced differences in magnitude among NDVI time series normalised to different SZAs. When using VI time series that allowed SZA to vary at local solar noon, the uncertainties in estimating start, peak, end, and length of growing season introduced by local solar noon varying SZA VI time series, were 7.5, 3.7, 6.5, and 11.3 days for NDVI, and 10.4, 11.9, 6.5, and 8.4 days for EVI respectively, compared to VI time series normalised to a constant SZA. Furthermore, the stronger SZA dependency of NDVI compared with EVI, resulted in up to two times higher uncertainty in estimating annual integrated VI, a commonly used remote-sensing proxy for vegetation productivity. Since commonly used satellite products are not generally normalised to a constant sun-angle across space and time, future studies to assess the sun-angle effects on satellite applications in agriculture, ecology, environment, and carbon science are urgently needed. Measurements taken by new-generation geostationary (GEO) satellites offer an important opportunity to refine this assessment at finer temporal scales. In addition, studies are needed to evaluate the suitability of different BRDF models for normalising sun-angle across a broad spectrum of vegetation structure, phenological stages and geographic locations. Only through continuous investigations on how sun-angle variations affect spatiotemporal vegetation dynamics and what is the best strategy to deal with it, can we achieve a more quantitative remote sensing of true signals of vegetation change across the entire globe and through time.

Global biophysical products at decametric resolution derived from Sentinel-2 imagery have emerged as a promising dataset for fine-scale ecosystem modeling and agricultural monitoring. Evaluating uncertainties of different Sentinel-2 biophysical products over various regions and vegetation types is pivotal in the application of land surface models. In this study, we quantified the performance of Sentinel-2-derived Leaf Area Index (LAI), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR), and Fractional Vegetation Cover (FVC) estimates using global ground observations with consistent measurement criteria. Our results show that the accuracy of vegetation and non-vegetated classification based on Sentinel-2 surface reflectance products is greater than 95%, which indicates the vegetation identification is favorable for the practical application of biophysical estimates, as several LAI, FAPAR, and FVC retrievals were derived for non-vegetated pixels. The rate of best retrievals is similar between LAI and FAPAR estimates, both accounting for 87% of all vegetation pixels, while it is almost 100% for FVC estimates. Additionally, the Sentinel-2 FAPAR and FVC estimates agree well with ground-measurements-derived (GMD) reference maps, whereas a large discrepancy is observed for Sentinel-2 LAI estimates by comparing with both GMD effective LAI (LAIe) and actual LAI (LAI) reference maps. Furthermore, the uncertainties of Sentinel-2 LAI, FAPAR and FVC estimates are 1.09 m2/m2, 1.14 m2/m2, 0.13 and 0.17 through comparisons to ground LAIe, LAI, FAPAR, and FVC measurements, respectively. Given the temporal difference between Sentinel-2 observations and ground measurements, Sentinel-2 LAI estimates are more consistent with LAIe than LAI values. The robustness of evaluation results can be further improved as long as more multi-temporal ground measurements across different regions are obtained. Overall, this study provides fundamental information about the performance of Sentinel-2 LAI, FAPAR, and FVC estimates, which imbues our confidence in the broad applications of these decametric products.

Substantial uncertainty exists in daily and sub-daily gross primary production (GPP) estimation, which dampens accurate monitoring of the global carbon cycle. Here we find that near-infrared radiance of vegetation (NIR _v,Rad ), defined as the product of observed NIR radiance and normalized difference vegetation index, can accurately estimate corn and soybean GPP at daily and half-hourly time scales, benchmarked with multi-year tower-based GPP at three sites with different environmental and irrigation conditions. Overall, NIR _v,Rad explains 84% and 78% variations of half-hourly GPP for corn and soybean, respectively, outperforming NIR reflectance of vegetation (NIR _v,Ref ), enhanced vegetation index (EVI), and far-red solar-induced fluorescence (SIF _760 ). The strong linear relationship between NIR _v,Rad and absorbed photosynthetically active radiation by green leaves (APAR _green ), and that between APAR _green and GPP, explain the good NIR _v,Rad -GPP relationship. The NIR _v,Rad -GPP relationship is robust and consistent across sites. The scalability and simplicity of NIR _v,Rad indicate a great potential to estimate daily or sub-daily GPP from high-resolution and/or long-term satellite remote sensing data.

The development of near-surface remote sensing requires the accurate extraction of leaf area index (LAI) from networked digital cameras under all illumination conditions. The widely used directional gap fraction model is more suitable for overcast conditions due to the difficulty to discriminate the shaded foliage from the shadowed parts of images acquired on sunny days. In this study, a new LAI extraction method by the sunlit foliage component from downward-looking digital photography under clear-sky conditions is proposed. In this method, the sunlit foliage component was extracted by an automated image classification algorithm named LAB2, the clumping index was estimated by a path length distribution-based method, the LAD and G function were quantified by leveled digital images and, eventually, the LAI was obtained by introducing a geometric-optical (GO) model which can quantify the sunlit foliage proportion. The proposed method was evaluated at the YJP site, Canada, by the 3D realistic structural scene constructed based on the field measurements. Results suggest that the LAB2 algorithm makes it possible for the automated image processing and the accurate sunlit foliage extraction with the minimum overall accuracy of 91.4%. The widely-used finite-length method tends to underestimate the clumping index, while the path length distribution-based method can reduce the relative error (RE) from 7.8% to 6.6%. Using the directional gap fraction model under sunny conditions can lead to an underestimation of LAI by (1.61; 55.9%), which was significantly outside the accuracy requirement (0.5; 20%) by the Global Climate Observation System (GCOS). The proposed LAI extraction method has an RMSE of 0.35 and an RE of 11.4% under sunny conditions, which can meet the accuracy requirement of the GCOS. This method relaxes the required diffuse illumination conditions for the digital photography, and can be applied to extract LAI from downward-looking webcam images, which is expected for the regional to continental scale monitoring of vegetation dynamics and validation of satellite remote sensing products.

The primary restriction on high resolution remote sensing data is the limit observation frequency. Using a network of multiple sensors is an efficient approach to increase the observations in a specific period. This study explores a leaf area index (LAI) inversion method based on a 30 m multi-sensor dataset generated from HJ1/CCD and Landsat8/OLI, from June to August 2013 in the middle reach of the Heihe River Basin, China. The characteristics of the multi-sensor dataset, including the percentage of valid observations, the distribution of observation angles and the variation between different sensor observations, were analyzed. To reduce the possible discrepancy between different satellite sensors on LAI inversion, a quality control system for the observations was designed. LAI is retrieved from the high quality of single-sensor observations based on a look-up table constructed by a unified model. The averaged LAI inversion over a 10-day period is set as the synthetic LAI value. The percentage of valid LAI inversions increases significantly from 6.4% to 49.7% for single-sensors to 75.9% for multi-sensors. LAI retrieved from the multi-sensor dataset show good agreement with the field measurements. The correlation coefficient (R2) is 0.90, and the average root mean square error (RMSE) is 0.42. The network of multiple sensors with 30 m spatial resolution can generate LAI products with reasonable accuracy and meaningful temporal resolution.

Physically-based approaches for estimating Leaf Area Index (LAI) using remote sensing data rely on radiative transfer (RT) models. Currently, many RT models are freely available, but determining the appropriate RT model for LAI retrieval is still problematic. This study aims to evaluate the necessity of RT model selection for LAI retrieval and to propose a retrieval methodology using different RT models for different vegetation types. Both actual experimental observations and RT model simulations were used to conduct the evaluation. Each of them includes needleleaf forests and croplands, which have contrasting structural attributes. The scattering from arbitrarily inclined leaves (SAIL) model and the four-scale model, which are 1D and 3D RT models, respectively, were used to simulate the synthetic test datasets. The experimental test dataset was established through two field campaigns conducted in the Heihe River Basin. The results show that the realistic representation of canopy structure in RT models is very important for LAI retrieval. If an unsuitable RT model is used, then the root mean squared error (RMSE) will increase from 0.43 to 0.60 in croplands and from 0.52 to 0.63 in forests. In addition, an RT model’s potential to retrieve LAI is limited by the availability of a priori information on RT model parameters. 3D RT models require more a priori information, which makes them have poorer generalization capability than 1D models. Therefore, physically-based retrieval algorithms should embed more than one RT model to account for the availability of a priori information and variations in structural attributes among different vegetation types.

A sampling strategy to define elementary sampling units (ESUs) for an entire site at the kilometer scale is an important step in the validation process for moderate-resolution leaf area index (LAI) products. Current LAI-sampling strategies are unable to consider the vegetation seasonal changes and are better suited for single-day LAI product validation, whereas the increasingly used wireless sensor network for LAI measurement (LAINet) requires an optimal sampling strategy across both spatial and temporal scales. In this study, we developed an efficient and robust LAI Sampling strategy based on Multi-temporal Prior knowledge (SMP) for long-term, fixed-position LAI observations. The SMP approach employed multi-temporal vegetation index (VI) maps and the vegetation classification map as a priori knowledge. The SMP approach minimized the multi-temporal bias of the VI frequency histogram between the ESUs and the entire site and maximized the nearest-neighbor index to ensure that ESUs were dispersed in the geographical space. The SMP approach was compared with four sampling strategies including random sampling, systematic sampling, sampling based on the land-cover map and a sampling strategy based on vegetation index prior knowledge using the PROSAIL model-based simulation analysis in the Heihe River basin. The results indicate that the ESUs selected using the SMP method spread more evenly in both the multi-temporal feature space and geographical space over the vegetation cycle. By considering the temporal changes in heterogeneity, the average root-mean-square error (RMSE) of the LAI reference maps can be reduced from 0.12 to 0.05, and the relative error can be reduced from 6.1% to 2.2%. The SMP technique was applied to assign the LAINet ESU locations at the Huailai Remote Sensing Experimental Station in Beijing, China, from 4 July to 28 August 2013, to validate three MODIS C5 LAI products. The results suggest that the average R2, RMSE, bias and relative uncertainty for the three MODIS LAI products were 0.60, 0.33, −0.11, and 12.2%, respectively. The MCD15A2 product performed best, exhibiting a RMSE of 0.20, a bias of −0.07 and a relative uncertainty of 7.4%. Future efforts are needed to obtain more long-term validation datasets using the SMP approach on different vegetation types for validating moderate-resolution LAI products in time series.

Many applications, including crop growth and yield monitoring, require accurate long-term time series of leaf area index (LAI) at high spatiotemporal resolution with a quantification of the associated uncertainties. We propose an LAI retrieval approach based on a combination of the LAINet observation system, the Consistent Adjustment of the Climatology to Actual Observations (CACAO) method, and Gaussian process regression (GPR). First, the LAINet wireless sensor network provides temporally continuous field measurements of LAI. Then, the CACAO approach generates synchronous reflectance data at high spatiotemporal resolution (30-m and 8-day) from the fusion of multitemporal MODIS and high spatial resolution Landsat satellite imagery. Finally, the GPR machine learning regression algorithm retrieves the LAI maps and their associated uncertainties. A case study in a cropland site in China showed that the accuracy of LAI retrievals is 0.36 (12.7%) in terms of root mean square error and R2 = 0.88 correlation with ground measurements as evaluated over the entire growing season. This paper demonstrates the potential of the joint use of newly developed software and hardware technologies in deriving concomitant LAI and uncertainty maps with high spatiotemporal resolution. It will contribute to precision agriculture, as well as to the retrieval and validation of LAI products.

How to obtain the “truth” of land surface parameter as reference value to validate the remote sensing retrieved parameter in heterogeneous scene and coarse-resolution pixel is one of the most challenging topics in environmental studies. In this paper, a distributed sensor network system named CPP-WSN was established to capture the spatial and temporal variation of land surface parameters at coarse-resolution satellite pixel scale around the Huailai Remote Sensing Station, which locates in the North China Plain. The system consists of three subnetworks that are RadNet, SoilNet, and VegeNet. Time series observations of typical land surface parameters, including UVR, PAR, SWR, LWR, albedo, and land surface temperature (LST) from RadNet, multilayer soil moisture and soil temperature from SoilNet, and fraction of vegetation cover (FVC), clumping index (CI), and leaf area index (LAI) from VegeNet, have been obtained and shared on the web. Compared with traditional single-point measurement, the “true” reference value of coarse pixel is obtained by averaging or representativeness-weighted averaging the multipoint measurements acquired using the sensor network. The preliminary applications, which validate several remote sensing products with CPP-WSN data, demonstrate that a high quality ground “truth” dataset has been available for remote sensing as well as other applications.

Earth’s reflectivity is among the key parameters of climate research. National Aeronautics and Space Administration (NASA)’s Earth Polychromatic Imaging Camera (EPIC) onboard National Oceanic and Atmospheric Administration (NOAA)’s Deep Space Climate Observatory (DSCOVR) spacecraft provides spectral reflectance of the entire sunlit Earth in the near backscattering direction every 65 to 110 min. Unlike EPIC, sensors onboard the Earth Orbiting Satellites (EOS) sample reflectance over swaths at a specific local solar time (LST) or over a fixed area. Such intrinsic sampling limits result in an apparent Earth’s reflectivity. We generated spectral reflectance over sampling areas using EPIC data. The difference between the EPIC and EOS estimates is an uncertainty in Earth’s reflectivity. We developed an Earth Reflector Type Index (ERTI) to discriminate between major Earth atmosphere components: clouds, cloud-free ocean, bare and vegetated land. Temporal variations in Earth’s reflectivity are mostly determined by clouds. The sampling area of EOS sensors may not be sufficient to represent cloud variability, resulting in biased estimates. Taking EPIC reflectivity as a reference, low-earth-orbiting-measurements at the sensor-specific LST tend to overestimate EPIC values by 0.8% to 8%. Biases in geostationary orbiting approximations due to a limited sampling area are between − 0.7 % and 12%. Analyses of ERTI-based Earth component reflectivity indicate that the disagreement between EPIC and EOS estimates depends on the sampling area, observation time and vary between − 10 % and 23%.

This paper presents a simple radiative transfer model based on spectral invariant properties (SIP). The canopy structure parameters, including the leaf angle distribution and multi-angular clumping index, are explicitly described in the SIP model. The SIP model has been evaluated on its bidirectional reflectance factor (BRF) in the angular space at the radiation transfer model intercomparison platform, and in the spectrum space by the PROSPECT+SAIL (PROSAIL) model. The simulations of BRF by SIP agreed well with the reference values in both the angular space and spectrum space, with a root-mean-square-error (RMSE) of 0.006. When compared with the widely-used Soil-Canopy Observation of Photochemistry and Energy fluxes (SCOPE) model on fPAR, the RMSE was 0.006 and the R2 was 0.99, which shows a high accuracy. This study also suggests the newly proposed vegetation index, the near-infrared (NIR) reflectance of vegetation (NIRv), was a good linear approximation of the canopy structure parameter, the directional area scattering factor (DASF), with an R2 of 0.99. NIRv was not influenced much by the soil background contribution, but was sensitive to the leaf inclination angle. The sensitivity of NIRv to canopy structure and the robustness of NIRv to the soil background suggest NIRv is a promising index in future biophysical variable estimations with the support of the SIP model, especially for the Deep Space Climate Observatory (DSCOVR) Earth Polychromatic Imaging Camera (EPIC) observations near the hot spot directions.

Spatial heterogeneity is present in the land surface at every scale and is one of the key factors that introduces inherent uncertainty into simulations of land surface processes and parameter retrieval based on remotely sensed data. Because of a lack of understanding of the heterogeneous characteristics of global mixed pixels, few studies have focused on modeling and inversion algorithms in heterogeneous areas. This paper presents a parameterization scheme to describe land cover heterogeneity quantitatively by composition and boundary information based on high-resolution land cover products. Global heterogeneity features at the 1-km scale are extracted from the ‘GlobeLand30’ land cover dataset with a spatial resolution of 30 m. The composition analysis of global mixed pixels shows that only 35% of pixels over the land surface of Earth are covered by a single land cover type, namely, pure pixels, and only 25.8% are located in vegetated areas. Pixels mixed with water are more common than pixels mixed with any other non-vegetation type. The fragmentation analysis of typical biomes based on the boundary length shows that the savanna is the most heterogeneous biome, while the evergreen broadleaf forest is the least heterogeneous. Deciduous needleleaf forests are significantly affected by canopy height differences, while crop and grass biomes are less affected. Lastly, the strengths and limitations of the method and the application of the land cover heterogeneity characteristics extracted in this study are discussed.

Topographic correction methods rarely consider the canopy parameter effects directly and explicitly for sloping canopies. In order to address this problem, the topographic correction method MFM-GOST2 was developed by implementing the second version of the Geometric-Optical model for Sloping Terrains (the GOST2 model) in the multiple forward mode (MFM) inversion framework. First, a look up table (LUT) was constructed by multiple forward modeling of the GOST2 model; second, the radiance of a remotely sensed image and its corresponding topographic data were used for searching potential canopy parameter combinations from the LUT; and third, the corrected radiance was determined by averaging potential radiances of horizontal canopies from the LUT according to the canopy parameter combinations. The MFM-GOST2 and twelve generally used topographic correction methods were evaluated via a case study by visual analysis, linear relationship analysis, and the rose diagram analysis. The result showed that the MFM-GOST2 method successfully removed most of the topographic effects of a subset image of the Landsat-8 image in a case study. The case study also illustrates that the rose diagram analysis is a good way to evaluate topographic corrections, but the linear relationship analysis cannot be used independently for the evaluations because the decorrelation is not a sufficient condition to determine a successful topographic correction.

The operational Moderate Resolution Imaging Spectroradiometer (MODIS) Leaf Area Index (LAI) and Fraction of Photosynthetically Active Radiation absorbed by vegetation (FPAR) algorithm has been successfully implemented for Visible Infrared Imager Radiometer Suite (VIIRS) observations by optimizing a small set of configurable parameters in Look-Up-Tables (LUTs). Our preliminary evaluation showed reasonable agreement between VIIRS and MODIS LAI/FPAR retrievals. However, there is a need for a more comprehensive investigation to assure continuity of multi-sensor global LAI/FPAR time series, as the preliminary evaluation was spatiotemporally limited. In this study, we use a multi-year (2012–2016) global LAI/FPAR product generated from VIIRS and MODIS to evaluate for spatiotemporal consistency. We also quantify uncertainty of the product by utilizing available ground measurements. For both consistency and uncertainty evaluation, we account for variations in biome type and temporal resolution. Our results indicate that the LAI/FPAR retrievals from VIIRS and MODIS are consistent at different spatial (i.e., global and site) and temporal (i.e., 8-day, seasonal and annual) scales. The estimate of mean discrepancy (−0.006 ± 0.013 for LAI and −0.002 ± 0.002 for FPAR) meets the stability requirement for long-term LAI/FPAR Earth System Data Records (ESDRs) from multi-sensors as suggested by the Global Climate Observing System (GCOS). It is noteworthy that the rate of retrievals from the radiative transfer-based main algorithm is also comparable between two sensors. However, a relatively larger discrepancy over tropical forests was observed due to reflectance saturation and an unexpected interannual variation of main algorithm success was noticed due to instability in input surface reflectances. The uncertainties/relative uncertainties of VIIRS and MODIS LAI (FPAR) products assessed through comparisons to ground measurements are estimated to be 0.60/42.2% (0.10/24.4%) and 0.55/39.3% (0.11/26%), respectively. Note that the validated LAI were only distributed in low domains (~2.5), resulting in large relative uncertainty. Therefore, more ground measurements are needed to achieve a more comprehensive evaluation result of product uncertainty. The results presented here generally imbue confidence in the consistency between VIIRS and MODIS LAI/FPAR products and the feasibility of generating long-term multi-sensor LAI/FPAR ESDRs time series.

Increasing attention is being paid on leaf area index (LAI) retrieval in mountainous areas. Mountainous areas present extreme topographic variability, and are characterized by more spatial heterogeneity and inaccessibility compared with flat terrain. It is difficult to collect representative ground-truth measurements, and the validation of LAI in mountainous areas is still problematic. A cost-constrained sampling strategy (CSS) in support of LAI validation was presented in this study. To account for the influence of rugged terrain on implementation cost, a cost-objective function was incorporated to traditional conditioned Latin hypercube (CLH) sampling strategy. A case study in Hailuogou, Sichuan province, China was used to assess the efficiency of CSS. Normalized difference vegetation index (NDVI), land cover type, and slope were selected as auxiliary variables to present the variability of LAI in the study area. Results show that CSS can satisfactorily capture the variability across the site extent, while minimizing field efforts. One appealing feature of CSS is that the compromise between representativeness and implementation cost can be regulated according to actual surface heterogeneity and budget constraints, and this makes CSS flexible. Although the proposed method was only validated for the auxiliary variables rather than the LAI measurements, it serves as a starting point for establishing the locations of field plots and facilitates the preparation of field campaigns in mountainous areas.