The leaf area index (LAI) is a key structural parameter of vegetation canopies. Accordingly, several moderate-resolution global LAI products have been produced and widely used in the field of remote sensing. However, the accuracy of the current moderate-resolution global LAI products cannot satisfy the requirements recommended by the LAI application communities, especially in heterogeneous areas composed of mixed land cover types. In this study, we propose a mixed-pixel correction (MPC) method to improve the accuracy of LAI retrievals over heterogeneous areas by considering the influence of heterogeneity caused by the mixture of different biome types with the help of high-resolution land cover maps. The DART-simulated LAI, the aggregated Landsat LAI, and the site-based high-resolution LAI reference maps are used to evaluate the performance of the MPC method. The results indicate that the MPC method can reduce the influences of spatial heterogeneity and biome misclassification to obtain the LAI with much better accuracy than the Moderate Resolution Imaging Spectroradiometer (MODIS) main algorithm, given that the high-resolution land cover map is accurate. The root mean square error (RMSE) (bias) decreases from 0.749 (0.486) to 0.414 (0.087), while the R2 increases from 0.084 to 0.524, and the proportion of pixels that fulfill the uncertainty requirement of the GCOS increases from 38.2% to 84.6% for the results of site-based high-resolution LAI reference maps. Spatially explicit information about vegetation fractional cover can further reduce uncertainties induced by variations in canopy density for the results of DART simulated data. The proposed method shows potential for improving global moderate-resolution LAI products.

High spatiotemporal resolution time series of leaf area index (LAI) are essential for monitoring crop dynamics and validating coarse-resolution LAI products. The optical satellite sensors at decametric-resolution have historically suffered from a long revisit cycle and cloud contamination issues that hampered the acquisition of frequent and high-quality observations. The 16-m/4-day resolution of the new generation Gaofen-1 (GF-1) and Gaofen-6 (GF-6) satellites provide an unprecedented opportunity to address these limitations. Here we developed an effective strategy to generate daily 16-m LAI maps combing GF-1/6 data and ground LAINet measurements. All high-quality GF-1/6 observations were utilized first to derive smoothed time series of vegetation indices (VIs). Second, a random forest regression (RF-r) model was trained to link the VIs with corresponding field LAI measurements. The trained RF-r was finally employed to generate the LAI maps. Results demonstrated the reliability of the reconstructed daily VIs (relative error2of 0.05, 0.59 and 0.75, respectively. The LAI time series well captured the spatiotemporal variation of crop growth. Furthermore, the continuous GF-1/6 LAI maps outperformed Sentinel-2 LAI estimates both in terms of temporal frequency and accuracy. Our study indicates the potential of GF-1/6 to generate continuous decametric-resolution LAI maps for fine-scale agricultural monitoring.

Vegetation index (VI) derived from remotely sensed images is a proxy of terrestrial vegetation information and widely used in land monitoring and global change studies. Recently, the prediction of vegetation properties has been an interest in related communities. With the accumulation of satellite records over the past few decades, the spatial–temporal prediction of VI becomes feasible. In this letter, we developed deep recurrent neural networks (RNNs) with long short-term memory (LSTM) and gated recurrent units (GRUs) to predict the short-term VI based on historical observations. The pixel-based fully connected networks GRU and LSTM (FCGRU and FCLSTM) and patch-based convolutional networks (ConvGRU and ConvLSTM) are established and compared with the traditional multilayer perceptron (MLP) model. Moderate Resolution Imaging Spectroradiometer (MODIS) and Sentinel-2 normalized difference VI (NDVI) data sets were used in the experiments. The prediction performance is evaluated globally in different regions, different vegetation types, and different growing seasons. Results demonstrate that the RNN models can predict VI with high accuracy (average root mean square error (RMSE) around 0.03), which is superior to the MLP model. In general, the pixel-based RNN models performed better than the patch-based models especially in regions with a larger proportion of outliers. And the prediction accuracy is stable over different vegetation types and growing seasons.

High-accuracy estimates of fractional vegetation cover (FVC) are vital for regional-scale vegetation growth monitoring. In this context, a question worth exploring is whether FVC estimation methods developed at 300-m to 1-km spatial resolution are suitable for finer spatial resolution satellite images. This study compared the performances of three types of algorithms [i.e., the pixel dichotomy model (PDM) based on either the normalized difference vegetation index (NDVI) or an index of near-infrared reflectance of vegetation (NIRv), the gap probability theory (GPT), and linear spectral mixture analysis (LSMA)] based on FVC ground measurements and fine resolution reference maps from the Validation of Land European Remote sensing Instrument (VALERI) project and the ImagineS field campaigns. For all vegetation types, the FVC estimates from the GPT method showed the best consistency with ground measurements of FVC [root mean square error (RMSE) = 0.17 and bias (BIAS) = 0.05]. For forest types, the PDM method based on NDVI also showed satisfactory results with ground measurements (RMSE = 0.17 and BIAS = 0.11). For sparse grasses, the PDM method based on NIRv showed better agreement with ground measurements (RMSE = 0.15 and BIAS = 0.01). This study provides a reference for selecting the method of FVC estimation with fine spatial resolution images.

Spatiotemporally continuous long-term leaf area index (LAI) products are urgently needed to monitor environmental changes. The current filter- or curve-fitting-based time series reconstructive algorithms fail to reconstruct the LAI time series with many continuous missing values or missing values in key phenological periods, which are common issues in high-spatial-resolution LAI time series. In this article, a meteorological data-driven backpropagation neural network (MBPNN) was proposed to reconstruct discontinuous LAI profiles with a two-step process using vegetation phenological information. As the basis of the strong dependence of vegetation growth on meteorological conditions, a reasonable growth trajectory of reconstructed LAI can be guaranteed by the algorithm even though if many observed values are missing. Validations for reconstructed LAI were conducted both spatially and temporally based on reference maps and field-measured long-term observations. The results showed that the LAI predicted by the MBPNN had a similar accuracy (RMSE = 0.4076) as the Landsat LAI inversions (RMSE = 0.4083) and a similar reconstructed trajectory as the field-measured LAI series even though over 100 days of continuous data were missing (RMSE = 0.1620). A comparison with the Harmonic ANalysis of Time Series (HANTS) algorithm showed that the accuracy of MBPNN was more stable regardless of the size/position of the missing data, and the proposed method performed much better when the data were continuously missing for 50 days or more.

The Leaf Area Index (LAI) is a key structure parameter of plant canopy and is a basic input variable in various terrestrial ecological models. Previous studies indicate that the spatial heterogeneity and the mixture of different land cover types in the moderate resolution pixels will cause large errors in the retrieval of moderate resolution LAI product. Therefore, the influence of spatial heterogeneity should be corrected to retrieve a more reasonable LAI. In this study, we propose a method to improve the accuracy of moderate resolution LAI retrievals based on the mixed-pixel clumping index. The data simulated by the LESS model are used to validate the proposed method. Results show that the LAI estimated by the MODIS operational algorithm become smaller with the increase in spatial heterogeneity of pixel. The proposed method can correct the influence of mixed land cover types and spatial heterogeneity to obtain a more reasonable LAI, and thus shows the potential in generating the global moderate resolution LAI product with improved accuracy.

Leaf chlorophyll content (Chlleaf) is a crucial parameter in carbon cycle modeling and agricultural monitor. Taking advantage of remotely sensed red-edge vegetation index (VI) is an easy approach to estimate Chlleafat a large spatial scale. However, the spectral signals of Chlleafand other canopy/foliar/background factors (e.g. leaf area, leaf angle, soil moisture, etc.) are always coupled together, leading to the relatively low accuracy in direct Chlleafestimation. A new chlorophyll sensitive index (CSI) based on the red-edge band is proposed to estimate Chlleaf, with minimal canopy structural influences. Validation results using in-situ measurements show CSI performed better to estimate Chlleafof winter wheat and soybean at canopy scale: RMSE=8.24μg/cm2for CSI; RMSE=10.04 μg/cm2for the best existing index, MTCI. CSI also has the potential ability to estimate Chlleafacross diverse structural species types with high accuracy. Therefore, CSI provides an effective and convenient way to estimate Chlleafover large areas using satellite data.

Understory vegetation has largely affected the accuracy of forest leaf area index (LAI) estimation. This study analyzed the components of overstorey and understory based on the forest reflectance and transmittance (FRT) model for six forest types in China: i.e. a planted fir forest, a mixed fir and pine forest, deciduous broad-leaved forest, evergreen broad-leaved forest, broad-leaved pinus koraiensis forest, and seasonal tropical rainforest. The contributions of understory were varied for different forest types within a year. The contributions of understory for the tropical rainforest and the coniferous and broad-leaved mixed forest were low due to the dense canopy with higher LAI of overstorey. The effect of understory was unneglectable for the whole wavelengths.

Rugged terrain, as a high percent of the Earth's terrestrial surface, can cause the directionality of the surface thermal radiation, and affect the retrieved land surface temperature (LST) and longwave radiation (SLR) from satellite measurements due to the limited instantaneous field of view and observation angles. New directional brightness temperature (DBT) and equivalent brightness temperature (EBT) models were established considering terrain effects. The biases between them were also analyzed based on a simulated scene using the Advanced Spacebome Thermal Emission and Reflection Radiometer (ASTER) LST, emissivity and topographic data. The results show that BTs at the valley and peak points are clearly anisotropic, while this directionality at the cropland point is not obvious. The DBT shows hotspot effects which is closely related to the solar position. The range of DBTs can reach up to about 9 K in the valley point and the standard deviation of this difference in all view directions is 1.05 K. Thus, it can be concluded that it is hard to meet the requirement of retrieval accuracy of LST or SLR over rugged terrain if ignoring the three-dimensional structure of mountainous region and its angular thermal radiation.

Modeling plays an important role to link the land surface physical/chemical properties with remotely sensed data. In the past decade of years, Multi-scale Remote Sensing Models have been developed (Chen, J. M. et. al. 1997, Huang, H. et. al. 2013). In Recent years, remote sensing communities tend to consider the complex terrain scenario more reasonable recently, such as mixed pixel, topography issues and so on. This paper present the newest progresses on how to and quantitatively measure the heterogeneity of the land surface. Spatial heterogeneity exists in the land surface at every scale, and it is one of the key factors that introduces inherent uncertainty into simulations of land surface radiative processes and parameter retrieval based on remotely sensed data. However, because of the 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 quantitatively describe pixel heterogeneity based on end member and boundary information and high -resolution land cover products (Li, X et. at 2011), which are used to characterize and quantify global land surface heterogeneity. Then, the recent BRDF modelling progresses for two typical mixed pixels are introduced.

Global leaf area index (LAI) generally produced based on the satellite sensors with 1 km spatial resolution, such as the advanced very high resolution radiometer (AVHRR), moderate resolution imaging spectroradiometer (MODIS) and VEGETATION. At present, there isn't a LAI product estimated from the Chinese Feng Yun No.3 (FY-3) images. This study aims to generate a 10-day composite LAI product from FY-3A with a medium resolution spectral imaging (MERSI) at global scale in 2011. Making use of the land cover type as priori knowledge, the LAI for pure vegetation types was inversed from a lookup-table (LUT) based on an stochastic three-dimensional radiative transfer model (3D RTM). For the mixed water and vegetation types, LAI was inversed based on an improved linear decomposition method. The accuracy of LAI inversion from FY-3A MERSI was assessed by LAI field measurements from the Chinese ecosystem research network (CERN) in 2011.

The state of the land surface and the water cycle over the South and East Asia can be determined by space observation. New or significantly improved algorithms have been developed and evaluated against ground measurements. Variables retrieved include land surface properties, i.e. NDVI, LAI, FPAR, albedo, soil moisture, glacier and lake levels. Based on these biophysical parameters derived from microwave and optical remote sensing observations, a hybrid remotely sensed evapotranspiration (ET) estimation model named ETMonitor was developed and applied to estimate the daily actual ET of the Southeast Asia at a spatial resolution of 1 km. The changes in glaciers and lakes on the Tibetan Plateau, and the drainage links between glaciers and lakes are determined in this climate-sensitive region.

Leaf area index (LAI) is a critical parameter in many terrestrial ecosystem models. Continuous LAI measurements from global sites are an important dataset for the validation of remotely sensed LAI products. However, the spatial scale mismatch between the site measurement and the product pixel grid hinders the utilization of multi-temporal ground LAI measurements. In this study, a pragmatic method is presented for spatial upscaling of ground LAI measurements to the product pixel grid. The method is divided into three parts: retrieving high-resolution LAI maps, spatial representativeness grading and spatial upscaling. The proposed method was applied to the Järvselja site in the VALERI project. Results show that this method can reduce the scale mismatch error between the site measurement and the product pixel grid well. Moreover, this method has the potential to be applied to global site LAI measurements, which consequently can improve the reliability of LAI product validation.

Landscape heterogeneity is a common natural phenomenon but is seldom considered in current radiative transfer models for predicting the surface reflectance. This paper developed an analytical Radiative Transfer model for heterogeneous Agro-Forestry scenes (RTAF). The scattering contribution of the non-boundary regions can be estimated from the SAILH model as homogeneous canopies, whereas that of the boundary regions is calculated based on the bidirectional gap probability by considering the interactions and mutual shadowing effects among different patches. The multi-angular airborne observations and Discrete Anisotropic Radiative Transfer (DART) model simulations were used to validate and evaluate the RTAF model over an agro-forestry scene in Heihe River Basin, China. The results suggest the RTAF model can accurately simulate the hemispherica-directional reflectance factors (HDRFs) of the heterogeneous scenes in the red and near-infrared (NIR) bands. The boundary effect can significantly influence the angular distribution of the HDRFs and consequently enlarge the HDRF variations between the backward and forward directions. Compared with the widely used dominant cover type (DCT) and spectral linear mixture (SLM) models, the RTAF model reduced the maximum relative error from 25.7% (SLM) and 23.0% (DCT) to 9.8% in the red band, and from 19.6% (DCT) and 13.7% (SLM) to 8.7% in the NIR band. The RTAF model provides a promising way to improve the retrieval of biophysical parameters (e.g. leaf area index) from remote sensing data over heterogeneous agro-forestry scenes.

The environment of Greater Mekong Sub-region (GMS) was highly payed attention to its economic development. Remote sensing technology was a useful tool for global and regional environment monitor. The fractional vegetation cover (FVC) with 30m spatial resolution for GMS was extracted from the HJ-1/CCD data in this study. The vegetation covers were highly for the entire GMS, and the spatial differences were influenced by vegetation types. Besides, FVC product with high temporal resolution (5 days) and 1km spatial resolution were used to analyze the vegetation damages by typhoon Haiyan from 2edto 10thNov., 2013 based on a change detection method. The damage extents of forest by typhoon were much seriously than cropland and grassland for GMS memberships, especially for Vietnam and China (Guangxi). The vegetation damages varied from -50% to 10% in the 300 km suffer areas on the typhoon Haiyan pathway.

Current bidirectional reflectance distribution function (BRDF) inversions using ordinary least squares (OLS) criterion can be easily contaminated by observations with residual cloud and undetected high aerosols, which leads to abrupt fluctuations in the normalized difference vegetation index (NDVI) time series. The OLS criterion assumes the noise has Gaussian distribution, which is often violated due to positive noise biases caused by clouds and high aerosols. A changing-weight iterative BRDF/NDVI inversion algorithm (CWI) based on a posteriori variance estimation of observation errors is presented to explicitly consider the asymmetrically distributed noise and observations with unequal accuracy in the BRDF retrieval. CWI employs a posteriori variance estimation and an NDVI-based indicator to iteratively adjust the weight of each observation according to its noise level. The validation results suggest CWI performs better than the Li-Gao and OLS approaches. The rmse was reduced from 0.074 to 0.028, and the relative error decreased from 13.4% to 3.8% at the U.S. Department of Agriculture Beltsville Agricultural Research Center site. Similarly, at the Harvard Forest site, the rmse was reduced from 0.086 to 0.031, and the relative error decreased from 9.5% to 2.7%. The average noise and relative noise of the CWI NDVI time series over ten EOS Land Validation Core Sites from 2003-2009 was smaller (0.028, 3.7%) than those of MOD13A2 (0.041, 5.2%), MYD13A2 (0.039, 4.9%) and MCD43B4 (0.030, 4.4%). The results demonstrate the robustness of the CWI approach in suppressing the influence of contaminated observations in BRDF retrievals by producing results that are less affected by undetected clouds and high aerosols.

Landscape heterogeneity is a common natural phenomenon but is seldom considered in current radiative transfer (RT) models for predicting the surface reflectance. This paper developed an analytical RT model for heterogeneous Agro-Forestry scenarios (RTAF) by dividing the scenario into nonboundary regions (NRs) and boundary regions (BRs). The scattering contribution of the NRs can be estimated from the scattering-by-arbitrarily-inclined-leaves-with-the-hot-spot-effect model as homogeneous canopies, whereas that of the BRs is calculated based on the bidirectional gap probability by considering the interactions and mutual shadowing effects among different patches. The multiangular airborne observations and discrete-anisotropic-RT model simulations were used to validate and evaluate the RTAF model over an agro-forestry scenario in the Heihe River Basin, China. The results suggest that the RTAF model can accurately simulate the hemispherical-directional reflectance factors (HDRFs) of the heterogeneous scenarios in the red and near-infrared (NIR) bands. The boundary effect can significantly influence the angular distribution of the HDRFs and consequently enlarge the HDRF variations between the backward and forward directions. Compared with the widely used dominant cover type (DCT) and spectral linear mixture (SLM) models, the RTAF model reduced the maximum relative error from 25.7% (SLM) and 23.0% (DCT) to 9.8% in the red band and from 19.6% (DCT) and 13.7% (SLM) to 8.7% in the NIR band. The RTAF model provides a promising way to improve the retrieval of biophysical parameters (e.g., leaf area index) from remote sensing data over heterogeneous agro-forestry scenarios.

Spatial heterogeneity of land surface induces scaling bias in leaf area index (LAI) products. In optical remote sensing of vegetation, spatial heterogeneity arises both by textural and contextual effects. A case study made in the middle reach of the Heihe River Basin shows that the scaling bias in LAI retrieval is large up to 26% if the spatial heterogeneity within low-resolution pixels is ignored. To reduce the influence of spatial heterogeneity on LA! products, a correcting method combining both textural and contextual information is adopted, and the scaling bias may decrease to less than 2% in producing resolution-invariant LAI products.

A unique method for investigating continental surfaces uses backscatter data measured by a radar altimeter at the nadir point of a satellite, in contrast to other microwave sensors designed to work at oblique angles, such as scatterometers and synthetic aperture radar. To improve the altimetry resolution over land, we generated 0.5° × 0.5° merged altimetry backscatter maps covering 66° N to 66° S over the global land surface every six days for the period from January 2002 to June 2009 by combining three altimeter data sets (Jason1, Envisat, Jason2) in the Ku-band. The four backscatter products of Envisat RA2 from different retracking algorithms were evaluated prior to merging with the Jason1 and Jason2 data. The global pattern and the seasonal variation of the merged altimetry backscatter were examined, which show the merged results have better spatial sampling for the regional to global geophysical process due to the combination of three altimeters. To understand how the altimetry backscatter is related to land surface parameters, the advanced integral equation model for bare soil and water cloud model for vegetation are used to simulate the Ku-band backscatter response to soil and vegetation parameters at an incidence of 0°. Furthermore, we compared the time series of merged altimetry backscatter with the leaf area index (LAI) determined by an optical sensor and the backscatter coefficients obtained from a scatterometer (QuikSCAT) over seven selected vegetated areas over six years. The results confirm the sensitivity of ocean altimetry to vegetation. Further studies relating altimetry backscatter to geophysical parameter are needed.

Evapotranspiration plays an important role in the surface-atmosphere interaction. Remote sensing has long been identified as a technology capable of monitoring ET. But spatial scale problem has a great effect on the accuracy of the ET retrieval by satellite. The objective of this paper is to reduce the uncertainty produced by spatial scale problem based on the spatial characteristic of HJ-1B data. Firstly, the temperature is sharpened to 30m spatial resolutions associated with visible-near-infrared bands. Then net radiation, soil heat flux and sensible heat flux of sub pixel are calculated at 30m resolution using one-source energy balance model. Finally, the fluxes are averaged to 300m resolution and latent heat flux is computed as residual of surface energy balance. The result shows that Temperature sharpening and flux aggregation (TSFA) method reduces the uncertainty produced by spatial scale problem, and it can capture the land surface heterogeneities and associated uncertainties in a way.

Shortwave (SW) radiative flux (usually within 0.3∼3μm) is the dominant energy source of our planet, which drives the climate as well as the matter and energy cycle of the Earth system. It is an indispensable component of surface total energy balance. Considering the importance of SW radiation, during the past decades, more and more studies have conducted for estimating surface SW radiation using satellite-based data, such as MODIS, CERES, GOES etc. Although great effort has been made, most researches neglect the topographic effect and mainly focus on the retrieval of SW radiation over ideal horizontal surfaces for both instantaneous and time-integrated radiation. For this point, we propose a topographic SW radiation model based on the existing studies. Based on this, the SW radiative flux components are derived from MODIS data by fully accounting for the surface topographic effect. The results show that the errors induced in the retrieved daily SW radiation can reach up to 400W/m2at 1km scale. For instantaneous radiation, the uncertainties of derived SW radiation can reach up to 300W/m2even at 5km scale due to topographic effect. The findings of this paper prove the importance of topographic modeling of surface radiation over rugged terrain.

Leaf area index is a key parameter to describe physical and biological processes of plants. Remote sensing technology offers a new method to obtain LAI at regional scales, but it generally records plants information in horizontal. Therefore, the canopy reflectance is easier to reach saturation when plants growth flourished. This paper firstly defines the issues of canopy reflectance saturation, and then analyzes the reflectance contribution of each layers based on multilayer SAIL and FRT model for continuous and discontinuous vegetation, respectively. Besides, this paper analyzes the factors influencing canopy LAI saturation. Results show that the lower part of plants has fewer contributions to canopy reflectance. The leaf angle distribution and view zenith angle are two mainly factors influencing canopy LAI saturation.

Leaf area index (LAI) is an important ecological and environmental parameter. Currently, most LAI inversion methods are based on single model. For continuous vegetation canopy, PROSAIL model used to retrieve vegetation biophysical properties based on lookup table or neural network methods. As for discontinuous vegetation canopy, vegetation canopy reflectance was simulated from Geometric-Optical model, and the relationship between LAI and vegetation indexes was established using simulated reflectance. This paper introduces a method combining the PROSAIL and four-scale model simulated the continuous and discontinuous vegetation canopy reflectance respectively, and then establishes lookup tables among multispectral reflectance, vegetation indices and LAI. The new method using land cover map has been employed to inverse HJ-1B CCD2 LAI in zhangye region of the middle reaches of the Heihe river basin. Results show that the new inversion results have much more deviation with observed dataset, but have good agreement with MODIS LAI in crops.

The backscatter values of altimeter and scatterometer have been investigated for possible use over land surface, especially for vegetation covered area. The spatial and temporal variations of backscatter coefficient of merged altimeter data (JASON1/ ENVISAT RA2), QuikSCAT, and CYCLOPES Leaf Area Index (LAI) over seven selected vegetated areas from 2002 to 2007 are calculated and compared. Initial results indicated that over vegetated areas with a strong seasonal cycle, the altimeter and scatterometer SigmaO measurements are strongly correlated to LAI. The Linear relation between VV/HH QuikSCAT backscatter and LAI is found at short and sparse vegetation, which is not valid at the dense vegetation. The VV/HH is not sensitive to low LAI value. The altimeter and scatterometer backscatter measurements of Ku band are useful to monitor vegetation dynamics as an independent data source compared to the products of optical sensors.