Search

Although generally given little attention in vegetation studies, ground-dwelling (terricolous) lichens are major contributors to overall carbon and nitrogen cycling, albedo, biodiversity and biomass in many high-latitude ecosystems. Changes in biomass of mat-forming pale lichens have the potential to affect vegetation, fauna, climate and human activities including reindeer husbandry. Lichens have a complex spectral signature and terricolous lichens have limited growth height, often growing in mixtures with taller vegetation. This has, so far, prevented the development of remote sensing techniques to accurately assess lichen biomass, which would be a powerful tool in ecosystem and ecological research and rangeland management. We present a Landsat based remote sensing model developed using deep neural networks, trained with 8914 field records of lichen volume collected for > 20 years. In contrast to earlier proposed machine learning and regression methods for lichens, our model exploited the ability of neural networks to handle mixed spatial resolution input. We trained candidate models using input of 1 x 1 (30 x 30 m) and 3 x 3 Landsat pixels based on 7 reflective bands and 3 indices, combined with a 10 m spatial resolution digital elevation model. We normalised elevation data locally for each plot to remove the region-specific variation, while maintaining informative local variation in topography. The final model predicted lichen volume in an evaluation set (n = 159) reaching an R2 of 0.57. NDVI and elevation were the most important predictors, followed by the green band. Even with moderate tree cover density, the model was efficient, offering a considerable improvement compared to earlier methods based on specific reflectance. The model was in principle trained on data from Scandinavia, but when applied to sites in North America and Russia, the predictions of the model corresponded well with our visual interpretations of lichen abundance. We also accurately quantified a recent historic (35 years) change in lichen abundance in northern Norway. This new method enables further spatial and temporal studies of variation and changes in lichen biomass related to multiple research questions as well as rangeland management and economic and cultural ecosystem services. Combined with information on changes in drivers such as climate, land use and management, and air pollution, our model can be used to provide accurate estimates of ecosystem changes and to improve vegetation-climate models by including pale lichens.

It is important to understand temporal and spatial variations in the structure and photosynthetic capacity of tropical rainforests in a world of changing climate, increased disturbances and human appropriation. The equatorial rainforests of Central Africa are the second largest and least disturbed of the biodiversly-rich and highly productive rainforests on Earth. Currently, there is a dearth of knowledge about the phenological behavior and long-term changes that these forests are experiencing. Thus, this study reports on leaf area seasonality and its time trend over the past two decades as assessed from multiple remotely sensed datasets. Seasonal variations of leaf area in Congolese forests derived from MODIS data co-vary with the bimodal precipitation pattern in this region, with higher values during the wet season. Independent observational evidence derived from MISR and EPIC sensors in the form of angular reflectance signatures further corroborate this seasonal behavior of leaf area. The bimodal patterns vary latitudinally within this large region. Two sub-seasonal cycles, each consisting of a dry and wet season, could be discerned clearly. These exhibit different sensitivities to changes in precipitation. Contrary to a previous published report, no widespread decline in leaf area was detected across the entire extent of the Congolese rainforests over the past two decades with the latest MODIS Collection 6 dataset. Long-term precipitation decline did occur in some localized areas, but these had minimal impacts on leaf area, as inferred from MODIS and MISR multi-angle observations.

Long-term ground LAI measurements from the global networks of sites (e.g. FLUXNET) have emerged as a promising data source to validate remotely sensed global LAI product time-series. However, the spatial scale-mismatch issue between site and satellite observations hampers the use of such invaluable ground measurements in validation practice. Here, we propose an approach (Grading and Upscaling of Ground Measurements, GUGM) that integrates a spatial representativeness grading criterion and a spatial upscaling strategy to resolve this scale-mismatch issue and maximize the utility of time-series of site-based LAI measurements. The performance of GUGM was carefully evaluated by comparing this method to both benchmark LAI and other widely used conventional approaches. The uncertainty of three global LAI products (i.e. MODIS, GLASS and GEOV1) was also assessed based on the LAI time-series validation dataset derived from GUGM. Considering all the evaluation results together, this study suggests that the proposed GUGM approach can significantly reduce the uncertainty from spatial scale mismatch and increase the size of the available validation dataset. In particular, the proposed approach outperformed other widely used approaches in these two respects. Furthermore, GUGM was successfully implemented to validate global LAI products in various ways with advantaging frequent time-series validation dataset. The validation results of the global LAI products show that GLASS has the lowest uncertainty, followed by GEOV1 and MODIS for the overall biome types. However, MODIS provides more consistent uncertainties across different years than GLASS and GEOV1. We believe that GUGM enables us to better understand the structure of LAI product uncertainties and their evolution across seasonal or annual contexts. In turn, this method can provide fundamental information for further LAI algorithm improvements and the broad application of LAI product time-series.

Canopy radiative transfer (RT) modeling is critical for the quantitative retrieval of vegetation biophysical parameters and has been under intensive research over the decades. RT models of discontinuous canopies, such as three-dimensional (3D) RT models, posed challenges for the early one-dimensional (1D) hypothesis. Although 3D RT models have higher accuracy, theoretically, they suffer from two problems: detailed scene parameters and complex computational steps. To overcome these problems, the stochastic radiative transfer (SRT) theory, which is known to have the accuracy of 3D RT while being as simple as 1D RT, has been adapted from atmospheric research to the study of vegetation canopies. While the SRT model has been adopted into the operational production of vegetation parameters, its accuracy needs further improvement because of the insufficient consideration of hotspot effects. Additionally, the evaluation and validation of SRT process are still preliminary, which hinders its further development and application. To provide the community with missing information and a scientific basis for subsequent model improvement, we modified, evaluated, and validated the SRT model in this study. First, we proposed the new version of SRT model to better achieve the coupling of SRT process and hotspot effect by dividing the previous SRT into four subproblems. Then, we evaluated the performance of the modified SRT by comparing multiple intermediate variables in the SRT process with 3D computer simulations, and analyzed the model sensitivity to key input parameters as well as the spatial distribution and conservation of radiation energy. Our findings reconfirmed that the SRT theory can well describe the radiation regime of discontinuous canopies with balanced efficiency and accuracy. Moreover, the newly proposed coupling scheme of hotspot effect further improves the model performance in the hotspot regions. Finally, the unmanned aerial vehicle (UAV) observations served as a reference to validate the modeled canopy reflectance, which shows a high concordance. These results provide a detailed theoretical basis for applications and further improvements of the SRT model.

Oil spills have adverse effects on the environment and economy. Near real time detection and response activities enable to better manage the required resources at the incident area for clean-up and control operations. Multi-temporal remote sensing (RS) technologies are widely used to detect and monitor oil spills on the Ocean surfaces. However, current techniques using RS data for oil spill detection are time consuming and expensive in terms of computational cost and related infrastructure. The main focus of this work is oil spill detection from voluminous multi-temporal LANDSAT-7 imagery using high performance computing technologies such as graphics processing units (GPUs) and Message Passing Interface (MPI) to speed up the detection process and provide rapid response. Kepler compute architecture based GPU (Tesla K40) with Compute Unified Device Architecture (CUDA), which is a parallel programming mechanism for GPU is used in the development of the detection algorithms. Oil spill detection techniques that were adapted to GPU based processing include band-ratio and Morphological attribute profile (MAP) based on six structural and shape description attributes namely, Gray mean, standard deviation, elongation, shape complexity, solidity and orientation. Experimental results show the significant gains in the computational speed of these techniques when implemented on a GPU and MPI. A GPU vs. CPU comparison shows that the proposed approach achieves a speedup of around 10 × for MAP and 14 × for band ratio approaches, which includes the data transfer cost. However, the MPI implementation using 64 cores outperforms the GPU, and executes the time intensive task of computing the above said attributes in only 18 min, whereas a GPU consumes around an hour.

Land cover mixture at moderate- to coarse-resolution is an important cause for the uncertainty of global leaf area index (LAI) products. The accuracy of LAI retrievals over land-water mixed pixels is adversely impacted because water absorbs considerable solar radiation and thus can greatly lower pixel-level reflectance especially in the near-infrared wavelength. Here we proposed an approach named Reduced Water Effect (RWE) to improve the accuracy of LAI retrievals by accounting for water-induced negative bias in reflectances. The RWE consists of three parts: water area fraction (WAF) calculation, subpixel water reflectance computation in land-water mixed pixels and LAI retrieval using the operational MODIS LAI algorithm. The performance of RWE was carefully evaluated using the aggregated Landsat ETM+ reflectance of water pixels over different regions and observation dates and the aggregated 30-m LAI reference maps over three sites in the moderate-resolution pixel grid (500-m). Our results suggest that the mean absolute errors of water endmember reflectance in red and NIR bands were both

The Amazon rainforest is a critical hotspot for bio-diversity, and plays an essential role in global carbon, water and energy fluxes and the earth's climate. Our ability to project the role of vegetation carbon feedbacks on future climate critically depends upon our understanding of this tropical ecosystem, its tolerance to climate extremes and tipping points of ecosystem collapse. Satellite remote sensing is the only practical approach to obtain observational evidence of trends and changes across large regions of the Amazon forest; however, inferring these trends in the presence of high cloud cover fraction and aerosol concentrations has led to widely varying conclusions. Our study provides a simple and direct statistical analysis of a measurable change in daily and composite surface reflectance obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) based on the noise level of data and the number of available observations. Depending on time frame and data product chosen for analysis, changes in leaf area need to exceed up to 2 units leaf area per unit ground area (expressed as m 2 m − 2 ) across much of the basin before these changes can be detected at a 95% confidence level with conventional approaches, roughly corresponding to a change in NDVI and EVI of about 25%. A potential way forward may be provided by advanced multi-angular techniques, such as the Multi-Angle Implementation of Atmospheric Correction Algorithm (MAIAC), which allowed detection of changes of about 0.6–0.8 units in leaf area (2–6% change in NDVI) at the same confidence level. In our analysis, the use of the Enhanced Vegetation Index (EVI) did not improve accuracy of detectable change in leaf area but added a complicating sensitivity to the bi-directional reflectance, or view geometry effects.

Accurate characterization of variability and trends in forest biomass at local to national scales is required for accounting of global carbon sources and sinks and monitoring their dynamics. Here we present a new remote sensing based approach for estimating live forest aboveground biomass (AGB) based on a simple parametric model that combines high-resolution estimates of leaf area index (LAI) from the Landsat Thematic Mapper sensor and canopy maximum height from the Geoscience Laser Altimeter System (GLAS) sensor onboard ICESat, the Ice, Cloud, and land Elevation Satellite. We tested our approach with a preliminary uncertainty assessment over the forested areas of California spanning a broad range of climatic and land-use conditions and find our AGB estimates to be comparable to estimates of AGB from inventory records and other available satellite-estimated AGB maps at aggregated scales. Our study offers a high-resolution approach to map forest aboveground biomass at regional-to-continental scales and assess sources of uncertainties in the estimates.

This paper summarizes the implementation of a physically based algorithm for the retrieval of vegetation green Leaf Area Index (LAI) from Landsat surface reflectance data. The algorithm is based on the canopy spectral invariants theory and provides a computationally efficient way of parameterizing the Bidirectional Reflectance Factor (BRF) as a function of spatial resolution and wavelength. LAI retrievals from the application of this algorithm to aggregated Landsat surface reflectances are consistent with those of MODIS for homogeneous sites represented by different herbaceous and forest cover types. Example results illustrating the physics and performance of the algorithm suggest three key factors that influence the LAI retrieval process: 1) the atmospheric correction procedures to estimate surface reflectances; 2) the proximity of Landsat-observed surface reflectance and corresponding reflectances as characterized by the model simulation; and 3) the quality of the input land cover type in accurately delineating pure vegetated components as opposed to mixed pixels. Accounting for these factors, a pilot implementation of the LAI retrieval algorithm was demonstrated for the state of California utilizing the Global Land Survey (GLS) 2005 Landsat data archive. In a separate exercise, the performance of the LAI algorithm over California was evaluated by using the short-wave infrared band in addition to the red and near-infrared bands. Results show that the algorithm, while ingesting the short-wave infrared band, has the ability to delineate open canopies with understory effects and may provide useful information compared to a more traditional two-band retrieval. Future research will involve implementation of this algorithm at continental scales and a validation exercise will be performed in evaluating the accuracy of the 30-m LAI products at several field sites.

article i nfo In this study we use the 500 m Moderate Resolution Imaging Spectroradiometer (MODIS) Bidirectional Reflectance Distribution Function (BRDF) product to develop multivariate linear regression models that estimate canopy heights over study sites at Howland Forest, Maine, Harvard Forest, Massachusetts and La Selva Forest, Costa Rica using (1) directional escape probabilities that are spectrally independent and (2) the directional spectral reflectances used to derive the directional escape probabilities. These measures of canopy architecture are compared with canopy height information retrieved from the airborne Laser Vegetation Imaging Sensor (LVIS). Both the escape probability and the directional reflectance approaches achieve good results, with correlation coefficients in the range 0.54-0.82, although escape probability results are usually slightly better. This suggests that MODIS 500 m BRDF data can be used to extrapolate canopy heights observed by widely-spaced satellite LIDAR swaths to larger areas, thus providing wide-area coverage of canopy height.

The generation of multi-decade long Earth System Data Records (ESDRs) of Leaf Area Index (LAI) and Fraction of Photosynthetically Active Radiation absorbed by vegetation (FPAR) from remote sensing measurements of multiple sensors is key to monitoring long-term changes in vegetation due to natural and anthropogenic influences. Challenges in developing such ESDRs include problems in remote sensing science (modeling of variability in global vegetation, scaling, atmospheric correction) and sensor hardware (differences in spatial resolution, spectral bands, calibration, and information content). In this paper, we develop a physically based approach for deriving LAI and FPAR products from the Advanced Very High Resolution Radiometer (AVHRR) data that are of comparable quality to the Moderate resolution Imaging Spectroradiometer (MODIS) LAI and FPAR products, thus realizing the objective of producing a long (multi-decadal) time series of these products. The approach is based on the radiative transfer theory of canopy spectral invariants which facilitates parameterization of the canopy spectral bidirectional reflectance factor (BRF). The methodology permits decoupling of the structural and radiometric components and obeys the energy conservation law. The approach is applicable to any optical sensor, however, it requires selection of sensor-specific values of configurable parameters, namely, the single scattering albedo and data uncertainty. According to the theory of spectral invariants, the single scattering albedo is a function of the spatial scale, and thus, accounts for the variation in BRF with sensor spatial resolution. Likewise, the single scattering albedo accounts for the variation in spectral BRF with sensor bandwidths. The second adjustable parameter is data uncertainty, which accounts for varying information content of the remote sensing measurements, i.e., Normalized Difference Vegetation Index (NDVI, low information content), vs. spectral BRF (higher information content). Implementation of this approach indicates good consistency in LAI values retrieved from NDVI (AVHRR-mode) and spectral BRF (MODIS-mode). Specific details of the implementation and evaluation of the derived products are detailed in the second part of this two-paper series.

Radiation reflected from vegetation canopies exhibits high spatial variation. Satellite-borne sensors measure the mean intensities emanating from heterogeneous vegetated pixels. The theory of radiative transfer in stochastic media provides the most logical linkage between satellite observations and the three-dimensional canopy structure through a closed system of simple equations which contains the mean intensity and higher statistical moments directly as its unknowns. Although this theory has been a highly active research field in recent years, its potential for satellite remote sensing of vegetated surfaces has not been fully realized because of the lack of models of a canopy pair-correlation function that the stochastic radiative transfer equations require. The pair correlation function is defined as the probability of finding simultaneously phytoelements at two points. This paper presents analytical and Monte Carlo generated pair correlation functions. Theoretical and numerical analyses show that the spatial correlation between phytoelements is primarily responsible for the effects of the three-dimensional canopy structure on canopy reflective and absorptive properties. The pair correlation function, therefore, is the most natural and physically meaningful measure of the canopy structure over a wide range of scales. The stochastic radiative transfer equations naturally admit this measure and thus provide a powerful means to investigate the three-dimensional canopy structure from space. Canopy reflectances predicted by the stochastic equations are assessed by comparisons with the PARABOLA measurements from coniferous and broadleaf forest stands in the BOREAS Southern Study Areas. The pair correlation functions are derived from data on tree structural parameters collected during field campaigns conducted at these sites. The simulated canopy reflectances compare well with the PARABOLA data.

The Multi-angle Imaging SpectroRadiometer (MISR) instrument provides global imagery at nine discrete viewing angles and four visible/near-infrared spectral bands. MISR standard products include green leaf area index (LAI) of vegetation and fraction of photosynthetically active radiation absorbed by vegetation (FPAR). This paper describes the research basis for transitioning the MISR LAI/FPAR products from provisional to validation status. The efforts included not only comparisons to field data but also analyses of relationships, consistency and complementarity between various MISR products derived by independent algorithms. For example, we show how the energy absorbed by the ground below vegetation can be estimated from two independent MISR products, FPAR and BHRPAR (bi-hemispheric reflectance at PAR wavelengths). Further, we show that this information can be used to derive at least three measures of canopy structure — Beer's law extinction coefficient, mean leaf inclination and the gap fraction or vegetation ground cover. The spatial and temporal coverage of the LAI/FPAR product is mainly limited by cloud contamination. However, when a successful aerosol retrieval is performed, typically 95% of pixels have surface reflectance retrievals suitable as input to the LAI/FPAR algorithm. The algorithm provides LAI/FPAR retrievals in 50–80% of these pixels with suitable input. The early versions of the algorithm overestimated LAI values in grasses and broadleaf crops. The MISR LAI product from the recalibrated algorithm (version 3.3) is assessed by comparison with field data collected in a 3 × 3 km agricultural area (grasses and cereal crops) near Avignon, France. LAI retrievals in other biomes are compared to MODIS LAI product of known accuracy. The MISR LAI product shows structural and phenological variability in agreement with data. Our results suggest that the product is accurate to within 0.66 LAI in herbaceous vegetation and savannas and is an overestimate by about 1 LAI in broadleaf forests. LAI retrievals over needle leaf forests remain at provisional quality level.

The concept of canopy spectral invariants expresses the observation that simple algebraic combinations of leaf and canopy spectral transmittance and reflectance become wavelength independent and determine a small set of canopy structure specific variables. This set includes the canopy interceptance, the recollision and the escape probabilities. These variables specify an accurate relationship between the spectral response of a vegetation canopy to the incident solar radiation at the leaf and the canopy scale and allow for a simple and accurate parameterization for the partitioning of the incoming radiation into canopy transmission, reflection and absorption at any wavelength in the solar spectrum. This paper presents a solid theoretical basis for spectral invariant relationships reported in literature with an emphasis on their accuracies in describing the shortwave radiative properties of the three-dimensional vegetation canopies. The analysis of data on leaf and canopy spectral transmittance and reflectance collected during the international field campaign in Flakaliden, Sweden, June 25¿July 4, 2002 supports the proposed theory. The results presented here are essential to both modeling and remote sensing communities because they allow the separation of the structural and radiometric components of the measured/modeled signal. The canopy spectral invariants offer a simple and accurate parameterization for the shortwave radiation block in many global models of climate, hydrology, biogeochemistry, and ecology. In remote sensing applications, the information content of hyperspectral data can be fully exploited if the wavelength-independent variables can be retrieved, for they can be more directly related to structural characteristics of the three-dimensional vegetation canopy.

Gridding artifacts between observations and predefined grid cells strongly influence the local spatial properties of MODIS images. The sensor observation in any grid cell is only partially derived from the location of the cell, with the average overlap between observations and their grid cells being less than 30%. This mismatch between grid cells and observations has strong implications for the use of reference data for the validation of MODIS products or the training of MODIS algorithms. When generating multidate composites, gridding artifacts introduce bias when spectral compositing criteria are used. Also, results indicate that the ability to generate consistent long-term remote sensing records is dependent on both consistent sensing scenarios (spectral bands, view angle distributions, geolocation error) as well as consistent compositing approaches. The band-to-band registration for the different spatial resolutions of gridded MODIS data can be poor if the different resolutions of data are gridded before aggregation. In all cases it is imprecise to characterize the subpixel properties of the coarser resolution bands using the finer resolution bands due to poor correspondence in the areas from which the observations are derived. All of the band-to-band registration problems are minimized when the MODIS data are aggregated to coarser resolutions. When validating algorithm accuracy, available data on the observation dimensions and the offsets between the grid cell and the observation should be included to ensure the quality of validation results. If this information is not available, MODIS data should be aggregated to coarser resolutions to improve the correspondence between the location of observations and grid cells.

A prototype product suite, containing the Terra 8-day, Aqua 8-day, Terra–Aqua combined 8- and 4-day products, was generated as part of testing for the next version (Collection 5) of the MODerate resolution Imaging Spectroradiometer (MODIS) leaf area index (LAI) products. These products were analyzed for consistency between Terra and Aqua retrievals over the following data subsets in North America: single 8-day composite over the whole continent and annual time series over three selected MODIS tiles (1200×1200 km). The potential for combining retrievals from the two sensors to derive improved products by reducing the impact of environmental conditions and temporal compositing period was also explored. The results suggest no significant discrepancies between large area (from continent to MODIS tile) averages of the Terra and Aqua 8-day LAI and surface reflectances products. The differences over smaller regions, however, can be large due to the random nature of residual atmospheric effects. High quality retrievals from the radiative transfer based algorithm can be expected in 90–95% of the pixels with mostly herbaceous cover and about 50–75% of the pixels with woody vegetation during the growing season. The quality of retrievals during the growing season is mostly restricted by aerosol contamination of the MODIS data. The Terra–Aqua combined 8-day product helps to minimize this effect and increases the number of high quality retrievals by 10–20% over woody vegetation. The combined 8-day product does not improve the number of high quality retrievals during the winter period because the extent of snow contamination of Terra and Aqua observations is similar. Likewise, cloud contamination in the single-sensor and combined products is also similar. The LAI magnitudes, seasonal profiles and retrieval quality in the combined 4-day product are comparable to those in the single-sensor 8-day products. Thus, the combined 4-day product doubles the temporal resolution of the seasonal cycle, which facilitates phenology monitoring in application studies during vegetation transition periods. Both Terra and Aqua LAI products show anomalous seasonality in boreal needle leaf forests, due to limitations of the radiative transfer algorithm to model seasonal variations of MODIS surface reflectance data with respect to solar zenith angle. Finally, this study suggests that further improvement of the MODIS LAI products is mainly restricted by the accuracy of the MODIS observations.

Climate change is predicted to alter the canopy phenology of temperate and boreal forests, which will affect carbon, water, and energy budgets. Therefore, there is a great need to evaluate remotely sensed products for their potential to accurately capture canopy dynamics. The objective of this study was to compare several products derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) to field measurements of fraction photosynthetically active radiation (FPAR) and plant area index (PAI) for a deciduous broadleaf forest in northern Wisconsin in 2002. MODIS products captured the general phenological development of the canopy although MODIS products overestimated the leaf area during the overstory leaf out period. Field data suggest that the period from budburst to canopy maturity, or maximum PAI, occurred in 10 to 12 days while MODIS products predicted onset of greenness and maturity from 1 to 21 days and 0 to 19 days earlier than that from field observations, respectively. Temporal compositing of MODIS data and understory development are likely key factors explaining differences with field data. Maximum PAI estimates differed only by 7% between field derived and MODIS-based estimates of LAI. Implications for ecosystem modeling of carbon and water exchange and future research needs are discussed.

The influence of foliage vertical distribution on vegetation gross primary production (GPP) is investigated in this study. A new photosynthesis model has been created that combines the standard sunlit/shaded leaf separation (two-leaf) and the multiple layer approaches and uses vertical foliage profiles measured by SLICER (the Scanning Lidar Imager of Canopies by Echo Recovery). Daily gross carbon assimilation rates calculated by this model were compared with the rates calculated by two other models, the two-leaf model and the combined two-leaf multilayer model utilizing uniform foliage profiles. The comparison was made over a wide range of profiles and weather conditions for two mixed deciduous forest stands in eastern Maryland, measured by SLICER in September 1995. Incident radiation pattern, environmental parameters and total amounts of sunlit and shaded leaves were the same for all three models. The difference was in the distributions of radiation and sunlit/shaded leaves inside the canopy. For the combined models, these distributions were calculated based on the vertical foliage profiles, while for the two-leaf model, empirical equations were used to account for the average amounts of absorbed radiation. The simulations showed that: (1) the use of a uniform foliage distribution instead of the actual one results in large differences in the calculated GPP values, up to 46.4% and 50.7% for the days with partial and total cloud cover; (2) the performance of the two-leaf model is extremely sensitive to the absorbed radiation pattern, its disagreement with the proposed model becomes insignificant when the average amounts of absorbed radiation are the same; (3) days with partial cloud cover and a greater fraction of diffuse radiation are characterized by higher GPP rates. These findings highlight the importance of vertical foliage profile and separate treatments of diffuse and direct radiation for photosynthesis modeling.

Leaf area index (LAI) collected in a needle-leaf forest site near Ruokolahti, Finland, during a field campaign in June 14–21, 2000, was used to validate Moderate Resolution Imaging Spectroradiometer (MODIS) LAI algorithm. The field LAI data was first related to 30-m resolution Enhanced Thermal Mapper Plus (ETM+) images using empirical methods to create a high-resolution LAI map. The analysis of empirical approaches indicates that preliminary segmentation of the image followed by empirical modeling with the resulting patches, was an effective approach to developing an LAI validation surface. Comparison of the aggregated high-resolution LAI map and corresponding MODIS LAI retrievals suggests satisfactory behavior of the MODIS LAI algorithm although variation in MODIS LAI product is higher than expected. The MODIS algorithm, adjusted to high resolution, generally overestimates the LAI due to the influence of the understory vegetation. This indicates the need for improvements in the algorithm. An improved correlation between field measurements and the reduced simple ratio (RSR) suggests that the shortwave infrared (SWIR) band may provide valuable information for needle-leaf forests.

The objective of this paper is to review research conducted over the past decade on the application of multi-temporal remote sensing for monitoring changes of Arctic tundra lands. Emphasis is placed on results from the National Science Foundation Land–Air–Ice Interactions (LAII) program and on optical remote sensing techniques. Case studies demonstrate that ground-level sensors on stationary or moving track platforms and wide-swath imaging sensors on polar orbiting satellites are particularly useful for capturing optical remote sensing data at sufficient frequency to study tundra vegetation dynamics and changes for the cloud prone Arctic. Less frequent imaging with high spatial resolution instruments on aircraft and lower orbiting satellites enable more detailed analyses of land cover change and calibration/validation of coarser resolution observations. The strongest signals of ecosystem change detected thus far appear to correspond to expansion of tundra shrubs and changes in the amount and extent of thaw lakes and ponds. Changes in shrub cover and extent have been documented by modern repeat imaging that matches archived historical aerial photography. NOAA Advanced Very High Resolution Radiometer (AVHRR) time series provide a 20-year

The Multi-angle Imaging SpectroRadiometer (MISR) instrument is designed to provide global imagery at nine discrete viewing angles and four visible/near-infrared spectral bands. The MISR standard products include green leaf area index (LAI) of vegetation and fraction of photosynthetically active radiation absorbed by vegetation (FPAR). These parameters are being routinely processed from MISR data at the Langley Atmospheric Sciences Data Center (ASDC) since October 2002. This paper describes the research basis for transitioning the MISR LAI/FPAR product from beta to provisional status. The quality and spatial coverage of MISR land surface reflectances that are input to the algorithm determine the quality and spatial coverage of the LAI and FPAR products. Therefore, considerable efforts have been expended to analyze the performance of the algorithm as a function of uncertainties of MISR surface reflectances and to establish the convergence property of the MISR LAI/FPAR algorithm, namely, that the reliability and accuracy of the retrievals increase with increased input information content and accuracy. An additional objective of the MISR LAI/FPAR algorithm is classification of global vegetation into biome types—information that is usually an input to remote sensing algorithms that use single-angle observations. An upper limit of uncertainties of MISR surface reflectances that allows discrimination between biomes, minimizes the impact of biome misidentification on LAI retrievals, and maximizes the spatial coverage of retrievals was estimated. Algorithm performance evaluated on a limited set of MISR data from Africa suggests valid LAI retrievals and correct biome identification in about 20% of the pixels, on an average, given the current level of uncertainties in the MISR surface reflectance data. The other 80% of the LAI values are retrieved using incorrect information about the type of biome. However, the use of multi-angle data minimizes the impact of biome misidentification on LAI retrievals; that is, with a probability of about 70%, uncertainties in LAI retrievals due to biome misclassification do not exceed uncertainties in the observations. We also discuss in depth the parameters that characterize LAI/FPAR product quality—such as quality assessment (QA) that is available to the users along with the product. The analysis of the MISR LAI/FPAR product presented here demonstrates the physical basis of the radiative transfer algorithm used in the retrievals and, importantly, that the reliability and accuracy of the retrievals increase with increased input information content and accuracy. Further improvements in the quality of MISR surface reflectances are therefore expected to lead to LAI and FPAR retrievals of increasing quality.

This paper presents the analysis of radiative transfer assumptions underlying moderate resolution imaging spectroradiometer (MODIS) leaf area index (LAI) and fraction of photosynthetically active radiation (FPAR) algorithm for the case of spatially heterogeneous broadleaf forests. Data collected by a Boston University research group during the July 2000 field campaign at the Earth Observing System (EOS) core validation site, Harvard Forest, MA, were used for this purpose. The analysis covers three themes. First, the assumption of wavelength independence of spectral invariants of transport equation, central to the parameterization of the MODIS LAI and FPAR algorithm, is evaluated. The physical interpretation of those parameters is given and an approach to minimize the uncertainties in its retrievals is proposed. Second, the theoretical basis of the algorithm was refined by introducing stochastic concepts which account for the effect of foliage clumping and discontinuities on LAI retrievals. Third, the effect of spatial heterogeneity in FPAR was analyzed and compared to FPAR variation due to diurnal changes in solar zenith angle (SZA) to asses the validity of its static approximation.

Estimation of canopy biophysical variables from remote sensing data was investigated using radiative transfer model inversion. Measurement and model uncertainties make the inverse problem ill posed, inducing difficulties and inaccuracies in the search for the solution. This study focuses on the use of prior information to reduce the uncertainties associated to the estimation of canopy biophysical variables in the radiative transfer model inversion process. For this purpose, lookup table (LUT), quasi-Newton algorithm (QNT), and neural network (NNT) inversion techniques were adapted to account for prior information. Results were evaluated over simulated reflectance data sets that allow a detailed analysis of the effect of measurement and model uncertainties. Results demonstrate that the use of prior information significantly improves canopy biophysical variables estimation. LUT and QNT are sensitive to model uncertainties. Conversely, NNT techniques are generally less accurate. However, in our conditions, its accuracy is little dependent significantly on modeling or measurement error. We also observed that bias in the reflectance measurements due to miscalibration did not impact very much the accuracy of biophysical estimation.

The problem of how the scale, or spatial resolution, of reflectance data impacts retrievals of vegetation leaf area index (LAI) is addressed in this article. We define the goal of scaling as the process by which it is established that LAI values derived from coarse resolution sensor data equal the arithmetic average of values derived independently from fine resolution sensor data. The increasing probability of land cover mixtures with decreasing resolution is defined as heterogeneity, which is a key concept in scaling studies. The effect of pixel heterogeneity on spectral reflectances and LAI retrievals is investigated with 1-km Advanced Very High Resolution Radiometer (AVHRR) data aggregated to different coarse spatial resolutions. It is shown that LAI retrieval errors at coarse resolution are inversely related to the proportion of the dominant land cover in such pixel. Further, large errors in LAI retrievals are incurred when forests are minority biomes in non-forest pixels compared to when forest biomes are mixed with one another, and vice versa. A physically based scaling with explicit spatial resolution-dependent radiative transfer formulation is developed. The successful application of this theory to scaling LAI retrievals from AVHRR data of different resolutions is demonstrated. These principles underlie our approach to the production and validation of LAI product from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Multi-angle Imaging Spectroradiometer (MISR) aboard the TERRA platform. (C) 2002 Elsevier Science Inc. All rights reserved.

The development of appropriate ground-based validation techniques is critical to assessing uncertainties associated with satellite data-based products. In this paper, the second of a two-part series, we present a method for validation of the Moderate Resolution Imaging Spectroradiometer Leaf Area Index (MODIS LAI) product with emphasis on the sampling strategy for field data collection. Using a hierarchical scene model, we divided 30-m resolution LAI and NDVI images from Maun (Botswana), Harvard Forest (USA) and Ruokulahti Forest (Finland) into individual scale images of classes, region and pixel. Isolating the effects associated with different landscape scales through decomposition of semivariograms not only shows the relative contribution of different characteristic scales to the overall variation, but also displays the spatial structure of the different scales within a scene. We find that (1) patterns of variance at the class, region and pixel scale at these sites are different with respect to the dominance in order of the three levels of landscape organization within a scene; (2) the spatial structure of LAI shows similarity across the three sites, that is, ranges of semivariograms from scale of pixel, region and class are less than 1000 m. Knowledge gained from these analyses aids in formulation of sampling strategies for validation of biophysical products derived from moderate resolution sensors such as MODIS. For a homogeneous (within class) site, where the scales of class and region account for most of the spatial variation, a sampling strategy should focus more on using accurate land cover maps and selection of regions. However, for a heterogeneous (within class) site, accurate point measurements and GPS readings are needed.

The development of appropriate ground-based validation techniques is critical to assessing uncertainties associated with satellite data-based products. Here we present a method for validation of the Moderate Resolution Imaging Spectroradiometer (MODIS) Leaf Area Index (LAI) product with emphasis on the sampling strategy for field data collection. This paper, the first of two-part series, details the procedures used to assess uncertainty of the MODIS LAI product. LAI retrievals from 30 m ETM+ data were first compared to field measurements from the SAFARI 2000 wet season campaign. The ETM+ based LAI map was thus as a reference to specify uncertainties in the LAI fields produced from MODIS data (250-, 500-, and 1000-m resolutions) simulated from ETM+. Because of high variance of LAI measurements over short distances and difficulties of matching measurements and image data, a patch-by-patch comparison method, which is more realistically implemented on a routine basis for validation, is proposed. Consistency between LAI retrievals from 30 m ETM+ data and field measurements indicates satisfactory performance of the algorithm. Values of LAI estimated from a spatially heterogeneous scene depend strongly on the spatial resolution of the image scene. The results indicate that the MODIS algorithm will underestimate LAI values by about 5% over the Maun site if the scale of the algorithm is not matched to the resolution of the data.

An algorithm based on the physics of radiative transfer in vegetation canopies for the retrieval of vegetation green leaf area index (LAI) and fraction of absorbed photosynthetically active radiation (FPAR) from surface reflectances was developed and implemented for operational processing prior to the launch of the moderate resolution imaging spectroradiometer (MODIS) aboard the TERRA platform in December of 1999. The performance of the algorithm has been extensively tested in prototyping activities prior to operational production. Considerable attention was paid to characterizing the quality of the product and this information is available to the users as quality assessment (QA) accompanying the product. The MODIS LAI/FPAR product has been operationally produced from day one of science data processing from MODIS and is available free of charge to the users from the Earth Resources Observation System (EROS) Data Center Distributed Active Archive Center. Current and planned validation activities are aimed at evaluating the product at several field sites representative of the six structural biomes. Example results illustrating the physics and performance of the algorithm are presented together with initial QA and validation results. Potential users of the product are advised of the provisional nature of the product in view of changes to calibration, geolocation, cloud screening, atmospheric correction and ongoing validation activities. D 2002 Published by Elsevier Science Inc.

We evaluate the operational MODIS Leaf Area Index (LAI) product using field-sampled data collected at five sites in southern Africa in March 2000. One site (Mongu, Zambia) was sampled monthly throughout the year. All sites were along the International Geosphere Biosphere Programme's (IGBP) Kalahari Transect, which features progressively lower annual precipitation, and hence, lower vegetation productivity, from north to south. The soils are consistently sandy. At each site, we sampled the vegetation overstory along three 750-m transects using the Tracing Radiation and Architecture in Canopies (TRAC) instrument. The resulting plant area index values were adjusted with ancillary stem area data to estimate LAI. Despite some instrument characterization and production issues in the first year of MODIS operations, our results suggest the first-year MODIS LAI algorithm correctly accommodates structural and phenological variability in semiarid woodlands and savannas, and is accurate to within the uncertainty of the validation approach used here. Limitations of this study and its conclusions are also discussed.

The insights gained from present land cover classification activities suggest integration of multiangle data into classification attempts for future progress. Land cover types that exhibit distinct signatures in the space of remote sensing data facilitate unambiguous identification of cover types. In this two-part series, we develop a theme for consistency among cover type definitions, uniqueness of their signatures, and physics of the remote sensing data. In the first part, Zhang et al.’s [Remote Sens. Environ., in press.] empirical arguments in support of the consistency principle were presented. This part provides a theoretical justification of the consistency requirements. Radiative transfer best explains the physics of the processes operative in the generation of the signal in the optical remote sensing data. Biome definitions given in terms of variables that this theory admits and the use of the transport equation to interpret biome signatures guarantee the consistency requirements. It is shown in this paper that three metrics of the biome angular signature in the spectral space—location, angular signature slope (ASSI), and length (ASLI) indices—are related to eigenvalues and eigenvectors of the transport equation. These variables allow a novel parameterization of canopy structure based on the partitioning of the incident radiation among canopy absorption, transmission, and reflection. Consistency between cover type definitions and uniqueness of their signatures with the physics of the remote sensing data is required not only to reduce ambiguity in land cover identification, but also to directly relate land cover type to biophysical and biogeochemical processes in vegetation canopies. D 2002 Elsevier Science Inc. All rights reserved.

The evaluation and prediction of changes in carbon dynamics at the ecosystem level is a key issue in studies of global change. An operational concept for the determination of carbon fluxes for the Belgian territory is the goal of the presented study. The approach is based on the integration of remotely sensed data into ecosystem models in order to evaluate photosynthetic assimilation and net ecosystem exchange (NEE). Remote sensing can be developed as an operational tool to determine the fraction of absorbed photosynthetically active radiation (fPAR). A review of the methodological approach of mapping fPAR dynamics at the regional scale by means of NOAA11-AVHRR/2 data for the year 1990 is given. The processing sequence from raw radiance values to fPAR is presented. An interesting aspect of incorporating remote sensing derived fPAR in ecosystem models is the potential for modeling actual as opposed to potential vegetation. Further work should prove whether the concepts presented and the assumptions made in this study are valid.

The state-of-the-art on radiative transfer modeling in vegetation canopies arul the application of such models to the interpretation and analysis of remotely sensed optical data is summarized. Modeling of top-of-the-atmosphere and top-of-the-canopy radiance field is developed as boundary value problems in radiative transfer. The parameterization of the constituent functions with simple models and/or empirical data is outlined together with numerical solution methods and examples of results of model validation. Caveats in the assignment of signal characteristics to surface properties are itemized and discussed with example results. Algorithms to estimate surface properties from remote observations are classified as spectral vegetation indices, model inversion, expert systems, neural networks, and genetic algorithms. Their applicability is also discussed.