Monitoring phenological cycles of desert ecosystems using NDVI and LST data derived from NOAA-AVHRR imagery

The potential of the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) for monitoring phenological cycles in semi-arid lands has been demonstrated in this study. Attention was focused on two areas located only a few kilometres apart but across the political border between the Negev (Israel) and Sinai (Egypt). Although the areas are identical from the pedological, geomorphological, and climatic points of view, due to different land management, the Negev is under a continuous rehabilitation process while Sinai is under a desertification process. Four years of digital data were used to compute the Normalized Difference Vegetation Index (NDVI) and Land Surface Temperatures (LST) over two sampling polygons. The NDVI was used to monitor the vegetation reaction to rainfall, while LST proved to be a good indicator of seasonal climatic fluctuations. Using these biological and physical variables, the potential for following the vegetation dynamics throughout the year was demonstrated. Through cluster analysis, it was shown that the movements of the Sinai desertified side in the LST-NDVI space are only due to seasonal climatic fluctuations. On the Israeli recovered side, on the other hand, three different parts of the annual ecological cycle of the indigenous vegetation are evident: the dry season in which plants reduce their activity, the rainy season, and a growing season characterised by relatively intense biological activity. Within the LST-NDVI space it was also shown that Sinai is positioned similarly to the Sahara biome and the Negev similarly to the Sahel biome. Finally, LST-NDVI data were used to estimate phenological parameters that can be exploited for defining protection policies or, on the long term, for climate change studies.     

What determines the spectral reflectance of the Negev-Sinai sand dunes

The difference in spectra reflectance between the north-western Negev desert and the adjacent Sinai sand dunes has attracted the attention of many scientists. Remote sensing analysis of three Landsat Multi-Spectral Scanner (MSS) images acquired in the summer of 1984, 1987, 1989, followed by intensive field work, indicate that the area is not homogenous and is undergoing a quick recovery from the intensive grazing that it suffered between 1968 and 1982. The outcome is a gradual decrease in the brightness of the Negev between 1984 and 1989 in all MSS bands except band 7 which shows very little or no change with time. This is due to the increase in biogenic crust and vegetation cover. The Sinai bare sand shows the opposite trend of little to no change in all bands but band 7, which shows an increase in reflectance with time. This is probably due to the effect of further destruction of vegetation in the Sinai after 1982.

In the northern part of the area, dunes are low, have a high percentage of fines (silt and clay) in the interdune areas, and stabilize quickly. This area is also covered by dense carpets of annuals which have a reflectance that is lower than that of the crusted area, but higher than the reflectance of Artemisia monosperma, the most widespread shrub. The contrast ratios, measured in the field between the Sinai bare sand and the Negev biogenic crust (on the same sand) is similar to the contrast ratios measured by Landsat MSS between the Sinai and the Negev.     

Ground and space spectral measurements for assessing the semi-arid ecosystem phenology related to CO2 fluxes of biological soil crusts

This paper reports on ranges of carbon dioxide (CO₂) activity in biological soil crusts (BSC) correlated with different ranges of the BSC's spectral reflectance throughout the phenological cycle of the year. Methodology is based on surface CO₂ exchange measurements, ground spectral measurements, and satellite images interpretation. Thirty-nine field campaigns, each of duration of 3 days, were conducted over the course of 2 years at a sand dunes and a loess environment of the northwestern Negev desert in Israel, in order to relate the CO₂ fluxes and the spectral signals to the seasonal phenology. The Normalized Difference Vegetation Index (NDVI) was derived from ground measurements of the BSC's reflectance and correlated with their CO₂ exchange data. A linear mixture model, incorporating the different contributions of the sites' ground features, was calculated and compared with SPOT-HRV data. From the ground measurements, fairly good correlations were found between the NDVI and the CO₂ fluxes on a seasonal scale. Hence, the NDVI successfully indicates the potential magnitude and capacity of the BSC's assimilation activity. The linear mixture model successfully describes the phenological cycles of the BSC, annual, and perennial plants and corresponds well to the satellite data. Moreover, the model enables annual changes of the phenology cycle and the growing season length to be distinguished. Both the linear mixture model and the derived NDVI values recorded the recovery of the BSC at the beginning of the wet season before annuals had germinated. Finally, it is concluded that a combination of CO₂ exchange measurements, linear mixture model, and NDVI values is suitable for monitoring BSC's productivity in arid regions.     

Temporal and spatial vegetation cover changes in Israeli transition zone: AVHRR-based assessment of rainfall impact

The ability of NOAA/AVHRR data to monitor vegetation response to rainfall in three different vegetation zones was assessed along a north-south transect in Israel. The NDVI database was developed from atmospherically- and radiometrically-corrected NDVI composites from observations spanning three years. Three vegetation zones, Mediterranean region, transition zone, and semiarid region, were geographically separated by means of NDVI values. Based on three years of AVHRR observations during a relatively dry year and two years with near average rainfall, the phenological characteristics for all three vegetation zones were very similar and stable. The results showed that only a few AVHRR observations are necessary to monitor the seasonal and spatial variability of vegetation cover in different climatic zones located in Israel. The NDVI of the Israeli transition zone was found to be very sensitive to rainfall. The difference between maximum and minimum NDVI values in rainy season in the transition zone was at least two times higher than that in the Mediterranean and the semiarid regions. This phenomenon can be used as an indicator of any environmental changes in this region.     

Temporal dynamics of soil and vegetation spectral responses in a semi-arid environment

This paper discusses several difficulties encountered in detecting and monitoring temporal changes in vegetation using multispectral imagery from airborne or spaceborne sensors. These difficulties are due to (1) temporal change in the vegetation state; (2) temporal change in the soil/rock signature; and (3) difficulty in discriminating vegetation from soil or rock background. The seasonal dynamics of soil and vegetation was investigated over two years on permanent sample plots in a natural fenced-off area in the semi-arid region (200 mm annual average rainfall) of the Northern Negev, Israel. Results show that temporal analysis of natural vegetation in semi-arid regions should take into account three ground features--perennials, annuals, and biological soil crusts; all having phenological cycles with the same basic elements--oscillation from null (or low) to full photosynthetic status. However, these cycles occur in successive periods throughout the year. The phenological cycle of perennial plants is related to the adaptation of desert plants to scarcity of water. Annuals are green only for a relatively short period during the wet season and turn into dry organic matter during the summer. The microphytic communities (lower plants) of the biological soil crusts are rapidly affected by moisture and turn green immediately after the first rain, in a timescale of minutes. In arid environments, where the higher plants are sparse, this type of plant has considerable importance in the overall production of the greenness signal. However, crust-covered areas are visually similar to bare soil throughout the dry period. This paper concludes that a priori knowledge of the phenological changes in desert plants (lower and higher) is valuable in the interpretation of remote sensing data of arid environments. It is shown that rainfall amount and regime are the keys for understanding the dynamic processes of the different ground features. Through polynomial fitting, simple functions describing the annual variations in the NDVI of the different cover types have been formulated and validated; showing the feasibility and viability of modelling the processes. Although fluctuations in the rainfall regime between years poses a problem to designing a unique model, it is believed that such a problem can be overcome with long-term observations.     

Desert scrub optical density and spectral-albedo ratios of impacted-to-protected areas by model inversion

Bidirectional surface reflectances measured from NOAA AVHRR over the Negev (southern Israel) and the Sinai are analysed to assess the impact on the surface characteristics of anthropogenic pressures of overgrazing. The impacted Sinai is assumed bare, while the Negev is vegetated by desert scrub. The Negev plants are known to be much darker than the underlying soil, and thus assumed to be absorbing (black). The leaf area distribution as a function of the zenith angle is modelled initially as that of small spheres, which specifies a pronouncedly vertical architecture. We infer from the Negev-to-Sinai reflectance ratios the optical thickness b of the plants (spheres) in the range 0.12 to 0.20 for channel 1 (band centre at 0.63 w m), with only weak seasonal variability. Evaluated from average values of b, the Negev-to-Sinai ratios of the spectral albedos (hemispheric reflectances) are 0.63 and 0.55 in channel 1 and 0.67 and 0.60 in channel 2, at solar zenith angles of 30° and 60°, respectively. These ratios indicate the severe climatic impact of overgrazing in the Sinai, inasmuch as a high albedo means reduced shortwave heat absorption (which is detrimental to rainfall-inducing convection). We subsequently proceed to invert the Negev-to-Sinai reflectance ratios assuming a plant-element distribution tending even more to the vertical. The values of b are reduced when derived for a greater tendency to vertical architecture. The Negev-to-Sinai ratios of the spectral albedos are also significantly lower in these cases, which means that the assessed impact of over-grazing in the Sinai is indeed extremely severe. We conclude that plant architecture (which controls the reflection anisotropy) should be considered when evaluating the albedos of vegetated versus bare (impacted) surfaces from satellite-measured bidirectional reflectances. Uncertainty in the zenith angle distribution of the leaf area produces significant uncertainty in the albedo assessment. Multidirectional reflectance measurements made near the ground would greatly reduce uncertainties about the surface-reflection anisotropy, and thus enhance the value of satellite measurements.     

Desert-scrub as the cause of reduced reflectances in protected versus impacted sandy arid areas

High contrast, commonly in the 1-3— 1-5 range, was observed from satellite multi-spectral radiometers between overgrazed sandy terrain and adjacent protected (fenced-off) areas. Approximately the same contrast was reported in the visible and near-infrared spectral bands. These observations were first conducted at the border of the bright overgrazed Sinai and the darker Negev (in israel). Extensive ground observations were carried out at this border and in the 6 km x 6 km exclosure in the northern Sinai, in which plant growth spontaneously recovered after the area was fenced off in 1974. Karnieli and Tsoar (1995) rejected the predominant role of plants in producing such contrast, concluding that the well-known contrast between Sinai and the Negev, that has drawn the attention of many scientists, is not a direct result of vegetation cover but is caused by an almost complete cover of biogenic crust', This conclusion does not account for these factors: (I) the contrast between dune sand and biogenic crust measured by Karnieli and Tsoar is at most 1-25, and therefore cannot create contrasts of 1-5 observed from satellites; (ii) the dune sand to crust contrast is appreciable only above 0-6 ixm, and the satellite sensor contrast measurements include the 0-5-0-6μ m band; and (iii) ground visual observations and published photographs show individual desert-scrub plants much darker than the soil interstices close to the observer, but at elevated view directions the plants merge, producing a nearly uniform darkening of the terrain. This limb-darkening, observed on the Israeli side along the Sinai/ Negev border, is a clear indication of the strong optical effects of the plants on ihe terrain reflectances.     

Seasonal characterization and discrimination of savannah physiognomies in Brazil using hyperspectral metrics from Hyperion/EO-1

The seasonal characterization and discrimination of savannahs in Brazil are still challenging due to the high spatial variability of the vegetation cover and the spectral similarity between some physiognomies. As a preparatory study for future hyperspectral missions that will operate with large swath width and better signal-to-noise ratio than the current orbital sensors, we evaluated six Hyperion images acquired over the Estação Ecológica de Águas Emendadas, a protected area in central Brazil. We studied the seasonal variations in spectral response of the savannah physiognomies and tested their discrimination in the rainy and dry seasons using distinct sets of hyperspectral metrics. Floristic and structural attributes were inventoried in the field. We considered three sets of metrics in the data analysis: the reflectance of 146 Hyperion bands, 22 narrowband vegetation indices (VIs), and 24 absorption band parameters. The VIs were selected to represent vegetation structure, biochemistry, and physiology. The depth, area, width, and asymmetry of the major absorption bands centred at 680 nm (chlorophyll), 980, and 1200 nm (leaf water) and 1700, 2100, and 2300 nm (lignin-cellulose) were calculated on a per-pixel basis using the continuum removal method. Using feature selection and multiple discriminant analysis (MDA), we tested the discriminatory capability of these metrics and of their combined use for vegetation discrimination in the rainy and dry seasons. The results showed that the spectral modifications with seasonality were stronger with the savannah woodland-grassland gradient represented by decreasing tree height, basal area, tree density and biomass and by increasing canopy openness. We observed a reflectance increase in the red, red edge, and shortwave (SWIR) intervals towards the dry season. In the near-infrared, the reflectance differences between the physiognomies were smaller in the dry season than in the rainy season. From the 22 VIs, the visible atmospherically resistant index (VARI), visible green index (VIg), and normalized difference infrared index (NDII) were the most sensitive indices to water stress and vegetation cover, presenting the largest rates of changes between the rainy (March) and dry (August) seasons in shrub and grassland areas. Absorption band parameters associated with the lignin-cellulose spectral features in the SWIR increased towards the dry season with great amounts of non-photosynthetic vegetation (NPV) in the herbaceous stratum. The opposite was observed for the 680 nm chlorophyll absorption band and the 980 and 1200 nm leaf water features. In general, the number of selected metrics necessary for vegetation discrimination was lower in the dry season than in the rainy season. The best MDA-classification accuracy was obtained in the dry season using nine VIs (79.5%). The combination of different hyperspectral metrics increased the classification accuracy to 81.4% in the rainy season and to 84.2% in the dry season. This combination added a gain higher than 10% for the classification of shrub savannah, open woodland savannah and wooded savannah.     

Relationships between percent vegetation cover and vegetation indices

In this paper, percent vegetation cover is estimated from vegetation indices using simulated Advanced Very High Resolution Radiometer (AVHRR) data derived from in situ spectral reflectance data. Spectral reflectance measurements were conducted on grasslands in Mongolia and Japan. Vegetation indices such as the normalized difference, soil-adjusted, modified soil-adjusted and transformed soil-adjusted vegetation indices (NDVI, SAVI, MSAVI and TSAVI) were calculated from the spectral reflectance of various vegetation covers. Percent vegetation cover was estimated using pixel values of red, green and blue bands of digitized colour photographs. Relationships between various vegetation indices and percent vegetation cover were compared using a second-order polynomial regression. TSAVI and NDVI gave the best estimates of vegetation cover for a wide range of grass densities.     

Spectral modelling of multicomponent landscapes in the Sahel

Abstract

Certain landscapes in the Sahel and elsewhere consist of a ‘checkerboard’ arrangement of vegetated and non-vegetated areas in which there may be several spectrally distinct vegetation and bare ground components. When individual components form large spatially coherent patches, and the vertical dimension of the vegetation is small, spectral interactions between components are negligible. The influence of any one component on the average reflectance of the landscape can then be described by its spectral properties and relative area using simple additive mixture models. These models can be extended to the vegetation indices. The spatial average normalized difference vegetation index (NDVI) is a function of the brightness (red plus near-infrared reflectances), the NDVI and the fractional cover of the components. In landscapes where soil and vegetation can be considered the only components, the NDVI-brightness model can be inverted to obtain the NDVI of the vegetation. Aerial photoradiometer data from Mali, West Africa were used to determine the red and near-infrared component reflectances of soil and vegetation. The derived soil component reflectances were well correlated with ground measurements. The relationship between the vegetation component NDVI and plant cover was better than between the NDVI of the entire landscape and plant cover. The usefulness of this modelling approach depends on the existence of clearly distinguishable landscape components. The method resolves the spectral properties of individual components, but the vegetation component, while free of the effect of bare ground components, is still affected by the underlying soil.     

Modelling spectral reflectance of open cork oak woodland: a simulation analysis of the effects of vegetation structure and background

This study analyses the influence of vegetation structure (i.e. leaf area index and canopy cover) and seasonal background changes on moderate-resolution imaging spectrometer (MODIS)-simulated reflectance data in open woodland. Approximately monthly spectral reflectance and transmittance field measurements (May 2011 to October 2013) of cork oak tree leaves (Quercus suber) and of the herbaceous understorey were recorded in the region of Ribatejo, Portugal. The geometric-optical and radiative transfer (GORT) model was used to simulate MODIS response (red, near-infrared) and to calculate vegetation indices, investigating their response to changes in the structure of the overstorey vegetation and to seasonal changes in the understorey using scenarios corresponding to contrasting phenological status (dry season vs. wet season). The performance of normalized difference vegetation index (NDVI), soil-adjusted vegetation index (SAVI), and enhanced vegetation index (EVI) is discussed. Results showed that SAVI and EVI were very sensitive to the emergence of background vegetation in the wet season compared to NDVI and that shading effects lead to an opposing trend in the vegetation indices. The information provided by this research can be useful to improve our understanding of the temporal dynamic of vegetation, monitored by vegetation indices.     

Evaluating the species energy relationship with the newest measures of ecosystem energy: NDVI versus MODIS primary production

Ecosystem energy has been shown to be a strong correlate with biological diversity at continental scales. Early efforts to characterize this association used the normalized difference vegetation index (NDVI) to represent ecosystem energy. While this spectral vegetation index covaries with measures of ecosystem energy such as net primary production, the covariation is known to degrade in areas of very low vegetation or in areas of dense forest. Two of the new vegetation products from the MODIS sensor, derived by integrating spectral reflectance, climate data, and land cover, are thought to better approximate primary productivity than NDVI. In this study, we determine if the new MODIS derived measures of primary production, gross primary productivity (GPP) and net primary productivity (NPP) better explain variation in bird richness than historically used NDVI. Moreover, we evaluate if the two productivity measures covary more strongly with bird diversity in those vegetation conditions where limitations of NDVI are well recognized.Biodiversity was represented as native landbird species richness derived from the North American Breeding Bird Survey. Analyses included correlation analyses among predictor variables, and univariate regression analyses between each predictor variable and bird species richness. Analyses were done at two levels: for all BBS routes across natural landscapes in North America; and for routes in 10 vegetation classes stratified by vegetated cover along a gradient from bare ground to herbaceous cover to tree cover. We found that NDVI, GPP and NPP were highly correlated and explained similar variation in bird species richness when analyzed for all samples across North America. However, when samples were stratified by vegetated cover, strength of correlation between NDVI and both productivity measures was low for samples with bare ground and for dense forest. The NDVI also explained substantially less variation in bird species richness than the primary production in areas with more bare ground and in areas of dense forest. We conclude that MODIS productivity measures have higher utility in studies of the relationship of species richness and productivity and that MODIS GPP and NPP improve on NDVI, especially for studies with large variation in vegetated cover and density.     

Evaluating the species energy relationship with the newest measures of ecosystem energy: NDVI versus MODIS primary production

Ecosystem energy has been shown to be a strong correlate with biological diversity at continental scales. Early efforts to characterize this association used the normalized difference vegetation index (NDVI) to represent ecosystem energy. While this spectral vegetation index covaries with measures of ecosystem energy such as net primary production, the covariation is known to degrade in areas of very low vegetation or in areas of dense forest. Two of the new vegetation products from the MODIS sensor, derived by integrating spectral reflectance, climate data, and land cover, are thought to better approximate primary productivity than NDVI. In this study, we determine if the new MODIS derived measures of primary production, gross primary productivity (GPP) and net primary productivity (NPP) better explain variation in bird richness than historically used NDVI. Moreover, we evaluate if the two productivity measures covary more strongly with bird diversity in those vegetation conditions where limitations of NDVI are well recognized.Biodiversity was represented as native landbird species richness derived from the North American Breeding Bird Survey. Analyses included correlation analyses among predictor variables, and univariate regression analyses between each predictor variable and bird species richness. Analyses were done at two levels: for all BBS routes across natural landscapes in North America; and for routes in 10 vegetation classes stratified by vegetated cover along a gradient from bare ground to herbaceous cover to tree cover. We found that NDVI, GPP and NPP were highly correlated and explained similar variation in bird species richness when analyzed for all samples across North America. However, when samples were stratified by vegetated cover, strength of correlation between NDVI and both productivity measures was low for samples with bare ground and for dense forest. The NDVI also explained substantially less variation in bird species richness than the primary production in areas with more bare ground and in areas of dense forest. We conclude that MODIS productivity measures have higher utility in studies of the relationship of species richness and productivity and that MODIS GPP and NPP improve on NDVI, especially for studies with large variation in vegetated cover and density.     

Sub‐pixel reflectance unmixing in estimating vegetation water content and dry biomass of corn and soybeans cropland using normalized difference water index (NDWI) from satellites

Estimating vegetation cover, water content, and dry biomass from space plays a significant role in a variety of scientific fields including drought monitoring, climate modelling, and agricultural prediction. However, getting accurate and consistent measurements of vegetation is complicated very often by the contamination of the remote sensing signal by the atmosphere and soil reflectance variations at the surface. This study used Landsat TM/ETM+ and MODIS data to investigate how sub‐pixel atmospheric and soil reflectance contamination can be removed from the remotely sensed vegetation growth signals. The sensitivity of spectral bands and vegetation indices to such contamination was evaluated. Combining the strengths of atmospheric models and empirical approaches, a hybrid atmospheric correction scheme was proposed. With simplicity, it can achieve reasonable accuracy in comparison with the 6S model. Insufficient vegetation coverage information and poor evaluation of fractional sub‐pixel bare soil reflectance are major difficulties in sub‐pixel soil reflectance unmixing. Vegetation coverage was estimated by the Normalized Difference Water Index (NDWI). Sub‐pixel soil reflectance was approximated from the nearest bare soil pixel. A linear reflectance mixture model was employed to unmix sub‐pixel soil reflectance from vegetation reflectance. Without sub‐pixel reflectance contamination, results demonstrate the true linkage between the growth of sub‐pixel vegetation and the corresponding change in satellite spectral signals. Results suggest that the sub‐pixel soil reflectance contamination is particularly high when vegetation coverage is low. After unmixing, the visible and shortwave infrared reflectances decrease and the near‐infrared reflectances increase. Vegetation water content and dry biomass were estimated using the unmixed vegetation indices. Superior to the NDVI and the other NDWIs, the SWIR (1650 nm) band‐based NDWI showed the best overall performance. The use of the NIR (1240 nm), which is a unique band of MODIS, was also discussed.     

Seasonal landscape and spectral vegetation index dynamics in the Brazilian Cerrado: An analysis within the Large-Scale Biosphere-Atmosphere Experiment in Amazônia (LBA)

The Brazilian Cerrado biome comprises a vertically structured mosaic of grassland, shrubland, and woodland physiognomies with distinct phenology patterns. In this study, we investigated the utility of spectral vegetation indices in differentiating these physiognomies and in monitoring their seasonal dynamics. We obtained high spectral resolution reflectances, during the 2000 wet and dry seasons, over the major Cerrado types at Brasilia National Park (BNP) using the light aircraft-based Modland Quick Airborne Looks (MQUALS) package, consisting of a spectroradiometer and digital camera. Site-intensive biophysical and canopy structural measurements were made simultaneously at each of the Cerrado types including Cerrado grassland, shrub Cerrado, wooded Cerrado, Cerrado woodland, and gallery forest. We analyzed the spectral reflectance signatures, their first derivative analogs, and convolved spectral vegetation indices (VI) over all the Cerrado physiognomies. The high spectral resolution data were convolved to the MODIS, AVHRR, and ETM⁺ bandpasses and converted to the normalized difference vegetation index (NDVI) and the enhanced vegetation index (EVI) to simulate their respective sensors. Dry and wet season comparisons of the measured biophysical attributes were made with the reflectance and VI data for the different Cerrado physiognomies. We found that three major domains of Cerrado could be distinguished with the dry and wet season spectral signatures and vegetation indices. The EVI showed a higher sensitivity to seasonality than the NDVI; however, both indices displayed seasonal variations that were approximately one-half that found with the measured landscape green cover dynamics. Inter-sensor comparisons of seasonal dynamics, based on spectral bandpass properties, revealed the ETM⁺-simulated VIs had the best seasonal discrimination capability, followed by MODIS and AVHRR. Differences between sensor bandpass-derived VI values, however, varied with Cerrado type and between dry and wet seasons, indicating the need for inter-sensor VI translation equations for effective multi-sensor applications.     

Dynamics of the relationship between NDVI and SWIR32 vegetation indices in southern Africa: implications for retrieval of fractional cover from MODIS data

Fractional cover of photosynthetic vegetation (F_PV), non-photosynthetic vegetation (F_NPV), and bare soil (F_BS) has been retrieved for Australian tropical savannah based on linear unmixing of the two-dimensional response envelope of the normalized difference vegetation index (NDVI) and short wave infrared ratio (SWIR)32 vegetation indices (VI) derived from Moderate Resolution Imaging Spectroradiometer (MODIS) reflectance data. The approach assumes that cover fractions are made up of a simple mixture of green leaves, senescent leaves, and bare soil. In this study, we examine retrieval of fractional cover using this approach for a study area in southern Africa with a more complex vegetation structure. Region-specific end-members were defined using Hyperion images from different locations and times of the season. These end-members were applied to a 10-year time series of MODIS-derived NDVI and SWIR32 (from 2002 to 2011) to unmix F_PV, F_NPV, and F_BS. Results of validation with classified high-resolution imagery indicated major bias in estimation of F_NPV and F_BS, with regression coefficients for predicted versus observed data substantially less than 1.0 and relatively large intercept values. Examination with Hyperion images of the inverse relationship between the MODIS-equivalent SWIR32 index and the Hyperion-derived cellulose absorption index (CAI) to which it nominally approximates revealed: (1) non-compliant positive regression coefficients for certain vegetation types; and (2) shifts in slope and intercept of compliant regression curves related to day of year and geographical location. The results suggest that the NDVI–SWIR32 response cannot be used to approximate the NDVI–CAI response in complex savannah systems like southern Africa that cannot be described as simple mixtures of green leaves, dry herbaceous material high in cellulose, and bare soil. Methods that use a complete set of multispectral channels at higher spatial resolution may be needed for accurate retrieval of fractional cover in Africa.     

Remote sensing analysis of the land surface temperature anomaly in the sand-dune region across the Israel-Egypt border

In the sand-dune region across the Israel-Egypt border, an anomalous phenomenon of thermal variation was observed on remote sensing images: the Israeli side with much more vegetation cover has higher surface temperature than the Egyptian side, where bare sand surface prevails. The study intends to examine the phenomenon using NOAA-AVHRR and Landsat TM data. The focus is to analyse the seasonal and spatial change of land surface temperature (LST) in the border region, to verify it through ground truth measurements and to simulate the average LST change on both sides according to surface composition structure. A split window algorithm containing only two parameters (transmittance and emissivity) has been developed for retrieving LST from NOAA-AVHRR data and a mono-window algorithm is proposed for computing LST from the only one thermal band of Landsat TM data. Application of these algorithms to the available AVHRR and Landsat TM data indicates that the LST anomaly does occur not only in one day but almost all the year. In hot dry summer the Israeli side is usually about 2.5-3.5°C hotter. In wet cool winter the LST difference between the sides is not large but the Israeli side still has higher LST. The Egyptian side may have slightly higher LST when surface temperature is below 20°C, several days after heavy rain, which leads to very wet surface conditions. The sharp LST contrast disappears on night-time images. Ground truth measurements indicate that the LST contrast mainly can be attributed to the surface temperature difference on the two typical surface patterns: biogenic crust and bare sand, which have above 3°C difference in surface temperature during summer. Experiments on soil samples from the field indicate that biogenic crust and sand have emissivity values of about 0.972 and 0.954, respectively, in hot dry conditions that match the environment of the region in summer. Surface composition determination based on three methods indicates that more than 72% of the ground on the Israeli side is covered with biogenic crust and more than 80% on the Egyptian side is bare sand. Actually, the LST anomaly can be understood as the direct result of surface composition difference, especially in biogenic crust and sand cover rate. Simulation with this surface composition difference shows that the Israeli side has steadily higher LST when the temperature of the biogenic crust is more than 1°C higher that of the sand surface, which usually occurs at moderate to high temperature levels (>30°C). When temperature is between 15 and 25°C, such as at about midnight, the two sides will have no obvious LST difference. This result is in agreement with the remote sensing observation. Therefore, it can be concluded that the vegetation cover does not contribute much to the LST contrast in comparison to the effect of the biogenic crust and sand cover.     

Analysis and optimization of NDVI definitions and areal fraction models in remote sensing of vegetation

Variations in the definition of the Normalized Difference Vegetation Index (NDVI) and inconsistencies in vegetation areal fraction models prejudice the understanding of long‐term variability and change in land cover. We analysed the consequences of using NDVI definitions based on the digital number (DN), spectral radiance and spectral reflectance for six active and high spatial resolution multi‐ and hyperspectral satellite sensors (ALI, ASTER, ETM+, HRVIR, Hyperion and IKONOS) and optimized the NDVI definitions, and then examined the performance of three vegetation areal fraction models: the linear reflectance, linear NDVI and quadratic NDVI models. The examination was performed for three plots chosen from two biomass zones: a short and small leaf area index (LAI) creosote shrub zone, and a tall and large‐LAI piñon‐juniper zone. The results show that: (1) the difference in NDVI values among the NDVI definitions is sensor dependent and always significant; spectral reflectance should be used in NDVI calculations, and using radiance or DN values in calculating the NDVI should be avoided; (2) in deriving vegetation areal coverage, the linear reflectance model outperforms the other two models in the shrub biomass zone; and (3) the linear NDVI model outperforms the other two models in the piñon‐juniper biomass zone. These observations are consistent with the fact that the non‐linear effect is less important in shrubland than in piñon‐juniper woodland and that the linear NDVI model is more capable of capturing non‐linearity in the spectral analysis.     

Assessing the biomass dynamics of Andean bofedal and totora high-protein wetland grasses from NOAA/AVHRR

Biomass measurements of totora and bofedal Andean wetland grasses in the Bolivian Northern Altiplano were correlated over a growing season to vegetation indices derived from 1-km visible and near-infrared bands of the advanced very high resolution radiometer (AVHRR) instrument flown on the NOAA-14 polar-orbiting meteorological satellite. This article discusses the potential and limits of these indices for the assessment of the spatial and temporal variation of biomass and of the fraction of the photosynthetic active radiation absorbed by these herbaceous native forages growing in water-saturated environments. Bidirectional reflectance distribution function (BRDF) normalization was also investigated based on simple kernel-driven models. BRDF normalized difference vegetation index (NDVI) performed the best for both totorales and bofedales vegetation associations, followed by the uncorrected maximum-value composite NDVI. BRDF normalized NDVI was shown to be sensitive to the green leaf or photosynthetically active biomass.Estimation of biomass production after Kumar and Monteith (1982) was used to determine the efficiency of solar energy conversion into biomass (ε_b) for the main phenological periods, corresponding to the rainy and dry seasons. Two approaches were investigated for the biomass production estimation: the first one is based on monthly field biomass measurements; the second one is based on estimates from the regression computed previously using Roujean's BRDF normalized NDVI. The values found for these efficiencies for the rainy season agree with those of the literature for grasslands of temperate regions. For the dry season, more accurate information on totora and bofedal senescence and on animal consumption is required to get a reasonable efficiency value. This is not surprising, as other workers have reported biomass estimation with remotely sensed data to be most relevant to the growing season.