In the ESA study GeoBIRD (Geo- and Biophysical Information from Remote Sensing Data) the process-based vegetation growth model PROMET-V was coupled to the GeoSAIL canopy reflectance model for the retrieval of spatially distributed biophysical parameters from surface reflectance images derived from Landsat TM data. The raster based PROMET-V model calculates plant growth, water and nitrogen fluxes using meteorological data, a land use map, and a soil map. In this study the total leaf area index (LAI), the fraction brown leaf area and surface soil moisture, as modelled in PROMET-V, were used in conjunction with GeoSAIL, to model the surface reflectance spectra. Besides modelling the canopy reflectance GeoSAIL, also accounts for the soil reflectance using the dry soil spectrum and surface moisture as inputs. By minimising the difference between observed reflectance spectra derived from LANDSAT images and the modelled surface reflectance spectra, the total LAI, fraction of brown leaves and surface soil moisture were estimated. This optimisation process is constrained by physically and plant physiologically meaningful boundaries in order to obtain valid results and accelerate the process. By re-initialising the PROMET-V model, the retrieved and simulated LAI were matched. This leads to improved model results regarding biomass and yield.

The spatially distributed PROcess oriented Modular Environment and Vegetation model (PROMET-V) is presented. PROMET-V models water, carbon and nitrogen fluxes, their interactions and feedback. The model is based on physical and plant physiological equations. Remote sensing data are used to provide spatially distributed model inputs, to update model parameters and to validate model results. Dedicated interfaces were developed to use remote sensing data and to estimate key model parameters. Using LANDSAT TM data, model parameters such as plant density and frequency and date of cutting of meadows were estimated. Comparisons with ground truth measurement showed a significant improvement of the model results by using remote sensing data. Within-field biomass and LAI heterogeneities could be successfully detected. Through interactions and feedback mechanisms, the spatial patterns detected by the TM images also affect model parameters and fluxes which cannot be measured directly. The combined analysis of remotely measured and modeled parameters bears the potential to determine parameters which cannot be derived directly with either method alone, such as weed infestation or soil texture. Remote sensing data are also used to provide independent validation data. Examples are shown using soil moisture maps derived from ERS SAR data.

A prototype system for farm decision making was developed allowing satellite images to be properly displayed and combined with land parcel data. As a part of this prototype system, an Algorithm provides tools for the partial automation of the decision process to formulate N-fertilisation recommendations, in the form of a Nitrogen Application Map. The most important part of the algorithm is the work unit based on very high resolution remotely sensed data. To develop this work unit, field measurements were taken parallel to satellite images during two growing seasons (1999-2000) on winter wheat parcels with different plant densities and nitrogen treatments. The parameters measured included reflectance and Leaf Area Index (LAI) during the growing season and yield, protein content and number of spikes, at harvest. Several vegetation indices (VI’s), calculated from the satellite and ground data, were studied with respect to their sensitivity to the different nitrogen doses and to low noise production. The correlation between VI’s and important crop parameters for nitrogen fertilisation management (i.e. LAI) was tested as well.

In the south-west of Western Australia, the clearing of land for agricultural production has lead to rising saline ground water, resulting in the loss of previously productive land to salinity; damage to buildings, roads and other infrastructure; the decline in pockets of remnant vegetation and biodiversity; and the reduction in water quality. The region in question comprises some 24 million hectares of land. This has resulted in a wide variety of stakeholders requesting quantitative information regarding historical, present and future trends in land condition and use. Historically, two methods have been widely used to obtain information: (1) surveys requesting land managers to provide estimates of land use and condition; and (2) human interpretation of aerial photography. Data obtained from the first approach has in the past been incomplete, inaccurate and non-spatial. The second approach is relatively expensive and as a result is incomplete and is not regularly updated.In this paper, we describe an approach to land use/condition monitoring using remotely sensed and other data such as digital elevation models (DEMs). We outline our methodology and give examples of mapping and monitoring change in woody vegetation and salinity.

We present a simple algorithm to predict canopy chlorophyll content from directional reflectance spectra. The purpose is the compensation of the branches, soil and multiple reflection effect in the reflectance in order to obtain the single leaf spectra that are more closely correlated with chlorophyll content, as already assessed in literature. The analysis is limited to leaves with a chlorophyll content higher than 25-30 µg/cm2 since we verified that their spectra are little affected by multiple leaf reflection in the green spectral region, where the leaf reflectance is mostly dependent on chlorophyll concentration. In order to understand better the nature of the canopy reflectance, we measured the reflectance spectrum of both single and multiple leaves of two kinds of plants (Pittosporum tobira and Nerium oleander), of some branches and of some brown soil. Finally, we developed an algorithm able to translate the canopy reflectance to the single leaf level. We tried the algorithm on a canopy of Pittosporum. We found that it was able to retrieve the true chlorophyll content with an error ~ 10%.

Remote sensing instruments of the present and future generations include a variety of multi-angular off-nadir measuring facilities. In order to fully exploit their possibilities a thorough understanding of the anisotropic angular reflection properties of terrestrial surfaces is required. These are generally quantified in terms of the bi-directional reflectance distribution function (BRDF). We report on near infrared BRDF measurements of various vegetative surfaces including several crops performed with a CCD camera. While many of the investigated vegetation types show a rather similar reflection behavior, there are also distinct differences observed in some cases. For a quantitative analysis of the results we introduce several statistical measures which describe the characteristic properties of the reflectance distribution. We use these parameters as the input for an unsupervised cluster analysis algorithm. As a result the method provides suggestions for grouping different vegetation types into classes according to their angular reflection properties. This is helpful for evaluating which properties of the plants or the plant canopy structure cause recognizable reflectance features. The results can therefore be used to develop adapted observation strategies for the retrieval of biophysical parameters in agricultural or environmental studies.

Remotely sensed reflectance from stressed and non-stressed crop vegetative cover can be predicted from two combinations of spectral bands as a ratio or as normalized vegetation indices. The most common spectral bands used lie in the red and infrared region (350-800 nm) and are dominated by the absorption of chlorophyll and other accessory pigments. In addition, reflectance in the middle infrared is dominated by absorption from liquid water contained in plant’s tissues. The objectives of the present work are: 1) to develop characterization model to evaluate the reflectance data from frequently irrigated and water stressed alfalfa, Sudan grass, and other crops such cotton as using a handheld radiometer and assess the spectral correlation with the ground-truth and; 2) The model will be better model to evaluate the stressed crops. The experiment was designed to collect reflectance data from cotton crops planted at the Blythe area, California. The fields are planted with cotton crops in different stages of maturity at Longitude of -114°32.79 and -114°32.80 and Latitude 33° 39.64. With a field spectrometer, the scan over each treatment was made at 1 hr intervals between 10:00 a.m. and 2:00 p.m. Pacific Day Time (PDT). Vegetative samples were taken from the two treatments (i.e. stressed and unstressed vegetation) during the initial sampling for purposes of conducting chemical analysis. Soil samples were collected to determine the amount of available soil moisture differences in the two treatments. The suggested model in the present paper is called the Model of Water Stress (MWS) where it include in it the statistical values and parameters, indicates that the stressed crops have values higher than unstressed crops in MWS scale. This means that the model is differentiating between the stressed and unstressed vegetation. Additional work will evaluate the reflectance peaks and their relationship to other parameters that were collected and are relevant to the applications of the model. The model will be tested against the AVIRIS data that were collected at the same time of the collection of ground-truth data.

A methodology for deriving spatially averaged emitting layer temperature from high frequency microwave observations is presented. Microwave brightness temperature is a function of the emissivity and the physical temperature of the emitting soil layer, thereby possessing a strong physical basis for estimating soil temperature. Field observations have shown that maximum and minimum daily air temperatures are strongly related to midday and midnight surface soil temperature. Field measurements of surface temperature are also compared to METEOSAT thermal observations. Long term daily maximum and minimum air temperatures are used to derive data sets of daytime and nighttime surface temperatures. Results indicate that 37 GHz vertical polarization brightness temperature provides a reasonable estimate of the emitting layer soil temperature. This technique is especially useful for normalizing microwave brightness temperatures at longer wavelengths for soil moisture retrieval algorithms. It could provide a useful tool for climate modelling, land surface processes investigations, and other energy balance applications by providing consistent and independent long term estimates of daily global surface temperature.

The potential of satellite passive microwave remote sensing for the monitoring of surface moisture in semi-arid areas is generally recognized. While the largely unknown behavior of a.o. the vegetation parameters in the radiative transfer algorithm requires the validation of the satellite derived surface moisture for a restricted calibration time period, it often proves difficult due to the lack of adequate surface moisture field measurements on the regional scale and/or for the corresponding lifetime of the space platform. The present study describes the indirect validation of surface moisture derived from historical Nimbus/SMMR data (1978-1987) collected over the Upper Guadiana Catchment in central Spain using recharge estimates from historical surface flow measurements. The recharge estimates are used to calibrate a soil moisture balance model, which is applied in the absence of surface moisture measurements. If the followed approach is successful, satellite passive microwave inversion techniques may be used to assess recharge estimates. Since the recharge rate corresponds to the maximum abstraction yield without depletion of the groundwater reservoir, the approach may provide a tool for the sustainable development of areas suffering from the overexploitation of water resources.

This paper deals with the development of a water budget model for Eucalyptus forest, using a conceptually simple onedimensional mass balance approach within the root zone of the forest. The model uses Leaf Area Index to quantify the forest structure important for mass and energy exchange, and this represents a key simplification for regional scale applications. Remote Sensing vegetation indexes and mixture modeling techniques were used to estimate LAI. A five-layered water balance model, with water movement between layers along hydraulic gradients, was developed and parameterized for a eucalypt plantation (Eucalyptus grandis Hill ex.Maiden hybrids) in Brazil. Available soil water controls stomatal conductance and hence transpiration, which is calculated by the Penman-Monteith equation. The remote sensing derived LAI was used to compute the canopy conductance that drives the Penman-Monteith formulation. The model accounts for changes in the depths of the water table. The test period was from October 1995 to September 1996 in a nine-year-old plantation in an experimental catchment in eastern Brazil. Total transpiration for the year was 1116 mm, with 119 mm intercepted and re-evaporated and another 79 mm soil surface evaporation, giving evapotranspiration of 1314 mm compared to rainfall of 1396 mm. The water balance was closed by net flow below the root zone of about 53 mm and an increase in water storage (in the first layer) of 29 mm. The model also estimated a water deficit of 135 mm (difference between the potential and current transpiration) for the period. Upward flux from the water table was around 81 mm and piezometric measurements showed 1.5 m recession for the same period. The upward flux into the root zone was about 1 mm day-1 at the end of a long dry season; that kept the water storage in that zone to about 15% of capacity and helped prevent complete stomatal closure. Comparison between estimated water storage and measurements confirmed that this model is a very promising tool for calculating water use by plantations. It can also provide water balance information and information about stomatal conductance for growth prediction models.

In this paper remotely sensed images and ancillary data are used to assess the spatial distribution and seasonal variation of the SCS - Curve Number (CN) parameter (USDA, 1985) within a watershed. Two watersheds were considered in Thirrenian Liguria (North - Western Italy): Bisagno and Entella river. Two Landsat - 5 Thematic Mapper scenes were acquired during different season: late winter and late spring. A Supervised Maximum Likelihood Classification technique was used to obtain a land use map for the two catchments. To characterize the effect of the different surface condition within a watershed in relation to their hydrologic condition, the status of vegetation cover was analyzed in the two seasons. Seasonal variation was described through the evaluation of Fractional Vegetation Cover using an empirical relationship tested in a typical Mediterranean area similar to the case study. Land use and status of vegetation cover information were integrated in a Geographical Information System with lithological maps, in order to estimate the CN parameter in a distributed way. Values of spatially distributed CN parameter were evaluated by comparing measured runoff volumes during a storm events with runoff volumes simulated by a physical - based spatially distributed hydrological model.

Airborne scanning laser altimetry (LiDAR) is an important new data source for environmental applications, mapping topographic and surface object heights to high vertical and spatial accuracy over large areas. We present results of a segmentation system for LiDAR data for a reach of the river Severn, UK. The system has been developed to improve the 3 main data required by a leading numerical flood model predicting inundation extent, namely (i) a map of topographic height providing model bathymetry. A comparison with ground control points gives an accuracy of ±17cm (decreasing in the presence of steeply wooded slopes), (ii) the meandering location of the river channel and a suitable height contour which denote the extent of the model domain, and allow immediate finite element mesh generation, and (iii) a map of vegetation height (to an accuracy of ±14cm for grass and cereal crops) which is converted to friction coefficients. Errors due to overlapping swaths are significantly reduced. A 3-class segmentation of vegetation types (short, intermediate and tall) allows optimal height extraction algorithms to be separately applied, and enables realistic conversion to friction coefficients. Short (grass and cereal crops) and intermediate (hedges) vegetation are assumed to be flexible and either emergent or submerged during a flood cycle. Trees (tall vegetation) are modelled as rigid, emergent, stems.

Wind Scatterometers are active microwave instruments with low spatial resolution and high sampling rate. Recent studies have shown high potentials of these data to monitor land surface parameters over semi-arid areas, including the soil moisture and the vegetation herbaceous mass. The objective of this study is to evaluate the potentialities of the ERS Wind Scatterometer to retrieve land surface parameters. After a brief presentation of the model used for the interpretation of ?° time series, the inverse problem aiming at estimating herbaceous mass and soil moisture time series given the ERS WSC data is analysed. Due to the strong spatial and temporal variability of the soil moisture, the inverse problem appears to be a priori under-determined. We then solve the inverse problem with a “brute force” approach that consists in systematical exploration of the parameter space. This method does not only allow to obtain the optimal solutions like more classical method (generalised least square, simplex), but also the whole domain of admissible solutions. Analysis of this domain provides interesting results for the inverse problem subtle understanding

In developing and tuning passive microwave algorithms, which are used to estimate snow extent, snow water equivalent and snow depth, much of the effort has been directed towards better accounting for the effects of snow crystal size on the microwave response, and relatively little effort has been given to the role that crystal shape or orientation plays in this regard. Modeling using a discrete dipole scattering models has shown that the assumption used in radiative transfer approaches, where snow crystals are modeled as randomly oriented spheres, is adequate to account for the transfer of microwave energy emanating from the ground and passing through a snowpack. With this in mind and by having some knowledge of the size of the particles in the snowpack as well as the snowpack density, snow depth algorithms can be designed for specific basins to assess the snow water equivalent of the basin and to thus, estimate snowmelt runoff and seasonal streamflow. Work performed on an ongoing GCIP/GEWEX experiment for watersheds in the upper Mid West and the northern Great Plains (the Roseau river basin in Minnesota/Manitoba, and the Black river basin in Wisconsin) has shown that for each of these basins, a strong conelation exists between snow depth derived from SSMI passive microwave data and snow depth measured at meteorological stations and determined from airborne gamma overflights. For instance, for the Roseau basin in mid March (Julian day 75), during the period from 1992-1998, the coefficient of determination (R2) is a very strong 0.8975. Thus, ninety percent of the mid March snow depth variation in this basin, during these years, can be explained by the SSMI snow algorithm. Streamfiow has also been correlated with maximum seasonal snow depth for these two basins as well (Figure 3). Using only SSMI-derived snow depth as the predictor or dependent variable, the R2 value for the Roseau basin was 0.715 between the basin-wide snow depth on March 15 and ensuing streamfiow for the month of April. When there is a high degree of assurance that the satellite-derived estimates are reliable (the algorithms produce results which reflect the streamfiow — hydrographs), they can then be used to generate input to hydrologic models.

Active fires detected using remotely sensed data represent only a small fraction of the vegetation fires. Hence, several studies focused recently on directly mapping burnt areas using polar orbiting satellites, such as Landsat, NOAA and ERS- ATSR. European METEOSAT and Japanese GMS geostationary satellites were used to detect burned areas in different tropical environments, despite their low spatial and spectral resolutions. The methodology is based on a multiple threshold approach applied to the thermal radiance and to a spectral index specific for burned surfaces. The simple index for burned area (SIB A) makes use of the information contained in visible and thermal IR bands available on geostationary satellites. Such criteria take into account also the information concerning the background thermal characteristics. The results from METEOSAT were evaluated using AVHRR data acquired over the Central Africa forest-savannah areas during the EXPRESSO campaign of dry season 1996. For GMS imagery, AVHRR data acquired over the woodland savannah areas of Northern Territories in Australia during the Smoko-FRACTAL campaign (May-June 1999) were used. Accuracy is satisfactory in both tropical environments, and these maps provide a reliable input to available atmospheric models.

Legislation was introduced in Portugal to enable the Portuguese Forest Services (DGF) to implement a management strategy to control forest plantation in forest burned areas. Under this legislation, the forest owners have to notify DGF after a fire in order to replant the same species, or to seek DGF’s permission if the previous species is to be replaced. We developed a methodology that uses Earth Observation (EO) data to identify the species composition of the forests before being burned, and to yearly supervise the burnt areas in order to identify new reforestations and check if the forest owners are in compliance with the Portuguese legislation on burnt areas. Our methodology is based on vegetation indices differencing and map algebra leading to the production of maps where potential illegal areas are identified. These areas are then field checked to identify the forest species that was planted and compare it with the pre-fire forest cover to detect legal and illegal situations. The methodology was successfully tested in a study area in central Portugal with an extension of 640 km2. The results are encouraging for an operational implementation of the methodology by DGF, leading to an efficient application of this specific legislation.

Early remote sensing of forest fires from specifically dedicated low cost satellites has recently been proposed as one of the most promising techniques for the improvement in the efficiency of forest fire fighting on a global basis. Efficient forest fire remote sensing requires a high probability of detection for small fires combined to a low' false alarm rate. In this paper, a very simple algorithm based on the so-called fire index (FI) has been implemented in an acquisition system developed within the framework of the UE-DGXII Project FUEG02. This system is composed by tw;o infrared cameras operating in the mid and thermal infrared spectral regions and acquire simultaneous digital images of the scene that are calibrated in radiometric units. An image of the FI is then computed improving greatly the discrimination of false alarms. A new approach using only the mid infrared band is also suggested, and spectral intraband processing is studied as a basis for this approach.

The states of Honduras, Nicaragua, and El Salvador have been toughly hit by the “Mitch” hurricane in the first days of November 1998. The extent of damages due to this hurricane, as well as their impact upon local economy were exceptional. In this framework, a remote sensing project was scrambled to provide a large scale evaluation of the damages suffered by the three countries. To reach this objective, new remote sensing products called DYNAMIC products have been designed. These products are, based on using, either RADAR satellite images, or Optical/RADAR satellite data fusion. The most efficient techniques have been applied to produce these products, thus enabling change detection at a very fine spatial scale (10x10 meters). Project schedule and operations, as well as the validation of its products are reported in this paper.

A study has been carried out using MODTRAN 3.5 simulations of the Along-Track Scanning Radiometer-2 (ATSR-2) data at 3.7, 11 and 12 µm wavelengths to give a great range of algorithms for estimating sea surface temperature (SST) and land surface temperature (LST). Algorithms based on split-window, dual-angle and mixed structure have been considered. The coefficients of the algorithms are derived by regression analysis using the MATLAB code. The results show that, in general, dual-angle algorithms give better results than split-window ones, retrieving LST with a standard deviation as low as 0.4 K and 0.6 K respectively if the satellite data are error free. The introduction of 3.7 µm channel involves less error in the estimation of surface temperature. Water vapor dependence supposes an improvement of the accuracy of the results.

The characteristics of snow backscattering in microwave band are considered. It is shown that they depend upon many parameters describing the state of snow cover. The basic dependences obtained from theory of radar backscattering and experimental data are discussed and compared. Real radar remote sensing data for different dates in fall/winter period are presented for Eastern regions of Ukraine. It is demonstrated that the effects predicted theoretically are observed for remote sensing images obtained from side-look radar installed on the satellite “Ocean”. It is also shown that remote sensing means in some conditions permit to get the maps of snow cover thickness but in this case the reference data from several on-land meteorological stations are needed for calibration purposes.

In the field of remote sensing applications, more than 40 vegetation indices have been developed in recent years with the aim of minimizing the influence of internal and external factors (such as soil properties and atmosphere) which can affect the radiometric response of vegetation canopies. However, although those indices have showed good performances from laboratory and simulated data, most of them are difficult to be implemented from satellite data because of their complex definition that frequently requires the knowledge of different parameters besides the reflectance itself. That is the case of the generalized soil-adjusted vegetation index (GESAVI). The GESAVI was developed on the basis of a simple vegetation canopy model. It is defined in terms of the near-infrared NIR and red R reflectances and the soil line parameters (A and B) as: GESAVI = (NIR-BR-A)/(R+Z), where Z is related to red reflectance at the cross point between the soil line and the vegetation isolines in the NIRJR plane. This new index showed a better normalization of soil background effects when compared to the traditional NDVI using different reflectance data sets (acquired under laboratory conditions as well as by means of a simulation procedure). At present, a methodology is proposed to implement the GESAVI from satellite data. We will focus our attention mainly on semiarid landscapes, where the perturbance introduced by soil optical properties is very important. It would be desirable that the application of this new vegetation index to satellite images would require only information contained in the image itself. This is the main goal of the present research. Results show that GESAVI can be easily obtained from NDVI.

The BRDF (Bidirectional Reflectance Distribution Function) of vegetation canopies exhibits an anisotropic behaviour that is related to illumination and viewing geometries. However, some other aspects such as the optical properties and the structural parameters of the targets should be taken into account for an adequate explanation of the bidirectional phenomenon. The present investigation examines the anisotropic behaviour of the homogeneous canopies reflectance from laboratory data as a function of viewing geometry, structural parameters and optical properties of the samples in order to obtain relevant information to improve biophysical parameters retrieval and discrimination of vegetation canopies from optical spectral data. Airborne data acquired in Daisex-99 campaign over Barrax test site (Albacete/Spain) with the HyMap instrument are also included. The HyMap concept is able to record hot spot effect, and moreover, the different flight tracks carried out in Daisex-99 allow us to complete anisotropic behaviour shown in laboratory experience, where illumination was fixed, with airborne data acquired under different solar zenith angle. Results confirm initial hypothesis that anisotropy reflectance is related to structural parameters of the vegetation and show anisotropic behaviour usefulness to study vegetation canopies increasing data dimensionality, varying both illumination and view angles. The anisotropy factor, ANIF, has resulted a simple relationship to provide us with relevant information about vegetation canopies structure. Keywords- Vegetation Canopies, Anisotropy, Reflectance, Hot Spot, Hymap.

This paper explains the methodology and results of research on the growth of large forest fires, using information from the NOAA-AVHRR sensor. Numerous algorithms have been developed for discriminating pixels of active fires from the rest. These can be divided into three groups: a) those based on information from channel 3 (single channel); b) those using algorithms from several channels (multi-channel); c) those that, as well as using this information, compare the pixel with those around it (contextual). This research used the contextual algorithm of Flasse and Ceccato1 to discriminate between pixels corresponding to active fires and the rest, and to interpret the growth of the fire. Two fires were selected for the purpose: those in the Sierra de Cazulas (Granada) in August 1999, and in Valencia, in July 1994. The results provided useftil information on the growth of fire in general, the area, speed at which it spread and prevailing direction. This suggests that this method can be recommended for studies on fire behaviour, as well as in the provision of resources for fighting the fire at both the regional and national level.

The paper presents different polarimetric and interferometric inversion radar approaches to extract the characteristics of vegetation and bare soil surfaces. The electromagnetic waves interactions with scatterers in a random medium are a complex process, which is sensitive to a much larger number of target parameters. Consequently, the estimation of surface and vegetation parameters can not be achieved by using fixed polarization single frequency interferometric SAR data, even when a simple homogeneous one-layer model is considered. In this way, some multibaseline SAR interferometric inversion algorithms have been proposed in order to extend the number of independent parameters. However, these models are based on the first-order scattering approximation, ignoring the effect of multiple scattering. The introduction of polarimetry provides an important tool to minimize multiple scattering effects by selecting the optimum scattering mechanisms, making parameter estimation more robust. Some indoor and outdoor measurements have been carried out to test these inversion techniques. The measurements have been conducted on bare rough surfaces and sorghum fields at different growth stages with a very accurate gound-truth. The results are compared to those obtained by applying other existing techniques, such as the Frecuency Correlation Function (FCF) and empirical inversion methods.

The approach to the detection of fires suggested here is based on methods of pattern recognition in spaces of the informative parameters from information contained in indirect measurements, which in this case are five-channel videodata recorded with the AVHRR instrument placed onboard NOAA satellites. A problem of preliminary integrated normalization of satellite videodata, including a transition to constant dimensions of scanning spot projections on the Earth’s surface, an increase in the spatial resolution of images for a model of integration within the spot, and correction of the radiobrightness characteristics of the images, is considered. Normalized images are subsequently used to solve the problem of detecting small-sized fires with the help of a three-stage procedure by an algorithm of pattern recognition in space of the informative parameters. A natural criterion for estimating the information content for the class of detection and pattern recognition problems is the functional of the average risk. In this case, the informative set of parameters and the decision rule are found by minimization of this functional. Because conditional probability densities, being mathematical models of stochastic images, are unknown, the problem of reconstructing distributions based on teaching samples with the use of nonparametric estimates with modified Epanechnikov kernel is solved. Unknown parameters of distributions are determined by minimization of a functional of the empirical risk. A comparison between the results of operation by the algorithm and the operator work demonstrates high efficiency of the algorithm of detecting thermal anomalies of fire types.

Remote sensing is an important tool for the spatial description of urban green, it is often used in landscape ecology to detect habitat fragmentation and the alterations caused by man on the territory; furthermore the use of remote sensing is important for the detection of vegetation functional conditions. In this study some vegetation indices have been analysed and a land use map has been created using Landsat TM images of the city of Rome and of some of its outskirts. Remote sensing data have been integrated with field data (leaf area index and leaf fluorescence measurements) carried out on holm oak and coniferous woods. The values of the functional vegetation indices indicate a physiological alteration for the communities inside the city; the fluorescence data have confirmed this result. From a structural point of view, the NDVI map highlights the presence of some green corridors. The good correlation between the values of NDVI and LAI has allowed us to estimate LAI for the entire area analysed. This research has furnished quantitative data on structural and functional characteristics of the study area. The upscaling approach has shown a good performance of analysing and monitoring vegetation subjected to a different anthropogenic impact.

The analysis of LAI is of great importance to evaluate the structural characteristics of a plant community. Vegetation types remote sensed analysis is a useful tool for the investigation of structural and functional characteristics of ecosystems. The integration of data acquired at these two different spatial scales contributes to the management of the land use and in particular of protected areas for the conservation of biodiversity. Field measuraments (LAI) and remote sensed data (Landsat 5 TM) were performed in the Circeo National Park and in the Presidential Estate of Castelporziano (located along the coast of Lazio, Italy, and characterised by Mediterranean climate). In this study the relationships between LAI and vegetation indices (NDVI, PVI, SAVI, GVI) were analysed; the NDVI showed the best correlation with the LAI. Using NDVT/LAI low power we estimated the LAI for the entire study areas on a multitemporal basis and plotted different LAI maps. The results obtained permitted to define a useful approach to analyse the canopy conditions of Mediterranean ecosystems on a multitemporal scale and also to highlight the effects of biotic and abiotic causes on vegetative growth as phytophagous attac and climatic stress due to local annual variability and to potential climatic change.

Annual cycles of greenness, as observed from coarse satellite imagery, have proven to be a useful proxy of phenology of light interception. VEGETATION imagery has superior radiometric and geometric specifications compared to the commonly used NOAA-AVHRR images, which should result into a more accurate description of phenolgy. We have analyzed the annual cycles of VEGETATION reflectance quotients from 36 S10 products for a number of very detailed habitat types in the NW Mediterranean basin. We selected all VEGETATION pixels (1 km2) that were included within one single habitat patch, extracted their multi-temporal reflectance data and ancillary information, and analyzed the time series of normalized differences of near-infrared minus red and of near-infrared minus medium-infrared reflectance values. The distribution of habitats on the principal components plane indicated that individual time series tended to cluster by habitat type and that the main sources of variation were the average value of the time series and the position of the maximum value. The distribution of observations on the plane defined by both indices indicates a general linear correlation except if surface water is present. Our work shows the interest of the multi-temporal analysis of satellite imagery at detailed scales to fully understand the dynamics of light interception and the response of mixed multi-temporal VEGETATION image pixels.

This paper details an effort to develop an operational methodology to distinguish lacustrine, palustrine and riverine wetlands from irrigated agriculture in a continental area using archived multi-temporal/multi-spectral Landsat TM data.. Archival Landsat TM data were acquired over the Little Wood River Valley of Idaho in April, August and September of 1985. All dates of imagery were subjected to a Kauth-Thomas transformation and then stacked into a single 9-band image and submitted to a supervised classification. DEM data was used to remove spectral confusion with mountain vegetative systems with similar temporal signatures to the wetlands of interest. Field checks and comparison to National Wetland Inventory (NWI) maps completed in 1984 revealed a 98.3% agreement in classification of non-wetland areas. 54% of the areas classified as wetland on the NWI were classified as wetland using our method. This is attributed to practice of generalization of the NWI maps in which several small wetlands are circumscribed into a single large area. The digital method correctly identified the wetland patches and classified the interstices as dry land. Confusion with irrigated agriculture was almost completely absent.

Vegetation indices (VI's) are important tools in the seasonal and inter-annual monitoring of the Earth's vegetation. In this study, the vegetation index products from the Moderate Resolution Imaging Spectroradiometer (MODIS) are evaluated over a preliminary set of validation test sites, including a cerrado and rainforest site in Brazil, and two grass/ shrub sites in Arizona and New Mexico, U.S.A. Ground and airborne validation experiments were conducted to assess the performance of the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) for vegetation monitoring. Calibrated spectroradiometers were flown for top-of-canopy reflectance retrievals. Vegetation sampling provided the data needed for a biophysical validation of the VI's. Both single-day and 16-day composited MODIS data were processed and corrected for atmosphere at 500m and 1 km resolutions. The MODIS data compared quite well with the validation data with most of the uncertainty associated with the compositing process. Results show the MODIS VI products to offer enhanced sensitivity for land use discrimination and monitoring at both regional and global scales. The EVI was fairly well resistant to residual cloud and aerosol contamination and had a good range of sensitivity over the high biomass, forested areas.

The spatial extent of Australia's forest estate lends itself to the application of an integrated multi-scale approach for the identification of koala habitat extent and condition. A multiple-scale methodology initially identifies forests using AVHRR, TM or predictive modelling to give a regional-state coverage. Then, the application of calibrated geometrical-optical models is used within forests to identify communities and structural properties such as projected crown cover, stem density and biomass. Finally, habitat quality at scale of individual trees is evaluated. This work reports on approaches used to map the location of trees that indicate high habitat quality for koalas and possibly other foliovores such as possums and greater gliders. An approach has been developed for using canopy level field spectrometer measurements to transform hyperspectral data into four components representing within-pixel proportions of target and non-target forest species, grasses and other non-tree components, and shade. A canopy detection routine is then used to produce canopy scale maps of individual species. This approach was tested using an Analytical Spectral Devices spectrometer in conjunction with CASI image data in a mixed coastal eucalypt forest at Koala beach, Pottsville, New South Wales. The maps of target species location were found to be 90.1% accurate when compared to field located species.

The objective of this study was to evaluate environmental impacts surrounding a freshwater reservoir in South Carolina using landscape ecology pattern and structure metrics calculated from high-resolution optical/infrared imagery. In preparation for satellite-based studies using platforms such as SPOT-5, IRS-P5, Orbview 3 & 4, and Ikonos satelites, digital high-altitude color infrared (NAPP) and Airborne Terrestrial Applications Sensor (ATLAS) data were analyzed for a shoreline surrounding a freshwater reservoir in South Carolina subject to degradation from urban encroachment. An index was developed using a genetic learning neural network to mimic the impact rating given each section of the shoreline by experts in the field. It is hoped that this index can be extended on global scale by using high resolution satellites to allow for a standard index that would reduce the often subjective nature of shoreline degradation evaluations.

Annual cycles of vegetation index calculated from time sequences of satellite images at resolutions ranging from 1 to 8 km. are a useful proxy of the phenology of light interception, and are being used as an input for bio-geospheric models and landcover charts at global, continental and regional scales. Measuring the annual course of light intercepted by vegetation at a finer scale would have important practical consequences. The core of human action on the environment is undertaken at the landscape and district scales and an important part of the spatial variance is within resolutions finer than the coarse (~1 km2) pixels of cunent multi-temporal imagery. Time series of vegetation index at landscape and district scales would provide dynamic information on vegetation at the scale that is required for management. High-resolution images have an appropriate spatial scale but do not offer the required acquisition frequency. The integration of coarse but frequent imagery and the highresolution imagery is necessary. We have approached the integration of multi-temporal VEGETATION images into products derived from the high spatial resolution images in a forested Mediterranean landscape. Our results show that high-resolution imagery and frequent, coarse-resolution imagery provide complementary information.

Direct gradient analysis, and other canonical community ordination techniques, have been most commonly used by plant ecologists and others attempting to analyse complex multivariate datasets. These multivariate statistical techniques can be applied to a variety of spectral analyses. Particularly useful is the ability to test significance of environmental variables based upon Monte Carlo permutations, allowing for a step-wise model of variance to be built. This technique has been now applied to hyperspectral remotely sensed data, within the overall context of ESA DAISEX-99 experiment. An extensive field campaign in La-Mancha (Spain) was carried out, simultaneously with the overflight of two airborne imaging spectrometers (DAIS, HYMAP) and other sensors (POLDER, LEANDRE).We use in this work data from the 128-channels HYMAP imaging spectrometer jointly with the ground truth data. Direct gradient analysis of the imagery spectra indicated an overall statistical significance when a model based upon three variables was used. Leaf moisture, LAI, and total chlorophyll were the most highly correlated variables, and all demonstrated statistically significant p-values. Hyperspectral remote sensing data requires new techniques to analyse the increasingly complex data. Application of ordination techniques, although not commonly applied within the remote sensing data processing, show good perspectives for more in depth analysis of the whole DAISEX-99 dataset.

Directional reflectance spectra from 400 to 780 nm were acquired at an angle-of-view of 30° for nadir illuminated leaves of four different kinds of plants (Codiaeum variegatum, Eleagnus angustifolia, Pittosporum tobira, Ficus benjamini). We found that, differently from hemispherical reflectance measurements, directional reflectance data do not correlate well with chlorophyll content. This is mainly due to the Fresnel reflections at the air-epidermis and epidermis-inner layer interfaces that gives the main contribute to the reflectance in 400-480 nm band, where chlorophyll absorption is maximum. The inner reflectance (RI), obtained by subtracting the outer reflectance from the measured spectra, was found to correlate to chlorophyll content. A good correlation (r2 = 0.98) of RI(?)/RI750 versus the log(Chl) was obtained for RI(?) in the reflectance maximum in the green band around 550 nm. The error in the determination of Chi content was of about 3.5 |µ/cm2, lower than that obtained by applying previously suggested vegetation indexes to our experimental data.

Ground displacement resulting from the Landers earthquake is computed from a pair of SPOT panchromatic images as described in. Other sources of geometrical deformations (offsets) resulting from local topography, artifacts of the imagery, variations in the satellite attitude during the acquisition process are compensated at the sub-pixel scale. Images are then resampled to be in a cartographic projection using a sine interpolator. Measurements are achieved from a dedicated sub-pixel correlator in the Fourier domain which provides a 2D offset map. Residual biases are compensated from calibration. Amplitude of the remaining high frequency noise is about 0.01 pixel (rms). Most of the low frequency noise can be compensated from modeling. Measurements enables the cartography of the Landers fault with an accuracy of 80 meters. Displacement on the fault is extracted with an accuracy better than 80 meters. Comparison with a model derived from GPS and other geodetic data validates our results. Preliminary results derived from images with a metric resolution provide far more accurate measurements.

The aim of this paper is to outline the potential of imaging spectrometry and GIS techniques as tools for the management of data rich environments, as complex fluvial areas, exposed to geology cal, geomorphological and hydrogeological risks. The area of study, the Pescara River Basin (Central Italy) is characterised by the presence of important industrial sites (Sulmona and Pescara) and by the occurrence of floods, lanslides and seismic events. Data were collected, during a specific flight, using an hyperpectral MIVIS sensor (102 channels between 0.43 and 12.7 µm). Images have been processed in order to obtain updated and accurate land-cover and land-use maps that have been inserted in a specific GIS database and integrated with further information like lithology, geological structure, geomorphology, hydrogeological features, productive plants location and characters. The processing of data layers was performed, using a dedicated software, through typical GIS operators like indexing, recoding, matrix analysis, proximity analysis. The interactions between natural risks, industrial installations, agricultural areas, water resources and urban settlements have been analysed. This allowed the creation and processing of thematic layers like vulnerability, risk and impact maps.

Soil erosion is a widespread problem in semi-arid Mediterranean countries where irregular, intense rainfall can result in rapid erosion in areas of steep slope, soft lithology and sparse vegetation. This paper presents an integrated remote sensing/GIS approach to identify areas vulnerable to rapid erosion in Almeria Province, Southeast Spain. ERS SAR multitemporal interferometric coherence imagery is used to extract information on the erosion process, which causes random changes in the micro-topography of the land surface. This results in reduced coherence of the radar signals between the initial and eroded states. A known case of rapid erosion within the study area has been investigated using coherence images with 70, 140 and 210 days temporal separation (Liu et al., 1999). A GIS is now used to derive criteria for a set of environmental conditions causing susceptibility to rapid erosion. Slope is extracted from SAR interferometr; vegetation and lithology from ETM+ imagery. The areas predicted to be most vulnerable to rapid erosion across the imagery are then identified. The methodology provided by our study will help local land management authorities to understand where this type of erosion may occur. The research is supported by the ESA ERS A03-113 project.

Arid and semi-arid areas are specially susceptible to soil degradation processes such as erosion and salinization and the influence of land use. The identification of soil degradation stages form an important basis for sustainable land use and land conservation. The key aim in this work is to combine multispectral with radar data and to evaluate their effectiveness for delineating soil degradation stages in semi-arid environments. Radar images have the advantage of being very sensitive to textural differences along land surfaces. Principal component analysis is performed on two data sets using the six reflective bands of Landsat ETM+ including the mean texture band of ERS-2 SAR and using only the six ETM+ bands. To evaluate the usefulness of the textural information for delineating the soil degradation stages an automated classification is performed on both PC data sets. The methodology for identifying soil degradation stages is performed on test sites located within an area in the Central region of Spain. Ground truth verification is carried out to confirm the results obtained. Different soil degradation stages, according to the soil surface characteristics, are successfully identified in the study area. The ERS/ETM+ based classification has significantly improved the separation of rugged landscape features along the slopes from those in the plateau area.

It is very important to analyse terrain changes in order to have information to perform land management, especially in crowded area as Lombardia (north of Italy). The easiest way to individualise changes is to compare different age maps; this paper describes an automatic way to do this. Lombardia flat area, about 18000 km2, has been mapped in 1984 and 1991 in 1:10000 scale. 450 sheets compose the CTR (Cartography Technical Regional) map, each of them describing a 40 km2 area. In this way 1984-1991 change detection requires to menage 900 map sheets and it is clear that is indispensable to define an automatic way to data analyse. In the first part of the work the way to automatically extract civil and industrial areas from map have been defined. A math composition of the obtained spare maps (1984 civil coverage, 1984 industrial coverage, 1991 civil coverage and 1991 industrial coverage) gives as result a map of the coverage evolution (showed as 9 classes). In the second part of the work the 450 sheets of the evolution map are analysed in order to obtain areal parameters. The obtained parameters are the amount of each class as total area and percentage area for single square kilometre of Lombardia and for each one of the regional municipalities.