This paper addresses the characterization of Land Surface Temperature (LST) variability according to Land Cover. It is a first step of a study which concerns the extraction of hydrological parameters in a semi-arid catchment applied located in Southern-Africa, and which includes image processing of satellite data. The main applicative interest of this work is to make available profiles of evapotranspiration (ET), which can be derived from LST, and to detect hydric stress by comparison between profiles of ET: potential ET simulated by an hydrological model and that estimated by satellite measurements. LST can be daily computed using the two thermal bands of NOAA/AVHRR. However, due to its coarse resolution (1.1 km at nadir), a NOAA/AVHRR pixel includes several land cover types and LST cannot be linked to a particular component. So, we process a data fusion between NOAA/AVHRR acquisitions and one high resolution land-use classification derived from Landsat-TM (30 meters at nadir), and consider a physical-based mixture model of the temperature pixel. Inverting this model on a learning area outputs individual temporal profiles of LST for each land cover type: bare soil, vegetated surface (grass, arable land, forest...). The obtained results with Landsat classification are then used to generate LST maps at spatial resolution of 30 meters and with a daily frequency.

A combined technique of genetic k-means and radial basis function neural network (RBFNN) is used in this study to process remote sensing data and classify soil basing on its moisture content. Radial basis function neural network is used for its advantages of rapid training, generality and simplicity over feed-forward backpropagation neural network. The genetic k-means clustering is used to choose the initial radial basis centers and widths for the RBFNN. An attempt is also made to study the performance of the RBFNN with the centers and widths chosen using the classical k-means clustering. The results showed that genetic algorithms give global optimal centers and widths for the RBFNN. The results also indicated that this hybrid technique can be used in soil moisture classification and prediction.

A database of long-term soil moisture was compared to satellite microwave observations over a test site in the Midwestern United States. Ground measurements of average volumetric surface soil moisture in the top ten cm were made several times per month at 19 locations throughout the state of Illinois. Nighttime microwave brightness temperatures were observed at a frequency of 6.6 GHz, by the Scanning Multichannel Microwave Radiometer (SMMR), onboard the Nimbus 7 satellite. At 6.6 GHz, the instrument provides a spatial resolution of approximately 150 km, and a temporal frequency over the test area of about 3 nighttime orbits per week. Vegetation radiative transfer characteristics, such as the canopy transmissivity, were estimated from vegetation indices such as the Normalized Difference Vegetation Index (NDVI) and the 37 GHz Microwave Polarization Difference Index (MPDI). Because the time of satellite coverage does not always coincide with the ground measurements of soil moisture, the existing ground data were used to calibrate a water balance for the top 10 cm surface layer in order to interpolate daily surface moisture values. Such a climate-based approach is often more appropriate for estimating large-area average soil moisture because meteorological data are generally more spatially representative than isolated point measurements of soil moisture. Passive microwave remote sensing presents the greatest potential for providing regular spatially representative estimates of surface soil moisture at global scales. Real time estimates should improve weather and climate modeling efforts, while the development of historical data sets will provide necessary information for simulation and validation of long-term climate and global change studies.

Two different TAMSAT methods of Rainfall Estimation were developed respectively for northern and southern Africa, based on Meteosat TIR images; northern Africa since 1987 and southern Africa since 1990. These rainfall estimates are used operationally for agricultural purposes and for predicting famines and floods. The two different methods have both been used to make rainfall estimates for the southern rainy season October 1995 to April 1996, and then compared with estimates produced by the CPC method. The latter are made more simply from TIR, but have the addition of GTS rainfall data and orographic rain. All these estimates were then compared with kriged data from over 800 raingauges in southern Africa. The detailed results were then compared for the whole season across the whole SADC region, and then two detailed cross- sections were studied, with different orography. The results show that operational TAMSAT estimates are better over plateau regions, with 59% estimates within 1 Std of the rainfall, but over the whole region the CPC estimates perform best. Over mountainous regions all methods under-estimate and give only 40% within 1Std. The two TAMSAT methods show little difference across a whole season, but when looked at in detail the northern method gives unsatisfactory calibrations. The CPC method does have significant overall improvements by building in real-time raingauge data, but only where sufficient raingauges are available.

The objectives of this study were: to examine the sensitivity of radar backscatter, to estimate soil moisture under a corn plot and to evaluate the effectiveness and sensitivity of a Radiative Transfer Model (RTM), adapted from the earlier work of Njoku and Kong, (1977) in predicting brightness temperature from a grass plot. Microwave radar measurements were collected from plots of different vegetation cover types, vegetation density, and moisture conditions during the Huntsville 1998 field experiment. A large amount of ground data on brightness temperatures, soil moisture, and vegetation characteristics (e.g., biomass, and water content) were collected. The experiments were conducted at Alabama A&M University's, Winfred Thomas Agricultural Research Station, located near Hazel Green, Alabama. Six plots, one 50 X 60 m smooth bare plot, one 50 X 60 m grass plot, and four 30 X 50 m corn plots at full, 2/3, 1/2, and 1/3 densities were used. Radar backscatter data were obtained from a ground based truck mounted radar operating at L, C, and X bands (1.6, 4.75, and 10 GHz) with four linear polarization HH, HV, VV, and VH and two incidence angles (15 and 45 degrees). Soil moisture values were determined using Water Content Reflectometry (WCR). Three types of soil temperature sensors (Infrared Thermometer, Thermistor, and a 4-sensor averaging thermocouple probes) were used. A discrete backscatter approach model and RTM were evaluated. Comparisons between model prediction and experimental observation for HH polarization indicated good agreement for a corn full plot. The direct-reflected scattering coefficient is found to be the most dominant term for both polarization and the backscatter is also highly sensitive to soil moisture. The trends in time variation of brightness temperature are in agreement with the experimental results and the numerical results are within a few percent of the experimental results. The vegetation corrections as estimated by the Jackson and Schmugge method are very small. Detailed examination of the vegetation canopy contribution including the geometry of the canopy, the various absorption and scattering mechanisms are necessary.

Four Landsat-5 Thematic Mapper (TM) images acquired between May 1996 and June 1998 were selected in order to define a procedure to compute the surface temperatures and the daily energy budgets of Lake Iseo, located in northern Italy. TM data were atmospherically corrected by using the radiative transfer code LOWTRAN 6 (Low Resolution Transmittance Code), and results in some locations were evaluated comparing satellite derived temperatures with in situ measurements collected simultaneously to the TM overpasses. The root mean square error (ESQM) between satellite and measured temperatures was lower than 1 degree Celsius. Then, a method combining modelled values of incoming short and longwave radiation, remotely sensed measurements of surface temperatures and albedo was used to retrieve the instantaneous surface radiation budget and the net radiation of the basin. Using hourly meteorological measurements and a model founded in literature the daily energy budget and its accuracy were finally evaluated and compared to reference values.

Eutrophication is one of the most common problems of water resources in developed and developing countries. Traditional measurement of water quality requires on-site sampling and laboratory work, which is expensive and time consuming. Due to these imitations, the sample size is often too small to have a high reliability of the corresponding results especially for a large water body. Remote sensing provides a new technique to monitor water quality over a wide area with a two-dimensional data distribution instead of sample points. In this research, French satellite SPOT was chosen as remotely sensed data source and provided images to derive chlorophyll concentration, Secchi depth, and phosphorous concentration for a water body. By comparing a set of on-site samples and the corresponding brightness values on a SPOT image taken on the same date, a regression model converting satellite data to water quality variables was defined. A systematic image process was developed to transfer SPOT data to water quality variables. This system provides not only an instantaneous and repetitive eurtrophic status assessment but also a visualizing water quality variation. An image process and GIS software IMAGINE was adopted to carry out the process in a case study of the Te-Chi Reservoir in Taiwan. The final product is a set of thematic maps of eutrophic status (represented by Carlson's TSI) of the reservoir.

In order to effectively protect coastal ecosystems, objective information on the structural and functional characteristics of these systems must be obtained through regular monitoring. In recent years, more advanced technologies, such as satellite remote sensing, have arisen which are being utilized for the monitoring and management of coastal ecosystems in general, and vegetation in particular. This paper presents the results from monitoring the vegetation biomass of the Koycegiz Lagoon area, declared as a Specially Protected Area by the Turkish Government, using multi-temporal satellite imagery. Several problems need to be overcome when using imagery from different dates or from different sensors. These include; image registration, atmospheric variability, and, often, the lack of historical ground data. Furthermore, for monitoring purposes, techniques to overcome these problems should be robust and automatic allowing the database to be upgraded easily. The procedures we have developed include automatic registration (to subpixel accuracy), atmospheric normalization, and vegetation index (VI) calibration components. This was tested on multidate (1984, 1988, 1991, 1995 and 1996) LANDSAT-TM data. From this adjusted data set the performance of different vegetation indices, in this coastal environment was examined, and the vegetation trends analyzed.

Rapid, reliable and objective estimation of Leaf Area Index (LAI) at various scales is of utmost importance in numerous studies on the Earth's ecosystem. The Licor LAI-2000 Plant Canopy Analyzer (PCA) correlates measured gap fractions to overall LAI by means of the inversion of a radiative transfer model. The PCA's model assumes a random distribution of foliage elements in the stand canopy. However, clumping is observed at different scales in nature. The objectives of this study were, first, the quantification of the LAI measurement error of the PCA due to foliage clumping at stand-level, and second, the derivation of an easily measurable correction factor. For this, foliage elements were simulated in a virtual 3D-space. PCA LAI measurements were simulated by applying the same PCA inversion model onto virtually taken hemispherical photographs resulting in both exact reference LAI values and corresponding PCA measurements. Fractal dimension, quantifying the deviation from a complete random foliage distribution, was tested as a correction factor for PCA measurements. Correction models for PCA measurements were build, based on the measured fractal dimension. A post validation as performed on field data obtained by means of littertraps (reference). A clear relation between fractal dimension and the proportion of underestimation of LAI by the PCA with increasing clumping of foliage was found. Implementation of the regression model resulted in significantly improved LAI measurements.

The necessity for accurate and real time crop monitoring is particularly felt in arid and semiarid environments, because temporal and geographical rainfall variability leads to high interannual variations in primary production and often increases the risk of severe famines. In these cases remotely sensed data, available for wide areas and with high temporal frequency, are an important tool for crop production monitoring and harvest forecasting. In particular, GAC NDVI images derived from the NOAA-AVHRR sensor have already been used for this aim in the Sahelian area of Africa, obtaining good results. In the present paper, a similar approach is tested for the early estimation of cereal crop yield in North African countries. The first results indicate that this estimation is possible especially when stratifying the land surface in ecologically homogeneous zones, identified by supervised or unsupervised clustering techniques.

Rainfall monitoring is of vital importance in many parts of Africa dependent on rain-fed agriculture. Rainfall monitoring algorithms using satellite imagery have been available for more than a decade. A number of countries make good use of these data but it remains disappointing that the methodology is not in more widespread use. A recent collaborative project between Reading University and the Zambian Meteorology Service has attempted to address this problem by focusing on two issues -- the accuracy of the estimates and the format of the final product. To improve accuracy, we have developed a geostatistical technique for merging satellite estimates with gauge data which takes account of the relative accuracies of the two data sets. To address the second issue we have developed a versatile software framework allowing the presentation of the rainfall information in a format tailored to the end-user. The final images are emailed as Word documents for easy incorporation into publications. The system has now run successfully for a full rainy season and response from the users has been positive. Crucial to successful implementation has been the involvement of local personnel and dialogue with the user community. Analysis of the first season's results show that the merged rainfall estimates are more accurate than either the satellite or the gauge data used separately.

In the last decades many forest areas are suffering from conventional and new types of damage, with a consequent loss of valuable ecological and economic resources. The monitoring of these damage has therefore become a primary application of satellite remotely sensed data, and particularly of Landsat TM imagery. Unfortunately, conventional mapping methods based on uni or multivariate regressions between ground measurement and remotely sensed spectral information have often led to unsatisfactory results, especially in complex environments where several disturbing factors can affect the forest spectral signatures. It is here proposed that a new, more flexible estimation method based on fuzzy classification of remotely sensed data can offer several advantages when used for this purpose. After a brief description of its basis, the method is applied together with conventional multivariate regression procedures in two case studies in Tuscany (Central Italy) representative of different forest types affected by damages of different origins. The results show that the new method produces higher accuracies in the estimation of forest damage, particularly in areas with complex environmental situations.

Fluorescence is a highly specific signal of vegetation function, stress and vitality. Solar induced fluorescence, a very weak signal, is detectable using the Fraunhofer lines of the solar spectrum, allowing the observation from a satellite. A space mission is proposed that would provide new data type in the field of Earth observation, addressing big scale screening of terrestrial vegetation in relation to agricultural, forestry and global change issues, such as solar irradiance, ozone depletion, water availability, air temperature, pollution in air/water/soils,..). The scientific, technical and general characteristics of the proposed mission, a demonstration/validation of the technique, are exposed. The scientific payload, from a LEO sun-synchronous orbit, will measure the fluorescence in the H(alpha) (656.3 nm) and at least one other Fraunhofer line in the blue-UV region with a CCD matrix type imaging spectrometer. A thermal infrared imaging radiometer, optional although highly desirable, would measure the vegetation temperature. An additional CCD camera will provide cloud detection and scene identification. The spatial resolution will be better than 0.5 X 0.5 km2, with a nominal FOV of 8.4 degrees; a steering mirror will allow plus or minus 4 degrees across track depointing, while allowing, if necessary, freezing the image to increase the S/R. The FLEX mission will provide florescence intensities and reflectances at the same wavelengths, plus temperature if available. Processing of the data will require atmospheric characteristics (aerosols) from other missions. Interpretation of the fluorescence signal will require reflectance data over the 400 to 800 nm region; biome characteristics (LAI, architecture, biomass density factors, APAR, etc.) from space mission providing high spectral resolution, directional reflectance measurements in the visible domain; ground data on environmental factors and plant physiology and in situ florescence measurements for satellite signal validation.

In this paper, a validation of a new UV-A laser-induced fluorescence imaging system implemented in an all-road car for near-field remote sensing of vegetation will be presented. It has been developed as a part of a European Community Program INTERREG II and is consisting of three main parts: excitation, detection and control units. The excitation source is a frequency tripled Nd:YAG laser and the laser spot size is adjusted via a variable beam expander. Fluorescence images are recorded at four characteristic fluorescence bands: 440, 520, 690 and 740 nm with a gated intensified digital CCD camera. The laser head and camera are situated on a directed in site and azimuth platform which can be high up to 6 meters. The platform positioning, localization and distance detection, spot size determination and adjustment, focus, sharpness, selection of the filter, laser and camera synchronization, gain of the intensifier, real time visualization of images, acquisition time are controlled by a newly developed software which allows also image storage, analysis and treatment. Examples of remote sensing fluorescence images from several plant species recorded at a distance of 10 - 30 m will be given and discussed further in this paper.

The work presented in this paper investigates the sensitivity of the hyperspectral remotely sensed data to the vegetation density under different soil moisture conditions. The research testbed comprised four corn plots with 4 different densities, one grass plot, and one bare soil plot. For this purpose, the hyperspectral data were recorded simultaneously as the field measurements, which included soil moisture and temperature, soil characterization (gravimetric soil moisture, bulk density, surface roughness), and vegetation measurements (biomass; plant height; leaf orientations, length, thickness; dielectric constant of stalks and leaves; stalk diameter and height). The findings of this study showed that physical and physiological aspects, as well as the structure of the vegetation, have noticeable effects on its spectral response. The results showed distinct spectral response among the different vegetation densities, thus biomass. They also showed that hyperspectral data are effective in detecting soil moisture variability and discriminating among vegetation densities and conditions. The hyperspectral data were in agreement with the ground data and discriminated among small variations in soil moisture and vegetation densities and conditions. This study also showed that the variation in the spectral variability from different vegetation densities becomes negligible when the vegetation leaves cover completely the ground surface.

Vegetation indices have emerged as important tools in the seasonal and inter-annual monitoring of the Earth's vegetation. They are radiometric measures of the amount and condition of vegetation. In this study, the Sea-viewing Wide Field-of-View sensor (SeaWiFS) is used to investigate coarse resolution monitoring of vegetation with multiple indices. A 30-day series of SeaWiFS data, corrected for molecular scattering and absorption, was composited to cloud-free, single channel reflectance images. The normalized difference vegetation index (NDVI) and an optimized index, the enhanced vegetation index (EVI), were computed over various 'continental' regions. The EVI had a normal distribution of values over the continental set of biomes while the NDVI was skewed toward higher values and saturated over forested regions. The NDVI resembled the skewed distributions found in the red band while the EVI resembled the normal distributions found in the NIR band. The EVI minimized smoke contamination over extensive portions of the tropics. As a result, major biome types with continental regions were discriminable in both the EVI imagery and histograms, whereas smoke and saturation considerably degraded the NDVI histogram structure preventing reliable discrimination of biome types.

Vegetation monitoring has been one of the major targets of remote sensing studies. Remotely sensed reflectance concerning the impact of environmental factors upon 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 evaluate the reflectance data from frequently irrigated and water stressed Sudan grass and other crops using a handheld radiometer and assess the spectral correlation with the ground-truth; (2) to evaluate the applications of a Hyperspectral Structure Component Index (HSCI) developed by Shakir and Girmay-Gwahid in 1998; and (3) to evaluate the application of Index of Relative Stress (IRS) proposed by Shakir and Girmay-Gwahid in 1998. The experiment was designed to collect reflectance data from Sudan grass and other crops planted at the Blythe Research Station, California in rows. The size of the plots for Sudan grass was in rows, the unstressed mature stands were 9 feet tall, and the stressed mature stands were 5 feet tall. The other fields are in nearby and planted with cotton crops in different stages of maturity. 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 DayTime (PDT). Vegetative samples were taken from the two treatments 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 results of this experiment showed that in the 850 - 1150 nm wavelength ranges, the stressed Sudan grass stands showed lower reflectance than unstressed stands. However, the reflectance of stressed Sudan grass stands was higher than the unstressed stands above the 1150 nm. This is probably due to the absorption from liquid water contained in the unstressed plant tissues. The same pattern was found in the cotton crop. The analysis of data using the (HSCI) model showed that the stressed Sudan grass stands have values less than 1 and under unstressed Sudan grass stands have the value greater than 1. 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. Furthermore; the Index of Relative Stress (IRS) showed that the unstressed vegetation stands is higher in values than in the stressed.

The main objective of this study was to investigate whether AVHRR data could be useful for crop model simulation steering by intrinsically taking the mixed pixel effects into account. The second objective was to determine if the application of a genetic algorithm could be an effective technique for crop model steering. The principles were tested for the Seville test site using synthetic data and AVHRR data from 1995 and 1996 because these years show a large contrast in crop development. The main conclusions are that a genetic algorithm is a very powerful technique for crop model optimization, but adaptations are needed to the current optimization scheme in order to be able to steer the WOFOST crop model on the basis of NOAA-AVHRR data.

Triana is an Earth remote sensing satellite to be located at the distant Langrange Point L-1, the gravity-neutral point between the Earth and the Sun. It will provide continuous fill disk images of the entire sunlit side of the Earth with 8 km pixel resolution. The primary remote sensing instrument on Triana is a calibrated multispectral imager with 10 spectral channels in the UV, VIS, and NIR between 317 and 870 nm (reflected solar radiation). Due to its unique location at the Lagrange L-1 point, in the direct line-of-sight between Earth and Sun, Triana will view the Earth always in and near the solar retro-reflection direction which is also known as the hotspot direction. The canopy hotspot effect has rich information content for vegetation characterization, especially indications of canopy structure and vegetation health and stress situations. Primary vegetation-related data are the hotspot angular width W, and a hotspot factor C that quantifies the magnitude of the hotspot effect. Both quantities are related to the structural parameters of canopy height, foliage size, shape, and leaf area index (LAI). The continuous observations by Triana will allow us to establish time-series of these ecological parameters for all land biomes by longitude, latitude, and wavelength, that form the basis data set for a new global hotspot land vegetation ecology. The hotspot factor C will allow the determination of the enhanced radiant flux reflected from the Earth into space due to the hotspot effect. The hotspot flux enhancement due to the vegetation hotspot effect is estimated to account for about 1% of the entire Earth radiative energy balance.

Land use has an important impact on the climatic and hydrological cycle. For modeling this impact detailed knowledge of the land use and land cover pattern is necessary. Optical remote sensing data are good information sources to derive land use classifications for large areas. But due to the fact that commonly used classification algorithms are solely based on the spectral information, this often leads to misclassifications, because different classes can show similar spectral signatures. This is especially true for areas where a high rate of cloudiness reduces the availability of data. These are often heterogeneous and rugged areas such as mountains and their forelands. Advanced knowledge-based classification approaches which integrate non-spectral geographical ancillary data (i.e. climatic and terrain data) can improve the classification accuracy drastically. Still the method fails if spatially distributed ancillary data is not available or show no influence on the land use structure. The major advantage of the approach described in this paper is that it uses data, which are solely based on remotely sensed images and is therefore independent from map sources. The lack of multitemporal satellite data is cleared by the synergistic use of ERS radar data and LANDSAT-TM optical data.

In the Sao Gabriel do Oeste area (Pantanal, Brazil), since the '60s, zootechnics and farming activities have developed and arable lands and pastures replaced shrubs and forests. The 1966 to 1996 land-cover change was investigated through Remote Sensing and GIS methodologies. The effect on soil loss was evaluated through the Universal Soil Loss Equation (USLE). By integrating supervised classification and visual interpretation techniques, geo-coded land-cover data bases were built from aerial photographs and Landsat TM images (years 1966, 1985, 1996). Multi-temporal land-cover data bases were produced through 'post-classification comparison.' The application of the USLE in the ARC/INFO$CPY Grid environment enabled to perform the multi-temporal analysis of the potential soil loss. The R, K, C and P factors of such equation were assumed from the literature. The flowdirection and flowaccumulation Grid functions and the DEM allowed calculating the L and S factors. The results show that from 1966 to 1985 large extent of forest and shrubs were deforested. After 1985, deforestation rate decreased and part of burnt areas and pastures changed to secondary forest. The land-cover transformations induced a meaningful growth of the computed average soil loss per unit area (A) from 1966 to 1985 ((Delta) A approximately equals 3.7 t(DOT)ha^-1(DOT)y^-1). On the contrary, the reduction of A from 1985 to 1996 ((Delta) A approximately equals 0.8 t(DOT)ha^-1(DOT)y^-1) suggests that more recently the human impact became steady.

The paper analyzes secondary phenomena of atmospheric radioactive pollution caused by activity of the nuclear cycle enterprises. These effects being as indicators for remote diagnostics of a radio-activity are discussed. Excitation of a molecular and gas component in the air and various chemical reactions under the action of radiation have been considered. As a result of these reactions, new aerosol and gaseous components in the form of the excited atoms and ions appear in the atmosphere and relax with emission including microwave and optical wavelengths. The observable luminescence of the air during the emergency events at the nuclea power stations are long enough to be dedected by modern receivers. Intensity of such radiation in a radioactive plume is estimated for ecological monitoring of the atmosphere. Aerosols appearing, as a result of UF₆ hydrolysis, in the atmosphere and their behavior have been also shown to be detectable with remote sensing.

Land cover-mapping gives base for complicated tasks and high accuracy maps are necessary for reliable global estimations. Land cover mapping involves classification; its accuracy is often affected by mixture presence, which increases with pixel size. Using hard classification approach mixed pixels are source of errors because are treated as pure ones and obtained maps describe scene as made up by mosaic of homogeneous areas. Soft classifications, computing partial belongings of pixels to several categories, are useful tools for dealing with mixture. Two soft classifiers, one based on fuzzy-statistical and one on fuzzy neural network approach, are applied to the classification of NOAA images for snow cover estimation. Results are compared with the traditional hard classifier maximum likelihood. The analysis shows that accuracy of hard classifiers is greatly affected by increasing of pixel size. Soft classifiers perform better accuracy then hard classifier in areal estimation. Particularly fuzzy-statistical classifier gives better results then fuzzy neural network but it requires mixture information that fuzzy neural network does not need. Fuzzy neural network results the best tool to classify low resolution images for the evaluation of snow covered area as it performs the best balancing between minimal requested ground truth and high accuracy.

Remotely sensed reflectance data are often acquired at variable view and solar geometric configurations. Vegetation change monitoring with the NDVI (Normalized Difference Vegetation Index) is sensitive to the effects of solar and view angle geometry. However, by using a BRDF (Bidirectional Reflectance Distribution Function) model, the view and sun angle variability in the NDVI can be standardized. If multi- sun angle data are not available, a second method allows us to extrapolate (nadir) satellite observations to a standard sun angle by using predetermined linear regression relationships between sun angle and ground-based nadir NDVI values for a range of vegetation types. Both methods were applied to one month of daily, atmospherically corrected, multi-angle SeaWiFS (Sea viewing Wide Field-of-view Spectroradiometer) land reflectance data, with promising results. The difference in NDVI due to a sun angle change from 20 degrees to 70 degrees can be up to 50%. The NDVI values for very dense vegetated and bare soil surface areas are less affected by changes in solar zenith angles. This research shows that the sun and view angle effects on the widely used spectral indices could be standardized to improve the accuracy of regional and global vegetation and crop monitoring efforts.

Remote sensing image contains much information about the objects on the earth's surface e.g. vegetation. Through some transactions, those images can be changed to be more clear and more propitious for observing and analyzing. This paper adopts the correspondence analysis algorithm to process the TM image for composition estimation of cotton. By using this method, cotton areas are extracted from other objects and the characteristic TM bands of cotton are found. It also validated the effectiveness of correspondence analysis algorithm. This method can be applied in vegetation estimation.

TM data have been used by many researchers for remote sensing of monitoring chlorophyll-a concentration, which is strongly correlated with trophic state of surface water. In this paper, principal component analysis (PCA) is employed for extracting alga production in lake water from TM data. Input images are chosen as radiance ratios TM2/TM1, TM2/TM3 and TM4/TM3, instead of four independent band data. Regression analysis gives relationship between the first principal component (PCA1) and chlorophyll-a concentrations as: Chl((mu) g/L) equals -56.69+135.68X(PCA1), where PCA1 equals 0.146 X(TM2/TM1)-0.274X(TM4/TM2)+0.951X(TM4/TM 3). Chlorophyll-a concentrations estimated from TM image are compared with field measurements. The chlorophyll-a algorithm is also applied to TM images of the same lake acquired at different time. Multi-temporal chlorophyll-a distributions in the lake are mapped.

A problem of early detection of newly burning fires whose sizes are small is extremely actual, especially for almost inaccessible and sparsely populated regions. An approach proposed here for the detection of fires is based on methods of a pattern recognition in informative parameter spaces using information contained in indirect measurements, which in this case will be five-channel observations with the AVHRR instrument. For a class of detection and pattern recognition problems a natural informative criterion is an average risk functional. In this case the informative parameter complex is determined by minimization of this functional. Because the conditional probability density functions being mathematical models of stochastic images are unknown, a problem arises of reconstructing distributions based on learning samples. If the learning material sample length is small, it is natural to use the nonparametric Rosenblutt-Parsen estimates to reconstruct these distributions. The unknown parameters of these distributions are determined by minimization of the risk functional, when the learning sample is substituted by the empirical risk. To implement the developed algorithm, we used the data of observations with the AVHRR instrument performed in summer (May - August 1998 - 1999) over the territory of the Tomsk region, when many fires were recorded. A comparison between the results of algorithmic implementation and the operator work have shown high performance of the algorithm of detecting thermal anomalies.

A mathematical model for computation of parameters of eyesafe lidar for detection of forest fire smoke has been developed. It is assumed that the lidar uses a wavelength of 1.54 micrometer. This wavelength can be obtained from Er:glass lasers, from Nd:YAG lasers with an optical parametric oscillator, or from Nd:YAG lasers with a Raman cell. It is assumed that receiver optics of 20 cm diameter and an avalanche photodiode are used. Particle size distributions in the smoke from experiments in the literature are utilized for calculation of backscattering efficiency. The backscattering cross section is calculated on the basis of Mie formulae. Diffusion of the smoke plume is estimated on the basis of an analytical solution of the relevant hydrodynamics equations. Results of the calculations show that for detection of forest fires with fuel mass burned in unit time 2 kg/s at a distance of 10 km it is necessary to have a laser pulse energy of 120 mJ.

High-quality, multispectral thermal infrared sensors can, under certain conditions, be used to measure more than one surface temperature in a single pixel. Surface temperature retrieval in general is a difficult task, because even for a single unknown surface, the problem is under-determined. For the example of an N-band sensor, a pixel with two materials at two temperatures will, in principle, have 2(N+1) unknowns (N emissivities and one temperature for each of two materials). In addition, the upwelling path and reflected downwelling radiances must be considered. Split-window (two or more bands) and multi-look (two or more images of the same scene) techniques provide additional information that can be used to reduce the uncertainties in temperature retrieval. Further reduction in the uncertainties is made if the emissivities are known, either a priori (e.g., for water) or by ancillary measurements. Ultimately, if the number of unknowns is reduced sufficiently, the performance of the sensor will determine the achievable temperature sensitivity. This paper will explore the temperature sensitivity for a pixel with two temperatures that can be obtained under various assumptions of sensor performance, atmospheric conditions, number of bands, number of looks, surface emissivity knowledge, and surface composition. Results on synthetic data sets will be presented.

FLEX is a scientifically driven space mission to provide demonstration/validation of the instrumentation and technique for measuring the natural fluorescence of vegetation in the Fraunhofer lines. The payload consists of high spectral resolution (0.1 - 0.3 nm) CCD imaging grating spectrometer with two channels: one in the red (648 - 664 nm) and one in the blue (391 - 438 nm) for working with several Fraunhofer lines. The across track FOV is 8.4 degrees; ground spatial resolution is better than 0.5 X 0.5 km2. To increase the S/N ratio a steering mirror will be used, if necessary, to 'freeze' the image and also to provide plus or minus 4 degrees across track depointing. Calibration is made by viewing the sun via a diffuser plate switched into the telescope field of view. A separate CCD camera will allow cloud detection and scene identification. A TIR radiometer will provide simultaneous surface temperature measurements. The spacecraft, overall mass estimated at 200 kg, is derived from the ASI-MITA bus which provides all the necessary subsystems and stabilized platform. By use of on-board storage, ground requirements for satellite control and data link are minimized; the possibility of local stations for real time reception/distribution is also envisaged. Provisional orbit characteristics are: LEO sun synchronous, 500 - 900 km altitude. Priority will be given to highest revisit frequency on a sufficient number of selected test sites.

This paper presents a comprehensive investigation of Canadian boreal forest fires using satellite measurements. Algorithms were developed for detecting active fires (hotspots), burned areas, and smoke plumes using single-day NOAA-AVHRR images and 10-day AVHRR NDVI composites. The algorithms were rigorously validated using conventional fire survey data. The hotspot algorithm identified almost all fire events, but cumulative hotspot area was significantly smaller (approximately 30%) than burned area reported by fire agencies. The hybrid, burn mapping technique provided estimates of Canada-wide burned area that were within 5 percent of official statistics. A neural-network classifier was also developed that allows smoke plumes to be effectively separated from cloud cover at a regional scale.

Forest fires are a major problem in Portugal, consuming thousands of hectares per year. A great number of fires are due to arson, which most of the times is related to land use change purposes. Different Government Agencies are responsible for checking if the forest owners are in compliance with the legislation that regulates land use change after fire occurrence. Earth observation data can play a very important role in monitoring land cover transitions occurring in burnt forest areas. An exploratory analysis of a Landsat 5 Thematic Mapper (TM) multi-temporal dataset was carried out to see if pre-defined land cover transitions, within a burnt forest area, could be separated and identified. Three vegetation indices (VI) were used for this propose: NDVI, MSAVI and ARVI. The capabilities of these VI were evaluated on test areas that had pine forest before the fire, followed by a transition into eucalyptus planted in the first or second year after fire or shrub land. The three VI were ranked, in terms of separability, between these three types of transition. ARVI was found to be the one that discriminated better between the two eucalyptus transitions and shrub land.

The main objective of this work has been to evaluate the potential of integration of satellite data and topographic factor, in order to achieve improved performance in forest fire danger estimation. Existing AVHRR-based fire danger estimation methods (a review is specifically made) aim at obtaining fire susceptibility classification exploiting, mainly, the temporal evolution of NDVI, and Surface Temperature (T_s). In this work fire danger estimation has been performed integrating satellite data with fuel type and topographic factors. In order to evaluate the reliability of the estimated indices, the time-space distribution of actual forest fires, provided by the Italian Forestry service, has been used. Preliminary results are very promising; they have shown that in the summer of 1996, a large number of forest fires occurred in the estimated higher danger areas.

A fully automatic approach for extracting 3-D information from simulated 1 m panchromatic digital stereo data was tested within the context of mapping the infrastructure of an open- cast coal mine, and the accuracy of the derived products assessed in comparison to that produced from analogue aerial photographs by standard interactive photogrammetric procedures.

The assessment of flash flood potential of wadi (dry river) systems in arid lands is often difficult because of the lack of sufficiently long rainfall and discharge records of the infrequent and spatially variable rainfall events. Characterization of active wadi systems by remote sensing may offer an alternative solution. A methodology is presented to characterize the channel infill of wadi systems based on the soil composition, and the geomorphic and geologic properties of drainage basins. Spectral mixture modeling is performed on a Landsat TM image to identify the source material (endmembers) and source upland area of the alluvial infill. Endmembers are determined by two methods: (1) identifying pure image pixels, and (2) using spectral libraries representing main rock units of the drainage basin. Both methods are evaluated in terms of their ability to establish the relative contribution of upland source rocks to the overall alluvium composition in the lowland. The identification of wadi systems that are presently most active (high stream-power values) and efficient in transporting sediments to the basin outlet may enable identification of areas prone to flash floods.

This work aims to detect slope failures caused only by mechanical force of earthquake. In many cases so far, the pure detection where the slope failure occurred is not completed because that slope failures occurred with mixing the other events accompanying as heavy rain. To isolate the slope failure due to the mechanical stress of earthquake is important to consider why the slope failed. Our idea is to use the temporal high-resolution data as ADEOS/AVNIR ones just taken before and after the earthquake shock. In this experimental study, we found three slope failures on the sites induced by the main shock, on the northern mountainous region in Kagoshima. Extraction of slope failures purely caused by the earthquake-force serves analyzing the dynamical mechanics of slope failure. Three slopes failed at that time are flat and long about a few hundred meters with an inclination of about 45 degrees. Most of the slope failures have a size of few meters to a few tens of meters. Accuracy of satellite image is required within half DN values in radiometric and half pixel size in spatial. This work showed a practice of 8- meter resolution image to be applied for the detection of slope failures frequently happened by usual magnitude of earthquakes.

Depositional environment mapping in alluvial plains is a basic step in geomorphological, pedological and archaeological studies, where remotely sensed data give an indirect contribution in assessing soil moisture, which could be correlated to sediment texture. However, a textural discrimination based on soil wetness is strictly season- dependent, and any procedure used to map different deposits from remotely sensed data fails when the acquisition time is not appropriate, and the appropriate time is generally different for the various sediments in a study area; hence the need for a multitemporal approach. In the present study a multitemporal Wetness (Tasseled Cap Transformation, TCT) analysis has been performed on the Pisa plain (Central Italy), in order to reconstruct the environment hosting a Roman harbour which seems to be one of the most important Roman harbors ever discovered, as is emerging from the archaeological excavation in progress. Four geocoded and atmospheric corrected images, acquired in March, July, October and December 1991, were processed to obtain just as many Wetness maps. Wetness multitemporal images were produced, and the seasonal changes of this parameter were correlated with grain-size characteristics in selected points in which the soil was bare at each flying over. A Principal Component Analysis on Wetness images was also carried out and synthetic images were produced. Out of all the images, a reliable textural discrimination in the study area was obtained, together with palaeo-geographical information useful in order for a better understanding of the role of the ancient harbor.

The purpose of this study was to assess the two modes of runoff dispersion in order to determine effective management practices to control the sediment delivery rate from a critical portion of the Flint River Watershed in Alabama, USA. The two modes of dispersion were natural storm flow and dispersed storm flow. A 1997 Landsat TM image of the watershed was classified to produce a land use map using a supervised classification technique. Based on the classification, a critical sub-watershed was selected for this study. Using a Geographic Information System (GIS), the storm flow directions under the two modes of dispersion were identified. The model, Areal Non-point Source Watershed Environment Response Simulator (ANSWERS), was then simulated for a one-year return period storm event to compare the sediment yield during the two modes of runoff dispersion. The sediment yield from the selected sub-watershed averaged a reduction of 20% during the dispersed mode.

The recent knowledge on Mars' lithosphere evolution does not find yet sufficient analogies with the Earth's tectonic models. The Viking image analysis seems to be even now frequently, rather fragmentary, and do not permits to express any coherent relationships among the different detected phenomena. Therefore, today it is impossible to support any reliable kinematic hypothesis. The Remote-Sensing interpretation is addressed to a Viking image mosaic of the known Tharsis Montes region and particularly focused on the Arsia Mons volcano. Several previously unknown lineaments, not directly linked to volcano-tectonics, were detected. Their mutual relationships recall transcurrent kinematics that could be related to similar geostructural models known in the Earth plate tectonic dynamics. Several concordant relationships between the Arsia Mons volcano and the brittle extensive tectonic features of earthly Etnean district (Sicily, South Italy), interpreted on Landsat TM images, were pointed out. These analogies coupled with the recently confirmed strato- volcano topology of Tharsis Montes (Head and Wilson), the layout distribution of the effusive centers (Arsia, Pavonis and Ascraeus Montes), the new tectonic lineaments and the morphological features, suggest the hypothesis of a plate tectonic volcanic region. The frame could be an example in agreement with the most recent interpretation of Mars (Sleep). A buried circular body, previously incorrectly interpreted as a great landslide event from the western slope of Arsia Mons volcano, seems really to be a more ancient volcanic structure (Arsia Mons Senilis), which location is in evident relation with the interpreted new transcurrent tectonic system.

Using remote sensing data for geological applications requires specific spatial and spectral resolution combinations for the different geological problems. We evaluate the most adequate combinations for (1) lithological analysis and (2) delineation of depositional bodies. Responses of various hyperspectral and multispectral sensors were simulated using 2000 - 2500 nm laboratory reflectance spectra of metamorphic rocks. MIVIS data (40 bands in the 2000 - 2500 nm region) of a littoral were resampled and used to produce images with various band width and centers and with different ground resolutions. All the images were classified using Spectral Angle Mapper. Heat capacity images of a recent delta plain were obtained using spatially resampled MIVIS data and checking various combinations of resolutions for coalbedo and temperature increment.

The reconstruction of the Urartu civilization (IX-VII centuries BC) has been the main issue of an interdisciplinary project carried on since 1994 by the 'Istituto per gli Studi Micenei ed Egeo-Anatolici' (ISMEA) of the National Research Council (CNR), in collaboration with the Institute of Archaeology of the National Academy of Sciences of Republic of Armenia (IANAS). The Urartian kingdom, at its maximum expansion, encompassed the present eastern Turkey, Armenia and Iranian Azerbadjan. One of project target is the region of Lake Sevan, as the northeastern periphery of the Urartian State. Lake Sevan lies, nowadays, in a closed basin, surrounded by mountain ranges. Between the 6^the millennium BC and today, four cycles of transgression and regression have been detected. In the Sevan basin these flows fluctuations are also due to local volcanic phenomena. Therefore, one of the key problems of the investigation has been the definition of the lake level in different periods, biasing the inhabitants distribution and the environment exploitation. To fulfil this task, contour lines (100 m step) of 1:100,000 scale topographic sheets were first interpolated obtaining as result a Digital Elevation Mode (DEM), analyzed to gain insight to the territorial relationships among the archaeological sites. Moreover, by processing ERS SAR images acquired in different dates, the geo-environmental setting of the southern Lake Sevan region has been investigated, by using also interferometry techniques to create a more detailed DEM.

To stop, or at least to reduce, the weathering of the stone surface its strength is often improved by treatments of different chemical composition. The uniform distribution of the treatment and the conservation of its integrity during the years are crucial factors for a good protection of the monument. As a consequence a technique which allows a remote mapping of the treatment and its characteristics would be very attractive. This paper describes the first experiments devoted to the use of fluorescence lidars in the monitoring of protective treatments on stone surface. Although further investigations are necessary the first results demonstrate the possibility of a detection and characterization of protective treatments by fluorescence lidars.

Different kinds of organisms can grow on stone substrata. Their presence causes effects, which range from low to severe chemical and physical alterations. Up to now, methodologies are not available to investigate in situ biodeteriogens on stone monuments. The paper discusses the use of the laser- induced fluorescence for the detection and characterization of biodeteriogens on stone monuments on the base of the experiments recently carried out by the authors. Dolomitic marble samples inoculated with different cell concentrations of two biodeteriogen have been analyzed in controlled conditions. The fluorescence spectra are compared in order to investigate the possibility of detecting the biodeteriogens at an early state of development, of identifying them by the spectral signature, and of monitoring the effect of biocidal treatments. Although further investigations are necessary the results confirm the potential of this technique in the non- destructive, remote monitoring of biodeteriogen.

A remote sensing reflectance model, which describes the transfer of irradiant light within a plant canopy or water column has previously been used to simulate the nadir viewing reflectance of vegetation canopies and leaves under solar induced or an artificial light source and the water surface reflectance. Wavelength dependent features such as canopy reflectance leaf absorption and canopy bottom reflectance as well as water absorption and water bottom reflectance have been used to simulate or generate synthetic canopy and water surface reflectance signatures. This paper describes how derivative spectroscopy can be utilized to invert the synthetic or modeled as well as measured reflectance signatures with the goal of selecting the optimal spectral channels or regions of these environmental media. Specifically, in this paper synthetic and measured reflectance signatures are used for selecting vegetative dysfunction variables for different plant species. The measured reflectance signatures as well as model derived or synthetic signatures are processed using extremely fast higher order derivative processing techniques which filter the synthetic/modeled or measured spectra and automatically selects the optimal channels for automatic and direct algorithm application. The higher order derivative filtering technique makes use of a translating and dilating, derivative spectroscopy signal processing (TDDS-SP^R) approach based upon remote sensing science and radiative transfer theory. Thus the technique described, unlike other signal processing techniques being developed for hyperspectral signatures and associated imagery, is based upon radiative transfer theory instead of statistical or purely mathematical operational techniques such as wavelets.

A comparative study of three terrain interpolation methods, Delaunay triangulation interpolation, Kriging interpolation and fractal interpolation is presented in this paper. The study uses simulated and real terrains. The simulated terrains are generated by the fractal method. Each terrain is subsampled and then interpolated by each of the three methods. The difference between the original terrain and the interpolated one yields the error surface. A comparison between the error distributions for the three methods of interpolation is presented.

Kernel-driven bidirectional reflectance (BRDF) models have recently been widely used for mapping albedo with multiangle remote sensing data such as ASAS or temporal AVHRR sequences. An Ambrals algorithm will be used to produce global BRDF and albedo products in the coming EOS era using multiangle reflectance data from the MODIS and MISR. Its operational version currently uses a combination of Ross-Thick and Li- Sparse-Reciprocal kernels which has been validated favorably over other kernels or combinations. To further improve the ability of extrapolation of the Ambrals kernel combination with better physical sense while keeping its data-fitting ability, a new kernel, Li-Transit, is derived and suggested to replace Li-Sparse-Reciprocal kernel in next version of Ambrals. We tested the new kernel combination against the current one and a few alternatives using 29 field collected BRDF data sets. The results show similar data fitting ability and more reliable extrapolation for albedo mapping. A test is also done by using the new combination and the current one to produce temporal albedo change maps of New England of U.S.A. using AVHRR images. Presented also is our recent study on scaling effect of Helmholtz principle of reciprocity, and discussion on application of a priori knowledge in kernel- driven BRDF model inversion.

In a modeling case study it has been investigated whether it would be possible to retrieve from optical remote sensing data the (bio)physical parameters of the coupled soil-vegetation- atmosphere system that have an effect on spectral radiances detected by spaceborne sensors. For this, optical data on single leaves generated by means of the PROSPECT model have been applied in the integrated optical soil-canopy-atmosphere radiation model OSCAR. The influences of 2 soil parameters, 2 leaf parameters, 4 canopy parameters and 3 atmospheric parameters on hyperspectral directional planetary reflectances have been simulated in a model inversion experiment. The simultaneous retrieval of the 11 parameters has been tested using classical model inversion by means of the Gauss-Newton method of non-linear least squares parameter estimation. Preliminary results indicate that this approach has some potential, as in a number of widely differing cases the retrieval of all model parameters from 10 nm resolution hyperspectral red edge planetary reflectance data under 5 directions was successful.

Error models of slope and aspect of a terrain are presented in this paper. Such data are often extracted from a GIS which may contain information from digital maps and remote sensing images. Although the sources of these data are usually of diverse resolution, all of them are usually re-sampled to refer to the same resolution. In this paper we shall examine the error which is associated with such data because of subsampling. The error distributions will be modelled empirically.

In the state-of-the-art of 3D extraction from SAR images, we can distinguish three main techniques: radarclinometry, interferometry and radargrammetry. Our project is to perform radargrammetry on high-resolution images recorded by the airborne sensor RAMSES designed and operated by ONERA. Such images allow the visualization of infrastructures and urban areas. First, we are interested in the geometric stereomodel and the location errors due to sensor parameter and disparity errors. We propose a rigorous theoretical error model for every viewing configuration. We use it in order to study the geometric potential of RAMSES sensor for three configurations. We are then dealing with the pertinence of strong reflectors. Their study is based on the analysis of the 3D information extracted from matched points imaged as strong reflectors. For that purpose, we use RAMSES stereo images of an industrial site. Because of many metallic components, the image of this site presents a large quantity of strong echoes that makes difficult the visual interpretation of the scene. We show the first quantitative results on the potentiality of high- resolution radargrammetry on strong reflectors. This analysis allows us to conclude on the possibility of radargrammetric applications with high-resolution airborne sensors.

A novel vector filter called modified vector sigma filter is proposed for processing the multichannel remote sensing radar images. It is demonstrated through simulations and real data examples that the proposed filter is able to provide an excellent combination of properties. It possesses efficient suppression of multiplicative noise and good edge preservation. Moreover, it simultaneously ensures an ability to remove spikes from images and excellent preservation of fine details even if they are characterized by rather low contrasts. These features occur to be useful for further interpretation of multichannel radar images, e.g. for determination of bare soil characteristics like erosion state. For simulated images it is shown that the application of the modified vector sigma filter is preferable in respect with its componentwise counterpart as the former technique provides less misclassification errors.

The Halutza, Agur and Shunra sand dunes are the continuation of the great northern Sinai dunes. Playa sediments from the late Quaternary are found on upper parts of river channels in the area of the dunes of the Northwest Negev and represent disorder in the drainage system caused by wandering of the dunes during other periods. This research examines the ability to use Synthetic Aperture Radar data (SAR), Ground Penetrating Radar (GPR) and other shallow geophysical methods as ground truth in combination with optical and IR data, for detection, identification and mapping of buried channels under sand and exposed channels in the Negev Height, Israel. The buried continuation of Mubra channel, under the Shunra sand dunes towards the west can be seen at a distance of 285 meters from 30:58:42N, 34:36:03E, while it can not be clearly seen in the Landsat TM image in the IR range. The SIR-C image, having several frequencies and polarizations, exhibits meaningful differences in reflectance between Mubra channel and its surroundings. The wavelengths and polarizations that improve the contrast between reflectance from the river channel and the sand are (in descending order): L(HV), L(HH) and C(HH).

We evaluate the response of a passive microwave soil moisture retrieval algorithm to errors in the estimation of input variables and parameters. The model is run varying one parameter at a time within a specified range to quantify the effects individual parameters have on soil moisture retrieval. Although errors in the estimation of most parameters yield total variations in soil moisture of less than about 4% volumetric water content (vwc), variations in the estimates of vegetation water content, vegetation b parameter, percent clay, and surface roughness yield the greatest total variations in calculated soil moisture. The effects of these parameter variations on calculated soil moisture are greater for wetter soils (above 25% vwc) and can result in total variations in soil moisture retrieval up to 24% vwc. These same parameters have a compound effect on calculated soil moisture when they vary collectively; variations in soil moisture retrieval with errors in vegetation water content and surface roughness may be as high as 38% vwc (-12%, +26%). Even over more common conditions between 10% and 25% vwc, errors in vegetation water content, percent clay, and surface roughness result in total soil moisture variations of 9% to 15% (plus or minus 4.5% to plus or minus 7.5%), which are unacceptably high for many applications. When random errors are imposed on these three parameters of the Southern Great Plains 1997 (SGP97) Hydrology Experiment data set, the macrostructure of the soil moisture distribution remains intact compared to the original calculations, but the moisture field is significantly more heterogeneous. It is demonstrated that the distribution (plus or minus 2(sigma) ) of soil moisture for given values of brightness temperature ranges between plus or minus 5% vwc from random errors imposed on the same three parameters. Improvements in parameter estimation in SGP97 contributed to a decrease in the soil moisture uncertainty ((alpha) equals 0.05) by about 67% to plus or minus 3% vwc.

This study investigated the uncertainties inherent in the estimation of soil moisture from variations inherent in the measurement of soil temperature, texture, bulk density, surface roughness, and vegetation water content. Algorithms to derive soil moisture from brightness temperature have been developed, tested and validated at point, field and spatial scales. These algorithms consist of empirical components and their sensitivities to variations in input parameters have not been adequately assessed. We investigated the sensitivity of a passive soil moisture retrieval algorithm that incorporates the Wang and Schmugge dielectric mixing model. Parameters were tested both individually and in combination. Sensitivity is quantified as the difference in observed volumetric soil moisture calculated with the mean parameters and plus or minus 3(sigma) of the individual parameter or their combinations. The algorithm is relatively insensitive to variations in surface temperature, soil temperature at 15 cm depth, as well as clay and sand contents. The algorithm is, however, sensitive to variations in bulk density (maximum volumetric moisture deviation of 11.4%) and surface roughness (9.0% deviation) and highly sensitive to vegetation water content (17.6% deviation). The algorithm is most sensitive to surface roughness and vegetation moisture content under wet conditions. The algorithm is insensitive to the simultaneous variations in surface temperature and clay content within the same textural class. Permutations of variations in bulk density, surface roughness and vegetation moisture content produce a significant compounding effect on sensitivity of the algorithm. Deviations in volumetric soil moisture resulting from the combination of variables often exceed the sum of the deviations due to the individual parameters tested alone. This study suggests that large variations in the determination of surface temperature will have negligible effect on the soil moisture derived using the algorithm. Slight deviations from the mean of the clay content will not significantly affect the precision of soil moisture calculated. On the other hand, precise values of bulk density, surface roughness, and vegetation moisture content are necessary in the algorithm for precise derivation of soil moisture from microwave remote sensing. Further studies on other mixing models at multi- frequencies are recommended.

Soil erosion is a widespread problem in Mediterranean countries. Irregular and often intense rainfalls in this semi- arid region can result in rapid erosion in areas where the slope is steep, the lithology is soft and the vegetation is sparse. The erosion process randomly changes micro-topography of slope surface and thus results in reducing coherence of the radar signals between the initial and the eroded state. This paper reports a case study of rapid erosion using ERS SAR coherence imagery. Three ERS SAR scenes of Almeria Province, Spain with 70, 140 and 210 days separation were processed to produce three coherence images. The coherence images were analyzed in conjunction with Landsat TM images, topographic and geological maps. This study reveals hard evidence of rapid erosion on steep slopes along a motorway cut through strata of soft marls. The slopes subject to rapid erosion appear in intermediate level of coherence in 70 days and gradually lose coherence in 140 and 210 days. In contrast, the dense vegetation in the valley lost coherence almost completely in 70 days and afterwards while more resistant rocks (gypsum and limestone) above and beneath the marls show high to intermediate coherence in the three coherence images.

Data collected by the CZCS (79-85) and the SeaWiFS (98) highlight the relations between bio-geo-chemical and physical processes of the Mediterranean Sea. A comparison of the pigment concentration historical record with surface temperature and wind speed, from the AVHRR (82-91), GEOSAT (86-89), ERS-1 (92-95) and TOPEX (93-97) archives, differentiate between geographical provinces corresponding to (1) areas under the direct influence of coastal interactions; (2) regions of enhanced characteristics (i.e. Ligurian/Provencal/Balearic sub-basin, Adriatic Sea, Aegean Sea), linked to interactions with (northern) continental margins and prevailing winds (i.e. Mistral Bora, Etesians); and (3) open sea, oligotrophic areas, characterized by frontal structures and, in the eastern basin, by a permanent mesoscale eddy field. The trends in these provinces indicate two main seasons, with extreme conditions in winter and summer, and transition periods in spring and autumn. The general pigment cycle is similar to that of a subtropical basin, where light is never a limiting factor, but nutrients always are. Some near-coastal provinces have a distinct seasonality, e.g. that of a subpolar basin in the north-west, with enhanced spring and fall blooms. This view is supported by current SeaWiFS data, which show that features such as river plumes, filaments and permanent gyres are recurrent and maintain their characteristics over the long term.

The objectives of the national Italian research programme PRISMA 2 are presented for the Adriatic in which remote sensing is indicated not only as the tool to study and monitor the basin, but also to support related oceanographic activities. Particular attention was taken in making technologies operational and more effective in relation to complex phenomenologies of this sea after an evaluation of instruments, platform and measurement methods. The results, not yet completely available, contribute to defining a model to monitor the Adriatic Sea.

Advanced Very High Resolution Radiometer (AVHRR) and Sea viewing Wide Field of view Sensor (SeaWiFS) data from January 1998 to June 1999 are used to examine spatial and temporal variability of sea surface temperature (SST) and apparent chlorophyll (AChl) in the Adriatic Sea. Flows long the Albanian coast and the Italian can be distinguished year-round in the monthly averaged AChl, but only in the colder months in the monthly averaged SST's. The AChl averaged fields supply less information on circulation features away from coastal boundaries and where conditions are generally oligotrophic except for the early spring bloom in the Southern Adriatic Gyre. The winter-spring SST and chlorophyll distributions are very different between the two years, particularly in the Northern Adriatic shelf and the Southern Adriatic Gyre. It is hypothesized that this difference may be related to dense water formation that occurs only in the northern and southern Adriatic Seas. The time series of daily SST indicate that dense water formation was favored in 1999 by episodes of cooler winter temperatures in the southern gyre (less than 13.5 degrees Celsius) and on the northern Adriatic shelf. Blooms in 1999 may have been delayed due to surface replacement flows driven by sinking of dense water.

The monitoring of organic matter, suspended or dissolved in the water column, is relevant for the study of the aquatic environment. Actually, the Dissolved Organic Matter (DOM) represents a major reservoir of reactive carbon in the global carbon cycle, thus influencing significantly the marine ecosystem. Due to the strong absorption in the near ultraviolet, DOM reduces considerably the extinction path of solar light in the water column, affecting phytoplankton population and its vertical distribution. The measurement of the DOM absorption coefficient has to be regarded as a good parameter for the monitoring of water quality. This paper deals with the measurements carried out during the oceanographic campaign 'Marine Fronts,' which took place in the Western Mediterranean Sea and Atlantic Ocean from July 14 to August 5, 1998. In this measurement campaign, a high spectral resolution fluorescence lidar (FLIDAR) was installed on the rear-deck of the O/V 'Urania,' acquiring remote fluorescence spectra both in ship motion and in stations. A particular attention was devoted to the monitoring of DOM distribution in the different water masses in marine frontal areas. The lidar data were compared and integrated with SST satellite data and biological samplings. The results show that FLIDAR data agree with satellite imagery, particularly for marine front detection. The comparison with water sample data gave indications for retrieving the DOM absorption coefficient directly from fluorescence remote spectra. In addition, a protein like fluorescence band was detected in the measurements carried out on total suspended matter filtered from the water samplings.

A new method to estimate radiation budgets at air-sea interface by means of AVHRR data has been developed and tested in the framework of the Arctic Radiation and Turbulence Interaction Study (ARTIST). Main goal of the ARTIST project is assess the effects of clouds and of Arctic Haze on the radiative fluxes at the surface and in the atmospheric column in the European Arctic. One month of Advanced Very High Resolution Radiometer (AVHRR) data relative to the period March - April 1998 has been processed and analyzed in order to evaluate short and long wave radiation budgets in the Arctic Sea during the experiment. Remote sensing data (NOAA 14 satellite) have been acquired at the Tromso station and then processed at the Istituto Fisica dell'Atmosfera (IFA) to produce maps of surface albedo and brightness temperature of the experiment's zone. These maps were used to develop a new cloud detection algorithm for the region. Image pixels then have been classified as ice, clouds or water. The method was applied to 151 available AVHRR scenes. The pixels classification performance was verified against the analysis of an expert in satellite image. The cloud classification results to be quite accurate. In fact 99 images are classified as 'very good' by the expert and 37 images the accuracy is a little lower. The radiation budgets are then estimated using several available empirical formulae for clear sky and overcast conditions. The results were compared with in situ measurements made during the ARTIST experiment in order to define the best parameterization of the fluxes. The best estimates of shortwave incoming radiation results from the Bennet (1982) formula for clear sky condition with the Laevastu (1960) correction for overcast condition. The more accurate estimate of incoming long wave radiation in clear sky has been obtained using the Swinbank (1963) parameterization. Finally, averaged map of total radiation budget are calculated for the time period of the ARTIST experiment in the Arctic region.

A Landsat TM image of the Arno River mouth (Tuscany, Italy) was geocoded and atmospherically corrected in order to obtain suspended sediment concentration (s.s.c.) values. For the same day in which the image was a acquired, the river plume was simulated with MIKE 21 modeling system, given river discharge and wind speed. The suspended sediment concentration maps obtained through the two procedures were compared and eddy viscosity and dispersion coefficients were modified in order to gain the plume shape which best fits the remotely sensed one. After additional calibrations, the modeling system will be used to predict pollutant dispersion along the coast, which is intensively used for tourism.

A model which describes the transfer of irradiant light in water is used to predict the fluorescence response of the water surface reflectance under solar induced or an artificial light source such as a laser. Formulations for the estimation of wavelength dependent fluorescent coefficients. The techniques allows the description of a fluorescence reflectance response in deep and shallow waters with various bottom reflectance signatures such as submerged vegetation, corals and sand. Recent advances in the model are presented for obtaining wavelength dependent fluorescence spectrum responses from the solutions of the two flow equations following the procedures developed by Bostater. Synthetic or modeled signatures are presented using in-situ data from the Space Coast of central Florida, USA and the southeastern Atlantic waters near Beaufort, South Carolina. The synthetic or modeled signatures are also dependent upon the attenuation length of the water based upon knowledge of the diffuse attenuation coefficient (k), the beam attenuation (c) or the absorption coefficient (a). The model has potential applications for helping to select remote sensing optimal channels or bands useful in near nadir viewing geometry of estuarine or coastal water columns overlying shallow sand, submerged vegetation, or coral reefs. The analytical solution to the two-flow equations developed by Bostater have transferability to complex but important water quality detection problems that can be assisted using fluorescence processes.

This paper summarizes the current activities of atmospheric and marine data acquisition in the Adriatic Sea and their analysis, carried out at the CNR - ISDGM (Venice, Italy) on behalf of the 'Coastal region long-term measurements for color remote sensing development and validation' (COLORS) project. The main objective of COLORS is to establish an archive of long-term atmospheric and marine measurements in atmosphere and water, carried out on three sites representative of European coastal waters. The subsequent analysis and interpretation of the acquired data has the purpose of detecting time variability of atmospheric and oceanic parameters, in order to develop improved atmospheric correction schemes and better inversion methods to derive water constituents. This paper describes the activities of data processing actually being made; the former part deals with the analysis of in situ indirect and direct measurements. The latter describes the retrieval of water parameters to be used in radiative transfer models. The results currently achieved are summarized and commented.

A fractal model for sea surfaces is proposed, discussed and validated by fractal analysis on synthetic 2D reflectivity maps derived from live recorded data, and on a real ERS-1 SAR image. The proposed dynamic fractal model of the surface is based on the use of the band-limited WM functions. The model takes into account for the multistructure of the sea, it includes the time-evolution of sea waves, and it gives a realistic representation of the sea wave surface since it uses the sea hydrodynamic differential equations describing the sea wave motion. A suitable data analysis technique is defined for the estimation of the fractal model parameters. This methodology is based on fractal and multifractal analysis of the SAR image and allows to directly estimating the sea model parameters from a fractal analysis of the radar data. In particular, a method to obtain the fractal dimension D, based on the multiscale decomposition provided by the normalized Laplacian pyramid (NLP) is used. Multifractal analysis is also used to evaluate the 'local regularity' of the model by means of the Holder exponent spectra. Experiments on measured sea spectra, simulated radar reflectivity maps and real SAR images of the sea surface corroborate the underlying modeling assumptions and show promising results to strictly validate the proposed fractal model on a large class of SAR images of the sea surface.

In this paper, a method for the extraction of in-water structures (e.g., internal waves) and underwater structures (e.g., vertical transfer of sediments along the water column), observed in Multispectral Infrared and Visible Spectrometer (MIVIS) airborne images is discussed. Images of sea surface structures are captured by a variety of airborne or spaceborne remote sensing instruments, such as photographic cameras, optical scanners, synthetic aperture radars. Investigations on the generation process of these structures and on the local effects are of interest to biological oceanography, to fisheries biology as well as to hydrodynamics. Detection method is based on an interference created among low and mid- high frequencies of water-body images, acquired in spectral bands with different water penetration band characteristics. Features of sea-structures, semi-automatically identified, are mapped in image format.

Three different one dimensional upper ocean boundary dynamics models are used to predict the concentration of a short-lived natural photochemical tracer as a function of time and depth. The dynamics models are coupled with parameterizations of the production and destruction rates of the tracer using chemical rate constants from published experimental studies. The model predictions of the upper ocean concentration the tracer do not agree closely with the measured concentrations obtained during one expedition, probably mostly because of errors in the rate constants used in the chemical model (published values of the in situ rate constants show much variability). An optimization scheme is described to find the values of the production and destruction rate constants which minimize the difference between the modelled and measured upper ocean concentrations of the photochemical tracer and the uncertainty is assessed using a Monte Carlo simulation. Using the optimized estimates of the chemical production and destruction rate constants, the simulated upper ocean concentrations from the three dynamics models show close agreement with each other. However, it is still difficult to determine which of the models is the best description for actual observed tracer concentrations because of weaknesses in the existing in situ data sets and models.

This paper presents results of modeling of oil film contrast on the sea surface. Modeling consists of two stages: (a) determination of angular dependencies of transmittance and reflectivity of light at oiled seawater surface; (b) determination of upward light flux when wind velocity is changeable, with other parameters fixed. Stage (b) includes phases: (ba) modeling of light flux running perpendicularly upward from clean water (using Monte Carlo method); (bb) similar modeling like in (ba) but with surface covered by oil; at this stage results from stage (a) are utilized; (bc) calculation of contrast of oil smudges based on (ba) and (bb). In relation to (a) this paper shows results of modeling of optical parameters of an oiled seawater surface. Regarding the stage (b), figures show number of photons entering the atmosphere -- when sea surface is clean and when it is covered by oil film -- as a function of wind speed. When sea surface is relatively calm contrast of oil films becomes lowest (negative). On the contrary, when sea surface is strongly waved, contrast of oil film increases and becomes positive.