A simulation is defined and tested for oceanic constituent estimation in case II waters, for the future medium resolution imaging spectrometer (MERIS) oceanic remote sensing instrument, using singular valued decomposition (SVD) and artificial neural networks (ANN) inversion techniques. The SVD technique, which bears a close resemblance to multivariate statistic techniques has previously been successfully applied to the problem of chlorophyll estimation from case I waters. In this study, a model is developed for the calculation of oceanic surface reflectance, as a function of the three major constituents which contribute to the optical properties of the water, (chlorophyll like pigments, yellow substance and sediments). The oceanic models have been validated using optical data acquired in the North Sea (1994) using the MARAS instrument. This surface reflectance is used to predict top of atmosphere radiance, which is then inputted to the MERIS instrument model. The algorithms are implemented on the simulated data to provide robust algorithms for the estimation of chlorophyll, sediment and yellow substance concentrations. The results of this investigation are presented with emphasis on recommendations for algorithm development, pre-processing and sampling strategies.

Water absorption signatures were measured from water samples placed in a 50 cm pathlength cylindrical cuvette. Quantitative analysis of chlorophyll-a and dissolved organic matter (DOM- humic acid, fulvic acid, or tannic acid) was conducted using second derivative spectra followed by computation of double inflection ratio (DIR) spectra for all possible combinations of bands (from 362 - 1115 nm with 252 channels). A specially designed instrument system is described which allows measurements of absorption of particulate and dissolved organic matter (chlorophyll-a and DOM) in a water sample. The ability of the system to allow measurement of absorption signatures and relating the data to observed or in-situ water reflectance signatures measured from a moving or in-situ platform is described. The methods demonstrate the value of high spectral resolution signatures to estimate concentrations of the water quality parameters and an analytical technique using optimal ambient correlation spectroscopy for selecting bands or channels for estimating concentrations directly from spectral absorption signatures.

This paper presents the methodology and the results of a theoretical work aiming to evaluate the potential of sea-viewing wide field-of-view sensor (SeaWiFS) visible channels for marine parameter estimate, and to propose the relevant algorithms. A model of ocean color, briefly recalled, was used for the study. With the model, a large number of reflectance curves for different values of the input parameters (the concentrations of optically active components) were generated. The curves were then integrated over the visible channels of SeaWiFS radiometer. A regression analysis between integrated reflectances and input parameters yielded the algorithms. Results are presented for case 1 and case 1 plus 2 waters, and for various kinds of algorithms, i.e. combinations of reflectance ratios and/or reflectance logarithms, and the so-called 'hyperbolic' algorithms. The values of the correlation coefficient put in evidence the effectiveness of SeaWiFS data for the retrieval of pigment, suspended sediment and yellow substance concentrations.

A simplified two-flow model derived from the radiative transfer problem applicable to a water column is described and applied. The application of the model to reflectance signatures measured in Delaware Bay estuarine waters is described. The shapes of the reflectance signatures compare nicely with field data, although the absolute magnitudes of reflectance are different. The model could be improved if a more realistic non-homogeneous model is applied or actual absorption coefficients were measured. An analytic two-flow model was developed and the sensitivity analysis runs for estimating radiance reflectance (nadir view geometry) above the water surface suggest that the most important in-situ parameters are absorption coefficient, the internal diffuse reflectance coefficient, and the backscatter coefficient.

The lidar fluorosensor, built at ENEA Frascati to remotely monitor the sea-water quality by collecting the water Raman backscattering and induced fluorescence from dispersed oils, suspended matter and chlorophyll, has been employed in a marine campaign in the lower Adriatic sea. Extensive calibration measurements have been undertaken by analyzing with the lidar, both in the laboratory and during the campaigns, sea water samples taken at several places along the Italian coasts. Absolute values of organic matter chlorophyll concentrations have been obtained by calibrating the lidar data with standard physical-chemical methods. This system has been recently upgraded for detecting the phytoplankton photosynthetic activity, by means of the pump-and-probe technique, which has been assessed to monitor the process in microalgae during laboratory experiments.

The FLIDAR, built at IROE-CNR in Florence, was the first of a new generation of sensors operating with high spectral resolution both as fluorescence lidar and passive spectrometer. This research includes the sensor development and the study of a suitable data processing for an extensive monitoring of the marine environment. As a result, both the laboratory and field experiments allowed the monitoring of: (1) water column temperature, by the water-Raman spectral shape; (2) oil pollution (oil class identification and thickness), by both the oil fluorescence spectral features and the water-Raman signal intensity; (3) phytoplankton and phytobenthos (identification and stress), by their fluorescence spectral signatures and their spectral behavior; (4) mucillagine, its observation has been carried out directly in the field, addressing a method for its detection. Finally, the FLIDAR has been operating since 1991 from different platforms, such as ship, helicopter, and airplane.

The NOAA advanced very high resolution radiometer (AVHRR) multichannel sea surface temperature (SST) images of the Japan Sea are analyzed with the orthonormal wavelet transform (WT) technique. The decoding-encoding WT allows better representation of the subpolar front, mesoscale eddies, and streamers that appear in the SST patterns. The analysis shows remarkable difference in the typical SST patterns in October 1993 and April-May 1994. While SST patterns were fuzzy in autumn, they appeared sharp in spring. The plumes of irregular shapes with characteristic horizontal scale of less than 30 km developed over the SST field at the beginning of the cooling season when the horizontal temperature gradients were essentially eroded. In contrast, very narrow and elongated up to 100 km streaks appeared at the beginning of the heating season when the temperature gradients were sharp. The spectral analysis of SST results in the power law behavior of k^-2.0 in October and k^-2.8 in May, for the scales ranging from 10 km to 100 km. These observations suggest that the surface cooling in the autumn favors the enhancement of three dimensional mixing, while during the onset of heating the horizontal stirring becomes more pronounced at geostrophic turbulence scales.

Twenty-six daytime NOAA-11/AVHRR images covering the Danish waters are analyzed together with 99 cloud-free bulk temperature measurements. The images are cloud screened with a cloud-detection algorithm, which among other tests is based on a correlation test between channels 2, 4, and 4-5. This correlation test is capable of distinguishing cold water pixels from cloud pixels. According to the split-window hypothesis, the channel 4-5 brightness temperatures should increase, and the channel 4 brightness temperatures should decrease with the radiation path. These split-window assumptions are examined, and the channel 4-5 difference does not seem to carry significant information about the true sea surface temperature.

Using scatterometer data from ERS-1 we have shown that realistic wind fields may be obtained with the VIERS scatterometer algorithm. The VIERS algorithm determines for given parameters such as wind speed and direction, sea state and radar incidence angle, the radar backscatter at the ocean surface. Thus, by inverting the algorithm the wind field over the ocean may be determined from the observed radar backscatter provided the sea state (given by the WAM model) is known. The inversion of the algorithm was achieved by means of minimization of a cost function that penalizes deviations between modeled and observed radar backscatter and deviations between retrieved and atmospheric model wind direction. The retrieved wind fields are in good agrement with analyzed ECMWF winds. Compared to the present operational CMOD4 algorithm we found that the VIERS algorithm performs better in extreme conditions; in the high wind speed regime wind waves are in a relatively early stage of development and are therefore so steep that the radar backscatter from the short waves is affected. From a synoptic study we have concluded that the VIERS winds are of high quality. The VIERS algorithm performs well in cases of frontal systems, for example, and only occasionally problems with the wind direction ('the ambiguity problem') occur. The end result is that we have developed a scatterometer algorithm based on our present knowledge of the physics of the relevant processes. From the observed radar backscatter this algorithm enables us to obtain reliable wind fields.

In this paper we propose a different approach to the wind field retrieval over the sea surface from the scatterometer data. This approach is based on the exploitation of different properties of the wind field which allow the coupling of the wind field reconstruction in nearest neighbor points of the scatterometer measure grid (not pointwise method). The results obtained by the reconstruction procedure on ERS-1 scatterometer data over the Mediterranean sea are shown.

The wind stress is the primary force driving the tropical oceans from daily to interannual time scales. Conventional measurements from ships of the wind vectors are not available with a sufficient quality regarding data accuracy as well as their coverage. Satellite observations of the surface wind over the sea are now available on a routine basis at the Institut Francais de Recherche pour l'Exploitation de la Mer (IFREMER), derived from the European Remote Sensing-1 (ERS-1) scatterometer and altimeter and from the radiometer special sensor microwave/imager (SSM/I). More than 3 years of weekly stress fields (1991-1994) with a resolution of one degree in latitude and longitude are produced using an objective analysis method. The accuracy of these gridded winds was evaluated by comparison with TAO buoys in the tropical Pacific area (Riou, 1995). The root mean square differences are of the order of 1.2 m/s and 15 degrees. The greatest differences are observed in the TOGA/COARE region where the wind variability is largest on the weekly scale. The low frequencies (monthly to interannual) of the wind variability are discussed and compared to those obtained from the TAO buoys. In this paper the time and space scales of the sea surface wind are described using a complex EOF analysis. One of the most interesting results is that the weekly averaged wind fields derived from ERS-1 scatterometer are useful to depict a 30-50 day oscillation over the tropical Pacific Ocean.

The usefulness of the low resolution ERS.SAR imagery for the detection of ice in the waters off Cape Farewell is evaluated by comparing of the images with manual observations, photographs and video recordings, obtained from the ice reconnaissance aircraft from May- June 1994. All the images that were analyzed contained large backscatter due to capillary waves; even in moderate wind conditions (less than 5 m/s). Areas of open pack ice with concentration less than 6/10, which was common in the images, individual floes, which are small and medium (less than 200 m) in these waters, or groups of such floes, were completely masked out by the sea clutter. In some SAR images the backscatter from the water areas had large variability (perhaps due to local wind conditions); it was stronger or weaker than, or even same as, the backscatter from the neighboring regions of ice. In many cases, therefore, it was difficult, if not impossible, to identify the regions of ice and water using the ERS.SAR imagery. Nevertheless the images appeared generally good for monitoring the edges of fast and compact ice (with ice concentration of about more than 7/10).

The object of this work is to solve the problem of the ocean hydrophysical and ecological parameters evaluation from the multispectral optical sensing data. The presented procedure enables the time characteristics of the vertical and horizontal stratifications of the sea water ecological parameters everywhere over the region under investigation with the resolution defined by parameters of the devices in use to reconstruct. This procedure makes use of both the seas surface optical remote sensing data measured everywhere over the region and the vertical stratification of the sea water optical properties received in a limited number of points on that region. Also the correction algorithm is elaborated. It allows the contribution of the radiation reflected by both the atmosphere and the sea surface to reduce.

Natural organic compounds are present in significant concentration in all types of water. The spectroscopic study of dissolved organic matter and its major components is important for applications of lidar remote sensing techniques for water quality measurements. In this paper, fluorescence emission and excitation spectra, as well as absorption spectra for commercially available analogues of major components of gelbstoff are analyzed and compared with spectra of natural water samples of different origin.

Biomass burning of forests and savannas is a phenomenon of continental or even global proportions, capable of causing large scale environmental changes. Satellite space observations, in particular from NOAA-AVHRR GAC data, are the only source of information allowing documentation of burning patterns at regional and continental scale and over long periods of time. This paper presents some techniques, such as clustering and rose- diagram, useful in the spatio-temporal analysis of satellite derived fire maps to characterize the evolution of spatial patterns of vegetation fires at regional scale. An automatic clustering approach is presented which enables us to describe and parametrize spatial distribution of fire patterns at different scales. The problem of geographical distribution of vegetation fires with respect to some location of interest, point or lines, is also considered and presented. In particular rose-diagrams are used to relate fire patterns to some reference point, as experimental sites of tropospheric chemistry measurements. Different temporal data-sets in the tropical belt of Africa, covering both Northern and Southern Hemisphere dry seasons, using these techniques were analyzed and show very promising results when compared with data from rain chemistry studies at different sampling sites in the equatorial forest.

Kernel-driven linear bidirectional reflectance models are gaining increasing attention for their potential use in operational processing of global remote sensing data. Nevertheless, the ability of these models to simulate actual reflectance anisotropy has not been completely explored with remote sensing data. To assess the suitability of linear models for the MODIS atmospheric correction system, we inverted a series of models with AVHRR and MODIS airborne simulator (MAS) data. For comparison, we also fit 2-stream turbid medium models to the respective data sets. Although the more complex models produced more accurate fits, the linear models were acceptably accurate and considerably faster. We conclude that linear models perform with sufficient speed and accuracy for atmospheric correction algorithms.

The linkage between sea surface temperature (SST) and surface winds from tropical and subtropical regions of the North Atlantic, and ground temperature and rainfall amount from southern and central Europe are studied using canonical correlation analysis (CCA). The data sets which consist of monthly means span the interval from January 1965 to February 1985. The effects of Atlantic processes on European temperature and rainfall anomalies are seasonal dependent. Two types of variability patterns are found: one in which the response to Atlantic forcing is rather local (Iberian area) and the other that seems to be related to some large scale phenomena. The North Atlantic oscillation signal is revealed by the Atlantic-European patters of winter variability.

Groundbased microwave measurements are well suited for the intercomparison and validation of satellite data. A microwave heterodyne spectrometer, which can be used for this purpose has been developed at our institute. It measures the 22.235 GHz water vapor spectral emission and supplies water vapor profiles in the altitude range from 35 to 85 km with a time resolution of one day. The instrument, consisting of a cooled heterodyne receiver frontend and a chirp transform spectrometer (CTS) backend is described and water vapor profiles are presented.

In microwave remote sensing of atmospheric trace gases a retrieval technique described by C. D. Rodgers as optimal estimation has widely been adopted during recent years. We found out some difficulties in applying this method to the analysis of long term groundbased observations due to the large variability of the tropospheric transmission. The corresponding changes of the data quality and the weight of the a priori profile in the inversion algorithm can lead to an artificial correlation between the retrieved profiles and the transmission. In addition an assessment of the inversion only from the error bars of the profile is impossible since they don't respond linearly to the errors of the spectra. It is shown, that an inversion algorithm according to Backus-Gilbert's philosophy will avoid these difficulties.

A groundbased microwave heterodyne spectrometer for the monitoring of the vertical distribution of middle atmospheric ozone has been developed at our institute. It has supplied data since the end of 1992. Because standalone operation has been planned, one important design goal of the instrument was to achieve a high degree of automation. In this paper an overview of the instrument is given and some of the long term data are presented.

The design and scientific objectives of the SAFIRE-A (spectroscopy of the atmosphere using far infra-red emissions-airborne) instrument are described. The instrument exploits technical developments achieved during the development of the SAFIRE concept for space. These include an interferometer of compact design with spectral resolution 0.004 cm^-1, and narrow-band filters (2 cm^-1 wide) to reduce photon noise. The instrument is capable of detecting ozone and trace gases that affect ozone depletion, for example OH, HO₂, H₂O₂, HOCl, HCl, ClO, HF, HBr, and also H₂O. The measurements include column content above the aircraft (above 20 km) and height resolved abundances at lower altitudes. The first flight is scheduled for July 1996.

According to long-term ultraviolet (UV) observations which have been carried out in the Meteorological Observatory of Moscow State University since 1968 influence of different cloud types and cloud amount on UV radiation has been analyzed. UV transmission by cloudiness (Cq) was shown to have distinct spectral peculiarities. Approximate formulas were carried out for different extended clouds Cq distributions. Cq values were shown to be independent of cloud amount (N) of high plus middle clouds. Nonlinear Cq versus N dependence of low clouds was approximated by analytical equation. It may be used in analysis of possible effects of cloud amount trends. Results of data analysis show positive trends of N in different regions of the world (e.g. dN equals 1 in the USA and Canada. dNlow equals 0.2 and dNlow equals 0.5 per decade for Estonia and Moscow). In Moscow Nlow growth caused 6-8% UV attenuation per decade. Due to nonlinear Cq dependence on N the effect of cloud trends on UV radiation may be stronger in climatic regions characterized by great cloud amount values.

Measuring the spectral signatures of under water coral reefs and mapping of coral reefs by satellite remote sensing are described. The spectral signatures of different species of the coral reefs were measured using a spectroradiometer off Kuroshima Island, Okinawa, Japan and investigated spectral difference between different species of the coral reefs. As well as the field experiments, laboratory experiments for measuring the spectral signatures of 9 different species of coral reefs were carried out with the same spectroradiometer. The spectral reflectance of each coral reef showed a significant result that a narrow absorption band exists in the spectral region between 66a0 and 680 nm, and very strong spectral reflectance from about 700 nm towards the longer wavelength range. On the other hand, absorption and the high reflectance region were not observed from the bottom sands or bare rocks underwater. These experiments suggest that there is a significant spectral difference between coral reefs and bottom sands or bare rocks and so the best spectral range for separating the coral reefs from other underwater objects in the ocean would be between 700 and 800 nm. As well as the basic spectral measurement either in the field or at the laboratory, SPOT satellite imageries were used to classify the underwater coral reefs. Classification methods used here were the principal component analysis, and the maximum likelihood. Finally, the evaluation of classification method for extracting the coral reefs was introduced.

The study of phenomena such as ENSO requires consideration of the dynamics and thermodynamics of the coupled ocean-atmosphere system. The dynamic and thermal properties of the atmosphere and ocean are directly affected by air-sea transfers of fluxes of momentum, heat and moisture. In this paper, we present results of turbulent heat fluxes calculated by using two years (1992 and 1993) monthly average TOGA data and ATSR SST data in TOGA area. A comparison with published results indicates good qualitative agreement. Also, we compared the results of heat flux exchange by using ATSR SST data and by using the TOGA bucket SST data. The ATSR SST data set has been shown to be useful in helping to estimate the large space scale heat flux exchange.

Climatological sea surface temperature (SST) is an initial step for global climate processing monitoring. A comparison has been made by using Oberhuber's SST data set and two years monthly averaged SST from ATSR thermal band data to force the OGCM. In the eastern Pacific Ocean, these make only a small difference to model SST. In the western Pacific Ocean, the use of Oberhuber's data set gives higher climatological SST than that using ATSR data. The SSTs were also simulated for 1992 using climatological SSTs from two years monthly averaged ATSR data and Oberhuber data. The forcing with SST from ATSR data was found to give better SST simulation than that from Oberhuber's data. Our study has confirmed that ATSR can provide accurate monthly averaged global SST for global climate processing monitoring.

Ground based total ozone column retrievals have been compared with NOAA satellite based total ozone column retrievals. The ground based retrievals were references in the intercomparison. For ground based retrievals Brewer and SAOZ retrievals are used. Brewer retrievals were used as references from March to September and SAOZ retrievals were used as references from October to February. The NOAA satellite based retrievals are CNRM, ITPP3 and NOAA/NESDIS retrievals. The NOAA satellite based retrievals have been used throughout the whole year in the intercomparison study. The intercomparison between satellite based and ground based retrievals has been carried out for the years 1993 and 1994, for the near polar location of Sodankyla (67.4N, 26.6E). Relative biases and correlations have been calculated between the satellite based and ground based retrievals. A seasonal variation in relative biases between the ground based and satellite based retrievals can be seen. For October to February the relative biases are with regard to the SAOZ references: 1.5% for CNRM, 5.3% for ITPP3 and 3.7% for NOAA/NESDIS retrievals. For March to September the relative biases are with regard to the Brewer references: 4.2% for CNRM, 17.0% for ITPP3 and 7.0% for NOAA/NESDIS retrievals. Relative biases between ITPP5 satellite based retrievals and the ground based retrievals have been included in figures for some months.