OPTICAL REMOTE SENSING

DYNAMICAL INTERPOLATION OF THE OCEAN COLOR IMAGES USING COUPLED PHYSICAL, BIOLOGICAL AND OPTICAL MODEL

Besiktepe, Sukru Turan<sup>1

1</sup>NATO Undersea Research Centre Viale S. Bartolomeo 400, La Spezia, --, 19126, Italy

Collocated physical and biogeochemical observations at near

synoptic resolution during the Spring of 2001 in a region of

the South West Black Sea overlapping coastal and deep waters

across the Rim current are combined with remotely sensed

data (AVHRR, MODIS, T/P), and historical data to generate

four dimensional physical and ecological fields via a

coupled physical and biogeochemical models of the Harvard

Ocean Prediction System with sequential data assimilation.

The dynamical model employed here is the Harvard Primitive

Equation Model. The bio-chemical model coupled to physical

model includes phyto-plankton, zoo-plankton, detritus, nitrate,

ammonium and Chlorophyll. The propagation of light in the water

column as well as the utilization of that light by the phytoplankton

are linked with a full spectral irradiance model coupled

with a spectral absorption-based photosynthesis model.

The model calibrated and validated for the Southwetern Black Sea and

then used for data driven simulations. MODIS derived chlorophyll-a

data assimilated into the model together with in-situ chlorophyll-a

measurements which allows dynamical interpolation of the satellite derived

chlorophyll-a in which regions with missing data due to the cloud

covers was filled. Furthermore, 3-d daynamical model also allows

extensions of the surface values obtained by satellites to subsurface.

Therefore accurate and realistic four dimensional chlrophyll fields based

upon satellite and near synoptic observations constructed.

COMBINING HYPERSPECTRAL AND ENVIRONMENTAL KNOWLEDGE USING PROBABILISTIC METHODS TO PRODUCE SHALLOW WATER HABITAT MAPS.

fearns, peter¹; Garcia, Rodrigo¹; Klonowski, Wojciech<sup>1

1</sup>Curtin University of Technology GPO Box U1987, Perth, --, 6845, Australia

Airborne hyperspectral data were collected over the Ningaloo Reef, Western Australia using the HyMap sensor during April 2006. These data have been processed using a shallow water reflectance model to infer benthic cover classes. The level of classification, in terms of the number of classes and confidence in applying specific species names to biotic cover, can in principle be extended to retrieve many classes. However, limitations in the ability to classify substrates can occur due to many factors, including when assumptions about water column properties are not accurate, reflectance signatures of substrates are not known well, or are variable within classes, SNR of the sensor is low, viewing conditions compromise data quality, water depths are large, or substrate reflectance is low.

An alternative, and older, technology for mapping shallow water habitats is manual interpretation of aerial photography. This method requires an “expert” to interpret a photograph, and based on information such as image brightness and colour, shapes of features, scene texture, locations relative to shore lines, outflows, reefs etc, local knowledge, ecological knowledge, and historical data, the expert determines a habitat class.

We have extended the purely spectral, pixel-by-pixel approach, by including scene and ecological information using Bayesian statistics. This probabilistic method is useful when spectral data are compromised to some extent, but when there is enough spatial information within the scene to extract ecological and physical information. Examples of preliminary research into this advanced classification method will be presented.

EXTENDING THE SATELLITE SURFACE OPTICS TO DERIVE THE 3D OPTICAL FIELD BY DEFINING THE UNCERTAINTY OF PHYSICAL – OPTICAL RELATIONSHIPS

Arnone , Robert ¹; Casey , Branndon ²; Ko, Dong S³; Ladner, Sherwin ²; Flynn, Peter ¹; Rowley, Clark ³; Gould , Richard <sup>1

1</sup>NRL NRL, Code 7330 - Ocean Sciences Branch , SSC, MS, 39529, United States; ²PSI, SSC, MS, 39529, United States; ³NRL Code 7320, SSC, MS, 39529, United States

Subsurface optical properties can be substantially different from the surface optical properties that are traditionally defined using satellite ocean color. We developed a method to vertically extend the satellite optical properties though relationships of physical and optical properties. Circulation models provide hindcast and forecast of the physical ocean properties, such as the temperature, salinity and current properties using advanced data assimilation. These physical properties such as the mixed layer depth (MLD) and the degree of stratification (IMLD) are many times coupled to the optical layers such as the chlorophyll maximum and the subsurface light field (1% light field).

The subsurface optical layers can be parameterized using a Gaussian profile shape which is constrained by 1) surface satellite optics at the surface 2) physical characteristics MLD and IMLD and 3) 1% light level. Using these constraints, we optimize the vertical profile shape based on insitu observations (gliders, profiles etc.) to determine optimal coefficients to define these physical – optical relationships for a Gaussian profile shape.

The optimized coefficients are used to derive the regional 3d optical field for satellite optical imagery combined with physical properties from circulation models. The uncertainty of the 3d optical field is defined by the variability of the optimized coefficients which represent the physical – optical relationships. We characterize the uncertainty of the 3d structure by creating ensembles of 3d profiles using variations in the optimized coefficients.

The results provide a unique method to combine satellite optical properties, insitu optical and physical measurements and ocean circulation models to represent the 3d bio-optical properties with some estimate of the uncertainty.