The spatial resolution (approximated by the pixel size) limits the spatial information that can be extracted from remotely sensed images. For a given set of sensor and image characteristics (e.g., pixel size, waveband combination etc.) the objective of remote sensing analysis should be to process the available imagery to maximise the amount of information in the final map. Where the variable to be predicted is categorical (for example, land cover) then it is well known that soft classifiers provide more information than hard classifiers. In this presentation, it will be demonstrated that soft classified imagery can be further processed to increase the amount of spatial information provided. The technique presented uses the concept of spatial dependence, either with or without a prior spatial covariance model, to produce a hard classification at a finer spatial resolution than that of the original image. The super-resolution mapping approach represents a large step forward in remote sensing classification research from (i) hard classification (1980s), through (ii) soft classification (1990s) to (iii) super-resolution classification (2000s).

terug naar programma

HYPERSPECTRAL REMOTE SENSING FOR COASTAL MONITORING.
Bart Deronde, VITO, België

The goal of this project is to determine the spatial distribution of different sand types along the Belgian coast (i.e. sand originating from different sources and/or locations). Airborne measurements using the hyperspectral CASI-scanner provide spectral signatures of the topsoil at the Belgian coast.  However, the recorded signals are signatures of wet and dry sand what complicates the classification of the different sand types. Therefore, lab-measurements on a statistically significant number of samples are performed using an ASD-Fieldspec Pro FR spectroradiometer to log the signatures of the sand under different relative humidity conditions. Subsequently, a humidity-index ('NDWI') is implemented to define the moisture content of the top soil. The lab-measurements and the humidity index are used to transform the CASI-data into dry sand signatures. These signatures are clustered into three to seven distinct classes. An airborne laser scanner is used to obtain an accurate DEM of the coast. This DEM allows an estimation of the available amount of each sand type. If the project proves to be successful the measurements will be repeated regularly and lead to a better insight in the processes of sand transport along the Belgian coast (cf. Coastal zone management).

terug naar programma

ONSHORE MUD VOLCANO CHARACTERISATION IN AZERBAIJAN USING ASTER AND LANDSAT TM
Klaas Scholte, TU Delft

Oil and gas fields in Azerbaijan (South Caspian Basin) are intimately associated with mud volcanism. Mud volcanoes naturally expel Jurassic, Cretaceous, Paleogene, Miocene and Pliocene sediments onto the surface together with traces of hydrocarbons (oil and gasses). Studies carried out on the different mud volcano outflow (breccia) components, i.e. gas, oil, mud and rock clasts of mud volcano breccia, show that mud volcanoes vary in terms of breccia type, activity and size.
In this study we propose a mud volcano characterisation based on the surface spectral properties of the mud volcano breccia by assessing Landsat and ASTER satellite image data in combination with detailed field survey. We hope to provide information on mud volcano dynamics as well as the source of the reservoirs and productive series.
Our analysis shows that we are able to describe and classify a variety of mud volcano flow materials, as well as crater materials from mud volcanism, in terms of endmember combinations by using remote sensing data, field data, image processing techniques and dynamic GIS map overlay. These variations in mudflow types may be related to different eruption events controlled by recent tectonic events.

terug naar programma

OPERATIONAL MONITORING OF WATER QUALITY PARAMETERS BY THE EPSA SCANNER
Hans van de Woerd, Robert Vos en Hans Hakvoort, IVM VU / Survey Department RWS

The hyperspectral optical scanner of Rijkswaterstaat, EPS-A, has been implemented in their remote sensing aircraft and is operational available. Technical and logistic problems that occurred in 2000 and 2001 currently are being taken care of. Dedicated software has been developed to offer a prompt service to a wide user community in the Netherlands. In this presentation the EPS_A and the processing chain for measured data are described. Particular attention is given to the atmospheric correction procedure and the matrix inversion methods for water quality parameters. Results of successful observations of lake IJsselmeer are presented.

terug naar programma

A HYPERSPECTRAL APPROACH TO STUDY COLOURED DISSOLVED ORGANIC MATTER (CDOM) IN WETLAND ECOSYSTEMS
Marnix Laanen, IVM VU

The VU has initiated a research programme concerning the effects of discrete water level changes ('vernatting') on the natural and societal values of wetland ecosystems. Part of this programme is the PhD study presented here, in which hyperspectral remote sensing is used to obtain spatial and temporal information on the physical-ecological status of the Vecht area wetlands. In this research the focus is on retrieving optical water quality variables. These variables are concentrations of chlorophyll and suspended matter (for which well defined algorithms exist) and CDOM concentration (for which novel algorithms must be designed). CDOM composition is seen as a potential indicator of wetland composition and trophic status because the water quality is predominantly influenced by fen degradation and/or eutrophication. Several new methods and ideas for measuring and modeling CDOM optical properties and composition are being pursued. Preliminary results of ground-based (PR650) and airborne (EPS-A) hyperspectral observations will be presented.