Naslov
Mapping Of Terrestrial And Marine Habitats Of Croatia

Autori
Kušan, Vladimir ; Križan, Josip ; Jelaska, Sven D. ; Petricioli, Donat ; Antonić, Oleg ; Bukovec, Dragan ; Bakran-Petricioli, Tatjana

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
24th EARSeL Symposium, New Strategies for European Remote Sensing : ABSTRACT BOOK / - Dubrovnik : Hrvatska akademija znanosti i umjetnosti (HAZU), 2004, 19-19

Skup
24th EARSeL Symposium "New Strategies for European Remote Sensing"

Mjesto i datum
Dubrovnik, Hrvatska, 25.05.2004. - 27.05.2004

Vrsta sudjelovanja
Poster

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
habitats; Croatia; satellite images; neural network

Sažetak
The Ministry of Environment Protection and Physical Planning of the Republic of Croatia is financing the project "Mapping of habitats of the Republic of Croatia", which is being implemented by Oikon Ltd. from Zagreb. The three-year project is in its final phase and will be finished by the end of 2003. For the terrestrial part of Croatian territory the data sources for mapping were Landsat ETM+ satellite images plus intense field samples. Classification and interpretation was done with minimum mapping units of 9 ha. Two sets of images of the whole country area were simultaneously used: spring and autumn set. In the first step, each Landsat ETM+ scene was classified using supervised classification to the basic landcover units. In the second step, each landcover unit (on each scene) was classified to the subunits using unsupervised classification supported with the optimisation of a number of clusters. Finally, the results of unsupervised classification were interpreted with the support of field samples and additional ancillary data (old vegetation maps, forest management maps, lithological maps and literature). Existing data for the mapping of sea bottom, as well as new data, which could have been collected during the project, were not sufficient to apply one uniform method, so a number of different methods were used. Supra-littoral and medio-littoral were mapped as a function of the coastal lithology, in a scale of 1:100 000. Infra-littoral was mapped with the aid of spatial modelling in the framework of raster-GIS. Neural networks were used as a modelling tool. Basic infra-littoral habitat types determined on known locations represented dependent variables. Independent variables were: 1) digital sea depth model produced by digitalisation and rasterisation of cartographic original (nautical maps) in scale 1:100 000, 2) sea bottom slope derived from digital model, 3) distance from the coast, 4) sea temperature and bottom sea current magnitude obtained through mathematical model and 5) specific spectral channels of Landsat ETM+ satellite image (assuming that they can be used for specific purpose within the photophilic zone). The lower border of infra-littoral was determined on the basis of sea depth, as well as borders of circa-littoral and bathyal. Circa-littoral and bathyal, for which accessible data were scarce, were mapped from: 1) lithological map of sea bottom in scale 1:1 000 000, 2) map of circa-littoral biocoenoses in the scale of 1:3 000 000 and 3) digital sea depth model.

Izvorni jezik
Engleski

Znanstvena područja
Biologija

POVEZANOST RADA