Recently, geological, geochemical, and biological data collection increased considerably in the marine environment together with ecological, economical, and scientific interests in marine coastal environments (e.g. North Sea) and ocean margins (e.g. Håkon Mosby Mud Volcano). The increasing amount of geodata results from new sampling devices, as in situ sensors, and mobile underwater platforms as ROVs (Remotely Operated Vehicles) and AUVs (Autonomous Underwater Vehicles), or from satellite-supported data transfer from moorings. Compared to the multitude of measured parameters and the quantity of information compiled during multidisciplinary research cruises, only few concepts and methods were developed for visualisation, distribution of data and thematic maps, efficient integration of the inhomogeneous data into existing database structures, management and spatial analysis of geodata.The identification of distinct provinces is currently an emphasis of marine research geosciences. A typological approach combining geological, biological and chemical properties is accomplished by geostatistical, multivariate statistical, and GIS techniques (Geographical Information System). Besides scientific needs as surface-related balances of geological and geochemical cycles, seafloor provinces support management decisions related to upcoming economic use of the seafloor (e.g. such as installation of off-shore wind parks or the declaration of protection zones) and bear up to model spatio-temporal connections and changes of coastal regions.Submarine mud volcanoes are considered as significant source locations for methane indicated by unique chemoautotrophic communities as Beggiatoa mats and pogonophoran tube worms. The Håkon Mosby Mud Volcano (HMMV) is located at the continental slope of the Barents Sea in a water depth of 1260 m. A large amount of georeferenced video mosaics and microbathymetric data, derived from a camera system and a multibeam echo sounding system mounted onto the ROV Victor6000 (Ifremer), are basis for a morphological as well as biogeochemical regionalisation of the HMMV. This regionalisation is accomplished due to visual inspection of video mosaics, a defined classification scheme, concerning the distribution pattern of the two occurring chemoautotrophic communities and uncovered mud areas, and due to a GIS-supported overlay of 13 resulting surface maps which were calculated geostatistically using indicator kriging. Furthermore, microbathymetry and slope inclination were included, defining and calculating areas of five biogeochemical habitats. These habitats indicate graduated methane consumption of microbial consortia, consisting of sulphate-reducing bacteria and anaerobic methane-oxidising archaea. These consortia represent an efficient methane biofilter, e.g. at gas hydrate bearing continental margins, and are thus an important sink for the global methane cycle.Approximately 16% of the flat centre of the HMMV is nearly void of any benthic communities. Therefore, this area is considered as a region of high methane discharge into bottom water. Source location and drainage direction for current mud flows were identified by trend surface computation of the comparably flat crater area (1.58° average slope angle), and consideration of temperature data, as well as the distribution pattern of the chemoautotrophic communities. This suggests that a present mud flow ascends close to the northern edge of the flat unit of HMMV, and that the drainage pattern of mud flows shifted from a westward to a south-south-eastern direction.The quality assessment of the surface maps is conducted by cross-validation evaluating the fit of the indicator kriging variograms by using statistical mean values of the deviations between estimated and measured values. Furthermore, the estimate was evaluated by a validation dataset of visual inspected analysis of video mosaics not included in the interpolation process, proving the interpolated surfaces independently.The large amount of video mosaics requires the development of an image analysis technique for the automated detection and quantification of the spatial distribution of Beggiatoa mats. In a first step it is differentiated between data, non-data and redundant (overlapping) areas on the mosaics. In the second step the data areas are pre-segmented into disjunctive homogeneous regions by a watershed transformation. A probabilistic approach, relaxation labelling, then realises the assignment of these regions as bacterial or non-bacterial on the basis of spatial correlation and defined contrast thresholds (comparing the grey values of neighbour regions). Comparison of the data derived by visual inspection with the automated image analysis revealed similarities better than 90%.Kriging methods were also applied and evaluated for selected parameters for the North Sea (bottom water measurements on salinity, temperature, silicate, ammonium, nitrate nitrite, phosphate as well as from punctual data on grain size ranges (0-20 µ, 20-63 µ, 63-2000 µ) creating surface maps from measured data as an assumption for multivariate statistics like Classification and Regression Trees (CART). The evidence of spatial autocorrelation by variogram analysis allowed calculation of raster maps by applying ordinary kriging.After intersecting, these raster maps with punctual data on benthic epifaunal communities a classification system is derived to predict the occurrence of these communities within the study area. The classification system is calculated from the intersected data by producing decision trees using CART. Since these decision trees correspond to hierarchically ordered sets of decision rules, they are applied on the geostatistically estimated raster data to predict benthic habitats.