Crossing natural and data set boundaries: coastal terrain modelling in the South-West Finnish Archipelago

Abstract

Geographical information systems (GIS) are important tools in coastal research and management. Coastal GIS applications involve special challenges, because the coastal environment is a complex transitional system between the terrestrial and marine realms. Also acquisition methods and responsibilities for spatial data (and thus their properties) change at the shoreline. This article explores the consequences of this land-sea divide for coastal terrain modelling. We study how methods designed for terrestrial environments can be used to create integrated raster coastal terrain models (CTMs) from coarse elevation and depth data. We focus on shore slopes, because many particularities of coastal terrain and the data which describe it as well as the resulting problems are concentrated in the shore zone. Based on shorelines, terrestrial contours, depth contours and depth points, we used the ANUDEM algorithm to interpolate CTMs at different spatial resolutions, with and without drainage enforcement, for two test areas in a highly complex archipelago coast. Slope aspect and gradient rasters were derived from the CTMs using Horn's algorithm. Values were assigned from the slope rasters to thousands of points along the test areas' shorelines in different ways. Shore slope gradients and aspects were also calculated directly from the shorelines and contours. These modelled data were compared to each other and to field-measured shore profiles using a combination of qualitative and quantitative methods. As far as the coarse source data permitted, the interpolation and slope calculations delivered good results at fine spatial resolutions. Vector-based slope calculations were very sensitive to quality problems of the source data. Fine-resolution raster data were consequently found most suitable for describing shore slopes from coarse coastal terrain data. Terrestrial and marine parts of the CTMs were subject to different errors, and modelling methods and parameters had different consequences there. Thus, methods designed for terrestrial applications can be successfully used for coastal terrain modelling, but the choice of methods and parameters and the interpretation of modelling results require special attention to the differences of terrestrial and marine topography and data.