PEI Geodatabase: a relational database for time series of factors controlling beach/dune interaction

The development of Arc Hydro data models opened the way to integrate time series with GIS data, and to expand GIS extensive capabilities to the temporal domain. A Geodatabase (GDB) built with ArcCatalog was developed to store and manage time series collected from short-term experiments and long-term monitoring of a section of the beach/dune system at Greenwich Dunes, Prince Edward Island (PEI) National Park, Canada. Long term monitoring was carried out using a remote sensing station including three digital cameras and instrumentation for measuring winds and aeolian sand transport. This permitted continuous measurements of winds, beach characteristics (including surface moisture, width and the presence of snow and ice) and vegetation cover on the backbeach and foredune over periods of months. Short-term experiments measured wind flow and sand transport using an array of anemometers, safires and sand traps. The information was processed with ArcGIS 9.2, PCI Geomatica 9.1, and Excel. The type of data sets and the expected use of the GDB directed its conceptual, logical and physical design, starting with the definition of seven thematic layers: wind, transport, beach boundaries (shoreline, vegetation), moisture, surface characteristics (ice/snow cover, etc.), and topography. Features were assigned the spatial representation that best described them (raster for vegetation and moisture, polyline for shoreline, etc.). Temporal data was stored in four types of time series: tables (e.g., percentage of snow cover), attribute time series (wind speed associated to different instruments), feature series (shoreline evolution), and raster series (moisture maps). Temporal geoprocessing tools provide transition between the spatial and temporal domains, allowing working with large amounts and patterns of data. Depending on the type of analysis, the user is able to select particular instruments, show evolution of different variables, and query the database in a number of ways. This provides an important tool for subsequent modelling.