In boreal catchments, stream water chemistry is influenced and controlled by several landscape factors. The influence of spatially distributed variables is in turn dependent on the hydrological scale. Headwater streams have larger variability of water chemistry, and thus together represent a large biodiversity, and therefore need to be monitored in official environmental assessments. One objective of this study was, using Geographical Information Systems (GIS), to analyse co-variation between landscape variables and water chemistry and to determine which of the landscape variables have a major influence on the concentration of dissolved organic carbon (DOC) in headwater streams. Another objective was to find a simple method for predicting sources of DOC, using official map data and publically available GIS applications.
Totally 85 headwater catchments (0.1-4 km2) in the county of Värmland, western south Sweden, were used in the study. Water chemistry was analysed for water sampled at low, medium and high flows, and landscape variables were extracted from official map data sources: topographic maps, a digital elevation model (DEM, 50 m grid), and vegetation data. Statistical analyses showed that topography (mean slope and mean topographic wetness index (TWI)) and wetland cover often correlated well with DOC in headwater catchments. Official map data could satisfactorily extract landscape variables (mean slope, mean TWI) that were useful in predicting stream water chemistry (DOC).
A high-resolution elevation model, which was generated by interpolation of photogrammetric data, was used to calculate and evaluate two different wetness indices and their ability to predict the occurrence of wetlands in six catchments of different sizes and topography. The SAGA (System for Automated Geoscientific Analyses) wetness index (SWI) gave substantially better results than the TWI. The effects of resolution of DEMs on calculations of the SWI were investigated using 5, 10, 25 and 50 m grids. The results showed that SWI values increased with increasing cell size. The near linear increment of mean values for resolutions 10-50 m suggests a independence of terrain type and catchment size, which supported previous findings that indicated that mean slope and mean wetness index calculated from coarse elevation models may be used for prediction of DOC in headwater streams.