Water quality modeling based on landscape analysis: importance of riparian hydrology

Sammanfattning: Several studies in high-latitude catchments have demonstrated the importance of near-stream riparian zones as hydrogeochemical hotspots with a substantial influence on stream chemistry. An adequate representation of the spatial variability of riparian-zone processes and characteristics is the key for modeling spatio-temporal variations of stream-water quality. This thesis contributes to current knowledge by refining landscape-analysis techniques to describe riparian zones and by introducing a conceptual framework to quantify solute exports from riparian zones. The utility of the suggested concepts is evaluated based on an extensive set of hydrometric and chemical data comprising measurements of streamflow, groundwater levels, soil-water chemistry and stream chemistry.Standard routines to analyze digital elevation models that are offered by current geographical information systems have been of very limited use for deriving hydrologically meaningful terrain indices for riparian zones. A model-based approach for hydrological landscape analysis is outlined, which, by explicitly simulating groundwater levels, allows better predictions of saturated areas compared to standard routines. Moreover, a novel algorithm is presented for distinguishing between left and right stream sides, which is a fundamental prerequisite for characterizing riparian zones through landscape analysis. The new algorithm was used to derive terrain indices from a high-resolution LiDAR digital elevation model. By combining these terrain indices with detailed hydrogeochemical measurements from a riparian observatory, it was possible to upscale the measured attributes and to subsequently characterize the variation of total organic-carbon exports from riparian zones in a boreal catchment in Northern Sweden. Riparian zones were recognized as highly heterogeneous landscape elements. Organic-rich riparian zones were found to be hotspots influencing temporal trends in stream-water organic carbon while spatial variations of organic carbon in streams were attributed to the arrangement of organic-poor and organic-rich riparian zones along the streams. These insights were integrated into a parsimonious modeling approach. An analytical solution of the model equations is presented, which provides a physical basis for commonly used power-law streamflow-load relations.