Flow directions of shallow groundwater in a boreal catchment

For most catchments, there is insufficient data to determine the location of the groundwater surface. For humid climates, it is, therefore, often assumed that the groundwater-surface follows the surface topography. This assumption allows using digital elevation models (DEMs) to estimate the flow directions and catchment boundaries. However, high-resolution elevation data also include many small-scale features that are unlikely to affect the direction of groundwater flow, or only affect it during specific conditions. Furthermore, flow directions may change during events or depending on the water level.

The optimal resolution of the DEM for determining groundwater flow directions is not known yet. Therefore, we studied how much DEM derived flow directions and catchment boundaries are affected by the resolution or smoothing of the elevation data for the Krycklan catchment in northern Sweden. We also measured the groundwater levels in two small sub-catchments to determine what DEM resolution best describes the actual groundwater-surface and flow directions.

For the topographic analyses, the LiDAR-based elevation data were first smoothed with various filters (e.g., Gaussian filters) and resampled to obtain lower resolution elevation data. We then determined the flow directions for these different DEMs. The aim was to determine where in the catchment the calculated flow directions are most sensitive to the resolution of the topographic data. The results of the topographic analyses show that for some areas, particularly flat areas, ridges, streambanks and locations where the local slope differs from the general slope, the calculated flow directions depend strongly on the resolution and smoothing of the elevation data.

To test how well the DEM based groundwater flow directions represent actual flow directions, we installed a dense (5-20 m spacing) network of shallow (1 to 6 m deep) groundwater wells (75 wells in total) in a 1 ha and a 2 ha gauged sub-catchment. The triangular nested design of the groundwater well network allowed us to determine the smaller (5 m) and larger scale (20 m) groundwater gradients. The recorded water levels were augmented and validated by manual measurements during the summers of 2018 and 2019. The high spatial and temporal resolution data allowed us to study the response of the groundwater level and the flow directions to different meteorological situations (e.g., large precipitation events after dry and wet conditions and during a very dry period in summer 2018). These observations indicate that the degree to which the groundwater-surface is a subdued copy of the surface topography varies throughout the year, and provides information on which DEM resolution most accurately represents the groundwater-surface and flow directions.