Eco-Hydrological Modelling of Stream Valleys

Predicting the effects of hydrological alterations on terrestrial stream valley ecosystems requires multidisciplinary approaches involving both engineers and ecologists. Groundwater discharge in stream valleys and other lowland areas support a number of species rich ecosystems, and their protection is prioritised worldwide. Protection requires improved knowledge on the functioning of these ecosystems and especially the linkages between vegetation, groundwater discharge and water level conditions are crucial for management applications. Groundwater abstraction affects catchment hydrology and thereby also groundwater discharge. Numerical hydrological modelling has been widely used for evaluation of sustainable groundwater resources and effects of abstraction, however, the importance of local scale heterogeneity becomes increasingly important in the assessment of local damage to these groundwater dependent ecosystems. This calls for new ways of combining measurement techniques and hydrological models working at different scales. The PhD thesis investigates the hydrological functioning of fen habitats in Denmark by combining detailed field investigations, hydrological modelling and statistical vegetative analyses from a subset of Danish fen habitats. Field investigations elucidated the hydro-geological settings creating the stable calcareous groundwater seepage required by specialised plant communities. The combination of sloping terrain, a high yield groundwater aquifer and relatively low-permeable deposits in the stream valley creates an upward flow direction and the vertical hydraulic gradient is pointed out as a key parameter controlling the groundwater seepage rate. Seepage is difficult to measure directly and an inverse modelling approach is suggested for accurate estimates. The model only requires that the vertical hydraulic gradient is measured over time and moreover it uses precipitation and potential evapotranspiration as inputs. The water level dynamics in groundwater dependent wetlands is concluded to contain valuable information on both soil parameters and seepage rate. Distributed three-dimensional modelling is required for hydrological impact assessment in relation to groundwater abstraction in wetlands. The heterogenic discharge patterns that often dominate in stream valleys lead to differences in the way areas are affected. Specifically full scale pumping tests have shown a flow reduction in the order of 20 % in a natural spring, whereas no effect could be measured in neither short nor deep piezometers in the river valley 50 m from the spring. Problems of measuring effects of pumping are partly caused by disturbances from natural water level fluctuations. In this aspect numerical models can clearly separate natural variations from a water table lowering induced by pumping. Water level is the most commonly measured hydrological variable in wetlands. Linkages between the water level regime and fen vegetation is studied for 35 Danish sites. All sites have piezometers installed with continuous registration of water levels, and plant registrations have been conducted in circles around the piezometers. The results show that water table conditions are directly limiting the occurrence of typical fen species, and bryophytes and moreover the analyses provide a rare quantification of the number of species supported by certain water level conditions. Statistical/empirical models are suggested for prediction of typical fen species and bryophytes based on water level and Ellenberg Indicator values which are derived from the vegetation composition. While the models are useful for identifying environmental conditions limiting the habitat quality, the underlying assumptions might not be valid for prediction of the vegetative response to water level changes, because internal eutrophication is not accounted for. The results of the study do not support threshold values for groundwater flows or water level conditions to determine significant damage from hydrological impacts. Such threshold values would be a strong simplification of ecosystems which depend on a number of interacting environmental conditions. Improved predictions of chemical changes and nutrient releases after lowering of water tables might be a way to approach meaningful thresholds.