Lewandowska, M., Hoffmann, C. C. & Kjeldgaard, A. 2020. Potential for further wetland restoration in the Odense River catchment and nitrogen and phosphorus retention. Aarhus University, DCE – Danish Centre for Environment and Energy, 60 pp. Scientific Report No. 396. http://dce2.au.dk/pub/SR396.pdf
The Odense River catchment has been subjected to intensive wetland restoration to mitigate the loss of nitrogen (N) and this restoration has led to a reduction of the N load to Odense River of 124 tons N per year. Restoration of additional areas would expectedly entail retention of more N in the catchment. In this study, potential wetland areas for restoration were found by creating an index model identifying suitable areas. The direct upland area to each wetland was calculated in ArcGIS and the N loss from each upland and the N removal from the direct upland were estimated based on soil type and drainage probability. N removal by flood inundation was founded on a flood estimate accounting for flooding 16% of the year (60 days), this estimate being calculated from measurements of the stream water level increase at monitoring stations and from predictions by the developed UPstream river length Model (UPM).
The UPM model was developed in the project with the aim of selecting areas topographically suited for flooding. The model only requires few input data and parameters as compared to models like e.g. MIKE 11, which needs cross sectional stream data for every 100 meters.
The UPM flood model fitted well both upstream and downstream in the river network. It performed best at the highest percentile (R2=0.85 at p0.9995), and the amount of explained variance decreased when longer time periods were included (R2=0.7 at p0.9, R2=0.65 at p0.84). The UPM-based flood calculations were comparable with the flood-based stream level elevations at the 0.9995 percentile calculated using a Mike11 stream model, but large differences in flood coverage occurred at the lower percentiles. Flood estimates from MIKE11 were based on a larger amount of input data suggesting that the MIKE11 model was the more accurate. However, also the MIKE11-based flood was overestimated in several areas, limiting its accuracy.
The calculated N removal in restored wetlands amounted to 91.7 tons N using the flood estimate for 60 days and 127 tons N using the flood estimate for 127 days. There is, though, high uncertainty regarding the latter. Additionally, several restored wetlands contained shallow lakes and N retention in these cannot be calculated without including the lake residence time. As result, there were discrepancies between the measured and the calculated values.
Restoration of an additional 3,617 ha of wetlands is estimated to remove 425 tons N, not including potential removal in lakes. Currently, many streams are not flooded, but they may be so after stream re-meandering. The amount of N removed in currently flooded areas is uncertain as upland drainage area expectedly affects denitrification. At high loads, N removal through irrigation by drainage water is below the standard rate of 50% but rises above 50% at low loads. It is therefore difficult to determine the accuracy of the calculated N removal. The current phosphorus (P) sedimentation amounts to 2,750 kg P per year in restored wetlands, which may be an underestimation, however. P sedimentation in potential wetlands is estimated to 3,551 kg P per year, which may be an overestimation as the calculation does not take into account the increase in upstream wetland areas, which would reduce the P loss in the catchment.
An uncertainty arises also as to how many of the potential wetland areas can actually be restored. Wetland restoration improves the natural value of an area, but an additional N input to protected nature types within the potential wetlands might deteriorate their ecological status.
