Journal of Conference Abstracts

Dynamic Watershed Modeling of Nitrogen Transport
in Sweden: Spatial Analysis, Source Apportionment
and Scenarios

Berit Arheimer (barheime@smhi.se), Maja Brandt (mbrandt@smhi.se)
& Hans B. Wittgren (hbwittgr@smhi.se)

Swedish Meteorological and Hydrological Institute, S-601 76 Norrköping, Sweden.

The eutrophication of the Baltic Sea in Northern Europe is considered a serious problem by surrounding countries. The prevalent opinion is that it is caused by high N loading, mainly from rivers. To find efficient ways to reduce this load, knowledge of the N sources, as well as retention in rivers and lakes, is crucial. The riverine N load is highly correlated with water discharge and weather, and may show large variation among different years and seasons. To analyse the N transport from land sources to the sea, we have developed and modified a dynamic conceptual model (HBV-N) for simulations of regional N transport. The poster and paper briefly presents the model and some results for southern Sweden for the period 1985-1994. The study include spatial distribution of leakage and retention, source apportionment and scenarios of the effect from various control measures.

The HBV-N model simulates nitrogen transport from the root zone through groundwater, rivers and lake systems towards the sea. The calculations are made for sub-basins that are coupled into larger river basins. Since Sweden has a rather thin till soil layer, the groundwater discharge is assumed to follow the surface topography. The water balance is estimated with the conceptual hydrological model HBV, which uses daily precipitation and temperature, and monthly evapotranspiration data from climate stations. It provides daily values for areal precipitation, snow accumulation and melting, soil moisture, groundwater level, and runoff from every sub-basin and routing through larger basins.

The N model is linked to the HBV-model and has separate routines for daily simulations of inorganic and organic N. Leakage concentrations for various land use categories are assigned to water discharge from the root zone to the groundwater. The soil leakage is mixed with discharge from rural households into the groundwater. Concentration variations in local runoff from open ditches and riparian zones, due to biological and chemical processes, are described conceptually with simple functions based mainly on temperature, concentration and hydrology. The local N runoff is then mixed with contributions from upper sub-basins and lake water, if present. In the lake routine, atmospheric N deposition on the water surface, as well as the load from industry and treatment plants, are included. Transformation of N concentrations may occur in rivers and, more importantly, in the lakes. The inorganic N may be reduced due to denitrification, sedimentation and biological uptake, while organic N may increase due to biological production, or decrease by sedimentation and mineralisation. These processes are also simulated with simple conceptual functions.

The studied area of southern Sweden (about 130,000 km²) is divided into 3,725 sub-basins and includes more than 60 drainage basins. For each sub-basin, geographical information has been gathered regarding; water divides, lake areas and depths, mean altitude, coupling to other subbasins, information of land use and management, soil type, forest fertility, point-sources (treatment plants, rural house-holds and industries), and if available, measured concentrations from monitoring programmes. The database is compiled at the Swedish EPA and the Swedish Meteorological and Hydrological Institute (SMHI). Leakage concentrations from arable land and pasture have been estimated with the SOIL-N model for more than 500 field types in the region by the Swedish University of Agriculture. Leakage from forests and other lands were taken from monitoring data. Atmospheric deposition on lake surfaces was simulated by the MATCH model at SMHI.

When modeling Southern Sweden, the N routine was calibrated in 722 sites and about 38,000 measurements were used. Sensitivity analyses were performed and the modeling efficiency was estimated for 32 sites with measurements of both water discharge and N concentrations. The explained variance (R²) of the N transport in these sites was on average 0.6. The quality of the land cover data was crucial for the results.

The simulation of biochemical processes affecting the N concentrations resulted in reduced total N load from the sources to the sea. The average gross N load from southern Sweden was estimated to about 120,000 tons yr^-1, but the net transport to the sea was only on 65 000 tons. The average net reduction in local discharge varied spatially between 5 and 35% of the gross load. However, most of the reduction occurred in lakes. Unfortunately, the areas of low reduction near the coast sometimes coincide with areas of very intensive farming, e.g. in the southern plain of Scania and Halland Counties, The fields there count for 1/3 of the total load.

The source apportionment showed that half of the N load on the sea originate from arable land. The western coast of the study area receives about twice as much N compared to the eastern coast, although the drainage area is about 25% smaller. This is partly due to differences in water discharge due to the meteorological gradient in the region, and the atmospheric deposition on the large lake Vänern. However, it also reflects the reduction capacities of the two drainage systems, as there are fewer lakes draining to the western coast.

Scenarios show that the measures taken in agriculture lately only reduce the total N transport to the sea by 10%. A case study on the effect of created wetlands in one river basin, shows that an optimal location of wetlands covering 1% of the watershed area reduces the riverine load by 20%. The optimal location of created wetlands is in coastal agricultural areas, downstream lakes, and where the summer load is significant.