Partner und Internationale Organisationen
(Englisch)
|
A, B, DK, FIN, F, D, GR, H, IRL, I, NL, N, PL, RO, E, S, CH, GB
|
Abstract
(Englisch)
|
Research and practices of the past years have shown, that in a country like Switzerland, where grassland is one of the dominating agricultural land use, agricultural Phosphorus (P) losses from non-point sources from fields to waters cause severe eutrophication problems in surface waters. In traditional intensively used grassland regions such as in the Lake Sempach region (Switzerland) with high stock densities topsoils are often enriched with P as a result of frequent manure applications and represent a high potential for P losses. Fresh additions of P in manure may generate additional P losses to that arising from the soil. P gets mobilized from these surface layers and is transported to surface waters mainly by surface runoff and preferential transport into subsurface drainage systems. In 1993 and 1999, ecological measures were introduced to reduce agricultural P losses among other things in Switzerland. To reasonably apply these measures one has to know the risk areas where P losses occur. Risk is high in areas where freshly applied manure and high soil P levels coincide with runoff. The identification of these areas is possible with model simulations but input data and parameters for common transport approaches are often missing at the catchment scale. We used a parsimonious rainfall-runoff model (RRM) for single runoff events and derived model parameters from hydrological properties of different catchments. The RRM was joint calibrated by runoff in four different catchments to give the areas contributing to discharge. These areas will be combined with the P contents in the topsoil, manure data from farmers and observed P concentrations from experimental runoff plots on grassland sites. Finally, the combination of discharge contributing areas with P enriched areas give the risk areas. First, we investigated the P export dynamic from two agricultural grassland catchments to understand the main mechanisms behind and to derive the perceptual model. The P export dynamic is often event-driven and variable in time. Accordingly, a proper monitoring requires high-temporal resolution sampling during flow events for concentrations of dissolved reactive Phosphorus (DRP) for periods of 1.2 and 5 y, respectively. The data set is complemented by measurements on total P (TP), total dissolved P (TDP) and particulate P (PP) obtained from routine sampling ongoing since 1984 in both catchments. In one area, we observed a pattern of the P-export dynamic as expected from other studies: DRP concentrations usually increased significantly over 1000 mg DRP m-3 during peak flow, while they remained low between 20 and 100 mg DRP m-3 at base flow. This pattern was partially confirmed in the other catchment in 1998 and 1999 but was partially replaced by a new one for the following years: In the growing season, DRP only weakly increased with flow rate or remained almost constant. During flow events values larger than 300 to 400 mg DRP m-3 were rare. In contrast to these relatively low peak flow concentrations, high values (200 to 300 mg m-3) were recorded at base flow during the years 2000 to 2002. During the same period, DRP concentrations exhibited strong daily oscillations. Point sources like leaks from manure tanks could be ruled out based on ammonia measurements. We suggest that biogeochemical processes within the brook are responsible for this change in the P export dynamics. Despite these changes in the pattern of the P export dynamic and ecological measures implemented on the farms no significant effect on the annual P loads could be observed during the last 10 years in the Lippenrütibach catchment. It may take several years before these measures show a positive effect on the P-loads in this catchment. Second, to identify those areas with discharge occurring relevant for P losses we developed a semi-distributed statistical RRM. One paper focus on the calibration and validation of this RRM in four catchments. We reproduced the catchment responses by dividing solely between well and poorly drained hydrological response units (HRU). Both HRU were coupled with a slow and a fast discharge component. We calibrated the model simultaneously in four catchments over an only 11-day period using Uniform Monte Carlo simulations to allow for parameter uncertainty. Basically, the joint calibration yielded robust global parameter sets for each HRU in each catchment to quantitatively describe the average soil type hydrologic response of each HRU. The simulation of different catchment-specific time variables were expressed as prediction bands that mostly bracket the discharge observations in validation periods that were more than 15 times longer than the calibration period. The model performed better for wetter conditions or longer stormflow periods but we also found limitations during baseflow or for isolated smaller stormflows. Third, we quantified the average hydrological contribution of well and poorly drained HRU in these four catchments in space and time. Fast flow contributions were consistently quantified in various catchments and were generally higher for poorly than for well drained HRU. However, well drained soils markedly contributed to peak flows in catchments as soon as these soils were wet enough. Using Uniform Monte Carlo simulations lead to noticeable variation in the fast flow components for each HRU and in the spatial extent of areas contributing to discharge. Spatially, we assigned probabilites to those areas with respect to the degree of contributing to discharge. Irrespective of soil wetness only a small fraction of those areas was identified as clearly contributing and non-contributing areas. The spatially-distributed observations of the P content in topsoils are linked with the RRM to calculate the contribution of P release due to surface runoff. The P-submodel is designed and first results are available. They indicate that the P-load from topsoil P-release is smaller compared with the observed load. This points towards the additional contribution of P losses from recently applied manure during inappropriate soil and weather conditions or P losses from sealed area. The model and data base to involve the P contribution from manure application is in its initial stage.
|
