University of Rostock models phosphorus removal in Belgian Kemmelbeek watershed
UROS successfully calibrated the monthly discharge and acquired a satisfactory Nash-Sutcliffe-Efficiency coefficient (used to assess the predictive skill of hydrological models) of 0.58 (Figure 1). The loads of dissolved reactive phosphorus (DRP) in the stream achieved a Nash-Sutcliffe-Efficiency of 0.66.
Figure 1: Calibration time curve of discharge (left panel) and DRP loads (right panel). The blue lines represent the observed values; the red lines represent the modeled values.
The Nash-Sutcliffe-Efficiency alone is not sufficient to evaluate the quality of a hydrological model. The distribution of flow constituents should likewise be considered in the evaluation process. During the discharge season, 40% of the water is transported to adjacent surface waters via tile drains, typical for intensively drained lowland catchments. Only 5% of the total flow originates from surface runoff.
Filter boxes with iron-coated sand
Field data for the P reduction scenarios were obtained from the NuReDrain project partners at Ghent University. They tested in-situ filter boxes at drainage outlets to reduce P loads in surface waters using iron-coated sand (ICS), a reused waste product from drinking water production. The reported filter efficiencies were implemented into the model. The filter boxes were applied to 5, 10, 15, 25, 50, 75, and 100% of tile drainage outlets of the drained agricultural areas within the watershed.
Figure 2: Base model (P) and P reduction potential of each reduction scenario (R5 to R100).
The filter installation's P reduction potential at outlets of drainage plots ranged from 254 kg to 6592 kg P for the 5% and the 100% reduction scenario, respectively (Figure 2). However, it should be emphasized that the modeling results are only a rough estimation of the P filter installation's possible effects.
Further work will include the calibration of nitrate loads and the implementation of N reduction scenarios.