Challenges in model based estimation of mitigation measures to improve water quality on catchment scale

Restoration of wetlands as well as other measures aimed on increasing water and nutrients buffering capacity along the streams are becoming increasingly proposed as solutions addressing problems of water pollution. We observed, however limited integration of this type of measures in  river basin management plans. We assume, that it is caused mainly by lack of knowledge (and tools) about   effectiveness assessment in case of numerous measures applied in a particular catchment. In other words, upscaling of restoration measures should be possible, when we can provide water managers with more accurate estimates of cumulative effects of different measures. It can be achieved by  hydrological model application in appropriate scale.

We have tested different approaches in modelling of several river basins where restoration activity or buffering  measures where applied or planned. This includes Kamienna River in upland landscape, Słupia River in costal settings and Pilica River in lowlands. There were also modelling experiments conducted on rivers in Finland, Sweden and Lithuania.   Based on the modelling  experience from these catchments using Soil and Water Assessment Tool (SWAT), where different types of river buffer zones and wetlands  implementation has been tested, a number of challenges can be emphasized. At a general level, three major sources of challenges dominate: (1) spatial extent and location of measures, (2) their accurate parametrization and (3) simplification of processes in modelling scheme. Most commonly, in semi-distributed models, principal calculation units for which water and nutrient balance is calculated, are lumped and non-contiguous geographic units within each sub-catchment. Particular model setup may consist of thousands of such units, and each of them may represent one field, a portion of a field, or, more likely, portions of many fields. It becomes problematic when we aim to simulate measures for particular river reach or wetland, but can only define it at coarser sub-catchment level. The second issue is related to proper parametrization of empirical/physical sub-models, simulating processes of nutrients adsorption and settling. In most cases, uncertainty related to parametrization of buffer zones and wetlands  is significant. No simple calibration procedure for setting the optimal parameters’ values can be applied, and the process itself is more expert-dependent. Another issue is related to simplifications of processes. For instance, in SWAT, nutrient transformations simulated in wetlands are limited to the removal of nutrients by settlings however transformations between nutrient pools are ignored. For the buffer zones, model only affects contaminants that are present in surface runoff and neglects the potential effects of buffer zones on shallow groundwater.

There is a big room for improvement by providing the best monitoring results of particular measures and applying them in a modeling scheme. This can be done based on wetland restoration projects which were monitored by our group in Kampinos National Park, Biebrza NP,  Słowiński NP as well as in Lithuania and Norway. Showing the results of our models we would like to rise discussion on inter-calibration of the particular measures in the modelling settings.