Modelling nutrient pollution in the Danube River Basin: a comparative study of SWAT, MONERIS and GREEN models

摘要

The Water Framework Directive requires the development and implementation of river basin management plans for improving the ecological status of freshwater bodies throughout Europe. The scientific community supports this process by developing decision-support tools for identifying the principal sources of water pollution. Models, however, are imperfect representations of the real world, and are conditioned by structural uncertainty, implicit in the description of biophysical processes, and data uncertainty, as well as in the various restrictions of the environmental data the models were developed. Hence, decision makers must plan management actions on the basis of the best available, however still incomplete, knowledge. The comparison of independent assessments may offer insights that are useful for decision-making, e.g. for identifying knowledge gaps, identifying data uncertainties, consolidating investigation results, and increasing stakeholders’ acceptance. The Danube River is the second largest and most international river of Europe. Its basin covers approximately 803,000 km2 of Central and South-Eastern Europe and is shared by 19 countries. Within the context of fostering scientific collaboration in the Danube region and under the auspices of the International Commission for the Danube River Protection (ICPDR), three independent model (SWAT, MONERIS and GREEN) were compared with the objective of reaching a shared appraisal of nutrient pressures and drivers in the Danube Basin. Annual water discharge (Flow, m3/s) as well as annual loads of total nitrogen (TN, ton/y) and total phosphorus (TP, ton/y) were compared at the outlet of 18 ICPDR regions for the decade 2000-2009. For each region, mean annual values, correlation, standard deviation, and root mean square error of model simulations were analysed. Good water discharge simulations across the basin confirmed that hydrology was correctly represented in all models. The nutrients comparison revealed for some assessment regions the need for a spatially and temporarily intensified monitoring especially for TN. Concerning TP, SWAT and MONERIS had comparable long mean annual TP loads, but differed for amplitude and phases; while GREEN generally overestimated TP loads. Despite differences in model approaches and considered input data, the three assessments were coherent, and all three models may be confidently used for river basin management of the region.