Attributiveness of a mass flow analysis model for integrated water resources assessment under data-scarce conditions.

Nutrients in river systems originate from multiple emission sources, follow various pathways, and are subject to processes of conversion and fate. One approach to tackle this complexity is to apply balance-oriented models. Although these models operate on a coarse temporal and spatial scale, they are capable of assessing the significance of the different emission sources and their results can be the basis for developing integrated water quality management schemes. In this paper we propose and apply a methodology to evaluate the attributiveness of such model results with regard to the modelled emission pathways. The MONERIS (MOdelling Nutrient Emissions in RIver Systems) model is set up, assuming plausible ranges of emission levels from four principal sources. The sensitivity of model performance is computed and related to the contribution from the pathways. The approach is applied for a case study in the upper Western Bug catchment (Ukraine). Coefficient of determination (R(2)) is found insensitive against the model assumptions, at levels around 0.65 for nitrogen and 0.55 for phosphorous emissions. Relative mean absolute error is minimized around 0.2 for both nutrients, but with equifinal combinations of the varied emission pathways. Model performance is constrained by the ranges of the emission assumptions to a limited extent only.