This paper describes the implementation of process-based models reflecting relative groundwater nitrate vulnerability of the shallow alluvial Lower Savinja Valley (LSV) aquifer in Slovenia. A ...spatially explicit identification of the potentially vulnerable priority areas within groundwater bodies at risk from a chemical point of view is being required for cost-effective measures and monitoring planning. The shallow LSV unconfined aquifer system consists of high-permeable Holocene and middle- to low-permeable Pleistocene gravel and sand, with a maximum thickness of about 30 m, mainly covered by shallow eutric fluvisoils or variously deep eutric cambisoil. The hydrogeological parameters, e.g. the depth to the groundwater, hydrological role of the topographic slope, etc. usually used in different point count schemes are, in the case of the lowland aquifer and shallow groundwater, spatially very uniform with low variability. Furthermore, the parametric point count methods are generally not able to illustrate and analyze important physical processes, and validation of the results is difficult and expensive. Instead of a parametric point count scheme, we experimentally used the Arc-WofE extension for weights-of-evidence (WofE) modelling. All measurement locations with a concentration higher than the value of 20 mg NO
3
−
per litre of groundwater have been considered as training points (173), and the three process-based models generalized output layers of groundwater recharge (GROWA), nitrate leached from the soil profile (SWAT) and groundwater flow velocity (FEFLOW), served as evidential themes. The technique is based on the Bayesian idea of phenomena occurrences probability before (prior probability) and after consideration of any evidential themes (posterior probability), which were measured by positive and negative weights as an indication of the association between a phenomena and a prediction pattern. The response theme values describe the relative probability that a 100 × 100 m spatial unit will have a groundwater nitrate concentration higher than the training points’ limit values with regard to prior probability value. The lowest probability of groundwater nitrate occurrence is in the parts of the LSV aquifer, which are known as anoxic condition areas with very likely denitrification processes. The cross-validation of the dissolved oxygen and dissolved nitrate response theme confirmed the accuracy of the groundwater nitrate prediction. The WofE model results very clearly indicate regional groundwater nitrate distribution and enable spatial prediction of the probability for increased groundwater nitrate concentration in order to plan the groundwater nitrate reduction measures and optimize the programme for monitoring the effects of these measures.
