Predicting lake alkalinity and depth for classifying Irish lakes under the EU Water Framework Directive typology

Abstract

The EU Water Framework Directive (WFD) typology for Ireland groups lakes based on their altitude, alkalinity, water depth and area. Alkalinity and depth data are, however, lacking for many unmonitored lakes on the Irish WFD programme. In light of legislative requirements to categorise these lakes, we examined relationships between the alkalinity and depth of lakes and the hydrological, morphological and geological characteristics of their catchments. We used a decision-tree learning approach to categorise lakes into their predicted Irish WFD typology categories. We then tested our model predictions using data from an independent set of 62 lakes. Alkalinity type was predicted correctly 87% of the time, with greatest success of classification in the low alkalinity category. The most important variable for predicting water depth was the range of slope within 50m of the lake shore. However, lake depth category was predicted correctly only 54% of the time, and predictions were particularly poor for shallow lakes. This latter model is therefore not proposed for use in typology assignment for WFD purposes. Our predictions allowed WFD alkalinity category to be assigned for 790 lakes on the Irish WFD programme and provide insights into relationships between catchment hydrological, morphological and geological characteristics and in-lake conditions in Ireland.