Predicting Freshwater Critical Loads from Catchment Characteristics using National Datasets

Maps of freshwater critical loads are used toguide emission strategies for sulphur and nitrogen bothnationally and internationally. Water chemistry data arerequired to calculate critical loads and the production ofnational maps therefore relies on the existence of extensivechemistry datasets. However, the data required to calculatecritical loads are not readily available for all sites. Thisarticle explores how empirical statistical models mightpotentially be used to predict critical loads using nationallyavailable datasets representing a range of catchmentcharacteristics. Initially a global regression model forexplaining freshwater critical load variation across a broadspectrum of catchment types (from lowland agricultural tomountain lakes) throughout mainland Britain is described. Whenattention is focused on more specific catchment types (i.e.upland and non-arable) it is shown that the global model hasless explanatory power. A regionalisation of Great Britain(based on 100 km grid squares) shows that the global modelcannot necessarily be applied successfully within a narrowerregional context. Separate analyses were undertaken on each ofthe regional subsets using backward selection regression. Thevariables emerging as significant predictors variedsubstantially across the regions, as did the explanatory powerof the models. This was also the case when the analysis wasconfined to upland and non-arable catchments. This approachcould be developed so that critical loads assessments can bemade for populations of standing waters rather than simplythose for which water chemistry is available.