RESUMEN
The protection of coldwater refugia within aquatic systems requires the identification of thermal habitats in rivers. These refugia provide critical thermal habitats for brook trout (Salvelinus fontinalis) and Atlantic salmon (Salmo salar) during periods of thermal stress, for example during summer high temperature events. This study aims to model these refugia using georeferenced thermal infrared images collected during late July 2008 and 2009 for a reach of the Cains River, New Brunswick, Canada. These images were paired with geospatial catchment variables to identify the driving factors for coldwater refugia located within tributaries to the main channel. Using Partial Least Square (PLS) Regression, results suggest that median temperatures of tributary catchments are driven by their position within the landscape including slope in addition to the density of wetlands and mixed forest within the upstream catchment. Similar results are presented when PLS models were developed to predict the magnitude of the cold water refugia (i.e. the difference between the mainstem water temperature and the thermal refugia). These results suggest that thermal infrared images can be used to predict critical summer habitats for coldwater fishes.