Abstract
Our database comprises daily minimum and maximum temperatures observed over 10 years at 859 pairs of meteorological stations throughout France. Each pairing associates a low and a high station. The influence of six predictors on the intensity, frequency, and duration of temperature inversions is measured by linear regressions. Five predictors are drawn from a 250 m-resolution DTM: elevation, depth of the valley where the low stations are located, magnitude of positive relief (ridge, hills), gradient of the slope of the hill or mountainside, and altitudinal amplitude between the high and the low station. The sixth descriptor used is the distance to the nearest sea. Topography exerts a major influence over the formation of thermal inversions. Three of the descriptors account for more than 80% of the variance of the inversion characters: distance to the sea, valley depth, and altitudinal amplitude. Elevation explains only 24% of that variance. The spatial distribution of the three characteristics of the inversions highlights several categorizations that fit into several nested scales. The 859 sites can be arranged into three classes relating to mountains, coastal areas, and plateaus. However, their distribution over the area under consideration is unclear and fails to indicate sharply delimited groupings.