Statistical analysis of the effects of forests on fog

Abstract

Fog is influenced by numerous factors, including forests. The aim of our study was to examine in detail the extent to which forests affect fog. We hypothesised that: (i) fog incidence is affected by forests, (ii) the forested area in the station's neighbourhood is a factor influencing fog incidence, (iii) the influence on fog differs between coniferous and broad-leaved forests and (iv) the effect of forests on fog occurrence differs with altitude. For this, we used long-term records of fog incidence measured in 1981–2017 at 56 professional meteorological stations in Romania, GIS-derived information on forested areas in the neighbourhood of these stations, and land-use data on the types of these forests. The analyses are based on a semiparametric generalised additive logistic model for the probability of fog occurrence with potentially nonlinear, smooth effects modelled via penalised splines. Our results indicated that forests do affect fog incidence significantly, though their effect is considerably lower than the effect of dominant factors we studied previously, such as seasonality and altitude. It was indicated that forested areas in the neighbourhood of these stations are a factor significantly influencing fog incidence, even when forest is not the dominant land-use factor. In this respect, a radius of 3 km was the most effective when considering the forested area in a circle around the meteorological station. Our model showed that the influence on fog occurrence differs between coniferous and broad-leaved forests, and the effect of forests on fog occurrence is modified by altitude. The hypotheses propounded were confirmed and the hypothesised effects were quantified. Our findings, relevant at least for temperate forests, will enhance further considerations related to fog formation and wet atmospheric deposition. Moreover, our study opens a new challenge for further research of water balance as related to deforestation in catchment areas.