The potential distribution of tropical lowland cloud forest as revealed by a novel MODIS-based fog/low stratus night-time detection scheme

摘要

Fog is a crucial driver of epiphyte richness in tropical montane cloud forests but its spatial occurrence and role in tropical lowland areas is poorly understood. Recent studies in French Guiana have reported high epiphyte richness in previously undescribed “tropical lowland cloud forest” (LCF) due to radiation fog. Here, we analyze the spatial extent of fog/low stratus (FLS) in lowland forests of French Guiana using the frequency distribution by means of night-time MODIS (Moderate Resolution Imaging Spectroradiometer) satellite data for the years 2007–2010. The analysis is based on a newly developed dynamic threshold-test method relying on brightness temperature differences between thermal and mid-infrared bands. Individual thresholds for the discrimination between fog/low stratus and cloud-free pixels were retrieved by radiative transfer calculations and validated using discriminant analysis. The thresholds dynamically depend on total precipitable water (TPW) and the terrain-induced maximum possible sub-pixel fog coverage. The results of the new retrieval were validated using in-situ data and compared to results from existing fog detection algorithms, showing an improvement of the new detection scheme regarding the capability to detect sub-pixel fog coverage under varying TPW. FLS frequency maps derived from the novel fog classification scheme indicate a widespread distribution of night-time fog in river valleys, marking a multitude of potential areas for LCF throughout French Guiana. LCF is probably not only a local phenomenon but also may be widely distributed in river valleys in the lowland tropics, with significant consequences for biodiversity mapping in tropical lowland areas.