Mapping Vernal Pools Using LiDAR Data and Multitemporal Satellite Imagery

Abstract

Seasonal depressional wetlands, or vernal pools, offer critical breeding habitats for diverse species (primarily amphibians and invertebrates) in northeastern America. Their seasonal aspect contributes to their uniqueness. Degradation and loss of these habitats are attributed to human activities such as urban development and some forestry activities. Thus, it is important to map vernal pools to plan conservation measures. We studied topographic depressions derived from airborne LiDAR Digital Elevation Models (3 m) combined with optical multitemporal satellite imagery provided by Pleiades (50 cm) to detect depressional wetlands with the objective of discriminating vernal pools from other small wetlands. We first set a hierarchical identification approach with five criteria to plan the field campaigns, but after using statistical analysis, we were able to reduce these to only one criterion, the temporal difference (May to September) of the Normalized Difference Water Index (NDWI). Our results suggest that vernal pool occurrence is highly correlated with the temporal difference of NDWI. The higher the difference, the higher the observed occurrence. Considering 81 field-validated vernal pools, the user’s accuracy reached 83% and the producer’s accuracy, 59%. Our study presents a simple approach to map vernal pools on large territories while considering their temporary status.