Georeferenced thermal infrared images from UAV surveys as a potential tool to detect and characterize shallow cave ducts

Abstract Caves are buried geological features of great scientific and engineering interest. Based on the well-known thermal inertia of cave settings with respect to the surface, we have tested the use of thermal infrared (TIR) cameras carried by unmanned aerial vehicles (UAVs) to identify and characterize shallow karstic conduits. A pilot area was selected in a limestone rock massif from the Betic Cordillera (Southern Spain). At surface, this area appears as a doline field with several explored shaft entrances. The main and deepest shaft was thermally monitored at different depths, showing two well defined seasonal stages: (i) air and temperature stratification in summer and (ii) upward air flow and temperature homogenization in winter. We carried out UAV surveys at the expected maximum thermal contrast hours, finding out that winter dawns yielded the most distinctive images. These images show new warm air exits (warm spots) apart from the known cave entrances, and emphasize some of the main entrances depicting large warmed areas around them, which, in turn, can be related to shallow voids and conduits in the cave system. Furthermore, TIR images have been georeferenced using a network of identifiable points that can be transferred from visible orthoimages. The resulting TIR orthoimage mosaics have allowed us to define a number of quantitative parameters to characterize the warm spots. The most important of these parameters are: (i) the characteristic temperature, which is the maximum significant temperature of the air escaping from cave openings; (ii) the warmed area, which is the area that appears on the TIR orthoimage with a temperature greater than the surface temperature mode; and (iii) the average temperature weighted to the affected area. The georeferenced TIR orthoimages and the derived parameters are a new and very valuable tool for both speleological exploration and engineering purposes.