Využití termální kamery na bezpilotním letounu při výzkumu sesuvného území v obci Březno u Postoloprt

Abstract

Landslides are complex geological phenomena that can cause significant damage to property and infrastructure and put public safety at risk. For mitigation of these risks, monitoring of landslides is essential. The application of thermal imaging on board unmanned aerial vehicles (UAVs) represents one of the recently developing methods for landslide monitoring. The thermal imaging can be used to detect changes in surface temperature that are associated with landslide deformation. In this study, the Workswell WIRIS 2nd gen 640 thermal camera on the UAV DJI Matrice 600 Pro (Fig. 1) was used to investigate a landslide in Březno u Postoloprt (Fig. 2). This small-scale active rotational landslide occurs in Cretaceous marlstones with a variety of deformation features, including cracks, gullies, and toppling. The flight was planned considering the local topography, vegetation season and weather. It took place in October 2022, during constant cloud cover, approximately at an altitude of 27 meters above the terrain using the pilot-controlled mode. The final spatial resolution was 2.92 cm per pixel with approximate image overlap of 90 %. The thermal orthomosaic was analysed using a standard deviation threshold-based classification algorithm (Fig. 3), where the threshold defining the deformation features was set as 2,5 times standard deviation from mean temperature. Based on the algorithm, pixels with temperatures below 13.4 °C were identified as potential landslide deformation features (Fig. 4). These features were then verified using the 3D photogrammetric model (Fig. 5). The results of the study showed that the thermal orthomosaic was able to successfully detect the main system of cracks that runs perpendicular to the slope of the landslide. These thermal anomalies may results from the air circulation in the fractures, cooling caused by evaporation in the higher-moisture zones or the shadowing effect of the terrain roughness. The findings of this study suggest that UAV thermal imaging represents a promising tool for landslide monitoring upon condition that a number of factors such as the vegetation cover, actual temperature, the aspect of the slope and its exposure to the sunlight during the day are taken in account before making conclusions. The results of UAV thermal imaging should be also interpreted in conjunction with other data, such as RGB and multispectral imaging and ground-based observations. Thanks to easy implementation, the UAV thermal imaging represents an effective option for landslide monitoring.