Operational fuel model map for Atlantic landscapes using ALS and Sentinel-2 images

Abstract

Abstract Background In the new era of large, high-intensity wildfire events, new fire prevention and extinction strategies are emerging. Software that simulates fire behavior can play a leading role. In order for these simulators to provide reliable results, updated fuel model maps are required. Previous studies have shown that remote sensing is a useful tool for obtaining information about vegetation structures and types. However, remote sensing technologies have not been evaluated for operational purposes in Atlantic environments. In this study, we describe a methodology based on remote sensing data (Sentinel-2 images and aerial point clouds) to obtain updated fuel model maps of an Atlantic area. These maps could be used directly in wildfire simulation software. Results An automated methodology has been developed that allows for the efficient identification and mapping of fuel models in an Atlantic environment. It mainly consists of processing remote sensing data using supervised classifications to obtain a map with the geographical distribution of the species in the study area and maps with the geographical distribution of the structural characteristics of the forest covers. The relationships between the vegetation species and structures in the study area and the Rothermel fuel models were identified. These relationships enabled the generation of the final fuel model map by combining the different previously obtained maps. The resulting map provides essential information about the geographical distribution of fuels; 32.92% of the study area corresponds to models 4 and 7, which are the two models that tend to develop more dangerous behaviors. The accuracy of the final map is evaluated through validation of the maps that are used to obtain it. The user and producer accuracy ranged between 70 and 100%. Conclusion This paper describes an automated methodology for obtaining updated fuel model maps in Atlantic landscapes using remote sensing data. These maps are crucial in wildfire simulation, which supports the modern wildfire suppression and prevention strategies. Sentinel-2 is a global open access source, and LiDAR is an extensively used technology, meaning that the approach proposed in this study represents a step forward in the efficient transformation of remote sensing data into operational tools for wildfire prevention.