Lightning-ignited fires in Portugal: A Comparison of satellite-derived and in-situ Databases

Fire databases typically contain information regarding the location, timing, and duration of fire occurrences, as well as the cause (natural or human-induced). These databases allow for the analysis and understanding of the circumstances surrounding the ignition and propagation of wildfires, being highly relevant when addressing fire suppression and management plans, or for improvement of prevention policies. In Portugal, the Mediterranean country with the highest burned area over the last five decades, a large number of fires in the official database have unknown causes, indicating that further investigation is required to understand to what extent lightning ignitions are a relevant fire cause over the country. In this context, we aim to investigate the spatial-temporal variability of lightning-induced fires in Portugal using the official fire database from the Portuguese Institute for Nature Conservation and Forests (ICNF) from 2003 to 2020. Then, we evaluate the potential of remote sensing data retrieved daily from the Moderate Resolution Imaging Spectroradiometer (MODIS) to act as a complementary and alternative information source when identifying fire causes. Our results revealed that lightning-ignited fires are less relevant to the fire regime than claimed by the current official report about fire causes in the country, accounting for only 1% of the fire occurrences instead of 2% and 4.6% of the total burned area, instead of 5.6% in the 18-year period under analysis. Despite accounting for a low percentage of the total amount of fires, lightning ignitions presented a higher relevance among larger fires (above 100 hectares) and contributed significantly to the total area burned during the unusual 2003 and 2017 fire seasons. Besides often occurring in remote areas, which delays the initial suppression and contributes to the larger fire size, as its importance in these extreme years suggests, lightning can trigger large fire events when in conjunction with extreme events such as droughts and heatwaves. Our results reveal that the use of remote sensing data resulted in reasonable precision when classifying the ignition cause of fires. However, due to its inherent limitations, the remote sensing data failed to account for the smaller fires, which represent the majority of the occurrences in the Portuguese database.