Integration of Lightning- and Human-Caused Wildfire Occurrence Models

Fire risk indices are useful tools for fire prevention actions by fire managers. A fire ignition is either the result of lightning or human activities. In European Mediterranean countries most forest fires are due to human activities. However, lightning is still an important fire ignition source in some regions. Integration of lightning and human fire occurrence probability into fire risk indices would be necessary to have a complete picture of the causal agents and their relative importance in fire occurrence. We present two methods for the integration of lightning and human fire occurrence probability models at 1 × 1 km grid cell resolution in two regions of Spain: Madrid, which presents a high fire incidence due to human activities; and Aragón, one of the most affected regions in Spain by lightning-fires. For validation, independent fire ignition points were used to compute the Receiver Operating Characteristic (ROC)-Area Under de Curve (AUC) and the Mahalanobis Distance. Results in Madrid are satisfactory for the human fire occurrence probability model (AUC～0.7) but less suitable for the lightning and the integrated models. In Aragón the fit for the human model is reasonable (AUC～0.7) whereas for the integration methods is practically useless (AUC～0.58).