Comparison of real and simulated lava flows in the Holocene volcanism of Gran Canaria (Canary Islands, Spain) with Q-LavHA: contribution to volcanic hazard management

Abstract

Lava flow simulations are valuable tools for forecasting and assessing the areas potentially affected by new eruptions, interpreting past volcanic events and understanding the lava flow behaviour's controls. Q-LavHA (v 2.0) plug-in of Mossoux et al. (Mossoux et al., Comput Geosci 97:98–109, 2016) combines and improves existing deterministic (FLOWGO) and probabilistic ("Maximum Length" and "Decreasing Probability") codes which allow calculating the probability of lava flow spatial propagation and terminal length. We investigate the Q-LavHA algorithm's effectiveness in twenty Holocene ʻaʻā lava flows of Gran Canaria (Canary Islands). Pre-eruptive and updated digital elevation models (DEMs) (25 m of resolution) and associated topographic and morphometric parameters have been used as essential input data to simulate the lava flows. Besides, thermo-rheological properties of the studied Holocene lavas have also been provided in the deterministic approach. The probabilistic lava flow maps produced by Q-LavHA and the fitness indexes calculated for assessing the simulated lava flow' accuracy indicate that the probabilistic "Maximum Length" constraint provides the best simulations. By using this method, many of the simulations in Gran Canaria almost overlap the real lava flow entirely even if overestimated areas are, in some cases, relatively high. By contrast, underestimated areas are generally low. The best results are those in which the highest inundation probability is observed within the main channel where the actual lava flow is emplaced, and even if overestimated areas are high, they are associated with low pixel inundation.