Detecting Anomalies in Volcanic Ashfall Forecast During Large Volcanic Eruptions: Sakurajima Taisho Eruption Case

In the form of fine ashes (i.e. ashfall), tephra affects vast, vulnerable areas with deadly consequences and long-term impacts. Moreover, both eruption scale and wind conditions exacerbate the catastrophe to a broader range. Accordingly, countries with volcanic risks have developed early warning systems to reduce the ashfall’s impact when a powerful explosive eruption occurs. Such systems utilize the prediction of expected eruption coupled with the numerical weather forecast to alert the people at risk and prepare the response activities. However, abrupt weather can influence the ashfall forecast to have inaccurate warnings. Here, we evaluate the quality of the forecast results as an early warning or an evacuation policy. Using the Sakurajima Taisho eruption as a case study, we simulate a volcanic ash dispersal process with two different wind data, predictive and historical data. This study found some alarming deviations (i.e. anomalies) in the ashfall forecast under specific weather conditions. The anomalies bring errors in defining the impact areas and unforeseen damages and losses. Furthermore, this study analyses the underlying pattern that causes a disturbance in the ashfall forecast and suggests improving the forecast results to enhance decision-making during crises.