2019 年 14 巻 5 号 p. 786-797
A high-resolution forecast methodology for the ash hazard at Sakurajima volcano, Japan, is presented. The methodology employs a combined modeling approach and utilizes eruption source parameters estimated by geophysical observations from Sakurajima, allowing for a proactive approach in forecasting. The Weather Research and Forecasting (WRF) model is used to downscale Japan Meteorological Agency (JMA) forecast data over the area of interest. The high-resolution meteorological data are then used in FALL3D model to provide a forecast for the ash dispersal and deposition. The methodology is applied for an eruption that occurred on June 16, 2018. Disdrometer observations of ashfall are used along with ash dispersal modeling to inform the choice of the total grain size distribution (TGSD). A series of pseudo-forecast ash dispersal simulations are then carried out using the proposed methodology and estimated TGSD, initialized with meteorological forecast data released up to ∼13 hours before the eruption, with results showing surprising consistency up to ∼10 hours before the eruption. Using forecast data up to 4 hours before the eruption was seen to constrain observation to model ratios within a factor of 2–4 depending on the timing of simulation and location. A number of key future improvements for the methodology are also highlighted.