Abstract
Volcanic ash emissions pose threats to aviation safety and socioeconomic activities, necessitating accurate dispersion modeling for effective early warning systems. The Mass Eruption Rate (MER) is a crucial parameter in ash transport models, yet its estimation remains a primary source of uncertainty. This study evaluates six MER formulations (MER1– MER4: non-wind-affected; MER5–MER6: wind-affected) for simulating ash dispersion from the Lewotobi Laki-Laki eruption using HYSPLIT, under two plume height scenarios (H = 1,000 m and 9,000 m). Results indicate that the plume orientation aligns well with Sentinel5P observations, though significant concentration discrepancies are evident. Case 1 shows a severe systematic underestimation (FB = -190.24%), likely due to uncertainties in column height and the high-resolution CAMS EAC validation constraints. Case 2 demonstrates much better agreement with CAMS EAC4 reanalysis data, with a minimal fractional bias (FB = -7.06%), a high index of agreement (IOA = 0.9653), and an RMSE of 0.7084 µg.m⁻ ³. Sensitivity analysis reveals that uncertainty in column height can increase MER variance roughly threefold, highlighting that source height precision largely determines MER accuracy. Although MER5 does not consistently outperform other formulations, it demonstrates relatively more consistent spatial dispersion patterns under wind-dominated conditions, though performance remains sensitive to source parameter accuracy. Consequently, this study recommends a multilevel operational approach that employs MER5 under favorable meteorological conditions, combined with rigorous uncertainty quantification to support operational volcanic ash advisories.