Reducing dependence of modeled resuspended volcanic ash on meteorological grid resolution

Remobilized volcanic ash from ground deposits can present significant hazards to human health, infrastructure, and aviation. Modeling of ash remobilization events is an important tool that can provide information on the timing and magnitude to assist in planning and response. We investigate how the horizontal resolution of meteorological data, specifically that of friction velocity provided by numerical weather prediction (NWP) models, affects the estimated vertical mass flux and modeled concentrations of volcanic ash. We then apply a method designed to reduce the influence of the resolution on these quantities. Resuspension of volcanic ash from a deposit in Iceland has been modeled with the HYSPLIT atmospheric transport and dispersion model (ATDM) driven by meteorological fields from the European Center for Medium-Range Weather Forecasts (ECMWF) ECMWF Reanalysis v5 (ERA5) dataset and the weather research and forecasting model (WRF) at different resolutions (27 km and 9 km). We tested several simple emission schemes: one widely used for both volcanic ash and dust emissions, one operationally used to forecast ash resuspension in Iceland, and one based on controlled measurements from prepared ash deposits. Scaling factors for emissions were estimated using a cumulative distribution function (CDF) matching technique. Friction velocity values varied significantly across meteorological datasets resulting in considerably different estimates of onsets and vertical mass flux. It is a common approach to compensate for these differences by applying a scaling factor and adjusting the threshold friction velocity. Here, we implement a scheme that utilizes a Weibull distribution for the friction velocity to reduce the dependence of emission estimates on meteorological data resolution. We find that all emission schemes and meteorological datasets can predict the timing of large resuspension events and subsequent transport of the resuspended material. Indeed, the coarser datasets of WRF 27 km and ERA5 perform better than the WRF 9 km in some respects. The use of Weibull distribution for friction velocity successfully reduces the dependence of emission estimates on grid resolution. Similar schemes have been used successfully for dust emissions. Reducing or eliminating this dependence is important in order to assess and compare the success of different emission schemes, threshold friction velocities, and calibration factors.