Mitigating hazards to aircraft from drifting volcanic clouds by comparing and combining IR satellite data with forward transport models.

摘要

Volcanic ash clouds in the upper atmosphere (>10km) present a significant hazard to the aviation community and in some cases cause near-disastrous situations for aircraft that inadvertently encounter them. The two most commonly used techniques for mitigating hazards to aircraft from drifting volcanic clouds are (1) using data from satellite observations and (2) the forecasting of dispersion and trajectories with numerical models. This dissertation aims to aid in the mitigation of this hazard by using Moderate Infrared Resolution Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR) infrared (IR) satellite data to quantitatively analyze and constrain the uncertainties in the PUFF volcanic ash transport model. Furthermore, this dissertation has experimented with the viability of combining IR data with the PUFF model to increase the model's reliability.;Comparing IR satellite data with forward transport models provides valuable information concerning the uncertainty and sensitivity of the transport models. A study analyzing the viability of combining satellite-based information with the PUFF model was also done. Factors controlling the cloud-shape evolution, such as the horizontal dispersion coefficient, vertical distribution of particles, the height of the cloud, and the location of the cloud were all updated based on observations from satellite data in an attempt to increase the reliability of the simulations. Comparing center of mass locations--calculated from satellite data--to HYSPLIT trajectory simulations provides insight into the vertical distribution of the cloud.;A case study of the May 10, 2003 Anatahan Volcano eruption was undertaken to assess methods of calculating errors in PUFF simulations with respect to the transport and dispersion of the erupted cloud. An analysis of the factors controlling the cloud-shape evolution of the cloud in the model was also completed and compared to the shape evolution of the cloud observed in the IR satellite data. An accurate eruption length of 28 hours--based on satellite imagery--resulted in an error growth rate that decreased by 50% from the original simulation. Using a dispersion coefficient that was calculated from satellite imagery further improved the PUFF simulation. Results show that using satellite-based information in the PUFF model decreases the error growth of the simulation by as much as 60%.;PUFF simulations were also compared to IR satellite data of four other eruptions: Augustine (2006), Cleveland (2001), Hekla (2000) and Soufriere Hills (2006). The Anatahan, Augustine and Cleveland eruptions produced clouds that were ash-rich. The Hekla and Soufriere Hills eruption produced clouds that were ice-rich. Mass retrievals performed on the satellite data for these eruptions were holistically compared to determine if the evolution of the ash clouds were dependent on the cloud species and the atmospheric environment into which they were ejected (arctic vs. tropical environments). Analysis show that the ice-rich clouds decrease in mass, area and optical depth more rapidly than the ash-rich clouds. Moreover, error growth rates of the simulated lowlatitude eruptions were linear, where as error growth rates of simulated high-latitude clouds were exponential. Results from this study provide some insight into possible implications for volcanic cloud simulations of ash and ice clouds in differing environments.;Finally, a sensitivity analysis of the PUFF model was implemented. This part of the research generated collaborative research with the University of Alaska, Fairbanks/Alaska Volcano Observatory, where the PUFF model is housed. Transport simulations of the May 10, 2003 Anatahan volcanic cloud were done from the PUFF model's command line. Unlike the web-based version of PUFF, the command line is not publicly accessible but provides substantially more control over the user's definition of certain variables. Results of this study show it is viable and practical to continually update the PUFF simulation of a volcanic cloud's evolution and location with satellitebased data in order to simulate the most reliable predication of volcanic cloud transport.