Concentration and dispersion modeling of the Kilauea plume
Annette Baerman, Steven Businger, John Porter and Duane Stevens, University of Hawaii 
Roland Draxler, NOAA/ARL


Active since January 1983, the Kilauea volcano on the island of Hawaii is the longest actively erupting volcano in the world. During the last 14 years Kilauea has remained in a quiescent outgassing stage. Emissions from the volcano chemically react with the ambient air resulting in a perpetual plume of volcanic smog, also known as vog. The vog plume is composed mainly of water vapor and sulfate particulates. Consequently, sulfate particulates are an effective size (0.1-0.6 µm) to reach down into the human lung (Morrow 1991) and in Haleakala Vogthe presence of high relative humidity may readily expand up to three times its original volume to further obstruct airways (Porter & Clarke 1997).Hawaiian Islands ImageThe presence of vog has been linked to numerous health problems (Worth 1995).The particulates may also dissolve in liquid water (i.e. in cloud) resulting in acid rain, which has the potential to destroy crops and leach lead from roofs and plum-bing. During episodes of increased sulfate production and atmo-spheric stability, vog may be thick enough to create a major visibility hazard to aircraft (Fig. 1) .In summary, vog is a significant threat to the island community and a need for a prognostic tool is evident.
Hawaii Volcanoes National Park measurements of SO2 often reveal average concentrations exceeding 290 ppb in a single hour, far exceeding EPA health standards.At present, SO2 emission rates average approximately 2,000 tons/day and have reached a maximum of 32,000 tons/day (Elias et al. 1998). Vog dispersion is primarily a function of synoptic and local wind patterns as well as stability of the environment. As SO2 converts to sulfate particulates near the main vent, the prevailing NE trade wind flow, along with island blocking effects and daily sea-breeze regimes, advect the vog past South Point and into Kona (Figs. 2a ,2b and 3).The majority of the vog pollutants are 
Kilauea Flows Image
trapped within the boundary layer due to the strong trade wind inversion (Fig. 3a and 3b ).Only during occurrences of weak inversions, strong trade winds or southerly winds associated with fronts, shearlines or Kona lows will the pollution be sufficiently advected away.
The primary goal of this research was to create a prognostic tool to aid in the prediction of vog plume concentration and dispersion. The windfields and
thermal data from the Meso-scale Spectral Model (MSM) were used as input into the Hybrid Single Particle Lagrangian Integrated Trajectory Model (HY-SPLIT) in order to produce vog simulations (Juang 2000; Draxler & Hess 1998).Validation of model results was con-ducted using aerosol concentrations derived from aircraft and ground-based data, as well as satellite imagery. Satellite imagery was also used to validate plume size, shape, and location.

Results and Discussion

Sun photometer measurements (Fig. 4a and 4b ) reveal low background concentrations in the Hilo and Puna areas. Modeled values in this area were several orders of magnitude smaller, consistent with their upwind placement.Small observed values down-wind of the vent are likely due to residual drainage flow off the mountain during late morning advecting the aerosol plume off-shore. Modeled values at these locations experience a strong spike up to 34 µg m-3.It appears the model plume is slightly misplaced, perhaps a reflection of a lack of representation of the diurnal land-breeze in the MSM wind field. A decreasing aerosol content is observed from South Kona to North Kona. Concentrations stay relatively constant for the duration of the journey, reducing slowly northward and inland. Leeward model output shows similar trends.
The simulation results (Fig. 5) show skill in reproducing general plume evolution based upon synoptic conditions.
The satellite image (Fig. 6a) shows a narrow plume off the SE coast. Modeled plume (Fig. 6b) orientation and concentration gradients are in good agreement with satellite observations.
A significant diurnal cycle in the wind field was observed during the field experiment due to light synoptic-scale trade winds and enhanced surface heating of the island.The thermally driven circulation that results is one of the key components enhancing aerosol build-up along the Kona coast, the MSM surface wind fields on the leeward coast tended to remain weak and offshore inhibiting this effect.
Aircraft data suggest that even in the area of aerosol pooling in the lee of Mauna Loa, model concentrations were under-forecast. In this study, background concentrations were not included. A reasonable explanation for the lower model concentrations near the leeward coast is the fact that the vog event in this research was unusually persistent. An extended period of SE winds caused aerosol to accumulate for a considerable amount of time, raising background concentrations prior to the field experiment.


Aircraft data confirm the importance of the trade wind inversion in trapping aerosols in the boundary layer, documented by the large drop in aerosol optical depth from 0.274 to below 0.023 (no units) when climbing from 151 meters to 2452 meters (Fig. 3).
Comparisons between model and ground-based data indicate that the model replicates observed trends reasonably well from North Kona to South Kona.Sun photometer measurements show an average concentration in Hilo of 4.55 µg m-3. Since Hilo remained upwind of the vent during the field experiment, these values reflect background concentration.
The model produced plume characteristics including size, shape, orientation, and concentration gradients consistent with those observed in satellite imagery.An interesting comparative result is the success of the model to reproduce the narrower-type plume and clear slots over Keauhou and South Point as seen in the satellite imagery (Fig. 6).Figures 5 and 6 suggest that the model performed best over the southern portion of the island and downwind over the ocean.As the MSM had difficulty resolving sea-breeze/land-breeze circulations and complex terrain effects on the leeward side, prognoses degraded with time and distance northward along the Kona coast.


  • Draxler, R., and G.D. Hess (1998): An overview of the HY-SPLIT_4 modelling system for trajectories, dispersion and deposition. Aust. Meteoral. Mag., 47. 295-308.
  • Elias, T., A.J.Sutton, J.B. Stokes, and T.J. Casadevall (1998): Sulfur dioxide emission rates of Kilauea Volcano, Hawaii, 1979-1997. USGS Open-File Report 98-462.
  • Juang, Hann- Ming Henry (2000): The NCEP mesoscale spectral model: A revised version of the nonhydrostatic regional spectral model. MWR, 128. 2329-2362.
  • Morrow, J.W. (1991c): Volcanic effects on the elemental composition of inhalable particulates in Hilo and Captain Cook. Vog and Laze Seminar, Hilo, Hawaii.
  • Nash, A.J., M.S. thesis, Dept. of Meteorology, School of Ocean and Earth Science and Technology, Univ. of Hawaii.
  • Porter, John N. and Antony D. Clarke (1997): Aerosol size distribution models based on in situ measurements. J. Geophys. Res. 102, D5. 6035-6045.
  • Worth, Robert M. (1995): Respiratory impacts associated with chronic vog exposure on the Island of Hawaii. Hawaii Department of Health Vog Symposium.
  • Acknowledgements

    This research was sponsored by NASA Solid Earth and Natural Hazards, Research and Application Programs: NRA 98-OES-13. We would like to thank Roland Draxler and NOAA/ARL for access and use of HY-SPLIT_4 code, Duane Stevens, Derek Funayama and Bruce Anderson for their work on the MSM data, MHPCC for computer time and John Porter for coordinating the aircraft observations.