WRF-Chem modeling of sulfur dioxide emissions from the 2008 Kasatochi Volcano ANNALS OF GEOPHYSICS, Fast Track 2, 2014; doi: 10.4401/ag-6626 1 WRF-Chem modeling of sulfur dioxide emissions from the 2008 Kasatochi Volcano SEAN D. EGAN1*, MARTIN STUEFER1, PETER W. WEBLEY1, CATHERINE F. CAHILL1, JONATHAN DEAN1 1 University of Alaska Fairbanks, Geophysical Institute *sdegan@alaska.edu Abstract We simulate the dispersion and chemical evolution of the sulfur dioxide (SO2) plume following the eruption of Kasatochi Volcano in Alaska, USA, on August 7th, 2008 with the Weather Research Forecasting with Chemistry (WRF-Chem) model. The model was initialized with the observed three distinct plumes, which were characterized by a total estimated SO2 mass of 0.5 to 2.7 Tg. WRF-Chem modeled output was compared to remote sensing retrievals from the Ozone Monitoring Instrument (OMI), and the modeled plumes agreed well in shape and location with the OMI retrievals. The calculated SO2 column densities showed comparable Dobson Unit values with higher densities especially in the center of the distal plume over northern Canada. We concluded from our analysis that WRF-Chem derived a 9.1-day lifetime of the SO2 when initialized with a 12km eruption height. Sensitivity tests with varying eruption plume heights revealed significantly in- creased lifetimes of SO2 up to 17.1 days for higher plumes. I. INTRODUCTION asatochi volcano [52.169°N, 175.511°W] is a small (2.7 x 3.3 km, 314 m above sea level, a.s.l.), uninhabited stratovolcano in the Aleutian Arc of Alaska (Scott, Nye, Waythomas, & Neal, 2010). On August 2nd, 2008 US Fish and Wildlife biologists reported small tremors and a sulfur odor while on as- signment (Waythomas et al., 2010). They were evacuated prior to a M5.8 earthquake on Au- gust 7th, 2008 at 2:00 pm AKDT (22:00 UTC), de- tected by instruments from the Great Sitkin seismic Network. Infrared satellite retrievals from the Advanced Very High Resolution Radi- ometer (AVHRR) confirmed the presence of a volcanic plume situated over the volcano’s vent during this time (Waythomas et al., 2010). Two additional eruptions followed at 01:50 UTC and 04:35 UTC (Scott et al., 2010). The ash and SO2 emissions dispersed in a complex pat- tern due in large part to a low-pressure cyclo- genesis situated nearly on top of the volcano (Krotkov, Schoeberl, Morris, Carn, & Yang, 2010). Coarse ash and fine lapilli deposited quickly while most of the fine ash and SO2 ini- tially dispersed to the southeast (Waythomas et al., 2010). The resulting SO2 plume eventually entered the jet stream and traveled into the con- tinental United States and Canada within a week (Krotkov et al., 2010). The Kasatochi eruption is unique for various reasons. It resulted in the largest injection of SO2 into the atmosphere since the Mount Hudson eruption in Chile, August 1991. Initial estimates were between 1.20 to 2.7 Tg (Krotkov et al., 2010; Prata, Gangale, Clarisse, & Karagulian, 2010). By using inverse transport modeling, Kristiansen et al. (2010) established a 1.7 Tg K ANNALS OF GEOPHYSICS, Fast Track 2, 2014 2 mass loading. In addition, plume altitudes ex- ceeded the tropopause (maxima near 7 to 12 km with smaller emissions up to 20 km) introduc- ing about 1.0 Tg of SO2 into the stratosphere (Kristiansen et al., 2010). Modeling SO2 emissions is useful for various reasons. SO2 is often collocated with volcanic ash and thus may be used as a proxy for ash where remote sensing is hindered by ice for- mation, water or cloud cover. Additionally, WRF-Chem studies of historical volcanic erup- tions are motivated to test and provide source data and model parameterization schemes ca- pable of predicting volcanic SO2 and ash erup- tions in an operational setting in near real-time. Here, we use the well-defined Kasatochi SO2 eruption to study WRF-Chem’s ability to model volcanic SO2 transport and conversion. II. BACKGROUND Sulfur dioxide emissions from Kasatochi have been modeled previously with particle disper- sion models (D’Amours et al., 2010; Kristiansen et al., 2010). Wang and others (2010) studied SO2 dispersion and aerosol formation plume height sensitivity using the Eulerian GEOS- Chem model by initializing the model domain with time-fitted SO2 column densities from the Ozone Monitoring Instrument (OMI) using the Extended Iterative Spectral Fit (EISF) method. This WRF-Chem study differs from the GEOS- Chem study in that it does not require plume column densities, only specific eruption source data such as location, height, emission rate and duration. These parameters are included in tab- ulated Eruption Source Parameters (Mastin et al., 2009), which may be used to initialize WRF- Chem for operational volcanic ash and SO2 fore- casts. Sulfur dioxide converts quickly (on the order of days) to sulfate aerosols. In the strato- sphere, where the majority of the Kasatochi SO2 converted to sulfate, the conversion process is dominated by the interaction of SO2 with the hydroxyl radical (OH). Production of OH be- gins with the generation of excited states of atomic oxygen from ozone and diatomic oxy- gen via photolysis. The hydroxyl radical oxi- dizes SO2 in the stratosphere according to Equa- tion 1. 𝑆𝑂2 + 𝑂𝐻 + 𝑀 → 𝐻𝑂𝑆𝑂2 + 𝑀 ∗ (1) M* is a third body quencher required to remove excess energy from the reaction. This process was initially proposed to decrease the ambient amount of OH, and thus a second order rate equation would be needed to model it. How- ever, as mentioned by McKeen, Liu, & Kiang (1984), there is a cycling of the hydroperoxy radical, HO2, and OH in the presence of nitro- gen oxide species, NOx (McKeen et al., 1984): 𝑂2 + 𝐻𝑂𝑆𝑂2 → 𝐻𝑂2 + 𝑆𝑂3 (2) 𝐻𝑂2 + 𝑁𝑂 → 𝑂𝐻 + 𝑁𝑂2 (3) This cycling ensures the regeneration of OH concentration. If we assume [OH] is constant and [M] varies only with pressure, we may solve the following differential equation to ana- lytically calculate the change in concentration of SO2 with time: 𝑑𝑆𝑂2 𝑑𝑡 = −𝑘3[𝑆𝑂2][𝑂𝐻][𝑀] (4) [𝑆𝑂2]𝑓 = [𝑆𝑂2]𝑖 𝑒 −𝑘3,𝑀[𝑂𝐻]𝑡 (5) Here, k3[M] is the pseudo first order rate con- stant based on Equation 3. Sulfur dioxide and OH also interact via aque- ous phase reactions. In such reactions, OH is produced by dissolved hydrogen peroxide in water, which then reacts with dissolved, aque- ous SO2. III. METHODS ANNALS OF GEOPHYSICS, Fast Track 2, 2014 3 The application of WRF-Chem for simulating the transport and effects of volcanic emissions within the atmosphere has been described in (Stuefer et al., 2013). Importantly, WRF-Chem has been proposed as an operational tool for volcanic emissions modeling. Here, we test the feasibility of using WRF-Chem to capture SO2 emissions using the well-studied 2008 Kasato- chi eruption. The choice of eruption initialization parameters greatly impacts the ability of the model to pre- dict volcanic ash and SO2 transport (Mastin et al., 2009; Webley, Stunder, & Dean, 2009). Table 1 lists the domain initialization parameters used in this study and Table 2 provides the initializa- tion parameters for the eruption. We utilized the Global Forecast System (GFS) Final Reanal- ysis (FNL) datasets as base meteorological fields (NOAA, 2014). WRF-Chem may use either default values for Eruption Source Parameters (ESP) or if availa- ble, source data from plume observations. Karagulian and others (2010) discovered a min- imum of 1.7 Tg SO2 from the Kasatochi eruption using remote sensing data from the Infrared At- mospheric Sounding Interferometer (IASI). Kristiansen and others (2010) utilized inverse transport modeling to establish a similar mass of 1.7 Tg based on measurements from UV, IR and Lidar data. In a recent GEOS-Chem study by Wang et al. (2013) a value of 2.0 Tg SO2 was used. Herein, we initialized WRF-Chem with a total of 1.7 Tg of SO2. This mass was gradually added to the model using a constant eruption rate of 23,600 kgs-1 over the course of the three eruptions, using eruption durations and times based on Waythomas et al. (2010) (compare Ta- ble . WRF-Chem initializes, by default, volcanic ash and SO2 plumes as an umbrella shape with 75% of erupted mass in the plume surrounding the specified plume height and 25% of the mass lin- early detrained underneath (Stuefer et al., 2013). For the eruption plume height, the ESP implemented within the WRF-Chem preproces- sor as a default includes a height of 11km for Kasatochi. However, in accordance with Kristi- ansen et al. (2010), we chose 12 km a.s.l ± 4 km for this study in order to test the sensitivity of the model to its plume height source. For the ex- ample of our mean plume height of 12 km a.s.l, the umbrella will include 75% of the mass be- tween 9 – 13 km a.s.l (peaking at 12 km a.s.l) and 25% below 9 km, linearly decreasing with height (Stuefer et al, 2013). Table 1: Domain parameters for WRF-Chem Domain Size 600 x 400 dx, dy 15 km x 15 km Vertical levels 40, terrain following Model Height 2,000 Pa Projection Lambert-Conformal Center Lat/Lon 50°N, -120°W Kasatochi erupted over half of the SO2 into the stratosphere. Therefore, it is important to cap- ture the gas-phase chemistry behind strato- spheric SO2 oxidation shown in Equation 1. The model simulations utilized the Second Genera- tion Regional Acid Deposition Model Mecha- nism (RADM2) for gas and aqueous phase reac- tions. The RADM2 model includes the oxida- tion of SO2 by OH as depicted in Equation 1 using the first order kinetics in Equations 4 and 5 (Stockwell et al., 1990) as well as the treatment of the NOx species in Equations 2 and 3. Gase- ous precursors, such as NO and OH, were loaded into the model using the Prep-Chem- Source 1.4 preprocessor (Freitas et al., 2011). As mentioned, SO2 also converts via aqueous phase chemistry with OH. This scheme is also parameterized within WRF-Chem RADM2. ANNALS OF GEOPHYSICS, Fast Track 2, 2014 4 Table 2: Eruption times and durations Eruption Date, Time Eruption Du- ration Eruption Rate 8/7 22:00 UTC 60 minutes 23600 kg/s 8/8 01:50 UTC 30 minutes 23600 kg/s 8/8 04:35 UTC 30 minutes 23600 kg/s Global ozone and other trace gases, such as SO2, are detected by the Ozone Monitoring Instru- ment (OMI), a nadir viewing, ultraviolet (UV)/visible spectrometer aboard the National Aeronautical & Space Administration’s (NASA) Earth Observing System’s (EOS) Aura satellite. OMI covers a spectral range of 264-504 nm, allowing measurements of ultraviolet and visible SO2 signals. It provides global coverage once per day with a nadir pixel size of 13 x 24 km2 and swath width of 2,600 km and has been used in previous research studies for volcanic emissions analysis (Kristiansen et al., 2010; Krotkov et al., 2010; Lopez et al., 2013; Wang et al., 2013). For spatial analysis, we utilized NASA’s Level 2 SO2 product, ColumnAmountSO2_STL (from here on STL). Column densities of SO2 in Dob- son Units (DU) for this product are shown in Figure 1. Since the STL derived data may under- estimate the total amount of SO2 in plume areas of high concentration, we used values from Krotkov et al. (2010) based on the Extended It- erative Spectral Fit (EISF) method for mass analysis. Applications of this method to Kasato- chi SO2 suggest that it may capture additional SO2 that other algorithms might miss (Krotkov et al., 2010; Yang et al., 2010). IV. RESULTS Figure 1 shows the dispersion of the plume as modeled by WRF-Chem and derived by the OMI STL product. The model captured the plume’s interaction with the meteorology well, as it dispersed over the North American conti- nent. Figure 2 shows a brief spatial analysis along two transects (105W and 145W) marked in red. Plume SO2 column densities were gener- ally collocated with OMI here, however the nor- malized masses peaked in different areas. To compare the change in mass in the domain, a linear correlation plot was constructed and pre- sented in Figure 3. We saw a high degree of cor- relation (> 0.9 r2) for all plume height test cases between the change in WRF-Chem predicted SO2 mass and those observed by OMI. Figure 1: Dispersion of the Kasatochi SO2 plume as mo- deled by WRF-Chem (left) and calculated by the OMI STL product (right). The transects used for spatial analy- sis are shown in red on the August 11th plot. Geopoten- tial height and wind vectors are plotted on WRF-Chem plots. A lifetime of SO2 was established using linear regression analysis; the lifetimes significantly ANNALS OF GEOPHYSICS, Fast Track 2, 2014 5 varied with height. The 12km eruption height yielded a 9.1 days lifetime (r2=0.74) while the 8km and 16km plumes resulted in longer life- times of 10.6 and 17.1 days (r2=0.72, 0.68), re- spectively. V. DISCUSSION WRF-Chem generally predicted a more dis- perse plume than was observed by OMI. In ad- dition, there is a higher mass bias in the model results. This is markedly different from the work of Wang et al. (2010) where GOES-Chem produced a low mass bias. In addition, the mass located in the distal plume trended higher than that in the proximal. This is likely a direct result of the chosen ESP used for the modeled case as this varied with the plume height. The rate of SO2 conversion agreed well with lit- erature values. We used values from Krotkov et al. (2010) using the EISF method and from Kristiansen et al., (2010) to test WRF-Chem out- put. In Figure 3 we see that all three initialized eruption heights produced r2 values above 0.9. As mentioned, a range of values was produced for the lifetime using linear regression analysis. The 12km eruption produced the shortest life- time, being about 9 days. This dependence of lifetime on plume height is most likely a direct result of different chemistry at the various lev- els of the atmosphere. The 12km eruption in- cluded SO2 mass located mostly in the strato- sphere where conversion based on Equation 1 dominated, yet also included enough SO2 in the troposphere where aqueous phase and hetero- geneous chemistry can also occur. WRF-Chem captured the dynamics and mass changes of the Kasatochi plume according to these results. It is, therefore, a robust candidate for volcanic emissions modeling, especially in the operational setting where multiple un- knowns such as a specific plume height are pre- sent. Figure 2: Spatial comparisons of SO2 column densities from WRF-Chem using 8, 12 and 16 km initialized plume heights and from the OMI STL product. Figure 3: Correlation between WRF-Chem (y-axis) and literature OMI (x axis) SO2 domain masses for 8, 12 and 16 km plume heights. REFERENCES [D’Amours et al., 2010] D’Amours, R., Malo, A., Servranckx, R., Bensimon, D., Trudel, S., & Gauthier-Bilodeau, J.-P. (2010). Application of the atmospheric Lagrangian particle dispersion model MLDP0 to the 2008 eruptions of Okmok ANNALS OF GEOPHYSICS, Fast Track 2, 2014 6 and Kasatochi volcanoes. Journal of Geophysi- cal Research: Atmospheres, 115(D2), D00L11. [Freitas et al. 2011] Freitas, S. R., Longo, K. M., Alonso, M. F., Pirre, M., Marecal, V., Grell, G., Sánchez Gácita, M. (2011). PREP-CHEM-SRC – 1.0: a preprocessor of trace gas and aerosol emission fields for regional and global atmos- pheric chemistry models. Geoscientific Model Development, 4(2), 419–433. [Karagulian et al., 2010] Karagulian, F., Clarisse, L., Clerbaux, C., Prata, A. J., Hurtmans, D., & Coheur, P. F. (2010). Detection of volcanic SO2, ash, and H2SO4 using the Infrared Atmos- pheric Sounding Interferometer (IASI). Journal of Geophysical Research, 115(D2), D00L02. [Kristiansen et al., 2010] Kristiansen, N. I., et al. (2010). Remote sensing and inverse transport modeling of the Kasatochi eruption sulfur diox- ide cloud. Journal of Geophysical Research, 115(D2), D00L16. [Krotkov et al., 2010] Krotkov, N. A., et al. (2010). Dispersion and lifetime of the SO2 cloud from the August 2008 Kasatochi eruption. Jour- nal of Geophysical Research, 115(D2), D00L20. [Lopez et al., 2013] Lopez, T., Carn, S., Werner, C., Fee, D., Kelly, P., Doukas, M., Schneider, D. (2013). Evaluation of Redoubt Volcano’s sulfur dioxide emissions by the Ozone Monitoring In- strument. Journal of Volcanology and Geother- mal Research, 259, 290–307. [Mastin et al., 2009] Mastin, L. G., et al. (2009). A multidisciplinary effort to assign realistic source parameters to models of volcanic ash- cloud transport and dispersion during erup- tions. Journal of Volcanology and Geothermal Research, 186(1–2), 10–21. [McKeen et al., 1984] McKeen, S. A., et al. (1984). On the chemistry of stratospheric SO2 from vol- canic eruptions. Journal of Geophysical Re- search, 89(D3), 4873–4881. [NOAA, 2014] NOAA. (2014, June 18). National Centers for Environmental Prediction Global Forecast System (Final) global gridded analysis archive - 2000 to Present. Retrieved June 18, 2014, from rda.ucar.edu/datasets/ds083.2 [Prata et al, 2010] Prata, A. J., et al. (2010). Ash and sulfur dioxide in the 2008 eruptions of Ok- mok and Kasatochi: Insights from high spectral resolution satellite measurements. Journal of Geophysical Research, 115, D00L18. [Scott et al., 2010] Scott, W. E., et al. (2010). Au- gust 2008 eruption of Kasatochi volcano, Aleu- tian Islands, Alaska—Resetting an island land- scape. Arctic, Antarctic and Alpine Res.,42(3),250–259. [Stockwell et al., 1990] Stockwell, W. R., et al. (1990). The second generation regional acid deposition model chemical mechanism for re- gional air quality modeling. Journal of Geo- physical Research, 95(D10), 16343–16,367. [Stuefer et al., 2013] Stuefer, M., et al. (2013). In- clusion of ash and SO2 emissions from volcanic eruptions in WRF-Chem: development and some applications. Geosci. Model Dev., 6(2), 457–468. [Wang et al., 2013] Wang, J., et al. (2013). Mod- eling of 2008 Kasatochi Volcanic Sulfate Direct Radiative Forcing: Assimilation of OMI SO2 Plume Height Data and Comparison with MODIS and CALIOP Observations. Atmos- pheric Chemistry and Physics, 13, 1895–1912. [Waythomas et al., 2010] Waythomas, C. F., et al. (2010). The 7–8 August 2008 eruption of Ka- satochi Volcano, central Aleutian Islands, Alaska. Journal of Geophysical Research: Solid Earth (1978–2012), 115(B12). [Webley et al., 2009] Webley, P. W., et al.,. (2009). Preliminary sensitivity study of eruption source parameters for operational volcanic ash cloud transport and dispersion models — A case study of the August 1992 eruption of the Crater Peak vent, Mount Spurr, Alaska. Journal of Volcanology and Geothermal Research, 186(1–2), 108–119. [Yang et al., 2010] Yang, K., et al. (2010). Direct retrieval of sulfur dioxide amount and altitude from spaceborne hyperspectral UV measure- ments: Theory and application. Journal of Geo- physical Research: Atmospheres, 115(D2).