A multi-sensor satellite assessment of SO2 emissions from the 2012–13 eruption of Plosky Tolbachik volcano, Kamchatka

    A multi-sensor satellite assessment of SO2 emissions from the 2012-13 eruption of Plosky Tolbachik volcano, Kamchatka J. Telling, V.J.B. Flower, S.A. Carn PII: DOI: Reference:

S0377-0273(15)00218-8 doi: 10.1016/j.jvolgeores.2015.07.010 VOLGEO 5589

To appear in:

Journal of Volcanology and Geothermal Research

Received date: Accepted date:

2 February 2015 2 July 2015

Please cite this article as: Telling, J., Flower, V.J.B., Carn, S.A., A multi-sensor satellite assessment of SO2 emissions from the 2012-13 eruption of Plosky Tolbachik volcano, Kamchatka, Journal of Volcanology and Geothermal Research (2015), doi: 10.1016/j.jvolgeores.2015.07.010

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT A multi-sensor satellite assessment of SO2 emissions from the 2012-13 eruption of Plosky Tolbachik volcano, Kamchatka

PT

J. Telling1, V.J.B. Flower1, S.A. Carn1 1

SC

RI

Michigan Technological University 1400 Townsend Drive 630 Dow Environmental Sciences Houghton, MI, 49931

NU

Corresponding Author J. Telling [email protected] Telephone: 1.201.835.8478

MA

Abstract

Prolonged basaltic effusive eruptions at high latitudes can have significant atmospheric and

D

environmental impacts, but can be challenging to observe in winter conditions. Here, we use

TE

multi-sensor satellite data to assess sulfur dioxide (SO2) emissions from the 2012-2013 eruption of Plosky Tolbachik volcano (Kamchatka), which lasted ~9-10 months and erupted ~0.55 km3

AC CE P

DRE. Observations from the Ozone Monitoring Instrument (OMI), the Ozone Mapping and Profiler Suite (OMPS), the Atmospheric Infrared Sounder (AIRS), and the Moderate Resolution Imaging Spectroradiometer (MODIS) are used to evaluate volcanic activity, SO2 emissions and heat flux associated with the effusion of lava flows. Gaps in the primary OMI SO2 time-series dataset occurred due to instrument limitations and adverse meteorological conditions. Four methods were tested to assess how efficiently they could fill these data gaps and improve estimates of total SO2 emissions. When available, using data from other SO2 observing instruments was the most comprehensive way to address these data gaps. Satellite measurements yield a total SO2 loading of ~200 kt SO2 during the 10-month Plosky Tolbachik eruption, although actual SO2 emissions may have been greater. Based on the satellite SO2 measurements, the Fast Fourier Transform (FFT) multi-taper method (MTM) was used to analyze cyclical behavior in the complete data series and a 55-day cycle potentially attributable to the eruptive behavior of Plosky Tolbachik during the 2012 – 2013 eruption was identified.

ACCEPTED MANUSCRIPT Keywords: Plosky Tolbachik Volcano; Remote sensing; Sulphur dioxide; Ozone Monitoring

AC CE P

TE

D

MA

NU

SC

RI

PT

Instrument; Ozone Mapping Profiler Suite; Volcanic gas

ACCEPTED MANUSCRIPT 1. Introduction

The 2012-2013 eruption of Plosky Tolbachik Volcano (Kamchatka, Russia) provided an unusual

PT

opportunity to examine a prolonged, high latitude volcanic eruption. Prolonged (months-to-years in duration) effusive eruptions at high latitudes have been relatively rare in recent times, with the

RI

2014-15 eruption at Holuhraun (Bárðarbunga, Iceland), which began in August 2014, a notable

SC

exception. The Kamchatka peninsula has a high density of active volcanoes, but activity in the subduction zone arc is predominantly explosive. While explosive eruptions can generate

NU

stratospheric plumes that are readily observed using satellite instruments (e.g., Rose et al., 2000; Carn and Lopez, 2011; Young et al., 2012), observations of effusive eruptions at high latitudes,

MA

such as the 2012 – 2013 eruption of Plosky Tolbachik, can be much more challenging due to low-altitude emissions and adverse meteorological and/or viewing conditions. Hence, there has been relatively little opportunity to study long-lived effusive eruptions at high latitudes (e.g., in

D

Kamchatka, Iceland, Alaska) using remote sensing techniques, making the 2012 – 2013 eruption

TE

of Plosky Tolbachik an important case study.

AC CE P

The increasing number and resolution of satellite instruments provide a comprehensive way to monitor volcanic activity, particularly in remote areas (e.g., Carn and Bluth, 2003). Plosky Tolbachik Volcano, located on the Kamchatka Peninsula (55.832°N, 160.326°E), is well known for the Great Plosky Tolbachik fissure eruption of 1975-1976, which produced an estimated 1.18 km3 of erupted lava (Fedotov and Markhinin, 1983; Fodotov, 1984). However, the 1975 eruption of Plosky Tolbachik occurred prior to the global, moderate resolution satellite remote sensing era. Activity at Plosky Tolbachik resumed in 2012, by which time numerous remote sensing satellites were available to monitor various aspects of the eruption, including volcanic gases and thermal infrared radiance.

Volcanic SO2 is easily distinguishable from the surrounding atmosphere due to the low background SO2 signal (away from large anthropogenic pollution sources), unlike other major volcanic gases such as H2O and CO2 (Carn et al., 2013), which have significantly higher background concentrations in the atmosphere. In addition, SO2 has strong absorption bands in both the ultraviolet (UV) and thermal infrared (TIR) spectral regions and each of these

ACCEPTED MANUSCRIPT wavebands provide certain viewing advantages. However, the late November onset of the 201213 eruption, combined with the high latitude of Plosky Tolbachik Volcano, make the use of multiple satellite instruments particularly crucial in the examination of this eruption. Solar

PT

radiation is the dominant source of UV therefore UV remote sensing techniques are limited during high latitude mid-winter when there is little daylight. IR remote sensing can be limited by

RI

thermal contrast (the temperature difference between the SO2 plume and the background) or by

SC

lower tropospheric water vapor and clouds, because water vapor absorbs strongly in the infrared. Hence, IR remote sensing of SO2 is best utilized for volcanic emissions above 3 km because of

NU

the abundance of H2O in the lower troposphere (e.g., Prata and Bernardo, 2007). Combining data

MA

from UV and IR instruments provides the most robust eruption coverage.

In addition to measuring volcanic SO2, satellite instruments can be used to study the development of surface thermal features, such as lava flows, which were the dominant type of

D

activity during the 2012-2013 Plosky Tolbachik eruption (Edwards et al., 2013; Global

TE

Volcanism Program, 2012, 2013c; Belousov et al., 2015). The extent and development of these features can be monitored by satellite-based remote sensing instruments such as the Moderate

AC CE P

Resolution Imaging Spectroradiometer (MODIS) due to the strong TIR emission from such phenomena (e.g., Wright et al., 2002, 2004, 2015). Through the analysis and comparison of particular spectral bands in the mid- (MIR) and TIR, hot features can be distinguished from cooler background, facilitating the analysis of lava flow development (Wright et al., 2002). The combination of different remote sensing techniques can provide a comprehensive picture of remote eruptions such as the 2012-2013 Plosky Tolbachik eruption.

Here we utilize multiple satellite remote sensing techniques, including UV and TIR instruments, to observe SO2 emissions and TIR radiance during the ten-month long eruption of Plosky Tolbachik Volcano from November 2012 to August 2013. Combining data from instruments that observe in different wavelengths and have variable spatial resolution and repeat cycles provides the most detailed eruption coverage possible for this remote, prolonged, high-latitude eruption. We use the IR Atmospheric Infrared Sounder (AIRS), UV Ozone Monitoring Instrument (OMI) and UV Ozone Mapping and Profiler Suite (OMPS) data to quantify SO2 emissions, and TIR Moderate Resolution Imaging Spectrodiometer (MODIS) data to measure TIR radiance. To

ACCEPTED MANUSCRIPT account for data gaps in OMI SO2 measurements and improve estimates of total SO2 emissions for the Tolbachik eruption, we test multiple gap-filling methods, including substituting near zero values, short term means and use a Kalman filter.Finally, we analyze the cyclicity in both the

PT

detected SO2 degassing and the surface thermal features during the eruption to elucidate eruptive processes during a long-lived effusive eruption, and also speculate on the eruption‟s potential

SC

RI

impact on the Arctic climate.

NU

2. The 2012-2013 eruption of Plosky Tolbachik Volcano

Following a week of increased seismic activity (Edwards et al., 2013), Plosky Tolbachik became

MA

active again on November 27, 2012 when a new fissure eruption began that continued through late August 2013. The active fissure region, located on the southern flank of Plosky Tolbachik, was approximately 6 km long (Gordeev et al., 2013) and had both northern and southern

D

segments (Edwards et al., 2013) though activity at the northern portion of the fissure had largely

TE

ceased by December 1, 2012 (Gordeev et al., 2013). Edwards et al. (2013) estimated that lava flows extended 9 km from the active vent region by a day after the start of the eruption. The

AC CE P

early eruption was characterized by small to moderate eruptions with gas plumes reaching 6-10 km above sea level (Global Volcanism Program, 2012). Following the first week of the eruption, the continued activity was predominantly effusive (Edwards et al., 2013; Gordeev et al., 2013) and ash and gas plumes were generally constrained to 1-3 km above sea level (Global Volcanism Program, 2012) However, Carn et al. (2015) has shown that volcanic SO2 injections can be decoupled from ash emissions, leading to occasionally misleading plume height reports. Edwards et al. (2013) report that 0.06 kt SO2 was produced during the early eruption phase. The entire eruption produced an estimated 0.55 km3 of magma (Belousov et al., 2015; Dvigalo et al., 2014) and radiated an estimated 1.9 x 1016 J over the ten month duration, the sixth highest radiant flux recorded in a list of 95 eruptions between 2000 and 2014 (Wright et al., 2015).

3. Data Collection

ACCEPTED MANUSCRIPT The instruments used to collect SO2 data and thermal radiance data are introduced in this section. AIRS (IR), OMI (UV) and OMPS (UV) were used to examine SO2 emissions from the eruption.

PT

MODIS was utilized to assess thermal features, such as lava flows, produced during the eruption. AIRS, which is aboard NASA‟s Aqua satellite, was used to examine SO2 in the IR using the

RI

strong SO2 absorption band at ~7.3 µm (Prata and Bernardo, 2007). AIRS has the best spatial

SC

resolution of the three SO2 remote sensing techniques (AIRS, OMI and OMPS) with nadir dimensions of 13.5 km but does not provide daily global coverage. AIRS is most sensitive to SO2

NU

near the tropopause, where temperature contrast reaches a maximum, and is less sensitive to lower tropospheric SO2 than UV sensors due to water vapor interference. As a result, smaller

MA

volcanic eruptions and SO2 emissions at low altitudes are often not captured in the AIRS dataset. In addition, emissions from larger events, such as the beginning of the Plosky Tolbachik eruption, may only be captured after the plume has been transported away from its source and

TE

D

require trajectory modeling to identify the source and release time.

Level 1B (L1B) AIRS data (infrared geolocated radiances) are publicly available from the

AC CE P

NASA Goddard Earth Sciences (GES) Data and Information Services Center (DISC; http://airsl1.gesdisc.eosdis.nasa.gov/opendap/Aqua_AIRS_Level1/).

Level

1B

AIRS

data

(AIRIBRAD version 5) were processed using the algorithm described by Prata and Bernardo (2007) (Figure 1) to retrieve partial SO2 column amounts. AIRS SO2 columns are referred to as „partial‟ since retrievals at 7.3 µm are not sensitive to SO2 present in the lower troposphere (Prata and Bernardo, 2007) and hence the reported SO2 columns effectively correspond to an uppertropospheric and lower stratospheric (UTLS) column.

Figure 1. AIRS SO2 data for November 28, 2012 shows the initial eruption plume, containing ~40 kt SO2, over the East Siberia Sea. The edges of the AIRS granules are shown with a blue dotted line. The movement of the plume from Plosky Tolbachik to the East Siberia Sea was verified using the HYSPLIT model (Draxler and Rolph; Rolph). OMI, launched aboard NASA‟s Aura satellite in July 2004, has a nadir spatial resolution of 13 x 24 km and a repeat cycle of 16 days. Aura follows Aqua in the A-Train satellite constellation, a

ACCEPTED MANUSCRIPT group of Earth observing satellites that closely follow each other along the same orbit track, by roughly 10 minutes, so measurements from AIRS and OMI are considered to be near coincident. Near coincident satellite measurements are important since they provide observations of multiple

PT

volcanic phenomena (e.g., SO2 emissions, thermal anomalies) in different spectral bands at roughly the same time, permitting comparison between retrieval techniques. We use OMI level 2

(http://disc.sci.gsfc.nasa.gov/Aura/data-holdings/OMI/omso2_v003.shtml).

SO2

SC

GESDISC

RI

sulphur dioxide data (OMSO2 collection 3), which are publicly available from the NASA

column amounts in the OMSO2 product are retrieved using the linear fit (LF) algorithm

NU

described by Yang et al. (2007), which requires an assumption of SO2 altitude. Here, a plume altitude of approximately 3 km (the OMI lower tropospheric or TRL SO2 product) was assumed

MA

throughout other than the first week of the Plosky Tolbachik eruption (Figure 2). OMI data were available for two days of the first week of the eruption and a higher plume altitude of roughly 8 km was used on these days (the OMI mid-tropospheric or TRM SO2 product). Prior to 2008,

D

OMI provided global daily coverage, but the development of the OMI Row Anomaly (ORA) in

TE

late 2008 has led to data gaps in the OMI swath believed to be caused by an obstruction in the sensor‟s field of view (Figure 2). Previous work by Spinei et al. (2010) and Carn and Lopez

AC CE P

(2011) validated OMI SO2 measurements using ground-based measurements coincident with satellite overpasses during the 2008 Okmok and 2009 Sarychev Peak eruptions. Due to its greater spatial resolution at nadir OMI is able to detect smaller SO2 plumes than other operational UV instruments including OMPS.

Figure 2. OMI SO2 data for Kamchatka on July 21, 2013. The 1.4 kt SO2 plume from Plosky Tolbachik (triangle) is drifting to the south of the vent. The grey region in the northeast corner of the image is a mask over the ORA. The OMPS instrument suite, launched aboard NOAA‟s Suomi-National Polar-orbiting Partnership (SNPP) satellite in October 2011, is also used to measure volcanic SO2 in the UV (Carn et al., 2015). The OMPS nadir mapper (OMPS-NM) has a lower spatial resolution than OMI with nadir pixel dimensions of 50 x 50 km (Figure 3), except when data are collected in high-resolution mode (once per week on Saturday) with a spatial resolution of 10 x 10 km (Seftor et al., 2014). While the spatial resolution of OMI is more sensitive to smaller plumes,

ACCEPTED MANUSCRIPT OMPS-NM is better able to distinguish plumes on days when the ORA impacts OMI‟s spatial coverage. OMPS typically lags the OMI overpass time by 10-30 minutes. Despite the lower spatial resolution of OMPS-NM (in standard resolution mode), it is an important tool in the

PT

assessment and tracking of volcanic SO2 emissions, since it provides daily global coverage, in contrast to OMI. OMI and OMPS-NM can measure SO2 from the boundary layer to the

RI

stratosphere, albeit with higher sensitivity in the upper troposphere and stratosphere (Yang et al.,

SC

2007). SO2 column amounts are retrieved from OMPS-NM UV radiances using the same LF algorithm applied to operational OMI measurements (Yang et al., 2007; Carn et al., 2015), which

NU

requires an a-priori assumption of the SO2 vertical profile. Errors on the OMPS-NM SO2 retrievals are assumed to be similar to those impacting OMI SO2 measurements, with an overall

MA

uncertainty of ~20% (Yang et al., 2007). Though OMPS offers daily global coverage, gaps in daily observations of the Plosky Tolbachik SO2 emissions still arose due to meteorological clouds, noise in the data preventing a clear definition of the volcanic plume and interference

TE

D

from other nearby volcanic SO2 sources.

Figure 3. OMPS SO2 data for Kamchatka on May 6, 2013. The plume is still located above and

AC CE P

directly around Plosky Tolbachik (triangle) with some drift to the SW. Roughly half of the SO2 in the image, 0.63 kt, is directly attached to Plosky Tolbachik and can be attributed to the eruption.

The MODIS instruments flown on NASA‟s Aqua and Terra satellites were used to examine TIR anomalies associated with the Plosky Tolbachik eruption (Wright et al., 2002; Wright et al., 2004). The MODIS instruments have a spatial resolution of 1 km at nadir in the TIR and provide twice daily coverage from each platform with a greater number of observations at high latitudes (NASA, 2013). The MODIS spatial resolution, which is necessary to identify local changes in surface volcanic features, and revisit frequency of these instruments make them important tools for compiling an accurate picture of lava flow effusion during the Plosky Tolbachik eruption (Murphy et al., 2013, Wright et al., 2015). An automated „hot-spot‟ detection algorithm known as MODVOLC (HIGP, 2010) is applied to the MODIS IR radiances (Wright et al., 2002). MODVOLC relies on spectral radiances recorded in 5 MODIS bands at 4 wavelengths (3.96 μm,

ACCEPTED MANUSCRIPT 1.64 μm, 11.03 μm, and 12.02 μm) to identify thermal anomalies although no differentiation can be made between features of volcanic or non-volcanic origin (Wright et al., 2002).

PT

MODVOLC data used here consists of TIR radiances obtained from MODIS Band 22 (3.96μm) unless the saturation temperature has been exceeded at which point Band 21, which measures at

RI

the same wavelength but has a higher saturation temperature, was utilized. Band 22 is given

SC

preference in this analysis if unsaturated as it has a finer radiometric resolution than that of Band 21 (0.07 K and 2.00 K respectively) (NASA, 2013). One variation of note in the MODVOLC

NU

algorithm exists in the classification of day- and night-time measurements. The added interference of reflected solar radiation during daytime requires a higher threshold in the

MA

classification of thermal anomalies and can result in the exclusion of less radiant features that would be detected at night from the daytime measurements. However, since the Plosky Tolbachik eruption was dominated by active lava flows we expect minimal impacts on this

AC CE P

4. Processing Methods

TE

D

analysis (Wright et al., 2002).

Cyclical analysis has proved to be a useful tool in the identification of the underlying dynamics of volcanic systems (Lopez et al., 2013; Odbert & Wadge, 2009; Nicholson et al., 2013) as well as the identification of cyclical artifacts introduced by the measurement techniques and correction procedures applied. To assess whether significant cyclicity existed in the 2012-13 eruption of Plosky Tolbachik, we implemented time-series analysis similar to that of Nicholson et al. (2013), Odbert & Wadge (2009) and Flower and Carn (in review) using the UV and IR SO2 and thermal IR data. Both methodologies employ the Fast Fourier Transform (FFT) multitaper method (MTM) (Mann & Lees, 1996) to estimate the Power Spectral Density (PSD: signal power as a function of frequency) of each of the incorporated datasets. Confidence levels were calculated for each of the analyses at 95% and 99% to allow statistically significant cycles to be distinguished from the raw output (Duchon & Hale, 2012). MTM analysis is favored due to its ability to distinguish cycles in systems where the underlying dynamics are unknown, as is the case with most geophysical data (Mann & Lees, 1996). MTM analysis differs from a standard FFT technique through the convolution of data with a set of complimentary taper functions

ACCEPTED MANUSCRIPT before the calculation of PSD, minimizing spectral leakage, and is particularly appropriate when the dynamics of the emission source is undefined, providing information on both the frequency and strength of cycles in the data (Percival & Walden, 1993). Following the calculation of the

PT

PSD, confidence limits were calculated (95% and 99%) allowing the identification of significant cycles in the data (Duchon & Hale, 2012). A continuous time-series of data is required for MTM

RI

analysis and therefore any gaps in the OMI and MODIS datasets, such as those caused by the

SC

ORA, meteorological clouds, or periodic instrument adjustments, must be appropriately accounted for before the method is used. The maximum cycle length that can be distinguished

NU

depends on the length of the original data series. In order for cyclicity to be identified with certainty it must be possible to resolve multiple complete cycles to discern between cyclical

MA

phenomena and individual discreet events. Therefore a limit of detection (LOD) was established where cycles longer than n/4 days, where n is the total number of samples analyzed, were disregarded. Four distinct methodologies (near zero values, short term means, a Kalman Filter

D

and multiple satellite data sets) were employed to fill gaps in the OMI data and were analyzed

TE

using MTM analysis following the initial compilation of the OMI SO2 and MODIS TIR data.

AC CE P

The near-zero value method used to fill OMI SO2 data gaps involved substitution of a near zero value (0.01 tons) for any missing data points. This produced a continuous time series, which is necessary for the application of employed statistical techniques; but does not reflect the true SO2 mass. This method of accounting for data gaps also gives the periodic ORA undue influence by inserting low values into the data on days when the ORA precludes data collection.

The second method utilized short term means (i.e., data point n is equal to the average of n-1 and n+1) to fill in missing values. Many of the OMI gaps were single days, where this method is more appropriate. However, longer gaps had to be partially filled with near zero values since short term means could not realistically be calculated over multiple days.

The third method to address gaps in the OMI data utilized a Kalman filter (KF) to estimate daily SO2 loadings in the absence of direct observations. KFs are recursive estimation algorithms which incorporate measurements to estimate the state of a dynamic system, originally developed in 1960 (Kalman, 1960). KFs have been used in spacecraft orbit estimation, radar tracking, and

ACCEPTED MANUSCRIPT prediction of target trajectories (Hayes, 1996), as well as in volcanic contexts, specifically InSAR and thermal anomaly monitoring (Shirzaei and Walter, 2010; Zakšek et al., 2013). However, KFs have not previously been applied to volcanic degassing data from ground or

PT

remote sensing sources. The KF method is an appropriate assessment tool for the analysis of noisy datasets and dynamic systems (Chui & Chen, 2009), which is often the case for volcanic

RI

emission measurements. It has also been used previously in an array of meteorological and

SC

geological applications to predict values for missing time series data (Alavi et al., 2006; Beyou et al., 2013). The KF was initiated using a built-in IDL subroutine from the unprocessed data mean

NU

and an a priori error value (a correction factor used to refine an a priori value to provide an a posteriori estimate of the state of the system). Where an a priori error is relatively small the new

MA

estimate generated by the KF is more influenced by the previous estimate than direct observation. Conversely if the a priori error is large then the previous estimate is disregarded and instead the current measurement is used to estimate the state of the system (Brown and Hwang,

D

1997). All KF analyses utilized an a priori error of 0.05, following an initial trial of multiple a

TE

priori values to prevent the continued dominance of the ORA while maintaining lower frequency cycles that were persistent in the MTM output. The KF estimate was used to fill gaps in the time

AC CE P

series where OMI data were obscured by the ORA or where factors such as meteorological clouds precluded direct measurement of SO2 emissions. The KF method was found suitable for filling data gaps of roughly a week where at least two OMI SO2 observations during the week were present.

Finally, SO2 data from OMPS and AIRS were utilized to supplement the OMI SO2 data. Rather than using one of the above gap-filling techniques, in this case available AIRS or OMPS data were used to cover OMI data gaps. AIRS data were favored in the early weeks of the eruption when plumes were generally both larger and higher in the atmosphere and when low UV levels restricted OMPS SO2 measurements. OMPS data were used more extensively during 2013 when longer daylight hours improved UV remote sensing of SO2 and the typically low-altitude plumes from Tolbachik were not ideal for IR techniques due to water vapor in the lower atmosphere or poor thermal contrast. An in-depth analysis of the differences between SO2 measurements from OMI, OMPS and AIRS is outside the scope of this investigation because numerous factors, including viewing angle, overpass time, plume altitude and pixel size, affect SO2 retrievals and a

ACCEPTED MANUSCRIPT significant amount of data, beyond that collected during a single eruption, is required to fully examine this issue (e.g., Prata and Bernardi, 2007; Carn and Prata, 2010; Thomas et al., 2011; Carn et al., 2015). However, SO2 values from OMI, OMPS and AIRS were not found to have

PT

any consistent bias between instruments.

SC

RI

5. Results

The 2012-13 Plosky Tolbachik eruption lasted for roughly nine months and, in that time, OMI

NU

provided 134 days of SO2 plume observations. AIRS and OMPS provided, respectively, an additional 28 and 101 days of eruption coverage (Figure 4a). No SO2 data were available from

MA

OMI, OMPS or AIRS on 17 days of the eruption. An additional interference in this analysis was intermittent activity from other Kamchatkan volcanic sources; predominantly Klyuchevskoy and Shiveluch (Global Volcanism Program, 2013a, b), therefore only SO2 plumes that could be

D

traced directly back to Plosky Tolbachik were included in the analysis after the initial period of

TE

eruptive activity. Using data from OMI, OMPS and AIRS, we estimate that Plosky Tolbachik released a minimum of 194 kt of SO2 during its nine-month eruption. This estimate is a

AC CE P

minimum due in large part to gaps in the dataset as well as reduced instrument sensitivity to plumes in the lower troposphere, conversion of SO2 to sulfate aerosol prior to the satellite overpass and the interference of meteorological clouds (Carn et al., 2013).

Figure 4. (a) The complete time series of OMI (black), OMPS (red) and AIRS (green) SO 2 data during the 2012 – 2013 eruption of Plosky Tolbachik. A seven day average of the three data sets has been overlaid on the raw data (black line). (b) The complete time series of OMI data (black) with the Kalman filter estimate overlaid (purple). (c) The total daily radiance at 3.96 µm detected by MODVOLC during the 2012 – 2013 eruption of Plosky Tolbachik.

The initial eruption plume was not captured by OMI and OMPS due to the ORA and the low UV radiance caused by the high latitude, wintertime eruption onset. AIRS data were not available over the vent during the initial eruption. However, a plume containing ~40 kt SO2 does appear along the northern border of Russia near the East Siberia Sea on November 28, 2012 at 17:00-

ACCEPTED MANUSCRIPT 18:00 UTC (5:00-6:00 local time) (Figure 1). The NOAA HYSPLIT trajectory model (Draxler and Rolph; Rolph) was used to determine whether this plume could be the initial product of the Plosky Tolbachik eruption based on local meteorological conditions and an initial plume altitude

PT

of 5000 m above sea level (Figure 5a), which is slightly lower than the initial estimate of 6000 m (Global Volcanism Program, 2012). The results show that a plume initialized early on November

RI

28, 2012 under these conditions would travel northwest and reach the border between Russia and

SC

the East Siberia Sea by 18 UTC on November 28, 2012, confirming that the SO2 cloud observed by AIRS (Figure 1) is likely the initial Plosky Tolbachik eruption cloud, emitted during an

NU

energetic early phase of the eruption. Since the initial conversion of SO2 to sulfate aerosol is likely to have begun during this time (Krueger et al., 2009), the total amount of SO2 emitted

MA

during the first stage of the eruption was most likely greater than 40 kt and was certainly higher than the initial estimate of 0.06 kt (Edwards et al., 2013). The movement of the initial Plosky Tolbachik eruption plume was tracked by AIRS for seven days following the eruption as it

D

moved north over the East Siberia Sea (Figure 1), west along the northern coast of Russia

TE

(Figure 5b) and finally south again over central Russia (Figure 5c). Utilizing the seven days of AIRS data covering the first and largest eruption plume, an SO2 e-folding time of approximately

AC CE P

2.5 days was calculated.

Figure 5. (a) Results of the HYSPLIT 50 hour forward trajectory model. The plume was initialized at the summit location of Plosky Tolbachik at 5000 m above sea level. An initial release at either 00 UTC or 06 UTC on November 28, 2012, resulted in a plume that reached the coast by approximately 18 UTC on November 28, 2012, which is coincident with the AIRS observation at 17-18 UTC (Figure 1). (b) AIRS granules showing an estimated 36 kt SO2 as the initial plume was transported across the Northern border of Russia on November 29, 2012. (c) AIRS granules showing the decay of the SO2 plume to approximately 28 kt by December 1, 2013. By December 4, 2012, only trace amounts of SO2 were detected by AIRS over central Russia. The last lava flows and gas emissions from Plosky Tolbachik reported by the Global Volcanism Program (2013c) occurred between August 16 – 22, 2013 and the last clear but small SO2 emission (~0.2 kt) was observed by OMPS on August 30, 2013. Detected degassing activity in

ACCEPTED MANUSCRIPT the final two weeks of the eruption amounted to no more than 1.7 kt of observed SO2 and only two days during this time were adversely affected by the ORA; however, plumes were likely at very low altitude during this waning stage of the eruption and this may have impacted

PT

measurements. Sections 5.1 through 5.5 describe the methods used to process the raw instrument

NU

5.1. OMI Data with near zero substitution for ORA

SC

meaningful temporal cycles in the satellite measurements.

RI

time-series, for OMI, OMPS, AIRS and MODIS, in order to determine whether there were

The first correction method used on the OMI data was the substitution of near zero values for all

MA

missing days of data. The specified LOD, 0.014 cycles per day (70 days) for this eruption, requires the repetition of any identified cycle at least four times within the analysis period in order to prevent the inclusion of artifacts of the analysis methodology. The PSD output analysis

D

associated with this correction method displayed a single peak at a frequency of ~0.438 (2.3

TE

days), which has been linked to the ORA (Flower and Carn, in review) (Figure 6a). This feature dominates the MTM analysis and results in the suppression of other peaks below the

AC CE P

implemented confidence limits of 95% and 99% and justifies the need for a different correction factor to be used in the replacement of null values.

Figure 6. The results of the FFT MTM analysis using four different infilling techniques to compensate for gaps in the OMI SO2 data for the 2012-13 Plosky Tolbachik eruption. Confidence limits of 95% (orange) and 99% (red) and the LOD (blue) for each plot are shown. (a) MTM analysis for OMI data with near-zero values filling data gaps shows only one significant cycle at a frequency of 0.438 cycles/day (period of 2.3 days). The LOD for this and all other analyses is 0.014 cycles/day (70 day periods).(b) MTM analysis for the OMI data with short term means used to fill data gaps. The only significant cycle above the LOD is at 0.018 cyles/day (55 days).(c) MTM analysis after a Kalman filter was applied to the OMI data to fill gaps in the time series. Significant peaks above the LOD can be seen at 0.018 cycles/day (55 day periods) and 0.438 cycles/day (2.3 day periods).(d) MTM analysis for the OMI, OMPS and AIRS composite SO2 data set shows that the only significant peak above the LOD is at 0.018 cycles/day (55 day periods).

ACCEPTED MANUSCRIPT

5.2. OMI Data with moving average substitution for ORA

PT

The predominant limitation of this method is that it can only be applied to the correction of single value data gaps. Longer gaps had to be filled with both near-zero values and short-term

RI

averages, which increases the influence of the ORA. However, Figure 6b shows that this

SC

technique successfully suppresses the ORA cycle identified in Section 5.1, which previously dominated the power spectra at a frequency of ~0.438 cycles/day (2.3 day periods). However, it

NU

is important to note that no cycles with a frequency greater than 0.33 cycles/day (< 3 day periods) can be assessed with confidence because short-term means are calculated for three day

MA

windows. Filling data gaps with moving averages facilitated the identification of two significant lower frequency cycles at 0.011 cycles/day (~90 day period) and 0.018 cycles/day (~55 day period). However, due to the LOD at 0.014 cycles/day (70 day periods), only the cycle at 0.018

TE

D

cycles/day (55 day periods) can be assessed with confidence.

AC CE P

5.3. OMI data with Kalman filter substitution for ORA

In contrast to the limited capability of the short-term average technique, Kalman filters can generate values with minimal observations (1-2 per week). Figure 4b shows the OMI data (black) with the results of the KF overlaid as a continuous time series (purple) over the data. Significant cycles in the KF MTM analysis appear at 0.011 cycles/day (90 day periods), which is below the LOD, 0.018 cycles/day (55 day periods) and 0.438 cycles/day (2.3 day periods). The reoccurrence of the 0.438 cycles/day (2.3 day period) cycle in this analysis suggests that the KF is faithfully recreating the missing data, supporting the selected a priori error value, since the ORA results in an increase in the 2.3 day period which is present in pre-ORA OMI and MODIS analyses, particularly in equatorial regions. This cycle is most likely caused by variations in the viewing angle of the instrument and therefore accurate reconstruction of the data using a KF would include this characteristic feature. The application of the KF here moderates the effect of this cycle and reduces its significance, preventing the saturation of the PSD by this peak and allowing the identification of longer period cycles. 5.4. OMI data with AIRS and OMPS substitution for ORA

ACCEPTED MANUSCRIPT

OMPS and AIRS SO2 data (Figure 4a) were used to supplement OMI SO2 data and the MTM analysis was run on the composite data set (Figure 6d). The addition of OMPS and AIRS data

PT

removed the effects of the ORA leaving only two significant cycles at 0.007 cycles/day (140 day periods), which is below the LOD, and 0.018 cycles/day (55 day periods). We conclude that the

RI

continued presence of the latter cycle in this analysis is significant since it is supported by all

SC

three possible SO2 data sources.

NU

5.5 MODIS Data with Kalman Filter, November 2012 to August 2013

MA

The MTM analysis was also run on the MODVOLC TIR radiance data. However, no significant cycles could be identified. Figure 4c shows that the radiance peaks soon after the start of the eruption and subsequently tapers off to a low level daily radiance of around 10 Wm-2sr-1µm-1 , in

D

the manner of a typical waxing-waning effusive eruption, which supports the lack of detectable

AC CE P

6. Discussion

TE

cycles in the MTM analysis.

The complete time series of SO2 loadings during the Plosky Tolbachik eruption (Figure 4a) was evaluated in order to identify volcanic features that might explain the 55-day cycle in the MTM analysis. Additional cycles, outside the LOD, may also be present in the data and influence the fit of the 55-day cycle to the raw data. Four periods of increased SO2 degassing activity were identified in the data (Figure 4a): during the initial eruption phase (November 2012), late February 2013, late April to mid-May 2013 and late June to early July 2013. The start of the eruption, in late November 2012, produced the largest discreet SO2 emission event detected, based on the AIRS minimum estimate of 40 kt SO2. Each of the three subsequent elevated degassing events produced 3-3.5 kt SO2 over a 1- to 2-day period. Limited studies conducted on the plumbing system of Plosky Tolbachik make it difficult to identify the potential source of the apparent 55-day cyclicity. However, comparisons with other volcanic systems suggest that it may be the result of a shallow magma storage region moderating the outflow of magma from the system. Similar cycles to those identified at Plosky Tolbachik have also been identified at

ACCEPTED MANUSCRIPT Soufriere Hills Volcano (Montserrat), where 50-60 day cycles were identified in heat flux (Odbert & Wadge, 2009; Sparks & Young, 2002), ground-based SO2 monitoring (Nicholson et al., 2013), and satellite-based SO2 and thermal IR radiance data (Flower and Carn, in review).

PT

Modeling studies conducted at Soufriere Hills Volcano (Costa et al., 2007; Costa et al., 2013) indicated the source of cyclicity on this timescale was pulsatory magma supply from an elastic

RI

walled dyke at depth controlling dome growth. However, if a similar process is indeed the source

SC

of the observed cyclicity at Tolbachik, then we would expect the MODVOLC TIR radiance data

NU

to show similar patterns.

We interpret the presence of cyclic behavior in the SO2 data, coupled with the lack of cycles in

MA

the TIR radiance data, to suggest that lava flows at Plosky Tolbachik were enclosed in lava tubes. Similar behavior has been observed at Kilauea (Koeppen et al., 2013). Koeppen et al. (2013) show that SO2 flux and thermal radiance were decoupled at Kilauea when lava flows

D

were predominantly contained in tubes and not flowing on the surface. It is possible that the late

TE

February 2013, mid-May 2013 and late June 2013 peaks in SO2 emissions from Plosky Tolbachik were caused by the injection of new lava into lava tube systems (Edwards et al., 2013;

AC CE P

Zelenski et al., 2014) that did not reach the surface and therefore did not cause an increase in TIR radiance. Reports of observed surface activity (Global Volcanism Program, 2013c) mention peaks in lava flow activity during the initial eruption phase in late November 2012, and in late January 2013 and mid-April 2013. After the initial activity, which corresponds to the period of strongest SO2 degassing, the observed peaks in surface activity at Plosky Tolbachik do not correspond to the SO2 peaks identified by the MTM analysis. Sulfur dioxide degassing appears to lag enhanced lava flow activity at the surface, further suggesting that enclosed lava tubes may have placed an important role in the transport of lava during the eruption.

The earliest SO2 plume from Plosky Tolbachik was observed by AIRS for a week following the start of the eruption as it was transported north towards the East Siberia Sea (Figure 1), west along the northern border of Russia (Figure 5b) and then south again over central Russia (Figure 5c) before dispersing below detection limits. Relatively little conversion of SO2 to sulfate was apparent during the first five days of transport and the calculated e-folding time for SO2 loss during this period is ~8 days. Early plume transport was predominantly over ocean or

ACCEPTED MANUSCRIPT undeveloped tundra. However, including the last two days that the initial plume could be clearly observed by AIRS, when the plume was almost entirely located over land, reduces the e-folding time to only ~2.5 days. The abrupt drop in e-folding time suggests that the conversion of SO2 to

PT

sulfate aerosol occurred at a much higher rate over central Russia than it did over the East Siberia Sea (perhaps due to a change in latitude), or it could indicate descent of remaining SO2 to

RI

lower altitude, e.g., in a weather system. Indeed, the decreasing plume altitude predicted by

SC

HYSPLIT (Figure 5a) means that some of the SO2 may not have been detected by AIRS, which is most sensitive to SO2 in the UTLS. Conversion rates of SO2 to sulfate aerosol can vary widely

NU

and are significantly faster in the troposphere than the stratosphere (e.g., Oppenheimer et al., 1998; Rose et al., 2000; von Glasow et al., 2009) and higher particulate loading in the

MA

atmosphere will further expedite the conversion of SO2 to sulfate aerosol (von Glasow et al., 2009).

D

The transport of a relatively large SO2 emission through the Arctic during wintertime becomes

TE

particularly important when considering the radiative impact that it may have on the environment (Young et al., 2012). The initial emission from Plosky Tolbachik occurred during the Arctic

AC CE P

wintertime and contained SO2 with little to no ash (Edwards et al., 2013; Gordeev et al., 2013). It is unlikely that much, if any, of the SO2 was transported into the stratosphere (Figure 5a), even taking into consideration the Arctic wintertime conditions. Young et al. (2012) predicts that an ash poor, SO2 rich plume transported through the Arctic troposphere during the winter would produce a small cooling effect in the atmosphere over both seawater and snow but would have a larger effect on the direct aerosol radiative forcing efficiencies (DARFE) at both the top of the atmosphere and the surface. On a regional scale, the Plosky Tolbachik SO2 plume could have produced a minor cooling effect during the first weeks of the eruption when both shortwave and longwave radiative effects are considered.

7. Conclusions OMI SO2 measurements for the 2012 – 2013 Plosky Tolbachik eruption were collected and analyzed using four different techniques to address gaps in the data set. Near zero values, short term averages, a Kalman filter and supplemental SO2 data sources were all tested to examine

ACCEPTED MANUSCRIPT how well they remove the effects of the ORA. Using near-zero values emphasized the effects of the ORA and reduced features that might be attributed to other cycles. Short-term averages limited the effects of the ORA, largely because the 2.3 day ORA cycle is shorter than the number

PT

of days used to calculate the averages are calculated across (i.e., 3 days). The short term averaging technique also highlighted a 55 day cycle that persisted in other analyses. A Kalman

RI

filter was able to fill longer (~7 day) data gaps and the continued presence of a significant ORA

SC

feature ( 2.3 day periods) indicated that the KF accurately represented the initial data set. A significant 55-day cycle persisted in the KF results. Finally, a combined OMI, OMPS, AIRS SO2

NU

data set was tested and found to be the most efficient in removing the periodic effects of the ORA while still maintaining the 55-day cycle. Four periods of increased SO2 degassing were

MA

identified in the time series data (Figure 4a). Similar cyclical behavior was not observed in the MODIS data, indicating that the increased degassing was not due to new observable surface activity. Further analysis of this periodic degassing behavior should be completed in conjunction

TE

D

with seismic data to assess additional causes of the four observed peaks in SO2 emissions. The tropospheric release of SO2 during the 2012 – 2013 eruption of Plosky Tolbachik may have

AC CE P

had a short-lived cooling effect on the regional Arctic environment during the week after the start of the eruption. Any cooling effect would have been amplified by the initially slow conversion of SO2 to sulfate in the high latitude winter conditions. A week after the start of the eruption, most of the initial SO2 emission was converted to sulfate, though trace amounts of SO2 may not have been observed by AIRS. However, most of this conversion took place during the last two days of the week, suggesting that increased aerosol loading and/or weather systems over central Russia may have played an important role in the dispersal of the plume.

ACCEPTED MANUSCRIPT Acknowledgements

The authors acknowledge Fred Prata (Nicarnica Aviation) for generously providing his AIRS

PT

SO2 retrieval code and Kelly Wooten for contributing to the data analysis. We thank the NASAfunded SNPP Ozone Product Evaluation and Algorithm Test Element (PEATE) for providing

RI

OMPS Level 1B data, and Kai Yang for OMPS-NM SO2 retrievals. The NOAA Air Resources

SC

Laboratory (ARL) is acknowledged for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://www.ready.noaa.gov) used in this publication. The

NU

suggestions provided by the editor, Ben Edwards, and two anonymous reviewers improved the manuscript and are greatly appreciated. This work was funded by NASA through the

MA

MEaSUREs program (grant number NNX13AF50G) and an Earth and Space Science Fellowship

AC CE P

TE

D

to VJBF (grant number NNX14AK94H).

ACCEPTED MANUSCRIPT References Alavi, N., J.S. Warland and A.A. Berg, 2006. Filling gaps in evapotranspiration measurements for water budget studies: Evaluation of a Kalman filtering approach, Agricultural and

PT

Forest Meteorology, 141, 1, pp. 57-66, doi: 10.1016/j.agrformet.2006.09.011. Belousov A., M. Belousova, B. Edwards, A. Volynets and D. Melnikov, 2015. Overview of the

RI

precursors and dynamics of the 2012-13 basaltic fissure eruption of Tolbachik volcano,

SC

Kamchatka, Russia, Journal of Volcanology and Geothermal Research SI Tolbachik. Beyou, S., A. Cuzol, S.S. Gorthi and E. Memin, 2013. Weighted ensemble transform Kalman

NU

filter for image assimilation, Tellus Series A- Dynamic Meteorology and Oceanography, 65, doi: 10.3402/tellusa.v65i0.18803.

MA

Brown, R.G. and Hwang, P.Y.C., 1997. Introduction to Random Signals and Applied Kalman Filtering, Available at: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.147.3436. Carn, S.A. and G.J.S. Bluth, 2003. Prodigious sulfur dioxide emissions from Nyamuragira

D

volcano, D.R. Congo, Geophysical Research Letters, 30, 23, doi: 10.1029/2003GL018465.

Soufrière

Hills

TE

Carn, S.A., and A.J. Prata (2010). Satellite-based constraints on explosive SO2 release from volcano,

Montserrat,

Geophys.

Res.

Lett.,

37,

L00E22,

AC CE P

doi:10.1029/2010GL044971.

Carn, S.A. and T.M. Lopez, 2011. Opportunistic validation of sulfur dioxide in the Sarychev Peak volcanic eruption cloud, Atmos. Meas. Tech., 4, doi: 10.5194/amt-4-1705-2011. Carn, S.A., N.A. Krotkov, K. Yang and A.J. Krueger, 2013. Measuring global volcanic degassing with the Ozone Monitoring Instrument (OMI), Geological Society, London, Special Publications, 380, doi: 10.1144/SP380.12. Carn, S.A., K. Yang, A.J. Prata, and N.A. Krotkov, 2015. Extending the long-term record of volcanic SO2 emissions with the Ozone Mapping and Profiler Suite (OMPS) Nadir Mapper. Geophysical Research Letters, 42, doi:10.1002/2014GL062437. Chui, C. K., and Chen, G., 2009. Kalman Filtering with Real-Time Applications. Book (p. 229). doi:10.1007/978-3-540-87849-0. Costa, A., Melnik, O., Sparks, R. S. J., and Voight, B., 2007. Control of magma flow in dykes on cyclic lava dome extrusion. Geophysical Research Letters, 34, doi:10.1029/2006GL027466.

ACCEPTED MANUSCRIPT Costa, A., Wadge, G., Stewart, R., and Odbert, H., 2013. Coupled subdaily and multiweek cycles during the lava dome eruption of Soufriere Hills Volcano, Montserrat, Journal of Geophysical Research B: Solid Earth, 118, 1895–1903, doi:10.1002/jgrb.50095.

Trajectory)

Model

access

via

PT

Draxler, R.R. and Rolph, G.D. HYSPLIT (HYbrid Single-Particle Lagrangian Integrated NOAA

ARL

READY

Website

RI

(http://www.arl.noaa.gov/HYSPLIT.php). NOAA Air Resources Laboratory, College Park,

SC

MD.

Duchon, C, and R Hale. 2012. Time Series Analysis in Meteorology and Climatology: An

NU

Introduction. Advancing Weather and Climate Science, Wiley. Dvigalo,V.N., I.Yu. Svirid, A.V. Shevchenko, 2014. The first quantitative estimates of

Observations.

Journal

of

MA

parameters for the Tolbachik Fissure Eruption of 2012-‐2013 from Aerophotogrammetric Volcanology

10.1134/S0742046314050029.

and

Seismology,

8,

5,

261‐268,

doi:

D

Edwards, B., A. Belousov, M. Belousova, A. Volynets, D. Melnikov, S. Chirkov, S. Senyukov,

TE

E. Gordeev, Y. Muraviev, P. Izbekov and Y. Demianchuk, 2013. Another “Great Plosky Tolbachik” Eruption?, Eos, Transactions American Geophysical Union, 94, 21, doi:

AC CE P

10.1002/2013EO210002.

Fedotov, S. A., Markhinin, Y. K. (eds), 1983. The Great Tolbachik Fissure Eruption, 1975-1976. Cambridge University Press, New York, p. 341. Fedotov, S. A. (Ed.), 1984. The 1975-‐1976 Large Tolbachik Fissure Eruption in Kamchatka. Moscow, Nauka, p. 637 (in Russian). Flower, V.J.B. and S.A. Carn, in review. Synergistic assessment of A-Train data for the identification of volcanic cycles at Soufriere Hills Volcano, Montserrat, Jounral of Volcanology and Geothermal Research. Global Volcanism Program, 2012. Report on Plosky Tolbachik (Russia). In: Wunderman, R (ed.), Bulletin of the Global Volcanism Network, 37:12. Smithsonian Institution. http://dx.doi.org/10.5479/si.GVP.BGVN201212-300240. Global Volcanism Program, 2013a. Report on Klyuchevskoy (Russia). In: Wunderman, R (ed.), Bulletin

of

the

Global

Volcanism

Network,

http://dx.doi.org/10.5479/si.GVP.BGVN201307-300260.

38:7.

Smithsonian

Institution.

ACCEPTED MANUSCRIPT Global Volcanism Program, 2013b. Report on Sheveluch (Russia). In: Wunderman, R (ed.), Bulletin

of

the

Global

Volcanism

Network,

38:4.

Smithsonian

Institution.

http://dx.doi.org/10.5479/si.GVP.BGVN201304-300270.

PT

Global Volcanism Program, 2013c. Report on Plosky Tolbachik (Russia). In: Wunderman, R (ed.), Bulletin of the Global Volcanism Network, 38:8. Smithsonian Institution.

RI

http://dx.doi.org/10.5479/si.GVP.BGVN201308-300240.

SC

Gordeev, E.I., Y.D. Murav‟ev, S.B. Samoilenko, A.O. Volynets, D.V. Mel‟nikov and V.N. Dvigalo, 2013. The Plosky Tolbachik Fissure Eruption of 2012-2013: Preliminary Results,

NU

Doklady Earth Sciences, 452, 2, doi: 10.1134/S1028334X13100103. Hayes, M.., 1996. Statistical Digital Signal Processing and Modeling, John Wiley & Sons, Inc. 2010.

“MODVOLC

Near-Real-Time

MA

HIGP,

http://modis.higp.hawaii.edu/.

Monitoring

of

Global

Hot-Spots.”

Kalman, R.E., 1960. A New Approach to Linear Filtering and Prediction Problems. Transactions the

ASME-Journal

of

Basic

Engineering,

82,

pp.35–45.

Available

at:

D

of

TE

http://fluidsengineering.asmedigitalcollection.asme.org/article.aspx?articleid=1430402. Koeppen, W.C., M. Patrick, T. Orr, A.J. Sutton, D. Dow and R. Wright, 2013. Constraints on the

AC CE P

partitioning of Kilauea‟s lavas between surface and tube flows, estimated from infrared satellite data, sulfur dioxide emission rates, and field observations. Bulletin of Volcanology, 75, doi: 10.1007/s00445-013-0716-3. Krueger, A.J., N.A. Krotkov, K. Yang, G. Vicente and W. Schroeder, 2009. Applications of Satellite-Based Sulfur Dioxide Monitoring. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2, 4, doi: 10.1109/JSTARS.2009.2037334. Lopez, T., S. Carn, C. Werner, D. Fee, P. Kelly, M. Doukas, D. Schneider, 2013. Evaluation of Redoubt Volcano‟s sulfur dioxide emissions by the Ozone Monitoring Instrument, Journal of

Volcanology

and

Geothermal

Research,

259,

290–307,

doi:10.1016/j.jvolgeores.2012.03.002. Mann, M. E. and Lees, J. M., 1996. Robust estimation of background noise and signal detection in climatic time series, Climatic Change, doi:10.1007/BF00142586. Murphy, S. W., Wright, R., Oppenheimer, C. and Filho, C. R. S., 2013. MODIS and ASTER synergy for characterizing thermal volcanic activity, Remote Sensing of Environment, 131, 195–205, doi:10.1016/j.rse.2012.12.005.

ACCEPTED MANUSCRIPT NASA, 2013. MODIS. Retrieved August 25, 2013, from http://modis.gsfc.nasa.gov/. Nicholson, E. J., Mather, T. A., Pyle, D. M., Odbert, H. M. and Christopher, T., 2013. Cyclical patterns in volcanic degassing revealed by SO2 flux timeseries analysis: An application to

PT

Soufriere Hills Volcano, Montserrat, Earth and Planetary Science Letters, 375, 209–221, doi:10.1016/j.epsl.2013.05.032.

RI

Odbert, H. M. and Wadge, G., 2009. Time series analysis of lava flux, Journal of Volcanology

SC

and Geothermal Research, 188, 305–314, doi:10.1016/j.jvolgeores.2009.09.005. Oppenheimer, C., P. Francis and J. Stix, 1998. Depletion rates of sulfur dioxide in tropospheric plumes,

Geophysical

Research

Letters,

25,

14,

2671-2674,

doi:

NU

volcanic

10.1029/98GL01988.

MA

Percival, D. B., & Walden, A. T. (1993). Spectral analysis for physical applications. Cambridge University Press.

Prata, A. J. and C. Bernardo, 2007. Retrieval of volcanic SO2 column abundance from

TE

doi: 10.1029/2006JD007955.

D

atmospheric infrared sounder data, Journal of Geophyscial Research-Atmospheres, 112,

Rolph, G.D. Real-time Environmental Applications and Display sYstem (READY) Website

AC CE P

(http://www.ready.noaa.gov). NOAA Air Resources Laboratory, College Park, MD. Rose, W.I., G.J.S. Bluth and G.G.J. Ernst, 2000. Integrating Retrievals of Volcanic Cloud Characteristics from Satellite Remote Sensors: A Summary. Phil.Trans. Mathematical, Physical and Engineering Sciences, 358, 1770, 1585-1606. Seftor, C. J., G. Jaross, M. Kowitt, M. Haken, J. Li and L. E. Flynn, 2014. Postlaunch performance of the Suomi National Polar-orbiting Partnership Ozone Mapping and Profiler Suite (OMPS) nadir sensors. Journal of Geophysical Research- Atmospheres, 119, 4413– 4428, doi:10.1002/2013JD020472. Shirzaei, M. and Walter, T. R., 2010. Time-dependent volcano source monitoring using interferometric synthetic aperture radar time series: A combined genetic algorithm and Kalman filter approach, Journal of Geophysical Research, 115(B10), B10421, doi:10.1029/2010JB007476. Sparks, R. S. J. and Young, S. R., 2002. The eruption of Soufriere Hills Volcano, Montserrat (1995-1999): overview of scientific results, Geological Society, London, Memoirs, doi:10.1144/GSL.MEM.2002.021.01.03.

ACCEPTED MANUSCRIPT Spinei, E., S.A. Carn, N.A. Krotkov, G.H. Mount, K. Yang and A. Krueger, 2010. Validation of ozone monitoring instrument SO2 measurements in the Okmok volcanic cloud over Pullman, WA, July 2008. Journal of Geophysical Research: Atmospheres (1984–2012),

PT

115, doi: 10.1029/2009JD013492.

Thomas, H.E., I.M. Watson, S.A. Carn, A.J. Prata, and V.J. Realmuto (2011). A comparison of

RI

AIRS, MODIS and OMI sulphur dioxide retrievals in volcanic clouds, Geomatics, Natural

SC

Hazards and Risk, 2(3), 217-232, doi:10.1080/19475705.2011.564212. von Glasow, R., N. Bobrowski and C. Kern, 2009. The effects of volcanic eruptions on chemistry,

Chemical

Volcanology,

263,

131-142,

doi:

NU

atmospheric

10.1016/j.chemgeo.2008.08.020.

MA

Wright, R., Flynn, L., Garbeil, H., Harris, A. and Pilger, E., 2002. Automated volcanic eruption detection using MODIS, Remote Sensing of Environment, 82, 135–155, Retrieved from ://000177475900012.

D

Wright, R., Flynn, L. P., Garbeil, H., Harris, A. J. L. and Pilger, E., 2004. MODVOLC: Near-

TE

real-time thermal monitoring of global volcanism, Journal of Volcanology and Geothermal Research, 135, 29–49, doi:10.1016/j.jvolgeores.2003.12.008.

AC CE P

Wright, R., M. Blackett and C. Hill-Butler, 2015. Some observations regarding the thermal flux from Earth's erupting volcanoes for the period of 2000 to 2014. Geophysical Research Letters, 42, doi: 10.1002/2014GL061997. Yang, K., N.A. Krotkov, A.J. Krueger, S.A. Carn, P.K. Bhartia and P.F. Levelt, 2007. Retrieval of Large Volcanic SO2 Columns from the Aura Ozone Monitoring Instrument: Comparison and

Limitations.

Journal

of

Geophysical

Research

D:

Atmospheres,

112,

doi:10.1029/2007JD008825. Young, C.L., I.N. Sokolik and J. Dufek, 2012. Regional radiative impact of volcanic aerosol from the 2009 eruption of Mt. Redoubt. Atmospheric Chemistry and Physics, 12, p. 36993715, doi: 10.5194/acp-12-3699-2012. Zakšek, Klemen, Shirzaei, M. and Hort, M., 2013. Constraining the uncertainties of volcano thermal anomaly monitoring using a Kalman filter technique. Geological Society, London, Special Publications, 380(1), 137–160, doi:10.1144/SP380.5. Zelenski, M., N. Malik and Yu. Taran, 2014. Emissions of trace elements during the 2012 - 2013 effusive eruption of Plosky Tolbachik volcano, Kamchatka: enrichment factors, partition

ACCEPTED MANUSCRIPT coefficients and aerosol contribution. Journal of Volcanology and Geothermal Research,

AC CE P

TE

D

MA

NU

SC

RI

PT

285, p. 136-149, doi: 10.1016/j.jvolgeores.2014.08.007.

Fig. 1

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Fig. 2

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Fig. 3

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Fig. 4

AC CE P

Fig, 5

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC CE P

TE

D

Fig. 6

ACCEPTED MANUSCRIPT

TE

D

MA

NU

SC

RI

Tolbachik produced at least 140 kt of SO2 during the 2012-2013 eruption. Combining OMI, OMPS and AIRS data provided the most complete eruption coverage. A 55 day cycle in the eruption dynamics is identified. A satellite based lava effusion rate corroborates with the ground based estimate.

AC CE P

1. 2. 3. 4.

PT

Highlights Telling, J., V.J.B. Flower and S.A. Carn A multi-sensor satellite assessment of SO2 emissions from the 2012-13 eruption of Tolbachik volcano, Kamchatka