eruptive and passive degassing of sulphur dioxide at nyiragongo ...

S. A. Carnmodels of degassing. From October 2002 until the timeof writing, toms has continued to detect discharge ofSO 2from the active lava lake established at Nyiragongofollowing the 17 January 2002 eruption, correlated withthe emission of a vigorous plume from the volcano.The implications of over 1 year of continuous SO 2emissions for magma supply at Nyiragongo are alsodiscussed.Volcanic sulphur dioxide measurements fromspacetoms is an ozone sensor but it can also measure SO 2asboth gases absorb at the ultraviolet (uv) wavelengthsdetected by the instrument. The first toms instrumentwas launched in 1978 but its ability to detect volcanicSO 2was not realized until 1982 (Krueger 1983). Followingthis discovery, and with the launch of 3 further tomsinstruments, toms data for all major volcanic eruptionswere analyzed to create the first long-term record ofglobal volcanic SO 2emissions (Bluth et al. 1993), whichnow covers the period from October 1978 onwards withthe exception of a 18-month data gap in 1995-1996(Carn et al. 2003). For the present paper the primarysource of data was Earth Probe (ep) toms, which waslaunched in July 1996 and continues to operate at thetime of writing, collecting data once per day at aroundlocal noon. ep toms does not provide complete dailycoverage at the equatorial latitudes of Nyiragongo, andhence data gaps often cover the volcano on alternatedays. However, the locations of the data gaps vary anddaily coverage is achieved at certain times. The techniquesused to retrieve SO 2from toms data and calculateSO 2burdens, along with an error analysis and thelimitations of the toms measurements, are given inKrueger et al. (1995). A typical uncertainty on toms SO 2cloud tonnage measurements is ± 30% (Krueger et al.1995).The minimum amount of SO 2detectable by the tomssensor depends on eruption, background and meteorologicalconditions at the time of the measurement. The35 across-track scan positions sampled by ep toms correspondto target areas of ~1550-12760 km 2 on theEarth’s surface, with the minimum SO 2mass consideredabove noise in one pixel varying from ~460-3820tons. However, to produce an unambiguous SO 2cloudthat is easily separable from background noise a clusterof several pixels close to the volcano is normally required,giving a practical SO 2cloud detection limit of~4-32 kilotons (kt) depending on scan position. Cloudaltitude is also a factor; toms was designed forstratospheric measurements and detection of troposphericSO 2is affected by weaker uv absorption by theSO 2molecule at higher temperatures, increased uvscattering by the dense lower tropospheric atmosphere,and possible interference by uv-reflecting water cloudssituated between the sensor and a low-level SO 2cloud.Nevertheless, targets such as the SO 2emissions fromNyiragongo from October 2002 onwards (discussed later)demonstrate that ep toms is eminently capable of detectingand quantifying lower tropospheric SO 2.In addition to ep toms there are several other satellitesensors currently in orbit that are capable of measuringSO 2, including the Moderate Resolution ImagingSpectroradiometer (modis; on the eos Terra and Aquasatellites), the High-resolution Infrared RadiationSounder (hirs/2; on noaa polar-orbiting satellites) andthe Atmospheric Infrared Sounder (airs; on eos Aqua).These instruments exploit infrared (ir) SO 2absorptionfeatures around 7.3 mm (e.g., Prata et al., 2003) and 8.6mm (e.g., Realmuto 2000) and provide greater temporaland spatial resolution (~2 images per day; ~1-20 kmpixel size at nadir) than ep toms (1 image per day atthe equator; 39 km pixel size at nadir). The relativeperformance of the various sensors is not elaborated onhere (see Rose et al. [2003] for an example of a multisensorstudy of a volcanic eruption), but in the tropicalenvironment of central Africa the ir sensors are generallyless effective than the uv toms instrument for SO 2measurements, largely due to high atmospheric watervapour loadings which interfere with SO 2retrievals inthe ir. Daily modis, hirs/2 and airs data are availablefor the period of unrest at Nyiragongo discussed in thispaper and future studies using these data may providefurther insights into the degassing activity, although aninitial scrutiny of modis and airs data for selected datesrevealed no significant SO 2signals. Hence all SO 2measurementspresented herein are derived from ep toms,with supplementary data on volcanic cloud characteristicsobtained from modis data. Observations of SO 2made by the Global Ozone Monitoring Experiment(gome), a uv instrument on the European ers-2 satellite,were also used to track degassing at Nyiragongofrom October 2002 until mid-June 2003.Historical activity and degassing atNyiragongoWhilst no prior remote sensing studies of gas emissionsfrom Nyiragongo exist, the volcano has been visited orobserved on a reasonably frequent basis throughout theperiod since its discovery. Table 1 provides a summaryof historical observations of degassing at Nyiragongo.These are largely qualitative, derived from sporadicactivity reports and examination of photographs inpublished works, but they help to place the 2002 eruptionand the subsequent degassing crisis in context.As would be expected, degassing vigour at Nyiragongoappears to correlate broadly with the activity of the lavalake, with increased degassing associated with more active,molten lakes and reduced emissions when the lake iscrusted over (Table 1). Reviewing his numerous observationsat Nyiragongo from 1948-1982, Tazieff (1994)notes «the difference between the thin, small fumes thatusually drifted above the volcano’s summit and the mightyplume towering high over it when a ‘lava tempest’ wasdeveloping». However, the often long periods betweencrater visits, coupled with the ability of the lava lake tofluctuate in size and level on short timescales (e.g., Tazieff1994) and frequent cloud cover obscuring the volcano,preclude a full assessment of temporal trends in degassingat Nyiragongo. Prior to 2002-2003, a previous peak in gasemissions appears to have occurred in the early to mid-2