Atmos. Chem. Phys., 21, 12189–12206, 2021 https://doi.org/10.5194/acp-21-12189-2021 © Author(s) 2021. This work is distributed under the Creative Commons Attribution 4.0 License. Geometric estimation of volcanic eruption column height from GOES-R near-limb imagery – Part 1: Methodology Ákos Horváth1, James L. Carr2, Olga A. Girina3, Dong L. Wu4, Alexey A. Bril5, Alexey A. Mazurov5, Dmitry V. Melnikov3, Gholam Ali Hoshyaripour6, and Stefan A. Buehler1 1Meteorological Institute, Universität Hamburg, Hamburg, Germany 2Carr Astronautics, Greenbelt, MD, USA 3Institute of Volcanology and Seismology, Far East Branch of the Russian Academy of Sciences (IVS FEB RAS), Petropavlovsk-Kamchatsky, Russia 4NASA Goddard Space Flight Center, Greenbelt, MD, USA 5Space Research Institute of the Russian Academy of Sciences (SRI RAS), Moscow, Russia 6Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany Correspondence: Ákos Horváth ([email protected], [email protected]) Received: 21 February 2021 – Discussion started: 23 March 2021 Revised: 2 July 2021 – Accepted: 6 July 2021 – Published: 16 August 2021 Abstract. A geometric technique is introduced to estimate atmospheric dispersion models, which require the eruptive the height of volcanic eruption columns using the gener- source parameters, especially plume height and the mass ally discarded near-limb portion of geostationary imagery. eruption rate (MER), as key inputs (Peterson et al., 2015). Such oblique observations facilitate a height-by-angle esti- Plume height and MER are related by dynamics, and the lat- mation method by offering close-to-orthogonal side views ter scales approximately as the fourth power of the former. of eruption columns protruding from the Earth ellipsoid. Thus, a small error in plume height leads to a large error Coverage is restricted to daytime point estimates in the im- in MER, estimates of which can consequently have a fac- mediate vicinity of the vent, which nevertheless can pro- tor of 10 uncertainty (Bonadonna et al., 2015). The mass vide complementary constraints on source conditions for the eruption rate is commonly estimated from plume height ob- modeling of near-field plume evolution. The technique is servations using semiempirical relationships, Sparks–Mastin best suited to strong eruption columns with minimal tilt- curves, derived from buoyant plume theory and historical ing in the radial direction. For weak eruptions with severely eruption data (Mastin et al., 2009; Sparks et al., 1997). An bent plumes or eruptions with expanded umbrella clouds alternative is to use simplified 1D cross-section-averaged the radial tilt/expansion has to be corrected for either visu- (Folch et al., 2016) or 2D Gaussian (Volentik et al., 2010) ally or using ancillary wind profiles. Validation on a large plume rise models, which can be inverted efficiently to esti- set of mountain peaks indicates a typical height uncertainty mate MER from plume height. of ±500 m for near-vertical eruption columns, which com- Many techniques have been developed over the years to pares favorably with the accuracy of the common tempera- measure volcanic plume height (for a comparative overview ture method. see Dean and Dehn, 2015; Merucci et al., 2016; Zakšek et al., 2013; and references therein). Ground-based methods rely on weather radars, lidars, or video surveillance cameras. Space- based methods include radar and lidar observations, radio oc- 1 Introduction cultation, backward trajectory modeling, geometric estimates from shadow length and stereoscopy, and radiometric esti- Volcanic eruptions pose significant hazards to aviation, pub- mates utilizing the CO2 and O2 absorption bands and infrared lic health, and the environment (Martí and Ernst, 2005). Risk (IR) channels. assessment and mitigation of these hazards is supported by Published by Copernicus Publications on behalf of the European Geosciences Union. 12190 Á. Horváth et al.: Geometric estimation of volcanic eruption column height – Part 1 The height retrieval technique offering the best spatial plumes compared to lidar or stereo heights generally consid- and temporal coverage globally is the spaceborne “temper- ered the most accurate (Flower and Kahn, 2017; Pavolonis et ature method”, which is based on IR brightness temperatures al., 2013; Thomas and Siddans, 2019). (BTs) routinely available from a large suite of imaging ra- Globally applicable near-real-time techniques, such as diometers aboard both polar orbiter and geostationary satel- the satellite temperature method, are nevertheless indispens- lites. In its simplest and still oft used single-channel form, able to support operations at volcanic ash advisory cen- the method determines plume height by matching the 11 µm ters and mitigate aviation and health hazards. The single- BT to a temperature profile obtained from a radiosounding or channel temperature method is part of the VolSatView infor- a numerical forecast, assuming an opaque plume in thermal mation system (Bril et al., 2019; Girina et al., 2018; Gordeev equilibrium with its environment. Both of these assumptions, et al. 2016) operated by the Kamchatka Volcanic Erup- however, can be invalid. tion Response Team (KVERT; Girina and Gordeev, 2007). The plume tops of the largest explosions, especially ones The multichannel retrievals form the core of the VOLcanic that penetrate the stratosphere, might be in thermal disequi- Cloud Analysis Toolkit (VOLCAT) and are produced from librium due to decompression cooling in a stably stratified Geostationary Operational Environmental Satellite-R Series atmosphere. Undercooling can lead to a cloud top that is tens (GOES-R) and Himawari-8 imager data by the National of degrees colder than the minimum temperature of the sur- Oceanic and Atmospheric Administration (NOAA) and the rounding ambient, in which case the satellite-measured BT Japanese Meteorological Agency (JMA). The pursuit of new cannot be converted to height (Woods and Self, 1992). Ther- techniques is still worthwhile though, given the large uncer- mal disequilibrium of the opposite sign might also occur be- tainty of existing retrieval algorithms (von Savigny et al., cause the increased absorption of solar and thermal radiation 2020). by volcanic ash can cause significant local heating (Muser et Our proof-of-concept study introduces a simple geomet- al., 2020), resulting in negatively biased height retrievals. ric technique to derive point estimates of eruption column A more common problem is that the plume, especially its height in the vicinity of the vent from side views of the plume dispersed part further from the vent, is semitransparent to captured in near-limb geostationary images. In planetary sci- IR radiation and deviates strongly from blackbody behavior; ence, topography is often estimated by the radial residuals hence, the 11 µm BT is warmer than the effective radiative to a best-fit ellipsoid along a limb profile. Such limb topog- temperature. Surface contribution to the measured BT leads raphy was derived for Io (Thomas et al., 1998), saturnian to underestimated plume heights (Ekstrand et al., 2013). This icy satellites (Nimmo et al., 2010; Thomas, 2010), and Mer- low bias can be somewhat reduced by using only the mini- cury (Oberst et al., 2011), to mention a few. Limb images mum (dark pixel) BT of the plume least affected by surface were also used to estimate the height of ice and dust clouds radiation. on Mars (Hernández-Bernal et al., 2019; Sánchez-Lavega A more sophisticated treatment of semitransparency ef- et al., 2015, 2018) and even the height of volcanic plumes fects, however, requires BTs from multiple IR channels. on Io (Geissler and McMillan, 2008; Spencer et al., 2007; Pavolonis et al. (2013) developed a volcanic ash retrieval Strom et al., 1979). Closer to home, near-limb images from based on the 11 and 12 µm split-window channels and the geostationary satellites were used to reconstruct the atmo- 13.3 µm CO2 absorption band, with the latter providing the spheric trajectory of the 2013 Chelyabinsk meteor (Miller et needed height sensitivity for optically thin mid- and high- al., 2013) and study the altitude of polar mesospheric clouds level plumes. The algorithm solves for the radiative tempera- (Gadsden, 2000a, b, 2001; Proud, 2015; Tsuda et al., 2018). ture, emissivity, and a microphysical parameter of the plume Apart from these two applications, however, the near-limb by optimal estimation. These parameters are then used to de- portion of geostationary images is completely unused for any termine plume height, effective ash particle radius, and mass quantitative geophysical analysis. loading for four different mineral types (andesite, rhyolite, Here we exploit exactly these oblique observations, which gypsum, and kaolinite) to account for uncertainty in chemi- provide side views at almost a right angle of volcanic erup- cal composition. tion columns protruding from the Earth ellipsoid. The pro- All brightness-temperature-based height retrievals are posed height-by-angle technique is based on the finest- however problematic near the tropopause due to the char- resolution daytime visible channel images, and it is analo- acteristic temperature inversion. Small lapse rates and gous to the astronomical height retrievals and the height esti- nonunique solutions lead to a significantly increased height mation methods from calibrated ground-based video camera uncertainty. Over- and underestimation are both possible footage (Scollo et al., 2014).