Problems and Shortcomings of Current Methods. Larry G. Mastin U.S. Geological Survey Cascades Volcano Observatory

Transcription

1 Problems and Shortcomings of Current Methods Larry G. Mastin U.S. Geological Survey Cascades Volcano Observatory

2 The two methods most commonly used 1. Empirical height-rate relationships Quick & easy But doesn t consider atmospheric effects Uncertainty is high (+/-10x) 2. 1-D plume models Considers atmospheric effects Almost as quick Uncertainty theoretically lower (but is it really?) Mastin et al., Sparks et al., Mastin, L. G., et al. (2009), JVGR, 186(1 2), 10-21, doi: 2 Sparks et al. (1997), Volcanic Plumes, Wiley & Sons, 574pp. Plumeria (Mastin) PlumeRise (Woodhouse)

3 To test, I use the Fortran version of Plumeria A 1-D steady-state model 1, which considers effects of Wind Humidity I then calculate MER, given a plume height, for specific, well observed eruptions where: Meteorology is known Plume height is measured MER is estimated independently by mapping the deposits I then compare its estimates with that of the empirical curve (Mastin et al., 2009). (Plumeria s results shouldn t differ dramatically from other models) Plume models: How much better can they estimate MER? MER Estimates by Plumeria are similar to other models Other 1-D models Plumeria Results of plume model comparison 2 1 Mastin, L. G. (2014), Testing the accuracy of a 1-D volcanic plume model in estimating mass eruption rate, Journal of Geophysical Research: Atmospheres, 119(5), 2013JD020604, doi: /2013jd Costa, A., et al. (2016), Results of the eruptive column model inter-comparison study, J. Volcanol. Geotherm. Res., doi:

4 1. The list of well-observed eruptions 1. I Start with those listed in Mastin et al. (2009) 1 (Table 1) 2. Then obtain wind, humidity, other met. Data from NCEP/NCAR Reanalysis 1 model 2 3. Remove pre-1948 eruptions for which met. data are not (easily) available. 1 Mastin, L. G., et al. (2009), JVGR, 186(1 2), 10-21, doi: 2 Kalnay, E., et al. (1996), The NCEP/NCAR 40-years reanalysis project, Bull Amer Meteor Soc, 77, doi:2.0.co;2.

5 2. Characterize the wind & meteorological conditions during these eruptions Wind profiles during these eruptions Characterize strength of wind field using a non-dimensional parameter, (e.g., Degruyter & Bonadonna, ) v = γ N =wind-speed gradient w/elevation N=avg. Brunt-Vaïsälä frequency 1 Degruyter, W., and C. Bonadonna (2012), GRL, 39(16), L16308, doi: /2012gl

6 Under No Wind H max =elevation where u z 0 Under High wind H min =max height of plume axis H max =H min +radius (R)? But in low-winds H min +R>H nowind (not real) So I define: H max =min(h min +R,H nowind ) I also define: Low wind: H min +R>H nowind High wind: H min +R>H nowind But wait! What is the top height of the plume? H eq Model representation No Wind Low wind High wind u z H obs H min H obs H obs z max H max=hnowind H max=h min+r H min r Gaps & overlaps In control volumes

7 Note: few plumes in this table are high wind (i.e. weak plumes ) Defining high-wind as H max +R>H nowind, or v > ~1 By these definitions, few plumes (4 of 25) are strongly affected by wind The weak plumes are the lowest (mostly H<10 km) Weak plumes Ruapehu 96

8 4. Given height, we calculate MER For Weak plumes Empirical curve (M obs ) underestimates MER for weak plumes Same for the plume model if it ignores wind (M nowind ) But the plume model with wind overestimates For other plumes (21 of 25) No systematic bias in any method Log calculated eruption rate, kg s M empir M nowind M max M min Weak plumes Log eruption rate based on mapping (M obs), kg s

9 5. Residuals: Which method is more accurate? 1. Both Plumeria and the empirical curve give mean errors near zero (no systematic bias) 2. But Plumeria w/o wind underestimates, for small (weak) plumes 3. Standard error for Plumeria (w/wind) and the empirical curve is about the same (not more accurate) Plumeria vs. observed MER overestimate underestimate Plumeria (nowind) vs. observed Empirical curve vs. observed

10 Most likely Answer: Empirical observations are not very accurate! Mastin et al., 2009

11 Many Plume height data have high uncertainties Uncertainty in height from radar at Eyjafjallajökull 1 In H, usually tens of percent, at least In deposit mass, usually +/-2x H can vary with time Visual Eyja, Radar El Chichón 2 1 Arason et al., 2011, Earth System Science Data, 4, Carey & Sigurdsson, 1986, BV., 48(2-3), Harris et al., 1981, USGS Prof. Paper 1250, Time variation in H at St. Helens, isopleths

12 Ambiguities in height are not considered For high plumes, max Plume heights are assumed But isopleth derived heights are more likely from the umbrella Low plumes come from days-long eruptions of Cerro Negro, Etna. H is a very approximate average or midpoint in the range. Time variations are oversimplified St. Helens, 5/18/80 H derived from isopleths Carey & Sparks, Pinatubo Top height used 1 Carey, S., & Sparks, R.S.J., (1986), Bull. Volc., 48:

13 Uncertainty in erupted mass can be +/-2x Deposits used In height-rate relation Differences in deposit volume calculated using different methods of extrapolation, by Bonadonna & Costa (2012) Bonadonna, C., and A. Costa (2012), Estimating the volume of tephra deposits: A new simple strategy, Geology, 40(5), , doi: /g

14 Can a 1-D plume model do better for a really well observed eruption? Mastin, 2014 For Eyjafjallajökull, yes model Devenish, 2016 empirical Plume height above vent, km Eruption rate, kg/s days since 14 April 2010, 0000 UTC a max + b Gudmundsson et al., 2012 Plumeria w/wind max Mmin-Mmax M nowind empirical 10 2 M empir days since 14 April 2010, 0000 UTC Devenish, B. J. (2016), Estimating the total mass emitted by the eruption of Eyjafjallajökull in 2010 using plume-rise models, J. Volcanol. Geotherm. Res., doi: /j.jvolgeores Mastin, L. G. (2014), Testing the accuracy of a 1-D volcanic plume model in estimating mass eruption rate, Journal of Geophysical Research: Atmospheres, 119(5), 2013JD020604, doi: /2013jd Plus others by Woodhouse et al. (2013), Bursik et al. (2012), Kaminski et al. (2011)

15 How can we improve? 1. We can add new (better?) data 2. These data shift the height curve down slightly 2009 curve Shifts height curve down At lower MER Calbuco Chaitén Eyja phases 1, 3 Redoubt 2009 Shinmoedake

16 How does a new empirical curve affect the estimated MER at Eyja? Increases erupted volume ~50%, but not dramatically Mapped Modeled Phase I erupted volume estimated by summing MER over 6-hour time periods using different MER estimates. Plume heights used for estimates are + for each 6-hour period.

17 2. We can examine (and perhaps remove) lower quality data Removing isopleth-based heights reduces constraints on the high end Should we work toward improvements? Yes! But the basic limitations will remain. Santa Maria Quizapu Nevado del Ruiz Novarupta El Chichón 2009 curve

18 Should we consider other indicators of MER? Umbrella cloud method is just now being studied 1,2 Differences of ~2x in MER are theoretically resolvable using this method But it requires An umbrella cloud A geostationary satellite Rapid image repeat times The Kelut umbrella cloud in VIIRS Ash3d simulation Theoretical Growth curves (r t 2/3 ) 1 Pouget, S., M. Bursik, C. G. Johnson, A. J. Hogg, J. C. Phillips, and R. S. J. Sparks (2015), Interpretation of umbrella cloud growth and morphology: implications for flow regimes of short-lived and long-lived eruptions, Bull. Volcanol., 78(1), doi: /s Pouget, S., M. Bursik, P. Webley, J. Dehn, and M. Pavolonis (2013), Estimation of eruption source parameters from umbrella cloud or downwind plume growth rate, J. Volcanol. Geotherm. Res., 258(0), , doi: Work in progress with Alexa Van Eaton

19 Conclusions 1. MER estimation methods must be evaluated, by comparing with independent data 2. Independent data are currently not good enough to provide convincing validation 3. Priority should be placed on acquiring high quality data on plume height, and erupted mass.

20 Thorstein V. Jonsson photo, IMO The End