DB 2172 © British Crown Owned Copyright 2011/MOD 1. AWE Blacknest, Brimpton, United Kingdom RG7 4RS ([email protected]) 2. CEA/DAM/DIF, F-91297 Arpajon, France Acoustic Observations of Stratospheric Solar Tides: Examples from the Eruption of Eyjafjallajökull, Iceland, April-May 2010 David N. Green 1 , Julien Vergoz 2 , Robin S. Matoza 2 , Alexis Le Pichon 2 200° 180° 160° 140° 100° 120° 70° 60° 50° 40° 30° –40° –20° 0° 20° 20° 30° 40° 1. Signal Detection • The April - May 2010 summit eruption of Eyjafjallajökull volcano, Iceland, was recorded across 14 infrasound arrays, including 4 arrays of the International Monitoring System (IMS) network (Figure 1). • Array processing techniques can successfully discriminate between volcanic infrasound and ambient coherent and incoherent noise (Figure 2). • Quasi-continuous timeseries of detections over a number of days provide an opportunity to study variations in detection parameters (backazimuth, apparent velocity, signal frequency content, and signal amplitude) on diurnal timescales. • Timeseries recorded at stations located along distinct propagation paths exhibit clearly varying detection parameters on diurnal timescales (Figures 2 and 3, Section 2). Figure 1. Location of infrasonic stations recording the April–May 2010 summit eruption of Eyjafjallajökull. 14 remote infrasonic arrays (green inverted triangles) recorded the summit eruption of Eyjafjallajökull (black dot, ‘Eyjaf’), at distances from ~1,745 km (BKNI, UK) to ~3,666 km (IS48, Tunisia). Blue inverted triangles show other IMS stations that did not record the eruption. Colorscale represents number of intersecting mean infrasonic signal backazimuths ±3°registered at each station and associated with Eyjafjallajökull [from Matoza et al., (2011)]. Figure 2. Eyjafjallajökull associated detections: identification and timeseries. The upper plots show results for BKNI, with the right hand panel showing a 2D histogram of detections in frequency-azimuth space over the period 2010/4/10 to 2010/6/15 inclusive. The area enclosed within the white dashed line indicates the detections identified as associated with the Eyjafjallajökull eruption. The larger timeseries plot shows 10 days of Eyjafjallajökull associated arrival amplitudes between 18/04/2010 and 28/04/2010, with the median amplitude in 30 minute bins provided as a red line. The inset timeseries shows the extent of the detections across April and May 2010, with the grey shaded area indicating a period of data loss. The lower three panels show results for IS18. The map shows the relative array locations and configurations. Figure 3. Identifying periodicities in the unevenly sampled detection timeseries at BKNI, using the CLEAN algorithm (an iterative deconvolution of the sampling function, e.g., Heslop and Dekkers, 2002). Methodology allows the significance of the spectral peaks (left-hand panels) to be determined. On the right-hand side reconstructed signals from the CLEAN spectra at the 80% and 95% confidence levels are overlain on detection density plots for (top to bottom) signal RMS amplitude, backazimuth, apparent velocity (V app ) and central signal frequency. Darker shades represent higher detection densities. At BKNI, signal amplitudes and backazimuth estimates exhibit clear diurnal variations. 2. Identifying Periodicities • To identify the frequency of variations in detection parameters, a technique that is applicable to non-evenly sampled timeseries is required (Figure 3). • Significant diurnal variations in signal amplitude and backazimuth are observed at BKNI, and in amplitude, apparent speed and signal frequency observations at IS18. • At BKNI variations in the signal amplitude are in phase with stratospheric along-path wind variations, whereas backazimuth variations are approximately in phase with cross- wind variations (Figure 4). 3. Cause of Variations - Stratospheric Tides? • Significant diurnal variations in detection characteristics are observed – what generates these variations? Two plausible sources: 1. Diurnal stratospheric wind variations caused by the solar diurnal tide leading to periodic variations in acoustic waveguide characteristics (e.g., Donn et al., 1975). 2. Variations in near-receiver (boundary layer) atmospheric stability leading to differences in detection characteristics and pressure noise levels. • Observed diurnal variations in backazimuth are indicative of a process that must act over a significant propagation path length – suggesting tidal motions in stratosphere as source of observed variations (Figure 5). 4. Propagation Modelling • 3D ray-tracing has been used to identify acoustic propagation paths through ECMWF meteorological profiles between 2010/04/18 and 04/30 (Figure 6). • On path from Eyjafjallajökull to BKNI the study of how stratospheric tides influence the propagation is complicated by the presence of strong tropospheric ducts. For time period studied, 33% of rays propagated through only a stratospheric waveguide. • On path from Eyjafjallajökull to IS18 93% of rays propagate within the stratospheric duct. The ray density reaching the station exhibits a diurnal variation (Figure 6) with the correct phase to explain the observed signal amplitude and detection density variations. 5. Conclusions • Infrasound signals clearly recorded from Eyjafjallajökull eruption at ranges of up to 3600km. • Detections exhibit diurnal variations in signal characteristics – and these are correlated with diurnal variations in ECMWF-model stratospheric wind speeds. • Identifying diurnal components in results of ray-tracing is difficult, due to complications arising from spatial and temporal meteorological variations. References Donn,W. L., N. K. Balachandran, and D. Rind (1975), Tidal Wind Control of Long-Range Rocket Infrasound, J. Geophys. Res., 80 (12), 1662–1664. Gudmundsson, M. T., et al. (2010), Eruptions of Eyjafjallajökull Volcano, Iceland, Eos, 91 (21), 190–191. Heslop, D., and M. J. Dekkers (2002), Spectral Analysis of Unevenly Spaced Climatic Time Series using CLEAN: Signal Recovery and Derivation of Significance Levels using a Monte Carlo Simulation, Phys. Earth Planet. Int., 130, 103–116. Matoza, R. S., et al. (2011b), Long-range Acoustic Observations of the Eyjafjallajökull Eruption, Iceland, April-May 2010, Geophys. Res. Lett., 38 (L06308), doi:10.1029/2011GL047019. Figure 6. Effects of temporal and spatial variations in the effective sound speed on acoustic propagation predictions, for paths from Eyjafjallajökull to a) BKNI and b) IS18. The large left-hand panels show the temporal variation in along- path effective sound speed, while the right hand panels show along-path spatial variability in effective sound speed between the source (range=0km) and the station (white triangle) at times shown on the left-hand panel. The times chosen for BKNI highlight the complications generated by tropospheric waveguides (time B) and elevated stratospheric waveguides (time C). The times chosen for IS18 show the strengthening of the stratospheric waveguide across one day, reflected in the number of rays reaching the station. Here, the changes in waveguide strength are related to the stratospheric tidal variations. ECMWF model wind speeds at 45km altitude 12:00 00:00 12:00 Hour of Day (U.T.) 12:00 00:00 12:00 Hour of Day (U.T.) Along-Path Wind Cross-Path Wind Wind speed - mean[Wind Speed] (m/s) –1.*Wind speed - mean[Wind Speed] (m/s) 0.05 0.04 0.03 0.02 0.01 0 0.04 0.03 0.02 0.01 0 RMS Amp (Pa) RMS Amp (Pa) RMS Amp (Pa) 04/18 04/20 04/22 04/24 04/26 04/28 04/10 04/30 05/20 Date 2010. (mm/dd) 04/18 04/20 04/22 04/24 04/26 04/28 Date 2010. (mm/dd) Date 2010. (mm/dd) 0.02 0 RMS Amp (Pa) 04/10 04/30 05/20 Date 2010. (mm/dd) 0.05 0 1 0.5 0 –0.5 –1 Log 10 Freq. (Hz) 1 0.5 0 –0.5 –1 Log 10 Freq. (Hz) 270 315 0 45 Azimuth (°) 45 90 135 180 Azimuth (°) 40 30 20 10 0 No. of Detections Modulus (x10 –3 ) 1 0.5 0 0 1 2 3 4 5 Modulus (x10 –3 ) 0.6 0.4 0.2 0 0 1 2 3 4 5 Modulus (x10 –3 ) 2 1 0 3 2 1 0 1 2 3 4 5 Modulus (x10 –3 ) 0.1 0.05 0 0.02 0.01 0 0 1 2 3 4 5 Freq (cpd) Freq (Hz) Amp (Pa) Azimuth (°) 04/18 04/19 04/20 04/21 04/22 04/18 04/19 04/20 04/21 04/22 04/18 04/19 04/20 04/21 04/22 04/18 04/19 04/20 04/21 04/22 Date 2010. (mm/dd) 336 334 332 330 328 0.38 0.36 0.34 0.32 V app (km/s) 95 80 95 80 95 80 95 80 a) BKNI b) IS18 120 100 80 60 40 20 0 120 100 80 60 40 20 0 Altitude (km) Altitude (km) Altitude (km) A B C 04/17 04/22 04/27 Date 2010. (mm/dd) Eff. Sound Speed (km/s) 0.29 0.32 0.35 A B C 04/17 04/22 04/27 Date 2010. (mm/dd) Eff. Sound Speed (km/s) 0.29 0.32 0.35 80 60 40 20 0 0 500 1000 1500 80 60 40 20 0 0 500 1000 1500 80 60 40 20 0 0 500 1000 1500 Range (km) Altitude (km) 80 60 40 20 0 0 500 1000 1500 80 60 40 20 0 0 500 1000 1500 80 60 40 20 0 0 500 1000 1500 Range (km) A :2010-04-17 21:00 B :2010-04-20 06:00 C :2010-04-25 12:00 A :2010-04-21 00:00 B :2010-04-21 09:00 C :2010-04-21 18:00 120 100 80 60 40 20 0 Altitude (km) 0 20 40 60 80 100 Max. C eff Occillation (m/s) 120 100 80 60 40 20 0 Altitude (km) 0 5 10 15 20 Time of C eff Maximum (Hour) a) b) Figure 4. Superimposed epoch analysis of amplitude and backazimuth estimates at BKNI for 2010/04/16-05/07, showing the diurnal cycle in both parameters (upper panels). For comparison, superimposed epoch analyses of ECMWF meteorological model winds are given for the same period at altitudes of 45km (lower panels). The along-path stratospheric winds (left-hand panel) vary in phase with the amplitude variations and the cross-path winds (right-hand panel) vary approximately in phase with the backazimuth variations. Amp. Azi. Data 12:00 18:00 00:00 06:00 12:00 Hour of Day (U.T.) 12:00 18:00 00:00 06:00 12:00 Hour of Day (U.T.) 6 4 2 0 –2 4 2 0 –2 x 10 –3 Amp.-mean[Amp] day (Pa) Azi.-mean[Azi] day (°) Figure 5. The (a) amplitude and (b) phase of the daily oscillations in the effective acoustic speed along the Eyjafjallajökull to BKNI path, taken from the ECMWF meteorological model for 15/04 - 30/04/2010. The diurnal oscillations are generated by solar tides: global-scale atmospheric oscillations excited by solar insolation, observed as periodic variations in temperature, density and windspeed.