Explosions in the Stratosphere Test Atmospheric Sound Propagation Models

P. Golden, E.T. Herrin, P. Negraru, H. Bass, W. Andre, M. McKenna

Southern Methodist University Department of Geological Sciences,
National Center for Physical Acoustics,
US Army Space and Missile Defense Command,
US Army Engineer Research and Development Center,

12 February 2007

  1. Introduction
  2. Historical Perspective
  3. Motivation
  4. The Experiments
  5. Future Directions
  6. Acknowledgements
  7. References

1. Introduction

An infrasound research team has been established to conduct large-scale experiments utilizing atmospheric explosions to study infrasound propagation phenomenology and to calibrate a network of infrasound arrays.In the last two years three calibration experiments were carried out at White Sands Missile Range (WSMR), during which explosions with yields of 70 pounds TNT were conducted at altitudes greater than 30 km.Infrasound data from the explosions were collected by team members at many locations in the southwest US.†† The resulting waveforms represent a truly unique infrasound dataset in terms of well-known source location and yield. The primary goal of the project is to improve understanding of the fundamental physics of the generation of infrasound from explosions and the atmospheric dynamics affecting propagation of infrasound signals.Results indicate that the phase of an infrasound signal can be strongly distorted by short-term (hours) dynamics of the atmosphere.Additionally, established period/yield scaling relationships fail for sources above 40 Km.


2. Historical Perspective

From 1960 to the early 1970ís, NASA conducted a series of rocket grenade experiments to produce infrasound as a means to study the mesosphere.In the initial experiment 28 sounding rockets were launched from 1960 to 1963.Temperature, pressure, density and wind measurements were derived from the received infrasound signals.During this early period, the grenades were the equivalent of 1 or 2 pounds of high explosives.The charges were ejected and exploded at 4 to 6 km intervals to profile the atmosphere from ~30 to ~90 km.Radar and other ranging systems were used to measure the location of the explosions.The time of the explosions was recorded by photocells on the payload.Later, charges from 5 to 10 pounds were used up to 50 km, and 2 to 3 pound grenades from 50 to 90 km.


The frequency of infrasound signals received varied with altitude of the explosions, signals below 1 Hz were received from the higher explosions.These experiments were all reported in a series of NASA Technical Reports from 1963 onward.The current experiments utilize explosions at high altitude with small charges to produce infrasonic waves for confirmation of the performance and calibration of infrasound arrays.Small charges exploded at high altitudes can produce infrasound very well.


3. Motivation

The Columbia disaster of February 1, 2003, was recorded at Lajitas TX when the space shuttle generated infrasound as it traversed the continent to land in Florida.Previous passes had given researchers at SMU distinct patterns of normal behavior.However, on February 1, 2003, the recorded signals were vastly different from previous passes.We concluded that the signals were multipath arrivals from a single explosive event aboard Columbia.


It became apparent that an investigation into similar sources, atmospheric explosions, would be useful in understanding the event and modeling similar results (McKenna and Herrin, 2006).In order to estimate the size of the explosions we used an empirical period/yield formula developed for surface explosions and corrected for altitude using the method of Armstrong (1998).NASA data confirmed that the Columbia event occurred at a height of just over 63 Km.Our analysis concluded that the acoustic yield of the shuttle explosion was between 2 and 3 lbs TNT equivalent.This result prompted conversations with current infrasound researchers concerning an effort to investigate the possibility of conducting explosion experiments with well-controlled source yield and location.†† The US Space and Missile Defense Command undertook responsibility for the efforts and worked with the Navy at WSMR to develop the explosive charges and rocket delivery systems.A team of University, National Laboratory, private corporation and Government researchers was put together to plan and execute the experiments.††

Team members include the authors from Southern Methodist University, The University of Mississippi, the US Space and Missile Defense Command and the Army Engineer Research and Development Center.Additional team members are The University of Alaska Fairbanks, Dr. John Olson and Dr. Daniel Osborne; The University of California San Diego, Dr. Michael Hedlin; The University of Hawaii Manoa, Dr. Milton Garces; BBN Technologies, Dr. David Norris; Los Alamos National Laboratory, Dr. Rodney Whitaker and the US Naval Research Laboratory, Dr. Douglas Drob.



4. The Experiments

Atmospheric modeling played a key role in determining the schedules and locations of portable deployments of infrasound arrays for the explosion experiments.Figure 1 illustrates atmospheric modeling showing seasonal trends in effective sound speed profiles at the WSMR source site.These profiles are based entirely on climatology and attempt to resolve mean seasonal and diurnal atmospheric trends.In the Northern Hemisphere, strong westward ducts form during the summer months, with a peak in July.To take advantage of the possibility of strong ducting, which could provide increased distance for recording signals to the west, the third experiment was conducted in July 2006.


Figure1. Top: Map showing locations and distances from the source for infrasound arrays and stations deployed during the WSMR1 explosion experiment. Bottom: Effective sound speed profiles used in selecting optimal test dates.


Based on the expected climatic conditions, three infrasound calibration experiments were conducted at White Sands Missile Range, New Mexico on September 9, 2005 (WSMR1), March 25 (WSMR2) and July 21, 2006 (WSMR3).During each experiment two rockets were launched at approximately 4 hour intervals.Although plans were for detonations at 50 km altitude, WSMR1 detonations were limited to approximately 30 km due to Range Safety concerns.†† WSMR2 detonations were allowed at an altitude of approximately 35 km after additional debris modeling was completed. WSMR3 detonations were at 42 and 49 km respectively.The main goal of the experiments is to provide further understanding of the atmosphere propagation of infrasound signals during different atmospheric conditions. A second goal was validating the yield/dominant period scaling relationship. Figure 1 shows the location of the permanent and temporary infrasound arrays deployed for WSMR1 in September 2005. In total there were 19 arrays or stations at ranges from 63 to 2049 km. For WSMR2, 22 stations were used, and 23 were used for WSMR3 covering approximately the same distance ranges.The temporary infrasound arrays were placed mostly west of the source for the WSMR1 and WSMR3 experiments, and east of the source for the WSMR2 experiment.This pattern was chosen in agreement with the direction of the zonal stratospheric winds.The zonal winds are predominantly westward at around 10 meters/second in July and September, while at the end of March winds are predominantly eastward, with variable strength.However, observations suggest the second experiment was carried out close to the time when the zonal winds were turning to the west.

The detection patterns of WSMR1 and WSMR3 were strongly dependent on the zonal winds, as the preliminary modeling suggested (see Figure 1).The WSMR3 experiment yielded detections to the west as far as Pinyon Flat CA (I57US) at a distance greater than 900 km.Station UM1 at 265 km was the farthest station to the east to record signals for either WSMR1 or WSMR3.

Several interesting observations have been made from the data. First, although the explosions of the individual experiments were carried out approximately four hours apart, the signals show significant waveform variations. This suggests that dynamics of the atmosphere can change quickly and affect the amplitudes and arrivals of signals.Figure 2 is an example of waveforms from the first and second explosions recorded from WSMR1 recorded at Camp Navajo AZ. The first shot is clearly more impulsive than the second one, while in the record for the second shot more arrivals can be identified. It is important to note that the dominant periods of the signals are almost the same. Therefore, only the phase of the signals appears to be strongly distorted by the short-term dynamics of the atmosphere.Second, WSMR3 analysis indicates the yield/dominant period relationship changes unexpectedly for explosions above 40 km.


Figure 2. Recordings of the WSMR 1 experiment at Camp Navajo, Arizona. Although the shots were only 4 hours apart, and the source is not believed to vary significantly, the differences in the phase of the signal is significant.

Yield Determinations

To estimate the dominant periods of short duration signals we calculated power spectrum density (PSD) estimates using an autoregressive (AR) process of order 16 with Burgís method(Burg, 1972). The data were pre-filtered with a 1 Hertz high pass, zero phase, Butterworth filter because the PSD is dominated by very long period noise.Figure 3 is the AR power spectrum density estimate of the signal from the first WSMR1 explosion recorded at Camp Navajo, AZ showing a clear predominant frequency.Table 1 gives the altitude of the explosions, the predominant frequencies/periods and calculated yields for each of the signals.The periods given in the table were derived by calculating the mean predominant period for each array of stations (at least four) and then the mean of all arrays.The arrays included in the analysis were chosen based on good signal-to-noise ratios. The yields were calculated using the same method as the Columbia event.The table shows that the dominant period is dependent on the altitude of the source as predicted by the scaling relationship.However, the first two sets of explosions, conducted between 30 and 35 Km. show similar results in yield but those above 40 Km are significantly different.The scaling relationships appear to fail at higher altitudes.



Figure 7. Left: An example Power Spectral Density of the WSMR1 experiment showing the obvious predominant frequency/period. Right: Launch of the Orion rocket with explosive on board for detonation at altitude.Table 1 gives the explosion altitudes, dominant frequencies/periods of signals and estimated yields of all explosions.

5. Future Directions

Successful calibration experiments were carried out at WSMR over the past two years for the purposes of understanding the temporal dynamics of the atmosphere and for the calibration of sensors and arrays. The calibration experiments involved high altitude atmospheric sources for which the locations and yields were known with a high degree of accuracy. The initial goal of the calibration experiments was achieved, and detection patterns of the signals relative to the source locations were in agreement with predicted atmospheric conditions.Future modeling will use atmospheric observations at the time of each explosion, and will try to relate the observed signals to the atmospheric variations. As a byproduct of the modeling technique, the current infrasound modeling codes developed by BBN Technologies will be tested and validated.


There are established procedures for estimating the yield of an atmospheric explosion from the recorded infrasound signal.The relationship of energy partition into kinetic, heat and infrasound is not well understood, but during these experiments, the period/yield scaling relations appear consistent for sources below 40 Km.However, for explosions above 40 Km. the scaling relations clearly fail.Additional experiments with known source parameters conducted at altitudes above 40 Km may yield new scaling relationships and validate new formulas for yield estimates at these higher altitudes.


6. Acknowledgements

We wish to thank the Naval Surface Warfare Center at WSMR for their efforts in preparing the rockets and successfully conducting the launches including Mr. John Winstead, Head, Test Planning Branch, Navy Project Engineer Ms. Kathie Hoffman, Navy Flight Engineer Mr. Sal Rodriguez and, for Missile Systems, Mr. Troy Gammill from New Mexico State University.

7. References

Armstrong W.T., (1998), Comparison of infrasound correlation over differing array baselines. Proceedings of the 20th Annual Seismic Research Symposium, U.S. Department of Energy.

Burg J.P., (1972), The Relationship Between Maximum Entropy Spectra and Maximum Likelihood Spectra. Geophysics, Volume 37, Issue 2, pp. 375-376.

McKenna, M. H., and E. T. Herrin (2006), Validation of infrasonic waveform modeling using observations of the STS107 failure upon reentry, Geophys. Res. Lett., 33, LXXXXX, doi:10.1029/2005GL024801.

Mutschlecner, J. P., and R. W. Whitaker (2005), Infrasound from earthquakes, J. Geophys. Res., 110, D01108, doi:10.1029/2004JD005067.


Author Information

Paul Golden, E.T. Herrin and P. Negraru, Southern Methodist University, Department of Geological Sciences, Dallas TX; E-mail: pgolden@smu.edu; Henry Bass, National Center for Physical Acoustics; University of Mississippi, University MS; William Andre, US Army Space and Missile Defense Command, Huntsville AL; Mihan McKenna, US Army Engineer Research and Development Center, Vicksburg MS

(c) 2007 Southern Methodist University
Last update: dpa 12 February 2007