The Infrasound Monitoring System for The CTBT Control
Dr Elisabeth Blanc
Under the implementation of the CTBT (Comprehensive Test Ban Treaty), monitoring systems must be set up to provide worldwide coverage for the detection of nuclear explosions with a yield of at least 1 kt. Methods based on infrasound measurements allow the detection and localisation of the explosions in the atmosphere. The international infrasound monitoring system comprises 60 stations located on all continents and several islands, especially in the southern hemisphere, in order to provide a global coverage (Figure 1).
Figure 1: The 60 stations Infrasound network
Monitoring technology
The sensors used for the detection of the infrasound at the surface of the ground are microbarographs. The sensitive part is a barometric aneroid below submitted to deformations under atmospheric pressure changes. The sensors design eliminates temperature induced drifts. Microbarographs generally provide the relative pressure. The possible measurement of the absolute pressure in parallel allows a direct calibration of the sensor.
The microbarographs used for the CTBT measure infrasound in the range 0.02 to 4 Hz. They are characterised by a good sensitivity (18 dB below the minimum acoustic noise), and by a large dynamic (80 dB) in order to detect both explosions at very close distances and explosions at distances up to several thousands of kilometres. The sensors are equipped with acoustic filtering systems (microporous hoses or pipes) to reduce the noise produced by the surface winds.
The infrasound stations are composed of at least 4 sensors located in a triangle with a basis of 1 to 3 km with a central point. This allows to determine the explosion azimuth by triangulation, to remove false signals, not coherent at the scale of the array, and to increase the station reliability.
	Figure 2: Infrasound station. The stations detect and give the azimuth of the nuclear explosions. The location is given by the intersection of different azimuths
Infrasound signals from atmospheric explosions
Pressure waves are an important component of the source signal produced by a nuclear explosion, consisting essentially of sound waves in the infrasonic frequency domain (periods of about one second to several minutes or more for large explosion yields).
Infrasound is not audible, its frequency is higher than that of sound waves. It propagates at the sound speed in the atmospheric sound channel formed by the atmospheric temperature gradients under the effects of the high altitude winds (Figure 3).
Figure 3: Ray tracing showing the different possible infrasound paths (right). Typical wind and temperature profiles are shown in the centre and left part of the figure.
The rays are reflected at the level of the temperature increases. High altitude winds produce an anisotropy in the propagation.
Because of the propagation effects, signals recorded at few hundreds of kilometres from the explosion may be formed of several wave trains corresponding to different paths in the atmosphere.
Figures 4 shows two examples of signals recorded during a nuclear test of a few kilotons, at distances of 440 and 450 km from the explosion to the north-west and east, respectively. The signal recorded in the east corresponds to a propagation in the direction of the high altitude winds while the signal in the north-west correspond to a propagation with contrary winds.
The explosion signal recorded at the East is composed of three different signals of 40 seconds, each separated by about 30 seconds. Their amplitudes range from about ±130 Pa (1st arrival) to ±70 Pa (3rd arrival). The signals obtained at the north west are weaker and longer and only two arrivals are recorded. The duration of the first signal is about 350 seconds, the amplitude is ±6 Pa, while the second signal lasts about 100 seconds with an amplitude of ±1 Pa.
The decrease of infrasound amplitude as a function of distance has been established empirically from the French nuclear explosion database. The first dominant effect is related to the high altitude winds. In some cases, they allow explosions of about 10 kt to be detected at distances of up to 7000 km windward from the explosion location, but they can make detection difficult at distances of about 1000 km in the opposite direction. The most significant impact comes from high altitude stratospheric winds, with velocities ranging from 60-90 m/s at altitudes of about 30-70 km (figure 3). These winds are essentially zonal winds, directed to the West in the summer in the northern hemisphere, and to the East in the southern hemisphere, the opposite occurring in winter. Atmospheric models based on world-wide measurements provide wind profiles as a function of time of year and latitude.
Figure 5 shows the amplitude (peak-to-peak), DP, of the pressure waves measured for tests of less than 400 kt for distances D ranging from 150 to 7000 km. The plotted amplitude is that of a 1-kt test, assuming a W1/2 variation, where W is the explosion yield. The high altitude wind effect, determined by using the CIRA atmospheric model, has been taken into account. The propagation law giving the amplitude DP (in Pa) as a function of distance D (in km) and projection of stratospheric winds Vp (in m/s) on the propagation direction is of the form .
Infrasound noise
The infrasound noise is produced by pressure turbulence induced by the winds at the ground surface. An empirical law (Figure 6) has been determined by using meteorological data performed at the infrasound stations for an open sensor and for different system of acoustic filtering (circular or linear hoses). The noise varies by a factor of 100 between quiet and disturbed conditions of wind. The noise can be reduced by a factor of ten or more by well adapted filtering systems.
	Figure 6: Empirical law of infrasound noise versus wind speed at the ground surface
Detection and location capability of the network
The network detection and location capabilities have been determined by modelling. The amplitude of the infrasound from nuclear explosions is given by the empirical propagation laws obtained from nuclear explosion data (Figure 5), the high altitude winds being determined by the CIRA atmospheric model. The infrasound noise is determined by the empirical law (figure 6) and a recent atmospheric model of the wind at the ground surface. The map which gives the network detection capability as a function of yield W is represented in figure 7.
The location precision depends on the dimensions of the network: the most extended the network, the better the precision. This precision depends also on the signal characteristics. For an impulse signal, the precision of the time measurement is the time increment (0.05 s for a sampling rate of 20 Hz). The azimuth error is then 0.7° and 2° for a triangle base of 3 and 1 km, respectively. The same computation, performed assuming the detection of a non-impulse signal with a time measurement precision of a tenth of the period (0.6 s), gives an azimuth error of 8° for a 3 km network base, and of 24° for a 1 km base [16].
Figure 7: Detection capability by two stations of the infrasound network in January at 00 UT (left) and 12 UT (right).
The detection capability varies significantly as a function of the time of the day because of the variation of the infrasound noise under the effect of the winds at the surface of the ground. The map shows that, in the Pacific ocean, the detection could be more difficult in conditions of high winds at the ground, specially during daytime. This appears in the Pacific ocean because the density of the station is lower than in the other parts of the word. In these areas the number of station sensors will be increased in order to improve the detection capability. The network will then detect 1-kt everywhere in the word. In addition, the new development of more efficient acoustic filtering systems such the STAR system improve the noise reduction and then the detection efficiency.
The location precision depends on the dimensions of the network: the most extended the network, the better the precision. The location capability of the network has been computed assuming an angle precision of 0.7° for a 3 km station array. The location precision is estimated within a circle of about 100 km or less everywhere in the word.
Infrasound events
An experimental prototype station developed by the CEA has been set up at Flers (Normandy) in France. This station is a four-sensors array whose characteristics correspond to the requirements of the CTBT monitoring network.
The PMCC (Progressive Multiple Cross Correlation) method permits the analysis of the data in a permanent way. Infrasound bulletins are edited every day, showing typical infrasound events detected at the station. For each event, the infrasound velocity and the azimuth are automatically determined.
Wave systems, sometimes highly complex in nature, are propagated through the atmosphere. Infrasound is produced by specific phenomena such as meteorological storms, earthquakes, volcanic eruptions, winds over the mountains, the ocean swell (microbaroms) or boreal aurora.
Well identified signals produced by chemical explosions, supersonic aircrafts, thunderstorms or microbaroms are used to test the data processing method and the detection and location efficiency of the station.
Figure 8 shows examples of signals obtained by the CEA several thousands of km from the volcanic eruption of Mt. St. Helens and Pinatubo. They are compared to a one-megaton explosion. Volcanic eruptions are powerful sources capable of producing low-frequency waves that a one-megaton nuclear burst could not generate.
Table 1 summarises the main types of infrasound events detected at the station Flers. The characteristics of these events have been determined and discrimination methods can be defined for the event identification.
Type characteristics complementary data quarry explosions specific explosions generally <1kt the signal frequency range depends on the source energy explosion characteristics supersonic airplanes several daily flights several signal phases generally observed trajectory details volcano few large explosive eruptions (St Helens, Pinatubo) local events possible seismic data microbaroms from ocean swell extended source regions very frequently observed meteorological data, satellites thunderstorms lightning, convective motions meteorological data infrasound from auroras source available in high latitude regions camera pictures meteorites high altitude sources satellite data
Table 1: Different infrasound signals of man-made or natural origin
Conclusion
Infrasound measurements are well adapted for detecting and locating atmospheric explosions. The explosion shock waves are characteristic, their low-frequency components are able to propagate at long ranges and can be detected by microbarographs. Previous measurements during nuclear and chemical explosions have provided a database for estimating the detection and location capability of the CTBT infrasound network. Experimental stations, such as the prototype station Flers in France, are very useful in the study of noise and natural disturbances. PMCC method is used to establish the daily infrasound bulletin and several well identified infrasound sources such as quarry blasts, ocean swell, supersonic aircraft or thunderstorms are permanently used to control the detection and location efficiency.