Seismic Warning System Using RF Energy Monitor

This invention relates to the prediction of the fugure occurrence of seismic activity, particularly to the advance notification of earthquakes through the monitoring of ambient radio frequency (rf) energy.

Heretofore, insofar as we are aware, seismology, the science of earthquakes, has not been able to make any near-term predictions of earthquakes.

While scientists have known that certain animals may have had some sort of advance knowledge of quakes, due to the fact that they exhibited peculiar behavior before quakes, and not at other times, this behavior has not been consistent and reliable enough to be of practical use.

Also, while scientists have also been able to predict thunderstorms in advance by monitoring the ambient electrostatic field (see, e.g., U.S. Pat. No. 3,611,365 to Husbyorg and Scuka, 1968; 3,790,884 to Kohl, 1974; and 4,095,221 to Slocum, 1978), they have not been aware of any corresponding system for earthquake prediction.

Scientists have been able actually to detect earthquakes during their occurrence by monitoring air pressure variations (e.g., as described in U.S. Pat. No. 4,126,203 to Miller, 1978) and by monitoring the earth's physical movement by seismographs but, again, science has not been aware of any system for short-term advance detection or prediction of quakes.

Due to the devastating effects of quakes to property, life, and limb, public and governmental authorities would derive great benefit from any system which could provide short-time advance notification of great earthquakes. As it is now, except for aftershocks, which seismologists know will occur after any large quake, all great and small quakes occur without warning. Because people in the vicinity of such quakes are unprepared, they often are in places of great vulnerability, such as beside or inside collapsible buildings, so that severe and human injury usually occurs during a quake. Also, property itself is left vulnerable, e.g., by leaving automobiles in or near collapsible buildings, leaving gas and electricity connected such that disruption of these facilities causes fires, and leaving other valuable property in vulnerable areas. If advance notification of a large quake could be provided to the public and civil authorities, people and valuable property could be evacuated and protected, thereby preventing deaths, injuries, and greatly reducing property damage. Further, advance notification of quakes would eliminate the severe psychological trauma which often affects large segments of the populace due to the surprise occurrence of quakes.

Accordingly several objects and advantages of the invention are to provide a reliable and effective method of earthquake prediction, to provide a method of preventing death, injuries, and reducing property damage in earthquakes, and to provide a method of reducing the psychological trauma which often accompanies quakes due to their surprise occurrence. Additional objects are to provide such a system which is easy to use, economical, reliable, and portable. Further objects will become apparent from a consideration of the ensuing description, taken in conjunction with the accompanying drawings.

The following is a discussion of the background theory of the invention. While we believe it to be technically accurate, we do not wish to be limited by this theory since the operability of the invention has been empirically verified, as will be apparent from the later discussion.

We have recently worked work with the reception and utilization of broadband radio-frequency reception, e.g., for low-power utilization applications, as discussed in the copending application Ser. No. 06/539,223 of Joseph B. Tate, filed Oct. 6, 1983. While doing this work, we have noted that the antenna's output voltage fluctuated with time due to certain, known causes.

First, we noted that the higher we placed an antenna above the ground, the the greater the output signal it provided. We have observed this by raising the physical height of an antenna and observing an increase in power output, and also by observing variations in the output of a fixed antenna near a body of ocean water as a function of the tides: the antenna's output was greatest at low tide and lowest at high tide. We believe that the the change in water level, which serves as a ground plane, effectively lowers or raises the height of the antenna above the ground.

We also noted that the antenna's output was affected by solar flares to a limited extent; these caused the antenna to produce a higher output voltage during their occurrence. We believe this phenomena is caused by an increase in the level of ambient ionization due to the flares.

Further, we noted that the antenna's output dropped at certain irregular times; at first we would not attribute any cause to these drops. However investigation enabled us to correlate these drops with the subsequent occurrence of seismic activity. We found that the magnitude of the drop was proportional to the size of the subsequent earthquake.

Certain phenomena have been discovered to precede earthquakes. These include an anomalous uplift of the ground, changes in the electrical conductivity of rock, changes in the isotopic composition of deep well water, changes in the nature of small earthquake activity (e.g., bunching of small foreshocks), anomalous ground tilt or strain changes, changes in physical properties, such as porosity, electrical conductivity, and elastic velocity in the hypocentral region. Earthquake, McGraw-Hill Encyclopedia of Science And Technology, 1960; Earth by F. Press, W. H. Freeman & Co., 1974.

Phenomena associated with rocks have attracted much recent attention. Wm. Brace of the Mass. Inst. of Technology has found that when rocks were squeezed or compressed, just before they fractured, they tended to develop hairline cracks, swell or dilate (dilatancy), become more porous and electrically conductive, and transmitted high frequency seismic-like waves more slowly. Two of Brace's former students, Amos Nur of Stanford University and Christopher Scholz of Lamont-Doherty furthered Brace's work, connecting the dilatancy theory with seismic P-wave velocity shifts and rock resistivity changes as a precursor for earthquakes. See. e.g., Brace, Orange, and Madden, J. Geophys Res., 70(22), 5669, 1965; A. Nur, Bull. Seis. Soc. of Amer., V 62, Nr. 5, pp. 1217-1222, 1972 Oct.; Earthquake by B. Walker, Time-Life Books, 1982.

Based upon the above background, we have developed a theory as to the cause of this drop in antenna output as a precursor or predictor of earthquakes. We believe that before a quake occurrs, the pressure within underground rock bodies temporarily increases greatly, causing the rocks to dilate and become conductive, in accordance with the works of Brace, Nur, and Scholz. This increase in conductivity effectively raises the ground plane, thereby causing the antenna's output to decrease temporarily.

Thus before the occurrence of a quake, the underground pressure increases greatly temporarily, causing underground rock bodies to swell and become more conductive, thereby raising the ground plane, which in turn causes the voltaic output of nearby antennas to drop.

We accordingly constructed an apparatus to automatically monitor antenna output and provide a suitable indication if the output level dropped significantly. The indication was calibrated empirically after much experimentation so as to filter ouo the effects of solar- and tide-caused variations. We did this by arranging the apparatus so that an output indication was provided only if the antenna output dropped a predetermined degree beyond its average level; we utilized statistical filtering techniques to accomplish this.

FIG. 1 shows the front panel of a Seismic Early Warning (SEW) apparatus according to the invention.

FIG. 2 is a plot of voltage (representing ambient rf level) v. time as measured by the apparatus of FIG. 1.

FIG. 3 is a schematic diagram of an ambient power module circuit (used in the SEW apparatus) for producing a DC output voltage proportional to the ambient rf energy

FIG. 4 is a block diagram of a computer in the apparatus of FIG. 1.

FIG. 5 is a flowchart which depicts the operation of the the SEW system.

FIG. 6 is a flowchart which depicts the operation of an optional alarm trigger system useable with the SEW apparatus.

In accordance with the invention, a seismic early warning apparatus is provided as shown in FIG. 1. The apparatus consists of a housing containing a general purpose computer (not shown), a disc drive 10, an analog system comprising a microampere meter 12 arranged to monitor direct current (which is proportional to the ambient rf energy), and a direct current strip chart recorder 14 arranged to provide a continuous indication of the current antenna output, which will be called the ambient power level. A hexidecimal keypad 16 is provided to enter data, such as time, for entering programs and changes and for operating the system according to preset codes. The time, date, and voltaic level of the antenna's output are continuously indicated by digital readouts 18, 20, and 22, respectively. A screen display 24 is provided to display graphic and alphanumeric information of the current status of the apparatus and previous data records.

Lastly the apparatus includes four status indicating lamps, which preferably are LEDs (light-emitting diodes) as follows: A green LED 26 indicates that the system is on and functioning normally. A yellow LED 28 indicates that the system has detected an event, namely the occurrence of a drop in ambient power below the preset level, which would be the prediction of an impending earthquake. A red LED 30 is provided as backup confirmation of the occurrence of the event; LED 30 is illuminated when a duplicate receiving system also detects an event. A blue LED 32 indicates initiation of operation of an automatic telephone dialer within the system, which has been preprogrammed to dial a predetermined number and provide a warning in the event of an occurrence of an alarm condition. Lastly the apparatus includes a hard copy output port 33 for providing printed graphic and numeric outputs of all system data.

FIG. 2 illustrates a reproduction of an actual plot of a voltage as a function of time, which voltage was proportional to the ambient RF (radio frequency) level, from the period from before to after a relatively large earthquake. This plot, which is typical of many we have observed before a quake, was made by deriving the voltage with a 30-meter, long-wire monopole antenna (not shown) which was mounted horizontally and which extended over San Francisco (Richardson) Bay easterly from Sausalito, California, 9 meters above sea level. The antenna thus intercepted and converted to an RF voltage the ambient RF energy, mainly from local (San Francisco area) AM radio stations. We rectified and filtered the output of the antenna using one-half of the circuit of FIG. 3 (described below) to provide a DC voltage which was plotted on a conventional ink-on-paper plotter. Note that on the section of the chart for Apr. 19 (1984), which begins at time 0:00 (midnight) and ends at 24:00, the voltage or ambient RF power level at the antenna increased and fell and then increased slightly in the 24-hour period. This wavelike variation typically occurs on a daily basis and is caused by tides: the peaks occurring at low tide when the effective ground plane provided by the water drops and the troughs occurring at high tide when the ground plane rises.

On Apr. 20, from about 8:00 to about 12:00, a sharp and constant-level dip in the ambient rf power occurred, as indicated. The magnitude of this pronounced dip is far greater than the normal tide-caused variations, as is its beginning and ending slope.

Thereafter, from Apr. 20 to Apr. 23, the plot (not shown) continued unremarkably, albeit with a slight variation from normal.

The same occurred on Apr. 24, with the plot actually being generally similar to a normal day. However at 13:15 on Apr. 24, as indicated, a large, Richter magnitude 6.0 quake occured near Hollister, Calif., about 340 km away from the antenna. No change in the plot occurred at this time.

Correllation of this quake with the plot's marked dip of Apr. 20 was made by the repeated observation of dozens of similar dips and subsequent quakes. Pronounced dips were always followed by a quake several days later. Thus we have empirically established causal and theoretical connections between pronounced dips of the type shown and the occurrence of subsequent seismic activity.

The circuit of FIG. 3 is used to convert the ambient RF energy to a direct voltage which can be used and handled by data processing equipment. Designated an ambient power module (APM), it is connected to an antenna 40, preferably a broadband monopole antenna of the type described in the preceeding section. The distal end of the antenna is free and its proximal end is connected to the circuit via two capacitors Cp1 and Cn1, each being in series with the signal line for coupling and each having a value of 0.047 microfarad. Taking the left or negative side of the circuit first, it comprises two rectifiers (diodes) Dn1 and Dn2 (1N34 type) and a filter capacitor Cn2 (40 microfarads). Rectifier Dn1 is connected in parallel to the signal path and rectifier Dn2 is connected in series, in the wellknown voltage multiplier arrangement. Capacitor Cn2 is connected in parallel across the output of the APM to smooth the rectified output. The right or positive side of the circuit is similar, except for the polarity of the diodes.

In operation, an RF voltage is developed across antenna 40; this voltage is voltage multiplied by the two rectifiers on each side of the circuit. The resultant direct voltages are smoothed or filtered by capacitors Cn1 and Cp2 and are supplied to output terminals 42 and 44. A positive version of this direct voltage is plotted in FIG. 2, as described above.

A computer for performing the monitoring and alarm functions of the invention and which is provided within the apparatus of FIG. 1 is shown in FIG. 4. The computer receives the positive voltage from the APM (FIG. 3) and processes this, providing an alarm if the voltage dips a predetermined amount from its recent average value.

The computer comprises an analog to digital converter (ADC) 50 which is arranged to convert the positive DC voltage from the AAPM to digital form, preferably in the form of a parallel signal at the output of ADC 50. The digitized voltage from ADC 50 is supplied to a central processing unit 52, which is a type 68000 microprocessor or computer on a chip. CPU 52 and ADC 50 are clocked by a clock 54 in conventional fashion.

CPU 52 operates on instructions from a program contained in an electricallyprogrammed read only memory (EPROM), the program being listed later. CPU 52 temporarily stores data in a read and write memory (RAM) 58. CPU 52 also supplies output data to display screen 24, disc drive 10, and hard-copy printer 26', each of which was already described in conjunction with FIG. 1.

CPU 52 can receive input data manually from hexidecimal keypad 16 (see FIG. 1) via a keyboard encoder 60.

CPU 52 can supply an alarm output to a radio transmitter or automatic telephone dialer 62 via a modem (modulator-demodulator) 64 for connecting the CPU to a phone (not shown).

As also indicated in FIG. 4, the negative output of the AAPM of FIG. 3 is connected to ammeter 12 and chart recorder 14.

In operation, the system of FIG. 4 operates under control of the program in EPROM 56 in accordance with the flowchart of FIG. 5 as follows:

Startup: Blocks 70 and 72: An initialization and start-up sequence is first initiated when the machine is turned on, as indicated by block 70; this sets all registers and counters to zero. The time and data are then set manually (using EPROM 56), as indicated by block 72.

Clock Reading: Blocks 74 and 76: Next, under automatic program control, the machine reads the elapsed time on its clock display register, as indicated by block 74. If the "seconds" register does not indicate the number one (#1), the machine continues to read the clock, as indicated by the "no" output of decision block 76.

Minute Sample: Block 78: When second #1 appears, as it will once per minute, the decision in block 76 will be "yes", so that the machine will take one sample of the rectified, smoothed, and digitized version of the antenna's output, i.e., the output of ADC 50 of FIG. 4, as indicated in block 78. This sample will be taken once per minute, i.e., whenever second #1 is displayed.

Running Average: Block 80: Next, as indicated by block 80, a running average of the samples taken in block 78 is calculated. This is done by accumulating the samples to keep a running total of their values, counting the number of samples accumulated, and dividing the running total by the latest number of samples each time a new sample is taken.

Store Hourly Average: Blocks 82 and 84: Next, as indicated in block 82, a test is made to see if the time display register indicates that minute number one (#1) has come up. If not, the decision is "no" and the clock is read again (block 74). If the decision is "yes", as it will be once per hour, the running average in the accumulator will be stored (block 84) and the accumulator will be cleared or reset to zero.

One Day Test: Block 86 ("No" decision) and Block 94: Next the machine makes a test to see if 24 hours have passed. If not, the machine will not be able to make any valid statistical determinations. Thus it must run at least 24 hours before being operative. Assuming the decision in block 86 is negative (24 hours have not yet elapsed) another test is made (block 94) to see if hour zero is indicated, which will occur once per day. If hour zero is not indicated, (decision in block 94 is negative), the clock will be read again (block 74) in the usual loop.

Calculate SD: Block 86 ("Yes") and Block 88: If a full day has elapsed, so that valid statistics can be calculated ("yes" from block 86), the standard deviation (SD) of the last 24 hourly averages is calculated, as indicated in block 88. This is done once per hour. The calculation is made using the usual SD formula

where SDDEV=SD; SQR=the square root; sum=the sum of; x=the individual hourly averages; X=the mean of the hourly averages; and n=the number of individual hourly averages. Essentially the SD is calculated by taking the mean of all of the hourly averages, taking the difference or deviation of each hourly average from the mean, squaring each deviation, taking the mean of the squared deviations, and then taking the square root of the mean of the squared deviations.

Evaluate SD: Block 90: The SD is then evaluated to see if it is greater than 0.3. This value has been empirically determined to be the level at which a the present apparatus will provide a reasonably positive indication that an earthquake will occur, while neglecting the effects of non-seismic-caused variations. If the SD is less than 0.3, (a "no" output from block 90), this indicates that the last hourly average was not greatly different from the average of the last 24 hourly samples, so that no alarm need be indicated. I.e., the antenna's output did not drop significantly to indicate an impending earthquake. Thereupon the program moves to block 94, where a test is made for the existence of hour zero, as described. If, however the SD exceds 0.3 ("yes" output of block 90), this indicates that the antenna's output has dropped significantly so as to affect the last hourly average, thereby to indicate an impending earthquake.

Alarm: Block 92: In response to the Yes output of block 92, an alarm is triggered (block 94). The alarm may be a bell, the dialing of a telephone to a location where personnel are present if the apparatus is placed at a remote or non-manned location, or the initiation of the further program of the Flowchart of FIG. 6, the alarm trigger sequence. To eliminate the possibility of equipment failure and to provide confirmation from another apparatus at another location, we prefer to provide an alarm only upon confirmation from another apparatus, as discussed in the description of FIG. 6 below.

Make Record: Block 94 ("Yes") and Block 96: If hour zero is being displayed when the operation of block 94 is performed, which occurs once per day at midnight, the operation of block 96 will be performed, i.e., the data in the registers will be stored to disc to create a permanent record and the registers will be cleared to create new data for the next day. However the previous 24 hourly averages are still stored at all times so that a valid SD can be calculated and tested every hour. After the operation of block 96, the clock is read again in accordance with the regular program (block 74).

The sequence of FIG. 6 is performed when the alarm is triggered in block 92 of FIG. 5 as an optional, but preferred backup confirmation of an impending earthquake. The operations in the backup confirmation system will be described briefly.

Beginning with blocks 100 and 102, the system is continually tested (hourly) for the occurrence of a SD of the hourly averages of greater than 0.3. If the SD is greater than 0.3, the alert indicator (28 of FIG. 1) is triggered (block 104) and the program initiates a test (block 106) to see if a backup apparatus (not shown) is present. If so (yes output of block 106) the backup apparatus is also checked (blocks 108 and 110). If the backup does not indicate an excess SD, the indicators are reset to normal (block 112), but if backup confirmation is received, the alarm indicator (30 of FIG. 1) is triggered per block 114 and a preprogrammed telephone number is dialed and indicator 32 is lit (block 116).

After the alarm condition is manually checked and the system is reset, the output of block 120 will be a "yes" and the system will be reset to normal (block 112). If a valid alarm condition is indicated and confirmed, civil authorities will have time (usually several days) to notify the populace, evacuate the area, or take any other needed precautions, depending on the size of the impending quake as indicated by the size of the standard deviation.

The attached computer programs will perform the calculations and operations above described. These programs are written in the BASIC programming language. Program "RECVOLT.AL" runs continuously and writes the information to disc every 24 hours. Program "GRASTAT.*" is manually run; it reads data from the disc and plots it on the screen or printer, as desired.

While the above description contains many specifications, these should not be construed as limitations on the scope of the invention, but merely as an exemplification of one preferred embodiment thereof. Many other variations are possible. For example, the programming language can be changed, or the calculations and operations can be performed with hard-wired conventional circuitry in lieu of a programmed computer. More than two corroboration receivers can be used, and these can be placed at various locations. In lieu of testing the antenna's output reception of the area's AM stations, a special, decicated transmitter with a special, dedicated frequency and a specially-tuned matching receiver can be used to avoid depedence on stations which are not under the control of the earthquake prediction system and its personnel. The transmitter and the receiver should be spaced apart geographically, preferably by at least several km, so that the ground plane conduction phenemon can operate. Also the transmitted signal can be a specially-coded or modulated signal, or it can be an auxiliary signal of a regular transmitter, e.g., a SSB or SCA signal, together with a matching receiver. In lieu of a test for an excess SD, the apparatus can be arranged to test for a predetermined drop in the value of the antenna output from its immediately previous value or its average value over a predetermined period, such as an hour or day, or for a drop having greater than a predetermined slope. Accordingly the full scope of the invention should be determined by the appended claims and their legal equivalents, and not by the examples given. ##SPC1##