Disaster alert mediation using nature

Abstract

A method for forecasting an environmental event/a type of environmental event includes acquiring at least one test data set of at least one behavioral and/or physiological parameter of a population of animals; generating a test profile based on said at least one test data set, representing behavior and/or physiological status of the population of animals; calculating a ratio between the test profile and a first reference profile; and setting an alert, if said ratio reaches a predefined threshold value. A system for forecasting an environmental event/a type of environmental event employs at least one data acquisition unit configured to acquire at least one data set of at least one behavioral and/or physiological parameter of a population of animals; at least one profile generation unit configured to generate at least one first reference profile and/or at least one test profile wherein each profile is based on at least one data set; at least one ratio calculation unit configured to calculate at least one ratio between the at least one test profile and one of the at least one first reference profile; at least one alert unit configured to raise an alert if at least one of the at least one calculated ratio reaches a predefined threshold value.

Claims

1. A method for use in forecasting an environmental event or a type of the environmental event, comprising: acquiring at least one test data set of at least one of a behavioural parameter and a physiological parameter, of a population of animals; generating a test profile based on said at least one test data set, representing at least one of a behaviour status and a physiological status of the population of animals; and setting an alert, when a ratio between the test profile and a first reference profile reaches a predefined threshold value, wherein the acquiring comprises acquiring the behavioural parameter of the population of animals, and the method further comprises attaching a data acquisition circuitry to an animal of the population of animals and acquiring the at least one test data set from the data acquisition circuitry, the data acquisition circuitry comprising a 3D-acceleration sensor configured to provide roll-pitch-yaw angle data and a global positioning system (GPS) receiver configured to generate location coordinates, wherein the at least one test data set is acquired based on 3D-acceleration activity obtained by the 3D-acceleration sensor and diurnal unidirectional movement obtained by the GPS receiver, and wherein the diurnal unidirectional movement is obtained based on a total daily distance in a linear direction travelled by the population of animals from a nocturnal resting location to a following evening resting location.

2. The method of claim 1, further comprising: acquiring at least one first data set of the at least one of the behavioural parameter and the physiological parameter of the population of animals in an absence of the type of the environmental event; and generating the first reference profile based on said at least one first data set, representing at least one of a normal behaviour status and a normal physiological status of the population of animals.

3. The method of claim 1 wherein the forecasted type of the environmental event is at least one event from among an earthquake, a marine earthquake, a tsunami, and a volcanic event.

4. The method of claim 1 wherein the population of animals comprises at least one of feral animals, semi-domestic animals, domestic animals, and animals in zoos.

5. The method of claim 1 wherein the population of animals comprises at least one of land animals, aquatic animals, and aerial animals.

6. The method of claim 1 wherein the population of animals comprises animals of the same species.

7. The method of claim 1 wherein the population of animals comprise at least one of goats, sheep, elephants, dogs, donkeys, monkeys, apes, and frogs.

8. The method of claim 1 wherein the population of animals comprises at least 5 animals.

9. The method of claim 1 wherein the behavioural parameter further comprises nocturnal activity.

10. The method of claim 1 wherein the threshold value is a ratio of 2.

11. The method of claim 1 wherein the threshold value is a ratio of 1.3.

12. The method of claim 1 further comprising: acquiring at least one second data set of the at least one of the behavioural parameter and the physiological parameter of the population of animals in a presence of the type of the environmental event; generating a second reference profile based on said at least one second data set, representing an abnormal behaviour status of the population of animals.

13. The method of claim 12 wherein the presence of the type of the environmental event is detected using methods comprising at least one of visual observation, acoustical observation, and seismological measurement.

14. The method of claim 12 wherein the threshold value is determined using the first and second reference profiles.

15. The method of claim 1 wherein the alert is raised at least 2 hours prior to the environmental event.

16. The method of claim 1 wherein the data acquisition circuitry further comprises a sensor configured to measure at least one physiological parameter of the animal using at least one of an electrocardiograph, an electroencephalograph, a clinical thermometer, an endocrinological measurement device, and an electromyograph.

17. A non-transitory computer program product comprising one or more non-transitory computer readable media having computer executable instructions for performing the method of claim 1.

18. A system, for forecasting a type of an environmental event, comprising: a data acquisition circuitry configured to acquire a data set of at least one of a behavioural parameter and a physiological parameter, of a population of animals; a profile generation circuitry configured to generate: at least one first reference profile, and at least one test profile, each based on at least one said data set; an alert mechanism configured to raise an alert when a ratio between the at least one test profile and the at least one first reference profile reaches a predefined threshold value, wherein the data acquisition circuitry acquires the behavioural parameter, and the data acquisition circuitry, comprising a 3D-acceleration sensor configured to provide roll-pitch-yaw angle data and a global positioning system (GPS) receiver configured to generate location coordinates, is attached to an animal of the population of animals, wherein the data acquisition circuitry is configured to acquire the data set based on 3D-acceleration activity obtained by using the 3D-acceleration sensor and diurnal unidirectional movement obtained by using the GPS receiver, and wherein the diurnal unidirectional movement is obtained based on a total daily distance in a linear direction travelled by the population of animals from a nocturnal resting location to a following evening resting location.

19. The system of claim 18 wherein: the profile generation circuitry is further configured to generate a second reference profile based on at least one data set, and wherein said threshold value is determined based on a ratio between the second reference profile and the at least one first reference profile.

20. The system of claim 18 wherein the data acquisition circuitry is attached to the animal of the population of animals by using a neck collar.

21. The system of claim 18 wherein the data acquisition circuitry comprises at least one of a data memory and a data transmitter.

22. The system of claim 21 wherein the data transmitter comprises a VHF transceiver configured to transmit the at least one of the acquired data set, the at least one generated profile, and the ratio between the at least one test profile and the at least one first reference profile to a remote server.

23. The system of claim 18 wherein the data acquisition circuitry comprises a power supply.

24. The system of claim 23 wherein the power supply comprises at least one of a battery, a rechargeable battery, and a solar cell.

25. The system of claim 18 further comprising a remote server having at least one of a data memory, the profile generation circuitry, and the alert mechanism.

Description

(1) The present invention is further described by reference to the following non-limiting figures and examples.

(2) The Figures show:

(3) FIG. 1. Schematic overview of magma within Mt. Etna, Sicily; Picture from Prof. Ulrich Schreiber, personal communication.

(4) FIG. 2. Screenshot from Movebank's acceleration viewer. The acceleration of one goat for 4 days and nights around a major volcanic event is shown. The X-axis shows time in days, minutes, seconds (depending on magnification), the y-axis shows the arbitrary acceleration units in relation to the overall average acceleration for an individual goat. The occurrence of the major volcanic event of January 4, 22:20 h local time, is indicated by a vertical line. Major unidirectional movement of the goat during the preceding day is indicated by an arrow.

(5) The photograph shows an eruption column and lava fountain from the New Southeast Crater seen from an airplane passing to the northeast of Etna during the Jan. 5, 2012 eruptive episode (photo taken by Gloria Guglielmo; original photo on Flickr).

(6) FIG. 3. Screenshot from Movebank's acceleration viewer. The acceleration of one goat for 8 days and nights around a major volcanic event is shown. The X-axis shows time in days, minutes, seconds (depending on magnification), the y-axis shows the arbitrary acceleration units in relation to the overall average acceleration for an individual goat. The occurrence of the major volcanic event of Mar. 4, 2012, 7:04 h local time, is indicated by a vertical line. Major activity of the goat during the preceding night is indicated by an arrow.

(7) The photograph shows an eruption column of the Mar. 4, 2012 paroxysmal eruptive episode seen from the Catania plain, about 40 km southwest of the summit of the volcano (photographed by da Elisabetta Ferrera; University of Catania).

(8) FIG. 4. Movements of one goat. The screenshot from Movebank's acceleration viewer demonstrating the movements of one goat around Mt Etna is shown. The location of the goat was measured by GPS.

(9) FIG. 5. Linking of acceleration data to location data of one goat. The screenshot from Movebank's acceleration viewer demonstrating acceleration and location of one goat is shown. The left panel shows the acceleration of the goat #1910 for 9 days and nights. The X-axis shows time in days, minutes, seconds (depending on magnification), the y-axis shows the arbitrary acceleration units in relation to the overall average acceleration for an individual goat. The right panel shows the movements of the goat around Mt Etna during this period (lower right in the Worldwind picture).

(10) FIG. 6. Diurnal unidirectional movement and Nocturnal activity of one goat around the 19.sup.th and 21.sup.st paroxysmal events. A time scale including the major and minor volcanic events is shown on the left. Some of the volcanic events also shown on photographs (photos from INGV). The large photograph indicated by the arrow as 19. Paroxysmal shows an eruption column and lava fountain from the New Southeast Crater seen from an airplane passing to the northeast of Etna during the Jan. 5, 2012 eruptive episode (photo taken by Gloria Guglielmo; original photo on Flickr). The large photograph indicated by the arrow as 21. Paroxysmal shows an eruption column of the Mar. 4, 2012 paroxysmal eruptive episode seen from the Catania plain, about 40 km southwest of the summit of the volcano (photographed by da Elisabetta Ferrera; University of Catania).

(11) The graphs show 3D-acceleration of one goat around the 19.sup.th (left graph) and the 21.sup.st (right graph) paroxysmal events. The occurrence of the major volcanic event is indicated by vertical lines. The graph on the left shows that on the day before the major volcanic event, the goat showed enhanced diurnal unidirectional movement (i.e. it runs 350 m down). The graph on the right shows that the night preceding the major volcanic event, the goat showed enhanced activity (i.e. it was active all night).

(12) FIG. 7. Nocturnal activity of 8 goats around the 21.sup.st and 22.sup.nd paroxysmal events. The major event starts at 7:04 a.m. (local time) after the 6.sup.th night. A and B: The graphs are showing the overnight-integrated activity of the population of 8 goats around a major volcanic event (21.sup.st paroxysmal event; A) and a minor volcanic event (22.sup.nd paroxysmal event; B) including the standard deviation as error bars in arbitrary units (a.u.).

(13) C and D: The graphs are showing the relative overnight-integrated activity of the population of 8 goats around a major volcanic event (21.sup.st paroxysmal event; C) and a minor volcanic event (22.sup.nd paroxysmal event; D) including the standard deviation as error bars.

(14) The graphs demonstrate that in the night before the major volcanic event, the population of animals shows an increase of nocturnal activity of at least 100% as compared to the activity of the five preceding nights. In contrast, the population of animals does not show such an increase in nocturnal activity in a night before a minor volcanic event.

(15) FIG. 8. Accumulation of -hourly activity of 8 goats during the night preceding a volcanic event. In 6 nights preceding different volcanic events, the average 3D-acceleration activity of 8 goats has been determined every 30 min and each measured value has been summated to the foregoing values, so that the graphs show an accumulation of the measured activity values. The accumulation curves of average acceleration behaviour of 8 goats for 5 control nights preceding a minor volcanic event, compared to the night preceding the major volcanic event (line which is above all other lines starting from 21:30) is shown. Please note the jump in acceleration of this line of at least 30% after 1:00 hours of the night preceding a major volcanic event. The major volcanic event occurred ca. 5 hours later. In the 5 nights preceding a minor volcanic event, no jump in acceleration of at least 30% occurs.

(16) FIG. 9. Diurnal unidirectional movement of 8 goats around the 19.sup.th and 20.sup.th paroxysmal events. The major event starts at 22:20 p.m. (local time) after the 6.sup.th day. A and B: The graphs are showing the daytime-integrated unidirectional movement of the population of 8 goats around a major volcanic event (21.sup.st paroxysmal event; A) and a minor volcanic event (22.sup.nd paroxysmal event; B) including the standard deviation as error bars.

(17) C and D: The graphs are showing the relative daytime-integrated movement of the population of 8 goats around a major volcanic event (21.sup.st paroxysmal event; C) and a minor volcanic event (22.sup.nd paroxysmal event; D) including the standard deviation as error bars.

(18) The graphs demonstrate that the day before the major volcanic event, the population of animals shows an increase of unidirectional movement of at least 100% as compared to the unidirectional movement of the five preceding days. In contrast, the population of animals does not show such an increase in unidirectional movement the day before a minor volcanic event.

(19) FIG. 10. Overview of the system of the invention according to one embodiment of the invention. The system comprises a data acquisition unit (100), a profile generation unit (200), a ratio calculation unit (300), and an alert unit (400).

(20) FIG. 11. Overview of the data acquisition unit (100) and the sensor unit (120) according to one embodiment of the invention. The data acquisition unit (100) comprises a fixing unit (110), a sensor unit (120), a data memory unit (130), a data transmission unit (140), and a power supply (150). The sensor unit (120) comprises a global positioning system receiver (121), a 3D-acceleration sensor (122), and means configured to measure at least one physiological parameter (123).

(21) FIG. 12. Overview of the data acquisition unit (100) according to one embodiment of the invention. The data acquisition unit (100) is a biologger (e-obs) comprising an acceleration sensor, optionally a location sensor (e.g. GPS), a controller (e.g. MicroController), a radiolink (e.g. GSM, GPRS, ARGOS, Bluetooth), and a control center.

(22) FIG. 13. Time line of volcanic events at Mt. Etna during a study period. Each dot represents a significant volcanic event starting on Sep. 1, 2011 until 25 Apr. 2013. The dates of major events are indicated and small insect pictures depict the magnitude of the event. Vertical lines indicate data readout periods. In Example 2, two major events have been detected. In Example 3 five additional major events have been detected, totalling seven major events.

(23) FIG. 14. Low seismic activity (i.e. below 5 on the Richter magnitude scale) at Mount Etna does not coincide with volcanic event. Each dot indicates an earthquake and it's magnitude. No data are available for the period from April to September 2012.

(24) The Examples illustrate the invention.

EXAMPLE 1

(25) Materials and Methods

(26) Study Site

(27) Fieldwork was carried out from April 2011 to October 2011 on the Mediterranean island of Sicily to determine whether feral goats show behaviors that could be used to anticipate and predict natural disasters. The study was conducted on the northern slopes of Mount Etna volcano, around the small town of Randazzo (37.8752 W, 14.9524 N). The study site consists of feral pastures on the outskirts of the town, as well as natural forests and openings in the vegetation along the slopes of the volcano at an altitude of ca. 1000 to 1900 meters above sea level. This altitude was chosen because Mt. Etna hosts magma chambers that horizontally extend from a central magma chimney towards the slopes of the volcano (see FIG. 1). It is expected that gases from the magma chambers may escape at these altitudes and potentially be detectable by organisms.

(28) Study Objects: Semi-domestic Goats

(29) We used adult female semi-domestic goats as study subjects because initial interviews with local naturalists and goat herders indicated that goats are the most sensitive animals towards natural changes in the area. The goats used in this experiment were chosen randomly from a captive herd of ca. 500 goats, all belonging to one farmer. All of these goats are locally adapted to the prevailing environmental conditions and herded in the area since presumably hundreds of years. These goats roam freely around the slopes of Mt. Etna for most of the year, but are brought down from the mountain during the time of calving (March/April) and harvesting (October) each year. The goats form small herds, usually three to a dozen goats, in a fission-fusion manner. Thus, the goats observed here were usually roaming around without immediate contact to other observed goats. However, at rare random times, collared goats were moving in the same herd and thus could not be considered independent units for statistical analysis. During the two times when Mt. Etna erupted in a substantial way (see below) within the observation period, all observed goats were roaming independent of each other, thus we considered all 8 individuals as independent.

(30) Biologging Tags and Attachment

(31) We used biologging tags from E-obs (www.e-obs.de) to determine the behavior and location of goats for up to 180 days (until the tag memory fills up). The tags were attached as neck collars to the goats in a simple procedure, i.e., one goat herder was holding the goat by the horns while standing above the goat with the legs pressing against the body of the boat, the other herder was putting the collar around the neck and tightening the self-tightening screws of the collar such that a two-finger opening remained between the collar and the goats' neck. In this way we ensured that the goats was minimally disturbed by the collars, similar to bell collars put on regularly by the goat herders to approximately every 20.sup.th goat in a heard.

(32) Measurements Taken by the Tags

(33) The e-obs tags recorded GPS position every 30 minutes as well as 3-D acceleration every 2 minutes for 3.6 seconds. The GPS signal was received with the help of a ceramic antenna and GPS position was calculated on board of the biologger using a commercial GPS chip. GPS timeout was set to 2 minutes, i.e., if the GPS receiver chip could not calculate a GPS position within 2 minutes, it would give up and try again 28 minutes later to get a GPS fix. This happened rarely whenever the goats were inside a concrete farmhouse with a metal roof. At all other times, the average time to a GPS fix was 28 seconds, ranging from 3 to 92 seconds.

(34) Acceleration was recorded in the z-axis only, to report the up and down movements of the goats, which we deemed sufficient to allow for an understanding of goat behavior. Only recording every 2 minutes for a short interval, and only recording the Z-axis of the accelerometer, massively reduces the amount of data that needs to be stored, transmitted and analyzed. The observation scheme represents the timed sampling method that is well established in behavioral analysis and is known to report the behavior of individuals with high accuracy (Altman 1965).

(35) The accelerometer used in the e-obs tags is a 3D-accelerometer. In general accelerometers have some kind of a piece of mass that is connected to a flexible material and a damping material. The mass is pushed against the flexible material by the acceleration, and the excitation is proportional to the acceleration. The damping material prevents oscillation. As a consequence the output is (for low frequencies) proportional to the acceleration. For higher frequencies the sensor becomes less sensitive (this is true for all types of sensors) and there will be some phase shift (this is also true for all types of sensors). The bandwidth used in e-obs tags is 150 Hz for the Z-axis and typically 350 Hz for the X- and Y-axis. The analog output signal is sampled with a user-adjustable sampling rate ranging from 10 Hz to 1778 Hz for all axes combined. Here we used 10 Hz.

(36) There is no anti-aliasing filter, which means that the user must be sure that acceleration doesn't oscillate with a frequency higher than half the sampling frequency. For example: If the sampling frequency is 10 Hz for one axis, then e.g. the jumping frequency of a goat should not be more than 5 Hz, otherwise the user will get the wrong jumping frequency during analysis. All XYZ axes are perpendicular to each other like a cartesian coordinate system. Acceleration can never be used to predict positions, because you have to mathematically integrate twice to retrieve position from acceleration. This, however, implies that you also integrate the errors. Additionally the axes are fixed relative to the goat, but the animal's orientation is not fixed relative to space, so you never know the direction of acceleration relative to space/earth. The data values are 12 bit-readings of the analog-to-digital converter and are not calibrated. Roughly, the two values 0 and 4096 (corresponding to a 12 bit value) represent 1.5 g and +1.5 g for high sensitivity setting (which we used here), whereby g is the earths acceleration (9.81 m/s^2). For acceleration we set an interval (similar to the GPS interval) every 2 minutes to record 54 Bytes of acceleration with a sampling frequency of 10 Hz on the Z axis. This means the samples are not evenly spaced in time, instead the data are collected in bursts. In our study, the acceleration sensor was turned on every 2 minutes for a certain time. This time was defined by the amount of data to be collected (which is user defined: here 54 Bytes) and the sampling frequency (also user defined: here 10 Hz). Since one axis was sampled, the sampling frequency per axis is full (here 10 Hz). One sampling point needs 1.5 Bytes, therefore the total amount of sampling points will be 54/1.5=36 i.e. 36/1=36 per axis. The total sampling time is 36/10 Hz=3.6 s. The next scheduled acceleration recording will be 2 min later (according to the interval). The required power is about 1 mA during acceleration recording.

(37) Data Download and Initial Data Handling

(38) We downloaded the stored data via an encrypted 868 MHz data download. During the download, a specific tag communicates with the handheld, battery powered base station exclusively and acknowledges and verifies the data packages that are being sent. Thus, all data are being transmitted fully and with perfect handshake recognition during the sending process. Once data are received by the base station, the base station acknowledges this receipt and programs the tag to erase this part of its memory. The base station records the data in a memory chip at a rate of ca. 1 Mb per minute. From each goat, we downloaded ca. 10 Mb of binary encoded data after 6 months of deployment of the tags. Data downloads stop after they are complete, which also means that all data on the tags are erased and the tags are ready to record new data for another 180 days, depending on the settings provided in the initial settings file.

(39) Data from the base station are then directly transmitted to a computer from an SD memory chip card. The binary file can then either be transferred into a regular text file on a Windows PC, or uploaded directly to Movebank (www.movebank.org), where the data are unpacked and double-checked against duplicates. Furthermore, the observations are linked to the absolute time of recording as determined from the GPS module and the GPS location settings. Thus, the accuracy of timing measurements in the tags is given by the precision of the GPS time. Movebank stores the data in a relational data base with the main fields of animal identification number (ID), time, GPS location and movement, GPS error, acceleration, as well as reports on the technical properties of the tag (battery voltage, GPS time to fix, memory status etc.).

(40) Data Inspection and Evaluation

(41) Once data are in Movebank or on the PC computer, they can be visualized by linking the acceleration data to location data. The visualization is conducted by plotting acceleration values (as described above, as values between 0 and 4096) in time and linking it to geographical location as displayed either on N ASA Worldwind or on Google Earth (see FIG. 5). Thus, the researcher can simultaneously watch the behavior of the goats (acceleration in the Z axis) and their locations on the slopes of Mt. Etna.

(42) The observations represent true timed samples of the goats' behavior (every 2 minutes for acceleration, every 30 minutes for GPS location), and as such are truly representative of the overall behavior of the goats. The acceleration behavior of the goats enables a quantitative determination of the movements of individuals in the Z-axis.

(43) For the analysis of acceleration and thus behavior, we used the average values of the 36 acceleration measurements during a 3.6 second burst as well as their statistical variance. These average values were taken as quantitative indicators of the goats' activity during a sampling interval. We compared the acceleration values measured in this way between different times, e.g. hourly before a major volcanic event (see below) or afterwards. We present the cumulative sum of half hourly acceleration averages or acceleration variances over night as an indicator of the sensing of goats of environmental conditions. We also used the entire sum of nocturnal activity, measured as the cumulative average of acceleration values, to compare activity levels between nights (defined as the time between 20:00 h local time and 6:00 h local time).

(44) To quantitatively analyze and compare the linear movements of goats during the day, we used the total daily distance travelled by goats from their nocturnal resting location to their evening resting location in a linear way to represent the unidirectional movements of goats during a day.

(45) Volcanic Activity

(46) We received official volcanological summary data from the Italian National Volcanological Institute (INGV) to characterize the magnitude of the volcanological events. The INGV runs at least 26 semi-automated measurement stations around Mt. Etna and also conducts visual observations and on-site chemical and geographical/geological measurements on Mt. Etna continuously. Measurements we used in our characterization of the overall magnitude of the volcanic event included the seismic activity of Mt. Etna as well as descriptions of the volcanological events such as the altitude of the volcanic eruption and the magnitude of material emitted during an event.

(47) Based on these data we highlighted 9 volcanic events during the study period. Most of these events were minor in the sense that only small amounts of ash were emitted or only local lava fountains were seen that did not produce lava flows to the outside of the volcano. Thus, people and animals on the outskirts of the volcano were not visually affected by these events, and no tremors of earthquakes were felt by humans in the area of Randazzo. Only two events were characterized as major during the current study period. The first one was the 19.sup.th paroxysmal event starting as a major event in the morning hours of January 5 and lasting until approximately midday of that day, the second one was the 21.sup.st paroxysmal event. Official descriptions of these events by the INGV are given below.

(48) The 19th Paroxysmal Event (Based on Official Information Published by the INGV):

(49) The 5 Jan. 2012 Paroxysmal Eruptive Episode at Etna's New Southeast Crater

(50) Following a quiet interval of 50 days, the New Southeast Crater (New SEC) of Etna reactivated on the evening of 4 Jan. 2012, and produced the first paroxysmal eruptive episode of the year (the 19th since the beginning of the series initiated on 12 Jan. 2011) on the morning of 5 January. The photo in FIG. 1 shows the acme of this paroxysm, shortly after 06:00 GMT.

(51) The reawakening was preceded by various signs of unrest recorded by the observation systems of the INGV-Osservatorio Etneo (INGV-OE) during the first few days of 2012; these included strong fluctuations in the volcanic tremor amplitude, an increase in degassing from the Bocca Nuova that culminated in an explosion quake accompanied by a minor emission of vapor and ash on the evening of 2 January, and finally by the resumption of weak explosive activity within the New SEC on 4 January. About 08:20 GMT on 4 January, small explosion signals started to be recorded by the EBELO infrasonic recorder, located about 0.9 km to the southeast of the crater.

(52) On the late evening of 4 January weak incandescence was visible in correspondence with the New SEC; however, observations were strongly hampered by inclement weather. From 22:30 GMT Strombolian activity was observed intermittently by INGV-OE staff from various sites on the southeastern and northeastern flanks of the volcano, and from 02:00 GMT on 5 January the activity was under continuous observation. Around 02:45 GMT, a small lava flow began threading its path across the deep notch curring the southeastern crater rim; this flow advanced very slowly following the same path of the lava flows emitted during the previous paroxysmal episodes.

(53) During the following hours, the Strombolian activity increased in intensity and from 04:00 GMT it increased more rapidly to become virtually continuous. Between 04:45 and 05:00 GMT, the Strombolian activity passed into discontinous, pulsating fountaining generating jets 100-150 m high.

(54) About 04:50 GMT, ash emission had become significant, and this was accompanied heavy fallout of scoriae, spatter, and bombs onto the flanks of the cone. From 05:15 GMT onward, lava fountaining was continuous, generating an eruption column of ash and vapor that rapidly rose in height, reaching an elevation of 7000-8000 m above the sea-level around 06:00 GMT (see FIG. 1).

(55) During the time interval between 05:35 and 05:45 GMT, incandescent pyroclastics completely covered the cone, which interacting with snow began to form avalanches and small pyroclastic flows extending for a few hundred meters. These flows repeatedly pushed far into the snow cover at the base of the cone, provoking phreatomagmatic phenomena and small lahars (mud flows), in particular on the northeastern, eastern, and southern flanks of the cone. The longest flows nearly reached the central portion of the eruptive fissure of 13 May 2008.

(56) The vents on the upper northern flank of the cone emitted a small lava flow that travelled a few hundred meters stopping before reaching the upper portion of the 13 May 2008 eruptive fissure.

(57) Around 06:00 GMT, several eruptive vents activated along the fracture that cuts the northern rim of the New SEC cone, producing small intermittent lava fountains. At 06:20 GMT, a powerful explosion marked the opening of a vent on the upper southeast flank of the cone, destroying a portion of the southeastern crater rim.

(58) Shortly after 06:30 GMT, the Bocca Nuova emitted a puff of ash, followed by weaker emissions of ash mixed with ash. At the New SEC, paroxysmal eruptive activity continued with full vigor until 06:57, and then terminated rather brusquely within the next few minutes. Only passive emission of ash continued after this until about 07:30 GMT at the New SEC, and lasted until 08:30 GMT at the Bocca Nuova.

(59) This paroxysmal episode has occurred after one of the longest repose intervals of the current eruptive sequence initiated one year ago; only the intervals between episodes #2 (18 Feb. 2011) and #3 (10 Apr. 2011) and between episodes #4 (12 May 2011) and #5 (9 Jul. 2011) were longer51 and 58 days, respectively. In terms of explosivity, this was one of the most violent events of the sequence, but the quantity of lava emitted was much inferior to that of previous episodes. The main lava flow toward southeast in the direction of the Valle del Bove, advanced little more than 2 km, flanking the northern side of the Serra Giannicola.

(60) For a few tens of minutes following the cessation of the paroxysm, the entire northern flank of the New SEC cone showed a wholesale gravitational movement due to the slow sliding of the abundant pyroclastic material deposited on that side. This process was accompanied by the release of abundant bluish gas, but did not result in the formation of a rheomorphic flow.

(61) The 21th Paroxysmal Event (Based on Official Information Published by the INGV):

(62) The 4 Mar. 2012 paroxysmal eruptive episode appeared at Etna's New Southeast Crater. The third lava fountaining episode at the New Southeast Crater (New SEC) of Etna in this yearthe 21st since the start of the current eruptive sequenceoccurred on the morning of 4 Mar. 2012. This event was more violently explosive, generating small pyroclastic flows and lahars (mudflows), due to the explosive interaction between lava flows and thick snow cover on the terrain (see FIG. 1).

(63) After the lava fountaining episode of 9 Feb. 2012, Etna remained quiescent for one week. On the morning of 16 February, small ash emissions resumed from the New SEC, and for 18 days, weak, sporadic Strombolian activity continued on the crater floor. Occasionally, faint glow was observed at night; there was also a conspicuous increase in the number of sources and in the volume of fumarolic emissions along the southern rim of the crater. During the last few days of February, this activity was accompanied by an increase of the explosive activity within the conduit of the Northeast Crater, producing loud bangs, which were well audible all over Etna's summit area. The volcanic unrest during the second half of February was accompanied by more accentuated fluctuations of the volcanic tremor.

(64) During the early morning hours of 4 Mar. 2012, the volcanic tremor amplitude showed a rapid increase; at the same time, the Strombolian explosions within the crater became more frequent and more intense. Shortly after 06:00 GMT (local time 1), lava started to overflow through the deep breach that cuts the southeastern rim of the crater. The lava flow reached the southeastern base of the cone after about 15 minutes and from there advanced toward the western rim of the Valle del Bove. In the meantime, the explosive activity was continuously waxing, and passed into continuous lava fountaining with development of an eruption column about 07:30 GMT. The abundant fall of large-sized pyroclasts onto the steep flanks of the cone led to the formation of rock and dust avalanches; around 07:50 small pyroclastic flows were generated by the partial collapse of the eruption column. These flows descended mainly on the northeastern flank of the cone, and to some lesser degree on the south flank.

(65) Also around 07:50 GMT, a lava flow was emitted from a new eruptive vent on the upper southwestern flank of the New SEC cone and started to descend in the saddle between the old and new SEC cones, interacting violently with thick snow covering the ground. This interaction provoked powerful explosions and small pyroclastic flows, the largest of which advanced rapidly across the flat terrain immediately to the east of the first eruptive fissure that opened on 17 Jul. 2001. Melting of the snow, in turn, led to the formation of a lahar, which descended toward the Belvedere monitoring station, on the western rim of the Valle del Bove, passing a few tens of meters to the north of the monitoring instruments.

(66) During the phase of maximum intensity in the eruptive activity, a lava flow was also emitted from an eruptive fissure on the upper northern flank of the cone. This flow descended a few hundred meters toward northeast, surrounding the northern base of the cone. The main lava flow, which was fed across the breach in the southeastern rim of the crater, followed a nearly identical path to that of the lava flow emitted during the 9 February eruptive episode. After descending the steep western slope of the Valle del Bove, the flow split into several branches on the more gently sloping terrain at the base of the slope. These branches exceeded in length those of the 9 February flow, reaching a total distance of about 3.5 km from the crater.

(67) The lava flow emitted from the fissure on the southwestern flank of the cone remained active for a few hours after the cessation of the paroxysmal activity, advancing slowly on the trace of the lahar that had occurred during the culminating events of 07:52 GMT.

(68) The advance of the lava flows on thick snow cover was often accompanied by phreatic explosions, which generated violent jets of vapor and launched rock fragments to several tens of meters away; these phenomena were observed along the southern lava flow and along the main lava flow descending into the Valle del Bove.

(69) Shortly after 09:00 GMT, the activity showed the first signs of diminishing in intensity; lava fountaining ceased at 09:32, two hours after the onset of the paroxysmal phase.

(70) This episode occurred 24 days after the preceding one, of 9 Feb. 2012, and was considerably more violent. The eruption column reached a height of several kilometers above the summit of Etna. Ash and lapilli were carried by the wind toward northeast, affecting the areas around Piedimonte Etneo and Taormina. Fine ash fell as far as the Messina area and southern Calabria. Once more, the pyroclastic cone of the New SEC has grown in height, mainly on its northern rim.

(71) Statistical Analysis

(72) Data were analyzed using Movebank and SPSS (2011) for Windows. We present data as averagesstandard deviation, except where indicated.

EXAMPLE 2

(73) Abberrant Nocturnal Activity of Goats in Anticipation of a Volcanic Event in the Morning

(74) We determined the nocturnal sum of the individual average acceleration variances for 8 goats during the 5 days leading up to a major volcanic event, during the night before the major volcanic event, and for 5 days after the major volcanic event. The data show a clear peak in the sum of average individual acceleration variances per night in anticipation of the major volcanic event (see FIG. 7A).

(75) For the prediction of a major volcanic activity, we suggest observers should use the doubling of the sum of acceleration variances above the 10-day population average.

(76) In contrast to a major volcanic event (as defined above), we did not find any significant aberrant nocturnal behavior of the goats during a minor event.

(77) Time Course of Aberrant Nocturnal Activity of Goats in Anticipation of a Volcanic Event

(78) To understand the temporal scale of predictability of a major volcanic event, we used the accumulation curve of average goat activity, measured as acceleration in the z-axis, during control nights preceding the major volcanic event. We compared these accumulation curves to the curve during the night before a major volcanic event.

(79) We found that approximate 5 hours before the major volcanic event, the average acceleration behavior of goats significantly increased above background levels (see FIG. 8).

(80) More specifically, the average acceleration behaviour of goats per night was plotted against time as an acceleration curve, i.e., adding all average acceleration values together over time (see FIG. 8). During control nights, there is a slow but constant accumulation of average acceleration behaviour. Only during the night with a major volcanic activity there existed a sudden increase in the accumulation of average acceleration behaviour. The slope of the accumulation curve remained steeper after the sudden increase, compared to control nights.

(81) For the prediction of a major volcanic activity, we suggest observers should use a more than 30% increase in the nocturnal accumulation curves of average acceleration behavior of goats. In contrast to a major volcanic event, we did not find any significant aberrant nocturnal accumulation of average acceleration behavior of the goats during a minor event.

(82) Linear Daily Movement as Predictor of Major Volcanic Activity

(83) To understand whether one could predict a major volcanic event from the diurnal behavior of animals, we determined the sum of daily uni-directional location movements of 8 goats (see FIGS. 9A and B).

(84) For the prediction of a major volcanic activity, we suggest observers should use the doubling of the daily unidirectional movements of a population of goats above the 10-day population average.

(85) In contrast to a major volcanic event, we did not find any significant aberrant daily unidirectional movements of the goats during a minor event.

EXAMPLE 3

(86) Forecasting Volcanic Events by using a Mixed Population of Animals

(87) Data over one year from 9 goats and 4 sheep at Mount Etna in Sicily, Italy have been gathered. The study site continued to be the northern flanks of Mt. Etna upwards from the small town of Randazzo. Exactly the same methods as described herein above have been used.

(88) Mt. Etna was seismically and volcanically active during the period of measurement. As in the foregoing Examples, the volcanic events have been classified into major and minor events. Major events consisting of volcanic eruptions with tephra stones thrown high into the atmosphere and ash clouds rising up to several kilometers. Minor events are those that only showed eruptions rising into low altitudes of a few 100 meters above the peak of Mt. Etna or volcanic eruptions that only occurred within the crater of Mt. Etna. In particular, during minor events only small amounts of ashes are emitted or only local lava fountains occur that did not produce lava flowing to the outside of the volcano (Volcanic Explosivity Index (VEI) of 1). Although minor events are volcanically significant, they do not comprise any danger to the lives of animals living along the slopes of Mt. Etna. Even major events most likely do not kill or harm animals, but could at least become very unpleasant for them, e.g. when hot tephra stones or acidic ash are falling down upon them.

(89) The classification into major and minor events has been confirmed by human visual and acoustical observation. In particular, a paroxysm occurred on Apr. 27, 2013. During the data readout during the paroxysm of Apr. 27, 2013, human witnesses observed the lava eruption and explosion. While reading out the data from the goats in Randazzo, black ash clouds rise from the volcanic caldera. No behavioral changes in the goats and sheep have been detected before or during the paroxysm. Based on the pre-set classification scheme described herein, the paroxysm of Apr. 27, 2013 has been classified as a minor event.

(90) TABLE-US-00001 TABLE 1 Major events that occurred Time when the threshold Date of the Time of the triggered an major event major event alert/notification Apr. 12, 2012 14:30 11:15 Apr. 24, 2012 1:40 22:25(on Apr. 23, 2012) Feb. 19, 2013 4:10 23:20(on Feb. 18, 2013) Mar. 5, 2013 23:12 20:55 Mar. 16, 2013 17:00 14:15

(91) Similar to the previous major events (that were part of Example 2), the 5 additional major events were predicted by the behavior of the 9 goats and 4 sheep ahead of time. The alert thresholds set forward herein were used in the additional events as well, and thus comprise general alert thresholds for disaster forecast (e.g when larger vertebrates are used).

(92) Again, unidirectional movement during the daytime events and 3D-acceleration-triggered activity events at night were used as described herein. In particular, the threshold value has been determined as follows.

(93) The threshold value was the double average value of the summated daytime-integrated unidirectional movement of the population. Accordingly, the test profile was generated in the same manner: the unidirectional movement of each animal was integrated over the whole day and daytime-integrated unidirectional movements of the test night of each animal of the population were added. Thus the alert was set, if the ratio between the test profile and the reference profile was 2, according to the aforementioned threshold.

(94) The threshold value was the double average value of the summated overnight-integrated activity of the population. Accordingly, the test profile was generated in the same manner: the activity of each animal was integrated over the whole night and the overnight-integrated activities of the test night of each animal of the population were added. Thus the alert was set, if the ratio between the test profile and the reference profile was 2, according to the aforementioned threshold.

(95) Compared to the volcanic events, low earthquake activity (i.e. seismic events below 5 on the Richter magnitude scale) at Mt. Etna did not allow for any predictions of volcanic activity. As shown in FIG. 14, seismic activity at Mt. Etna is quasi continuous, with few major events above category 4 on the Richter magnitude scale. These events do not coincide with the volcanic events as determined in FIG. 13. The goats and sheep did not react to or anticipate low seismic events (i.e. seismic events below 5 on the Richter magnitude scale), as expected because these events do not pose any threat nor are the highly visible for humans. As the population of animals did not forecast low seismic events (i.e. seismic events below 5 on the Richter magnitude scale), false-positive alerts were prevented.

(96) In summary, these data confirm that the described method and threshold can be used to predict major volcanic events at Mt. Etna, and potentially at other active volcanoes around the world, as well as potentially other natural disasters.

CONCLUSION

(97) By using remote sensing of behavior via 3D-acceleration and GPS location logging, and transmitting data via VHF telemetry, we showed that feral goats and/or a mixed population of animals on Mount Etna, Sicily, actively engage in behaviors that allow for the remote prediction of major volcanological events.