Sensor activities of internet user population in seismic hazard

Abstract

FIELD: geophysics.SUBSTANCE: invention relates to the field of geophysics and can be used to diagnose the activity of Internet users during seismogenic disturbances of geophysical fields. The method is based on the flow of electronic messages in Internet search engines for specified regions using a set of keywords that are indirectly related to disturbances of geophysical fields before earthquakes. Local extremes and situations with an abnormal distribution of the spectrum of variations with increased power levels during "signal" periods are identified in the time series of estimates of the repeatability of the set of keywords. Examples of sensor implementation in the diagnosis of seismic hazard are given.EFFECT: technical result is to increase the efficiency of assessing the activity of Internet users in seismogenic disturbances of geophysical fields.2 cl, 11 dwg, 2 tbl

Description

[0001] The proposed method for assessing Internet user activity during seismic hazard relates to ecology and geophysics, and can be used to assess the seismic hazard of regions.

[0002] Human perception is primarily visual. However, visual perception enables environmental assessment and the evaluation of population status in other species, which rely more heavily on other senses. The presence of a second sensory system in humans allows us to record these processes and phenomena. Internet resources allow for the automated collection and analysis of information on internet user activity.

[0003] The content of the proposed invention is based on the principles of general systems theory, fuzzy logic, information systems theory, information systems design theory, semantic information processing theory, and experiments in relation to artificial intelligence information technologies for collecting, processing, and analyzing information from Internet sources.

[0004] To record environmental changes, including through observations of the behavior of various populations of organisms, an analysis of the flow of recorded events on the internet is used. Each internet user can contribute to news feeds and queries.

[0005] The prototype of the method is proposed based on the capabilities of Internet search engines (machines) presented in [1] for marketing research (segmentation and positioning) based on the screening of information flows. The method for predicting the popularity of a search query [1] uses a search robot database, data from the search log, and browser log. Data from the search log represents the search activity of search engine server users based on the content of the search query. Data from the browser log represents the browsing activity of browser application users based on the content of the search query. Statistical web data is obtained from the search robot database, which represents embedded objects or links to a content element contained on web resources based on the content of the search query. Based on data from the search log, browser log, or statistical web data, the popularity of the analyzed query is predicted.

[0006] The prototype does not take into account that the popularity of some queries is related to changes in the natural environment to which internet users respond. Changes in internet user activity for a specific set of query words are used to assess environmental conditions, for example, during periods of seismic hazard before earthquakes. The influence of environmental changes during seismogenic disturbances of geophysical fields on internet user behavior is schematically shown in Figure 1.

[0007] Geophysical field disturbances during seismic hazard periods are described within the framework of biological and other types of earthquake precursors. A brief ontological model of precursors can be represented by a set of keywords. For each region, the keyword set may depend on geophysical conditions. Diagnostics of keyword query popularity allows us to assess the presence of geophysical field disturbances initiated by the evolution of a seismotectonic anomaly with a future earthquake source.

[0008] The purpose and technical result of the invention is to increase the efficiency of assessing the activity of Internet users in seismically active regions for diagnosing regional seismic hazard.

[0009] To implement this method, use:

[0010] 1) at least one personal computer with Internet access and access to search engines,

[0011] 2) sets of keywords,

[0012] 3) data processing, archiving and visualization tools,

[0013] 4) criteria for the activity of Internet users' behavior.

[0014] Technical implementation of the analysis of the flow of events in a given territory for given keywords and the behavior of populations of organisms is provided through the tools of search engine providers (e.g., Yandex™, Google™, Bing™, etc.), content aggregators (e.g., Digg™, Reddit™), content recommendation systems (e.g., StumbleUpon™, Pinterest™), etc. They often combine several functions. For example, large social networks such as Facebook™ and Twitter™ store billions of user messages and provide the ability to post new ones [2]. With the development of Internet technologies, the need to create server applications for search engines has disappeared. Attempts to use native server applications are given in [3, 4].

[0015] The set of keywords for assessing internet user activity during geophysical field disturbances should be oriented toward a given territory (e.g., Kamchatka), taking into account the ontological model of the diagnosed seismotectonic process and the nature of the populations. The technology and statistics for ranking keywords by precursors using internet search queries and the "analytic hierarchy process" method are presented, in particular, in [5]. Models of seismic activity precursors are presented in [6, 7].

[0016] Developing a region-specific set of keywords characterizing the manifestation of precursor disturbances of geophysical fields in seismically hazardous regions involves selecting the most significant terms based on the psychology of the region's internet users. For example, most internet users don't consider, for example, "hydrogeological precursors." However, the increasing popularity of the term "water" in information flows may indirectly indicate a change in the region's hydrogeological conditions. The internet population amplifies the appearance of weak signals in response to keyword queries [8].

[0017] Regional keyword patterns are also linked to the natural environment. Keyword formation can proceed from the general to the specific based on learning. For example, for Kamchatka, the terms "water" and "fish" would be particularly significant. Based on the general keyword set, an adaptive keyword set is created for each region, which best reflects internet user activity during seismic hazards. This adaptive term set should be refined after each strong seismic event.

[0018] Earthquake intensity determines the scale of precursor disturbances in geophysical fields and, accordingly, the ability of internet users to respond to them. Internet search engines provide the ability to define the boundaries of the region being studied. Additional geographic limitations are associated with the use of analyzed terms in the predominant national languages.

[0019] In the search for a basic set of query words to assess the activity of the Internet society in case of seismic hazard in [3] before almost two dozen strong earthquakes in various regions of the world, the following terms showed increased popularity: accident, health, water, air, radioactivity, and their English equivalents: Air, Crash, Health, Radioactivity, Water. The bilingual nature of this set is due to the fact that in the Russian-language Internet the share of Google search engines used is comparable to the share of Yandex search engines [2]. The term "water" showed the highest frequency in queries - at least 75% of the examined cases with a lead time of several days before the seismic event (Table 1). Therefore, in addition to using additive estimates for the repeatability of the set of specified keywords, which was tested in [3, 9-11], a search query for the term "water" was taken as the operational basis for diagnosing seismic hazard. And preferably in the language of Internet users in the region.

[0020] Differences in the frequency of language terms in English and the "national" language used in the analyzed region can be illustrated using the example of a Google search engine. Figure 2 demonstrates these differences by the different levels of the query frequency curves for the term "water" in Greece in English (curve 1) and Greek (curve 2); the ordinate scale is given as a percentage of the minimax of query frequency estimates for all Greek cities corresponding to queries for the term "water." The height of the bars to the left of the graphs demonstrates the average estimate for the analyzed quarter and the fact that the English term is more actively used in search queries (the different heights of the bars for the average estimates of curves 1 and 2).

[0021] Experience using this method has shown that using terms in the regional official language can be more informative. As an example of this, Figure 3 presents estimates of the frequency of the term "water" in Burma in three languages ​​(Russian, English, and Burmese). Before the strong earthquake in Myanmar on March 28, 2025. In the official language of Burma, clear manifestations of seismic hazard appeared several days before the earthquake, although in Fig. 3 they appear small against the background of the level of queries in Russian and English, which require an analysis of subtle effects.

[0022] So, for analyzing internet user behavior based on the frequency of keyword queries, there are software solutions in internet search engines [2], which are not only a tool for searching for information but also for analyzing internet user activity. They include programs for text processing, linguistic search, and information analysis. For example, Google Analytics

[12] allows for assessing the activity of queries used, related to the ontology of potential processes or phenomena, and offers a number of metrics for detailing internet user activity. In Russia, domestic hardware and software systems have been developed for operational assessments of internet user sentiment and activity, such as the "Prognoz" system for modeling and forecasting the development of situations (poisk-it.ru, where some examples have now been removed from the site).

[0023] Time series analysis of the obtained frequency estimates for selected terms in search engine results can be implemented based on the prototype

[11] , which utilizes parametric statistics and spectral analysis methods. These methods have been successfully tested in several regions of the world [11, 14, 15].

[0024] The main stages of the proposed method and their interaction are shown in Fig. 4. They include:

[0025] 1. Planning and management of reception, processing and transmission of information according to network plans for the operation of sensors;

[0026] 2. Selecting a region and keywords for screening, taking into account recommendations, with subsequent correction and adaptation to the geophysical conditions of the analyzed region;

[0027] 3. Analysis of the intensity of the flow of keywords in the Internet information flow using search engines;

[0028] 4. The formation of time series of the analyzed characteristics is carried out continuously and for different time periods (usually for average daily estimates from several days to several months);

[0029] 5. Processing of received data is performed using parametric and nonparametric statistics, spectral and fractal analysis. To improve the homogeneity and comparability of time series for query word frequency, it is advisable to use relative characteristics obtained, for example, by normalizing frequency estimates using the minimax method in the analyzed time period. This will allow for the correct aggregation of time series for each query word used, for example, using additive or multiplicative techniques (and their combinations), into a generalized time series of population activity.

[0030] 6. Based on the results of section 5, an assessment of internet user activity during seismic hazards is made using the existing "traffic light" indication criteria. Feedback from section 2 is required to adjust and validate the set of keywords specific to the region being studied and its detailing;

[0031] 7. The content of queries is adjusted when the compliance (entropy) indicators of Internet user activity improve after events with increased seismic hazard;

[0032] 8. Data transmission to data processing and analysis reception points via established communication channels and network schedules;

[0033] 9. Secondary data processing and comparison of results with models are necessary for validation of the proposed sensor of Internet user activity during seismic hazard;

[0034] 10. Archiving of received information as a mandatory stage and reporting on the functioning of the method for its verification and validation;

[0035] 11. Presentation of Internet user activity results and reference information on the computer display device.

[0036] The results of training and validation, as well as the feasibility of replicating the sensor, are presented, for example, in [3, 8, 9, 10, 11]. In addition, examples of diagnosing internet user activity before a series of earthquakes in 2025 are offered.

[0037] Santorini Earthquake. From late January to mid-February 2025, a surge of macroseismicity—nearly 20,000 underwater earthquakes—was recorded north of Crete near the island of Santorini. Many of these earthquakes exceeded magnitude 4.5. The strongest earthquake occurred on February 4 and had a magnitude of 5.3.

[0038] For the technical implementation of the proposed sensor, the resources

[16] and the term “vspo” (Greek, water) were used, which limits the screening region with the help of In the search engine, the search region was additionally indicated - Greece [https: / / trends.google.com / trends / explore?date=today%201-m&geo=GR&q=%CE%9D%CE%B5%CF%81%CF%8C&hl=ru]. The distribution of the total number of average daily queries for the specified term is presented in Fig. 2.

[0039] To perform a detailed search for internet user activity signals in the region, amplitude charts of the time series (Fig. 2) were calculated using a fast Fourier transform and assigned to the right boundary of a 32-day sliding window. The amplitude estimates for each period, calculated in the interval from January 18, 2024, to February 17, 2025, were normalized by its minimum and maximum values ​​(Fig. 5). Periods 0, 1, and 16 were not used for analysis.

[0040] Figure 5 uses a five-gradation display of the normalized amplitude intensity. Red and yellow colors are used for seismic hazard. Minimum amplitudes, indicated in blue, can also be used to compare the imbalance with hazardous amplitude levels.

[0041] The statistical characteristics of the calculated array are presented in Table 2. Potential signal periods of internet user activity during seismic hazard periods are identified. The following notations are used in the table: avg – mean value, sd – standard deviation, kvar – coefficient of variation, skewness – coefficient of asymmetry, ekc – coefficient of excess.

[0042] In the morphology of Fig. 5, three time periods are distinguished: 1) a quiet period until 2025, 2) increased activity of Internet users (seismic hazard), and 3) the period after a strong earthquake. In the first time period at the end of 2024, an increased amplitude appeared over a period of 9 days (horizontal dashed arrow). In the second period in 2025, bimodality is formed in the manifestation of increased amplitudes over periods of 10 and 6 days with minimum amplitudes over a period of 8 days, then the process of bimodal increase in amplitudes covers periods of 3 ± 1 days, that is, higher frequencies. For the earthquake of 04.02.2025, the main "signal" periods are 3 ± 1 and 6 days. These are the criteria for point 6 in Fig. 4. Following the earthquakes, stabilization and stagnation of variations across all periods occurred, which was consistent with Greek seismologists' assessments of a decrease in seismic activity in the Santorini region. The weak underwater volcanic earthquake of February 17, 2025.The island of Anidros has not shown any sign of Internet community activity.

[0043] These signal periods have been observed in internet user activity in other regions, such as the November 5, 2018, Japanese earthquake and subsequent eruption of Sakurajima volcano on the southern Japanese island of Kyushu on November 14, 2018

[10] .

[0044] In studies of Internet user activity, it was taken into account that the relationship between magnetic and seismic activity is weak [8, 17]. Increased solar activity, measured by the flux power at a wavelength of 10.7 cm, was recorded from January 16 to 23.

[0045] In Table 2, the largest variation coefficient was observed over a 3-day period. This period is a multiple of 6 days, for which Fig. 6 presents the cospectrum amplitude calculations over a 16-day sliding window, with the calculated amplitudegrams assigned to the right boundary of the window. A fast Fourier transform was used in the calculations.

[0046] The first (time) arrow in Fig. 6 corresponds to internet news feed posts for January 26, 2025, but according to the Geophysical Survey of the Russian Academy of Sciences

[18] , no earthquakes were recorded on that day. This example is given because the sensor can quickly perform the seismograph function when the term "earthquake" is used in a query, and there is an intensity level for earthquakes recorded by the Geophysical Survey of the Russian Academy of Sciences.

[0047] The sum of the amplitudes of the first five periods in Fig. 6 accounts for up to 95% of the sum of the amplitudes of all periods. It's worth noting that the amplitudes (internet user activity) reached their maximum well in advance of the strongest earthquake. The identified informational precursor manifested itself in the morphology of the calculated amplitudegrams and in the dynamics of periods 3, 4, 6, and their 12-day multiple.

[0048] The Istanbul earthquake of April 23, 2025 (09:49:10 (GMT) occurred in the Sea of ​​Marmara, several tens of kilometers from Istanbul, with a magnitude of M=6.3

[18] , the coordinates of the epicenter are latitude ϕ=40.92°, longitude λ=28.24° (Fig. 7), the depth of the hypocenter was estimated at 10 km). To analyze the activity of Internet users, the term "su" (Turkish, "water") was used using a search engine

[16] . The main focus in the flow of queries was on Istanbul (Fig. 8), since anomalous changes in the activity of Istanbul Internet users were detected even during macroseismicity near Santorini. But then Istanbul had lower activity compared to other Turkish corresponding cities. After the main earthquake near Santorini, Santorini Internet user activity in Istanbul has begun to increase.

[0049] The data processing procedure is the same. The normalized amplitudegrams of the time series in Fig. 8 over a 32-day sliding window are shown in Fig. 9. The amplitudegrams of the time series in Fig. 8 were calculated using the fast Fourier transform and assigned to the right boundary of the sliding window. In each period, the amplitude estimates calculated in the interval from April 10, 2025 to April 24, 2025 were normalized by its minimax.

[0050] The design of Fig.9 also uses five gradations of amplitude intensity.

[0051] In the morphology of Fig. 9 before the earthquake, dominant periods of internet user activity are identified at 3 and 6-7 days, as before earthquakes in other regions. During the 2-4 days, with a general trend of increasing amplitudes at higher frequencies, Fig. 9 shows a peak in activity during the 2-day period.

[0052] An elevated K-index of magnetic activity was recorded on April 16 (K=5+) and April 21 (K=4+)

[19] . Strong M-class solar flares were recorded on April 19, 21, and 22. The third quarter of the Moon occurred on April 20-21.

[0053] Thus, for the Istanbul earthquake, the increase in searches for the term by Istanbul internet users anticipated the evolution of seismic hazard. The method works.

[0054] Earthquake in Myanmar (Burma). The Burmese term for "water" (Fig. 3) was used to reliably predict the heightened activity of Burmese internet users more than a week before the strong earthquake in Myanmar (March 28, 2025, 12:50:52 UTC+06:30, magnitude Mw=7.7).

[0055] Kamchatka Earthquake on July 30, 2025. Figure 10 shows the results of activity diagnostics (in % relative to the quarterly minimax) of Internet users in Petropavlovsk-Kamchatsky and Yelizovo before the earthquakes on July 20 and 30, 2025. The latter event was anomalous and affected the whole of Kamchatka. It did not require diagnosing the subtle effects of Internet user activity using spectral analysis. Figure 10 clearly shows a clear and advance increase in Internet user activity before the earthquake on July 20, 2025, after which the activity of Internet users did not decrease, which is also anomalous, and was supplemented by the recording of events associated with the unusual behavior of the population of birds, fish, and animals, which is consistent with the descriptions of biological precursors of earthquakes [6]. The collection of this information using the proposed method is carried out promptly.

[0056] The term "su" in 2025 showed activity among Internet users in Dagestan in advance of weak coastal earthquakes in the Caspian Sea.

[0057] Earthquake in Turkey on August 10, 2025. Magnitude approximately 6.1. The epicenter was located in the Sındırı district of Balikesir province, at a depth of 11 kilometers at 19:53 local time. Using this earthquake as an example, Figure 11 shows how the proposed method was used to detect changes in internet user activity for the query "water pH" in the earthquake's epicentral zone. Changes in water quality are a manifestation of geochemical precursors.

[0058] Internet user activity for this search term has been increasing in Balikesir province since the end of July. This was not the case in other provinces.

[0059] The success of seismic hazard diagnostics using the proposed sensor is related to the scale of disturbances in geophysical fields, which are determined by the magnitude of an impending seismic event. A sufficiently dense network of internet users increases the possibility of detecting changes in the state of geophysical fields and the behavior of environmental populations. A preliminary assessment of the correspondence between internet user activity and seismic hazard based on a basic set of query words exceeds random guessing and, when the technology is adapted to a specific region, approaches the golden ratio. When abnormal internet user activity is detected using specified keywords, a signal is generated for official agencies responsible for seismic monitoring, confirming or refuting their inquiries about the level of seismic hazard.

[0060] The technological novelty of the proposed sensor is confirmed by the author’s publications, discussions of the contents of the proposed invention at scientific and technical events, and the presented materials.

[0061] The profitability of the proposed seismic hazard sensor is ensured by the active development of the Internet and social networks, accessibility (including cost) and the efficiency of obtaining data on the characteristics of information flows.

[0062] Literature

[0063] 1. Gusev G.G., Drutsa A.V., Serdyukov P.V. Method and server for predicting the popularity of a content element / Patent for invention. Application: 2015140585 dated September 23, 2015, application publication date March 29, 2017, Bulletin No. 10; Published November 16, 2017, Bulletin No. 32. Patent holder: YANDEX Limited Liability Company.

[0064] 2. Yudin A. World search engines, statistics 2018 / https: / / marketer.ua / search-engine-stat-2018 /

[0065] 3. Tertyshnikov AV The internet-community response to the Mexican Earthquake of February 1, 2019 / E3S Web of Conferences 196, 03007 (2020) STRPEP 2020 /

[0066] doi.org / 10.1051 / e3sconf / 202019603007.

[0067] 4. Tertyshnikov AB, Oltyan I.Yu., Milyukov VK, Vereskun AV Experimental studies on the development of a short-term seismic hazard index for the North Caucasus / Global and national strategies for disaster and catastrophe risk management. XX International Scientific and Practical Conference on the Problems of Protecting Population and Territories from Emergencies, May 19-21, 2015, Moscow.

[0068] 5. Tertyshnikov A.V., Pisanko Yu.V., Davydov V.E., Zinkina M.D., Konstantinova A.V. Expertise of the prospects of earthquake precursors / / Heliogeophysical studies. Issue 22, 2019. P.12-17. http: / / vestnik.geospace.ru / index.php?id=532.

[0069] 6. Tertyshnikov A.V. Earthquake precursors and features of their registration. - St. Petersburg: VIKA. 1996. 128 p.

[0070] 7. Tertyshnikov A.V. Seismo-ozon effects and the problem of earthquake forecasting. - St. Petersburg: VIKA, 2000. 258 p.

[0071] 8. Tertyshnikov A.V. Website traffic of the Federal State Budgetary Institution "IPG" and magnetic activity in 2018 / / Heliogeophysical research. 2019, Issue 21. Pp. 12-17. http: / / vestnik.geospace.ru / index.php?id=526.

[0072] 9. Tertyshnikov, A. V. (2023). The reaction of Internet Users to the Seismic Hazard. World Journal of Environmental Biosciences, 12(4), 14-17. https: / / doi.org / 10.51847 / VzkUqyDLWc.

[0073] 10. Tertyshnikov A.V. The eruption of Sakurajima volcano on November 14, 2018 according to semantic criteria of seismic hazard on the Internet / / Heliogeophysical research.

[0074] 2022. No. 35. P.12-20.

[0075] 11. Tertyshnikov A.V. Internet screening of seismic hazard signs / In the collection: Problems of military-applied geophysics and monitoring of the state of the natural environment. Proceedings of the VII All-Russian scientific conference. General editor Yu.V. Kuleshov. St. Petersburg, 2022. pp. 133-139.

[0076] 12. https: / / ru.wikipedia.org / wiki / Google_Analytics /

[0077] 13. Tertyshnikov A.V. Method for sensing seismo-orbital effects and variations in the density of the upper atmosphere / Patent for invention No. 2705161. IPC G01V 9 / 00 (2006 / 01). Application No. 2019112175 / 28 (023645) dated 04 / 22 / 2019. Published 12 / 05 / 2019. Priority until 2039.

[0078] 14. Tertyshnikov A.V. Fundamentals of monitoring emergency situations. Study guide. - Moscow-Obninsk, 2013. 278 p.

[0079] 15. Tertyshnikov A.V. Organization of forecasting of natural emergencies. - Moscow, 2013. 268 p.

[0080] 16. https: / / trends.google.com / trends / explore?date=today%201-m&geo=GR&q=%CE%9D%CE%B5%CF%81%CF%8C&hl=ru

[0081] 17. Tertyshnikov A.V. Assessment of the practical significance of geomagnetic precursors of strong earthquakes / / Heliogeophysical studies, 2013. Issue 3. Pp. 63-70. http: / / vestnik.geospace.ru / index.php?id=42.

[0082] 18. Unified Geophysical Service of the Russian Academy of Sciences, http: / / gsras.mmew / about.htrn

[0083] 19. http: / / www.celestrak.com / SpaceData / .

[0084]

[0085]

[0086] Figure Captions

[0087] (towards a method for assessing the activity of Internet users during seismic hazard)

[0088] Fig. 1 - Schematic diagram of the interaction of Internet users with seismogenic disturbances in the environment.

[0089] Fig. 2 - Flow of the number of requests (in %) for the terms "water" (1) and "vepo" (2, "water", Greek) in Greece during macroseismicity near the island of Santorini in early 2025.

[0090] Fig. 3 - Flow of the number of requests (in %) for Burma for the terms "water" (3) and " " (4), ray (5), before the earthquake (03 / 28 / 2025 by arrow 6) in Myanmar.

[0091] Fig. 4 - Main stages of the method.

[0092] Fig. 5 - Normalized amplitude charts of the time series of Fig. 2. The vertical arrow indicates the day with a strong earthquake. The dashed arrows indicate the trends in the activation of periods.

[0093] Fig. 6 - Change in the amplitude spectrum of the minimax-normalized amplitudes for a 6-day period in Fig. 3 over a 16-day sliding window. The numbers next to the curves correspond to the period number (in days). Black vertical arrows indicate days with earthquakes.

[0094] Fig. 7 - Epicenter and zone of maximum impact of the Istanbul earthquake of 23.04.2025.

[0095] Fig. 8 - Change in the number of queries (N, %) for the term "su" for Istanbul. The vertical arrow indicates the day with the earthquake.

[0096] Fig. 9 - Normalized amplitude charts of the time series of Fig. 8 over a sliding window of 32 days. The vertical arrow indicates the day with the earthquake.

[0097] Fig. 10 - Popularity dynamics (in % by minimax per quarter) of Internet user queries in Petropavlovsk-Kamchatsky and Yelizovo before the earthquakes of July 20 and 30, 2025 by terms: 7 - water, 8 - air, 9 - animals, 10 - birds, 11 - fish.

[0098] Fig. 11 - Dynamics of popularity (in %) of Internet user queries in Balikesir Province (Turkey) before the earthquake on August 10, 2025 for the term water pH.

Claims

1. A method for assessing the activity of Internet users in seismic hazard regions, which consists in the fact that in seismically hazardous regions a set of characteristics of geophysical fields is regularly measured, which are converted into a generalized time series, in which anomalies are diagnosed using the control chart method and in the spectrum of the generalized time series, an assessment is made of the identified anomalies relative to average values ​​and taking into account regional characteristics according to the criteria of high, medium hazard and calm conditions, the obtained results are archived in the center for receiving, processing and storing information, displayed in graphical or text form, and when new data is received after earthquakes, the characteristics of the anomalies of the generalized time series are clarified, while the analysis also uses time series formed based on the results of assessments of the frequency of query words passed through search engines,to which internet users respond during seismogenic disturbances of geophysical fields: water, air, accident, health, radioactivity, and the queries take into account the geographical location of the territory, while in order to increase the homogeneity of the time series of the frequency of query words, relative characteristics are used, obtained by normalizing the estimates of the frequency of query words by the minimax in the analyzed time period, and in the analysis, an additive estimate of the frequency of the used query words is used, statistical maps of the distribution of the frequency of query words are regularly updated, after strong earthquakes, validation of the assessments of the activity of internet users during seismic hazard is carried out based on a set of query words, and when abnormal activity of internet users is detected for the specified keywords, a signal is generated for the official structures responsible for monitoring the seismic regime,which notify after receiving a warning about a seismic hazard, or refute it.

2. The method according to paragraph 1, characterized in that the analysis uses only query words in the official language of the analyzed region.

Images