Seriously-damaged district estimation system

Abstract

To estimate, with satisfactory accuracy, a seriously-=damaged district with more persons required for rescue after occurence of a disaster.SOLUTION: The present invention is directed to a seriously-damaged district estimation system 1 for estimating, after occurrence of a disaster, a district with serious damages and with a large number of persons required for rescue. The system has a seriously-damaged district estimation unit 3 for estimating a district with serious damage and with a large number of persons required for rescue among a plurality of districts, based on population of each of the plurality of districts in a specific region as an estimation target, the number of posting to SNS relating to each of the plurality of districts during a predetermined period of time from occurrence of the disaster, and a disaster magnitude value indicating a magnitude of the disaster in each of the plurality of districts.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a severely damaged area estimation system that estimates areas that are likely to suffer severe damage and have a large number of people requiring rescue after a disaster occurs. [Background technology]

[0002] In order to quickly rescue as many people as possible when disasters such as earthquakes, tsunamis, and large-scale fires occur, it is necessary to estimate the areas where the damage is particularly severe and where there are likely to be many people in need of rescue within the specific disaster area. In this regard, for example, Patent Document 1 describes an earthquake response support system used to support earthquake response activities by response personnel inside and outside the disaster-stricken area when an earthquake occurs. This earthquake response support system includes: a data management means in which attribute information of each building in the disaster-stricken area is registered; a seismic motion intensity distribution acquisition means for acquiring seismic motion intensity distribution information for the disaster-stricken area immediately after the earthquake occurs; a damage possibility estimation means for estimating the damage possibility of each building based on the acquired seismic motion intensity distribution information and the attribute information of each building registered in the data management means; a building number counting means for dividing the disaster-stricken area into multiple areas so that buildings located close to each other are grouped together and for counting the number of buildings by damage possibility for each divided area; and a display means for displaying the total number of buildings counted for each area and subtotals by damage possibility in a predetermined graph in the corresponding area on a background map. Furthermore, Patent Document 2 describes an earthquake disaster recovery support device that predicts the degree of damage to multiple buildings that are the target of recovery support, outputs investigation instruction information that instructs buildings with higher predicted damage levels to be given priority in investigating the damage situation, inputs investigation result information that indicates the investigation result according to the output investigation instruction information, and stores the input investigation result information in a storage means. In Patent Document 2, the degree of damage to the buildings is derived based on basic building information about the buildings.

[0003] In Patent Documents 1 and 2, the possibility and extent of building damage are predicted based on information about the buildings. However, even if areas with severe damage are estimated based on the possibility and extent of building damage based on Patent Documents 1 and 2, these estimates are only areas with severe building damage, and do not necessarily indicate areas with a large number of people needing rescue. In order to estimate areas with particularly severe damage where there are likely to be many people needing rescue when a disaster such as an earthquake occurs, it is considered more desirable to focus on data related to people, such as population and the flow of people, rather than on buildings. Regarding population, for example, when an earthquake occurs, it is possible to estimate the areas that will suffer severe damage by calculating the population that will be exposed to a seismic intensity above a certain level using information on seismic intensity and data on population distribution. However, even if the population that will be exposed to a seismic intensity above a certain level is calculated in this way, the number of people who need rescue does not actually correspond to the calculated population, so it cannot necessarily be said that it is possible to accurately estimate the areas where there are likely to be many people needing rescue.

[0004] Furthermore, with regard to the flow of people, for example, if the population distribution in a certain area increases or decreases significantly from the normal population distribution as a standard, it may be determined that an abnormal crowd has occurred in that area. In this regard, for example, Patent Document 3 describes an information processing device that includes a request acquisition unit that acquires from a user's communication terminal a delivery request requesting delivery of information regarding the degree of abnormality that indicates the degree of abnormality regarding at least one of the population flow state and movement trends, and a screen information transmission unit that transmits to the communication terminal information for generating a screen including a distribution information display area in which distribution information that indicates the geographical distribution of the degree of abnormality is displayed and a fluctuation information display area in which fluctuation information that indicates the time fluctuation of the degree of abnormality is displayed.

[0005] However, even if a crowd abnormality is detected as described above, it is not necessarily the case that the number of people needing rescue corresponds to the detected crowd abnormality. Therefore, even in this case, it is not possible to accurately estimate areas where there are likely to be many people needing rescue. After a disaster occurs, it is desirable to accurately estimate areas that are likely to suffer severe damage and have a large number of people requiring rescue. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2020-91113 [Patent Document 2] Japanese Patent Application Laid-Open No. 2005-208893 [Patent Document 3] Patent Publication No. 2021-174410 Summary of the Invention [Problem to be solved by the invention]

[0007] The problem that the present invention aims to solve is to provide a severely damaged area estimation system that accurately estimates areas that are expected to suffer severe damage and have a large number of people requiring rescue after a disaster occurs. [Means for solving the problem]

[0008] The present invention employs the following means to solve the above-mentioned problems. That is, the present invention provides a severely damaged area estimation system that estimates areas that are likely to suffer severe damage and have a large number of people requiring rescue after a disaster occurs, the system comprising a severely damaged area estimation unit that estimates, among a plurality of areas within a specific area to be estimated, areas that are likely to have a particularly large number of people requiring rescue and have suffered severe damage, based on the population of each of the plurality of areas within the specific area to be estimated, the number of posts to social media related to each of the plurality of areas during a certain period after the disaster occurs, and a disaster scale value that indicates the scale of the disaster in each of the plurality of areas. After a disaster occurs, if a certain area within a specific target region has a particularly large number of people needing rescue and suffers severe damage, it is expected that the ratio of the number of social media posts about that area to the population of that area will be particularly high for a certain period after the disaster. However, if a certain area experiences a high disaster scale, such as seismic intensity, and as a result, many people are unable to move freely, for example, trapped under collapsed buildings, they will be unable to post about the disaster even if they try to do so on social media, despite the high disaster scale. Therefore, the ratio of the number of social media posts about that area to the population of that area may be particularly low for a certain period after the disaster. In this way, the number of social media posts after a disaster can be used as an indicator for estimating areas likely to suffer severe damage and have a large number of people needing rescue. Based on this idea, according to the above-described configuration, the severely damaged area estimation system uses the number of posts on social media as an indicator when estimating areas likely to have suffered severe damage and a large number of people needing rescue. More specifically, the severely damaged area estimation system estimates areas likely to have suffered particularly severe damage and a large number of people needing rescue based on the population of each of multiple areas within a specific area to be estimated, the number of posts on social media related to each of the multiple areas over a certain period after the disaster, and a disaster scale value indicating the scale of the disaster in each of the multiple areas. As a result, for example, based on the population of each of the multiple areas and the number of posts on social media related to each of the multiple areas over a certain period after the disaster, the system calculates the ratio of the number of posts on social media related to each of the multiple areas over a certain period after the disaster to the population of that area, and extracts areas where this ratio is particularly high, thereby making it possible to estimate areas likely to have suffered severe damage and a large number of people needing rescue. Alternatively, for example, by calculating the ratio of the number of posts on social media about each of multiple districts to the population of that district during a certain period after the disaster, and extracting districts where the ratio is too small despite the disaster scale value being large, it is possible to estimate districts that are likely to have suffered severe damage and have a large number of people requiring rescue. In this way, by using the number of posts on social media as an indicator, it is possible to provide a system for estimating severely damaged areas that can accurately estimate areas that are likely to suffer severe damage and have a large number of people requiring rescue after a disaster occurs.

[0009] In one aspect of the present invention, the severely damaged area estimation unit, for each of the plurality of areas, calculates a multiplier of the number of SNS posts by dividing the number of posts to the SNS related to the area by the population of the area, formulates a relational equation using regression analysis between the disaster scale value in the specific area when the disaster occurs and a value based on the multiplier of the number of SNS posts, and for each of the plurality of areas, multiplies the population of the area by the value obtained by applying the disaster scale value of the area to the relational equation to calculate a standard value for the number of posts to the SNS, which is the number of posts to the SNS that are expected to be made for the area when the disaster occurs, calculates the difference between the number of posts to the SNS and the standard value, calculates the difference between the number of posts to the SNS and the standard value, and estimates the areas where the damage will be severe based on the difference for each area. According to the above configuration, the severely damaged district estimation unit first calculates the SNS post count multiplier for each of the multiple districts by dividing the number of SNS posts related to that district by the district's population. Based on this, the severely damaged district estimation unit formulates a relationship between the disaster scale value for a specific region and a value based on the SNS post count multiplier through regression analysis. The resulting relationship represents the relationship between the disaster scale value after the target disaster occurs and the value based on the SNS post count multiplier, which is an expected representative value for all districts whose disaster scale corresponds to that disaster scale value. The severely damaged district estimation unit then multiplies the population of each of the multiple districts by the value obtained by applying the disaster scale value for that district to the above relationship, i.e., the expected value based on the SNS post count multiplier for that district. Since the SNS post count multiplier is calculated by dividing the number of SNS posts by the population, by multiplying the value based on the SNS post count multiplier for a given district by the district's population, we can obtain a reference value for the number of SNS posts, which is the number of SNS posts expected to be made for that district in the event of a disaster. Here, the difference between the number of SNS posts actually made for each district after a disaster occurs and the reference value for the number of SNS posts calculated as described above indicates the degree to which the number of SNS posts actually made for that district deviates from the expected number of posts for that district. Therefore, the severely damaged district estimation unit calculates this difference for each district and uses it to estimate districts that will suffer severe damage.

[0010] In another aspect of the present invention, the severely damaged area estimation unit estimates that the damage in each of the multiple areas is severe if the value based on the difference is greater than or equal to an upper threshold or less than or equal to a lower threshold. As already explained, when the number of SNS posts about a certain area is higher than expected, it can be inferred that the area is one where damage is severe and there are many people in need of rescue. Also, when the number of SNS posts about a certain area is lower than expected, it can be inferred that the area is one where damage is severe and there are many people in need of rescue, for example, where many people are trapped under collapsed buildings and are unable to move freely. Therefore, areas where damage is severe can be inferred based on the difference between the number of SNS posts and a reference value for the number of SNS posts, which indicates the degree to which the actual number of SNS posts about the area deviates from the expected number of posts about the area. Based on this idea, the above-described configuration calculates the difference for each district and determines whether the value based on this difference is above the upper threshold or below the lower threshold. This makes it possible to estimate districts that will suffer severe damage both when the number of posts to SNS is higher or lower than expected, as described above. [Effects of the Invention]

[0011] According to the present invention, it is possible to provide a severely damaged area estimation system that accurately estimates areas that are likely to have suffered severe damage and have a large number of people requiring rescue after a disaster occurs. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a block diagram of a severely damaged area estimation system according to an embodiment of the present invention. [Figure 2] FIG. 1 is an explanatory diagram of population distribution data. [Figure 3] FIG. 10 is an explanatory diagram of SNS data with location information. [Figure 4] FIG. 10 is an explanatory diagram of disaster scale data. [Figure 5] This is a scatter plot showing the relationship between the disaster scale value and the standardized SNS post count multiplier for each region. [Figure 6]This is a graph that plots the standardized difference between the number of posts on SNS and the standard value for the number of posts on SNS for all districts within a specific region. [Figure 7] This is a graph showing the cumulative frequency of standardized values ​​of the difference for all districts where the difference is a negative value. [Figure 8] FIG. 10 is an explanatory diagram showing the results displayed in different colors on a map. [Figure 9] 10 is a flowchart of a method for estimating a severely damaged area using the severely damaged area estimation system of the above embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] This invention is a severely damaged area estimation system that uses disaster scale data, population distribution data, and SNS data with location information to first estimate specific areas that will suffer severe damage after a disaster occurs, and then estimates areas within those specific areas with a large number of people requiring rescue. Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The severely damaged area estimation system of this embodiment estimates, after a disaster occurs within a specific target area, from among multiple areas within the specific area, areas that are likely to have suffered severe damage and have a large number of people requiring rescue. In this embodiment, the term "person requiring rescue" refers to anyone who has suffered some kind of disadvantage due to a disaster and needs rescue from a third party. For example, the term "person requiring rescue" includes anyone whose life or body is in danger due to a disaster and who is unable to eliminate the danger on their own. In this embodiment, the term "person requiring rescue" may also include those who require assistance in evacuating to a safe place after a disaster, such as elderly people, infants, and pregnant women. Furthermore, the term "person requiring rescue" may also include those who lack sufficient infrastructure and supplies to sustain their lives after a disaster and need relief supplies. Furthermore, the term "person requiring rescue" may also include those who, for example, need manpower to remove debris in order to maintain their livelihood after a disaster.

[0014] FIG. 1 is a block diagram of the severely damaged area estimation system according to this embodiment. The severely damaged area estimation system 1 is comprised of a computer terminal such as a server, a personal computer, or a tablet terminal, and performs the required functions by executing a pre-set program. Functionally, the severely damaged area estimation system 1 comprises an information collection unit 2, a severely damaged area estimation unit 3, and a result display unit 4.

[0015] The information collection unit 2 collects various information required to estimate which of multiple districts has a particularly large number of people needing rescue and has suffered severe damage from external sources via a network, etc. The information collection unit 2 collects population distribution data, SNS data with location information, and disaster scale data. First, the information collection unit 2 collects population distribution data 10. FIG. 2 is an explanatory diagram of population distribution data. More specifically, the information collection unit 2 acquires population distribution data 10, as shown in FIG. 2, from a computer device (not shown) on the side of an organization or business that provides information on population. The population distribution data 10 may be acquired before or after a disaster occurs, but it is desirable to acquire as up-to-date information as possible. As the population distribution data 10, for example, a service that acquires population distribution in real time, such as "Mobile Spatial Statistics" by Docomo Mobile Inc., can be used. The population distribution data 10 is acquired so as to cover the entire specific area A, which is the target area for estimating the areas with the most severe damage. The specific area A is divided into a plurality of areas M. In the population distribution data 10, each of the plurality of areas M is associated with the population of the area M. In this embodiment, the specific region A is divided into a plurality of districts M in a mesh pattern based on latitude and longitude. The specific region A can also be divided into a plurality of districts M, for example, based on addresses such as towns, villages, and street addresses. However, as will be described later, a process is executed in which it is determined which district M each post on the SNS relates to based on the SNS data with location information, and the number of posts on the SNS related to each district M is tallied. Here, in the SNS data with location information, the location information is expressed by latitude and longitude. Therefore, by dividing the specific region A into a plurality of districts M based on latitude and longitude as in this embodiment, the above-mentioned tallying process can be easily performed.

[0016] Next, the information collection unit 2 collects SNS data 11 with location information. FIG. 3 is an explanatory diagram of SNS data with location information. The location-information-attached SNS data 11 is data in which posts to an SNS are associated with the latitude and longitude of a location related to the post. The content of a post to an SNS is composed of text information constituting the post, and photo and video information attached to the post. The location-information-attached SNS data 11 can be created by assigning latitude and longitude information acquired based on the content of the post to the SNS, as described above. Therefore, strictly speaking, the location-information-attached SNS data 11 associates posts to an SNS with latitude and longitude related to the content of the post, rather than the latitude and longitude where the post actually was made. In FIG. 3, for posts made to an SNS within a certain period of time within a specific region A, points related to the posts are marked. The information collection unit 2 acquires the location-information-attached SNS data 11 from a computer device (not shown) of an appropriate institution or business that provides the location-information-attached SNS data 11.

[0017] After a disaster occurs, the severely damaged area estimation system 1 of this embodiment estimates areas M that are likely to have suffered severe damage and have a large number of people requiring rescue, based on the number of posts to SNS for each of multiple areas M within a specific area A. Therefore, the information collection unit 2 collects SNS data 11 with location information for a certain period after the disaster occurs. In the severely damaged area estimation system 1 of this embodiment, as described above, an area M where damage is severe and where there are many people requiring rescue is estimated based on the number of posts to SNS. To improve the accuracy of this estimation, a certain number of posts to SNS is required. Therefore, it is desirable to set the start point of the above-mentioned certain period for collecting SNS data 11 with location information as the time when the disaster occurred and the end point as several hours to one week after the start point, for example, so as to obtain a sufficient number of posts. On the other hand, the severely damaged area estimation system 1 may be able to appropriately determine the initial response, such as to which area M within the specific area A a rescue team should be dispatched to, in order to estimate the area M where the damage is likely to be severe and where there are many people needing rescue. From this perspective, it is more appropriate to shorten the fixed period for collecting the location-information-attached SNS data 11, for example, by setting the starting point at the time when the disaster occurred and the ending point several hours to one day after the starting point, and to estimate the area M where the damage is likely to be severe and where there are many people needing rescue as soon as possible. Alternatively, the severely damaged area estimation system 1 may be used to identify rescuers who are short of relief supplies after a disaster occurs. It is thought that rescuers start posting about the shortage of relief supplies on SNS not immediately after the disaster occurs, but after a certain amount of time has passed since the disaster occurred. Therefore, in this case, the certain period for collecting the location information-attached SNS data 11 may be set, for example, so that the start point is two to three days after the disaster occurs and the end point is several days to one week after the start point.

[0018] As described above, the information collection unit 2 acquires SNS data 11 with location information relating to the specific area A for a certain period of time after the occurrence of a disaster. The information collection unit 2 further identifies which of the multiple districts M each post on the SNS relates to, based on information relating to the latitude and longitude associated with each piece of SNS data 11 with location information. In this way, the information collection unit 2 tally up the number of posts on the SNS relating to each of the multiple districts M for a certain period of time after the occurrence of a disaster.

[0019] Next, the information collection unit 2 collects disaster scale data 12. FIG. 4 is an explanatory diagram of disaster scale data. The severe damage area estimation system 1 of this embodiment assumes that the disaster is an earthquake. However, as will be described in more detail later as a modified example, the disaster is not limited to an earthquake, and may be of other types, such as a tsunami, a landslide, liquefaction, or a large-scale fire. The severely damaged area estimation unit 3 acquires disaster scale data 12 for each area M within the specific area A from a computer device (not shown) on the side of an organization or business that provides information about the disaster. When an earthquake is assumed as the disaster, as in this embodiment, the disaster scale data 12 is data on the intensity of earthquake motion, such as the measured seismic intensity, for each area M. In this case, the disaster scale data 12 can be acquired from a computer device on the side of an organization that provides earthquake information (for example, the Japan Meteorological Agency). The information collection unit 2 acquires the disaster scale data 12 at an appropriate timing after the occurrence of a disaster.

[0020] In this way, the information collection unit 2 acquires population distribution data 10, SNS data with location information 11, and disaster scale data 12 in the specific area A. As a result, data on the population, the number of posts to SNS related to the area M for a certain period after the occurrence of a disaster, and a disaster scale value indicating the scale of the disaster (measured seismic intensity in this embodiment) can be obtained for each of a plurality of districts M within the specific area A. The severely damaged area estimation unit 3 estimates the areas M among the multiple areas M that have a particularly large number of people needing rescue and have suffered severe damage, based on the population of each of the multiple areas M within the specific area A to be estimated obtained in this manner, the number of posts on social media regarding each of the multiple areas M during a certain period after the disaster occurs, and the disaster scale value indicating the scale of the disaster in each of the multiple areas M. The severely damaged area estimation unit 3 performs processing based on the number of posts to SNS that are tallied after the disaster occurs. Therefore, basically, all processing in the severely damaged area estimation unit 3 is performed after the disaster occurs.

[0021] First, for each of a plurality of districts M within the specific area A, the severely damaged district estimation unit 3 calculates a SNS post number magnification by dividing the number of posts to SNS related to the district M by the population of the district M. Next, the severely damaged district estimation unit 3 calculates an average value γ across the multiple districts M of the SNS post count magnification calculated for each of the multiple districts M as described above. Then, the severely damaged area estimation unit 3 divides the SNS post count magnification of each area M by the average value γ and normalizes it by the average value γ to calculate a normalized SNS post count magnification for each of the multiple areas M. As a result, the average value of the normalized SNS post count magnification across the multiple areas M becomes 1. In this way, the severely damaged area estimation unit 3 calculates a standardized SNS post count multiplier for each of multiple areas M by dividing the SNS post count multiplier by the average value γ of the SNS post count multipliers for all areas M.

[0022] Figure 5 is a scatter plot showing the relationship between the disaster scale value and the standardized SNS post count multiplier for each district. The severely damaged area estimation unit 3 generates a graph with the horizontal axis representing the disaster scale value (measured seismic intensity in this embodiment) and the vertical axis representing the normalized SNS post count magnification (the value obtained by dividing the SNS post count magnification by the average value γ), as shown in Figure 5. The severely damaged area estimation unit 3 plots points corresponding to each of the multiple areas M in the graph based on the normalized SNS post count magnification calculated for each area M and the disaster scale value of each area M, to generate a scatter diagram. In Figure 5, when the scale of a disaster exceeds a certain disaster scale value, the number of posts to SNS increases sharply. The threshold value at which the number of posts to SNS increases sharply is set as the disaster scale value threshold β. In the example of Figure 5, the disaster scale value threshold β can be set to, for example, 4. The severely damaged area estimation unit 3 further expresses the relationship between the disaster scale value SI and the standardized SNS post count multiplier W as an exponential function using a coefficient α and a disaster scale value threshold β, as shown in the following relational equation (1). W = exp(α × (SI - β)) (1) The severely damaged area estimation unit 3 then performs regression analysis using the least squares method or the like to determine the coefficient α. In the example of Figure 5, the value of coefficient α is calculated to be 1.87. Therefore, more specifically, in the example shown in Figure 5, relational expression (1) is expressed as follows: W=exp(1.87×(SI-4))

[0023] In this way, the severely damaged area estimation unit 3 formulates the relational equation (1) between the disaster scale value SI in the specific area A when a disaster occurs and a value based on the SNS post count multiplier. In particular, in this embodiment, the severely damaged area estimation unit 3 formulates the relational equation (1) between the disaster scale value SI and the standardized SNS post count magnification W, using the standardized SNS post count magnification obtained by dividing the SNS post count magnification by the average value γ as a value based on the SNS post count magnification. The relational expression (1) obtained in this way represents the relationship between the disaster scale value SI after the occurrence of the target disaster and the expected representative value of the standardized SNS post count multiplier for the entire area M where the scale of the disaster corresponds to the disaster scale value SI.

[0024] The severely damaged area estimation unit 3 then calculates the value S for each area M as shown in the following formula (2). The severely damaged area estimation unit 3 calculates the value S by multiplying the population of the area M by the value W obtained by applying the disaster scale value SI for the area M to the relational formula (1) and the average value γ. S = population × value W × average value γ (2) Here, the above relational expression (1) is obtained by regression analysis based on the normalized SNS post count multiplier for each district M. Therefore, the value W obtained by substituting an arbitrary value for the disaster scale value SI into relational expression (1) is a representative value of the normalized SNS post count multiplier expected for all districts M whose disaster scale corresponds to the disaster scale value SI, as described above. The normalized SNS post count multiplier is obtained by dividing the number of SNS posts in each district M by its population, and then dividing this by the average SNS post count multiplier γ. Therefore, the representative value of the normalized SNS post count multiplier for each district M obtained by relational expression (1) is multiplied by the population of the district M by the average γ, as shown in the above equation (2). The value S is the estimated number of SNS posts for district M after the disaster occurs. The severely damaged district estimation unit 3 sets this estimated number of SNS posts as a reference value S and compares this reference value S with the actual number of SNS posts.

[0025] More specifically, the severely damaged area estimation unit 3 calculates, for each area M, the difference D between the number of posts to SNS that is obtained by tallying the number of posts to SNS after the occurrence of the target disaster in that area M and the standard value S of the number of posts to SNS calculated as described above for that area M. In this embodiment, the difference D is calculated using the following formula (3). Difference D = (actual) number of posts on SNS - standard value S of number of posts on SNS (3) The difference D calculated in this way is a value that indicates how much the actual number of posts on SNS in each district M deviates from the standard value S of the number of posts on SNS expected in that district M after the target disaster occurs.

[0026] For example, in district M with a population of 321, if after a disaster there are 36 posts on social media, the disaster scale value (measured seismic intensity) SI is 6.15, and the average value γ is 0.000594, the standard value S and difference D of the number of posts on social media are calculated as follows: The standard value for the number of posts to SNS is S = 321 × exp(1.87 × (6.15 - 4)) ×0.000594 =10.63 Difference D=36-10.64=25.37 In relational expression (1), if the value substituted for the disaster scale value SI is 4, which is equal to the disaster scale value threshold β, then the value of W will be 1. In contrast, for example, if the disaster scale value SI is 6.15 as described above, the value of W, exp(1.87 × (6.15 - 4)), will be approximately 56. Because the standard value S for the number of posts to SNS is calculated by multiplying by this value W, when the disaster scale value SI is 6.15, the standard value S for the number of posts to SNS will be approximately 56 times higher than when the disaster scale value SI is 4. This indicates that when the disaster scale value SI is 6.15, there is a possibility that the number of posts to SNS will be approximately 56 times higher than when the disaster scale value SI is 4.

[0027] In this way, for each of the multiple areas M, the severely damaged area estimation unit 3 calculates a standard value S for the number of posts on SNS by multiplying the value obtained by applying the disaster scale value SI for that area M to the relational expression (1) by the population of that area M, and calculates the difference D between the actual number of posts on SNS and the standard value S. Based on the difference D calculated in this way, the severely damaged area estimation unit 3 estimates an area M in the specific area A that will suffer particularly severe damage, as follows.

[0028] After a disaster occurs, if a certain area M within a specific target area A suffers particularly severe damage, it is expected that the number of posts on SNS related to that area will be particularly high for a certain period of time after the disaster occurs. Here, the difference D is calculated by subtracting the standard value S for the number of posts on SNS from the actual number of posts on SNS, as shown in equation (3). Therefore, the more severe the damage in a certain area M and the greater the actual number of posts, the larger the value of the difference D will be, a positive value. Based on this idea, the severely damaged area estimation unit 3 estimates that the damage in each of multiple areas M is particularly severe if the value calculated based on the difference D as described below is greater than or equal to a predetermined upper threshold.

[0029] On the other hand, after a disaster occurs, if damage is particularly severe in a certain area M within the target specific region A, there may be many people trapped under collapsed buildings, for example, and unable to move freely. In this case, even if the disaster scale value is large, people will be unable to post about it on social media, and the proportion of social media posts related to that area may be particularly low for a certain period after the disaster occurs. Here, the difference D is calculated by subtracting the reference value S for the number of social media posts from the actual number of social media posts, as shown in equation (3). Therefore, the more severe the damage in a certain area M and the more the actual number of posts falls below the reference value S, the smaller the value of the difference D will be, and the larger the absolute value will be, resulting in a negative value. Therefore, the severely damaged area estimation unit 3 estimates that the damage in each of multiple areas M is particularly severe if the value calculated based on the difference D as described below is below a predetermined lower threshold.

[0030] However, it is not easy to configure the severely damaged area estimation system 1 so as to compare the difference D calculated as described above with some threshold value to determine whether a certain area M is an area M that will suffer severe damage. In particular, it is not known in advance what the specific value of the difference D will be in an area M that will suffer severe damage from a disaster after the disaster occurs. Therefore, it is not easy to determine in advance the threshold value to which the difference D will be compared. For this reason, in this embodiment, in each district M, the difference D is not directly compared with a threshold value, but a standardized value of the difference D is calculated as a value based on the difference D, and this standardized value of the difference D is compared with an upper threshold value and a lower threshold value that are appropriately determined after the disaster occurs.

[0031] More specifically, the severely damaged area estimation unit 3 acquires the difference D of the area M with the largest value among the areas M where the difference D is a positive value as the maximum value of the difference D. For all areas M where the difference D is a positive value, the severely damaged area estimation unit 3 divides the difference D of that area by the maximum value of the difference D. In this way, the severely damaged area estimation unit 3 normalizes the difference D of all areas M where the difference D is a positive value so that it is a value greater than 0 and equal to or less than 1. Furthermore, the severely damaged area estimation unit 3 acquires the difference D of the area M with the smallest value (largest absolute value) among the areas M where the difference D is a negative value as the minimum value of the difference D. For all areas M where the difference D is a negative value, the severely damaged area estimation unit 3 divides the difference D of that area by the absolute value of the smallest value of the difference D. In this way, the severely damaged area estimation unit 3 normalizes the difference D of all areas M where the difference D is a negative value so that it is less than 0 and greater than or equal to -1. Figure 6 is a graph plotting the standardized difference between the number of posts to SNS and the standard value of the number of posts to SNS for all districts in a specific region. In Figure 6, the horizontal axis shows the districts M sorted by the standardized value of the difference D, and the vertical axis shows the standardized value of the difference D for each district M.

[0032] Next, the severely damaged district estimation unit 3 calculates the cumulative frequency of the standardized value of the difference D for all districts M where the difference D is a negative value. Figure 7 is a graph showing the cumulative frequency of the standardized difference values ​​for all districts where the difference is a negative value. In Figure 7, the standardized difference D value on the horizontal axis is the absolute value of the standardized difference D. As shown in FIG. 7, the cumulative frequency is represented as a graph, and multiple cumulative frequency thresholds are set so as to equally divide the possible values ​​of the cumulative frequency. In FIG. 7, four cumulative frequency thresholds are set: 0.2, 0.4, 0.6, and 0.8. The severely damaged area estimation unit 3 derives a value obtained by normalizing the difference D corresponding to each of these multiple cumulative frequency thresholds. For example, for a cumulative frequency threshold of 0.2, the value p1 obtained by normalizing the corresponding difference D is, for example, −0.0028. For example, for a cumulative frequency threshold of 0.4, the value p2 obtained by normalizing the corresponding difference D is, for example, −0.011. For example, for a cumulative frequency threshold of 0.6, the value p3 obtained by normalizing the corresponding difference D is, for example, −0.025. For example, for a cumulative frequency threshold of 0.8, the value p4 obtained by normalizing the corresponding difference D is, for example, −0.075. In this embodiment, the lower threshold value to which the normalized value of the difference D is compared is set to a normalized value of the difference D corresponding to these cumulative frequency thresholds. For example, if it is desired to carefully select and estimate districts M that have suffered particularly severe damage from among multiple districts M within a specific region A, the lower threshold value can be set to −0.075, which corresponds to a cumulative frequency threshold of 0.8. Alternatively, if it is desired to increase the estimated number of districts M that have suffered particularly severe damage from among multiple districts M within a specific region A, the lower threshold value can be set to −0.025, which corresponds to a cumulative frequency threshold of 0.6.

[0033] In addition, the severely damaged area estimation unit 3 calculates the cumulative frequency of the standardized value of the difference D for all areas M where the difference D is a positive value, in the same way as for all areas M where the difference D is a negative value. In this case, multiple cumulative frequency thresholds are set so as to equally divide the possible cumulative frequency values. As in the case of negative values, four cumulative frequency thresholds can be set: 0.2, 0.4, 0.6, and 0.8. The severely damaged area estimation unit 3 derives a standardized value of the difference D corresponding to each of these multiple cumulative frequency thresholds. For example, for a cumulative frequency threshold of 0.2, the standardized value of the difference D corresponding to this cumulative frequency threshold is 0.0016. For example, for a cumulative frequency threshold of 0.4, the standardized value of the difference D corresponding to this cumulative frequency threshold is 0.002. For example, for a cumulative frequency threshold of 0.6, the standardized value of the difference D corresponding to this cumulative frequency threshold is 0.0023. For example, for a cumulative frequency threshold of 0.8, the standardized value of the difference D corresponding to this cumulative frequency threshold is 0.0046. In this embodiment, the upper threshold value to which the normalized value of the difference D is compared is set to a normalized value of the difference D corresponding to these cumulative frequency thresholds. For example, if it is desired to carefully select and estimate districts M that have suffered particularly severe damage from among multiple districts M within a specific region A, the upper threshold value can be set to 0.0046, which corresponds to a cumulative frequency threshold of 0.8. Alternatively, if it is desired to increase the estimated number of districts M that have suffered particularly severe damage from among multiple districts M within a specific region A, the upper threshold value can be set to 0.0023, which corresponds to a cumulative frequency threshold of 0.6.

[0034] 7 is generated based on data collected after a disaster occurs, the correspondence between the standardized value of the difference D and the cumulative frequency will differ depending on the situation. Therefore, the upper and lower thresholds are determined as appropriate as described above when processing is performed in the severely damaged area estimation system 1 after a disaster occurs.

[0035] The result display unit 4 displays a map of the specific area A on a display device (not shown), and also displays the extent of damage in each of the districts M by displaying each of the districts M in different colors. FIG. 8 is an explanatory diagram showing the estimation results of the severely damaged area estimation system 1 displayed on a map in different colors. In Figure 8, each district M on the map is colored so that a different color is assigned to each interval of the standardized value of the difference D, with the standardized value of the difference D corresponding to each of the cumulative frequency thresholds set as described above as the boundary value. In Figure 8, the gray area indicated in the legend as "-0.0028 - 0.0016" is District M, where the number of posts to SNS expected from the population after a disaster is close to the actual number of posts to SNS, and the difference D is small. In contrast, the black area shown as "0.0046 - 1.00" is District M, where the damage was particularly severe and the actual number of posts on social media was particularly higher than the number of posts expected based on the population. Conversely, the light gray area shown as "-0.075 - -0.025" is District M, where the damage was particularly severe, as above, and where the actual number of posts on social media was particularly lower than the number of posts expected based on the population, and where it is thought that there are many people in situations where they are not allowed to act freely.

[0036] Next, a method for estimating a severely damaged area using the above-described severely damaged area estimation system will be described with reference to Figures 1 to 8 and Figure 9. Figure 9 is a flowchart of the method for estimating a severely damaged area. After a disaster occurs, the severe damage area estimation system 1 performs the following: First, the information collection unit 2 collects various pieces of information required to estimate which of multiple districts has a particularly large number of people needing rescue and has suffered severe damage from outside via a network, etc. The information collection unit 2 collects population distribution data, SNS data with location information, and disaster scale data (step S1). This allows for obtaining data on the population for each of multiple districts M within a specific region A, the number of posts on SNS regarding that district M for a certain period of time after the disaster occurs, and a disaster scale value indicating the scale of the disaster (measured seismic intensity in this embodiment).

[0037] The severely damaged area estimation unit 3 estimates, among the multiple areas M, the areas M with the largest number of people needing rescue and suffering the most severe damage, based on the population of each of the multiple areas M within the specific area A to be estimated, the number of posts on SNS regarding each of the multiple areas M during a certain period after the disaster occurs, and the disaster scale value indicating the scale of the disaster in each of the multiple areas M (step S2: severely damaged area estimation step). For this purpose, the severely damaged area estimation unit 3 first calculates the SNS post number magnification for each of a plurality of areas M within the specific area A by dividing the number of posts to SNS related to that area M by the population of that area M. Next, the severely damaged area estimation unit 3 calculates a normalized SNS post count magnification for each of the multiple areas M by dividing the SNS post count magnification by the average value γ of the SNS post count magnifications for all of the areas M. In addition, the severely damaged area estimation unit 3 formulates the relational equation (1) between the disaster scale value SI in the specific area A when a disaster occurs and a value based on the SNS post count multiplier (standardized SNS post count multiplier). The severely damaged area estimation unit 3 calculates a reference value S of the number of posts to SNS in each area M based on the relational expression (1). The severely damaged area estimation unit 3 calculates the difference D between the number of posts to SNS obtained by tallying the number of posts to SNS in each area M after the target disaster occurs in that area M and the standard value S of the number of posts to SNS calculated for that area M. The severely damaged area estimation unit 3 estimates an area M in the specific area A that will suffer particularly severe damage, based on the difference D calculated in this way.

[0038] The result display unit 4 displays a map of the specific area A on a display device (not shown), and also displays the extent of damage in each of the districts M by displaying each of the districts M in different colors.

[0039] The severely damaged area estimation system 1 as described above is a system 1 for estimating areas M that are expected to suffer severe damage and have a large number of people requiring rescue after a disaster occurs, and is equipped with a severely damaged area estimation unit 3 that estimates, among multiple areas M, areas M that have a particularly large number of people requiring rescue and have suffered severe damage, based on the population of each of multiple areas M within the specific area A to be estimated, the number of posts on SNS regarding each of the multiple areas M during a certain period after the disaster occurs, and a disaster scale value SI that indicates the scale of the disaster in each of the multiple areas M. After a disaster occurs, if a certain district M within a specific target area A has a particularly large number of people needing rescue and suffers severe damage, it is expected that the ratio of the number of SNS posts about district M to the population of district M will be particularly high for a certain period after the disaster. However, if the disaster scale value SI (seismic intensity, etc.), which indicates the scale of the disaster, is high in district M, and as a result, many people are unable to move freely, for example, buried under collapsed buildings, they will be unable to post about the disaster on SNS, despite the high disaster scale value SI. Therefore, the ratio of the number of SNS posts about district M to the population of district M for a certain period after the disaster may be particularly low. In this way, the number of SNS posts after a disaster can be used as an indicator for estimating district M, which is likely to suffer severe damage and have a large number of people needing rescue. Based on this idea, according to the above-described configuration, the severely damaged area estimation system 1 uses the number of posts on social media as an indicator when estimating an area M that is likely to have suffered severe damage and a large number of people requiring rescue. More specifically, the severely damaged area estimation system 1 estimates an area M among the multiple areas M that has a particularly large number of people requiring rescue and has suffered severe damage, based on the population of each of the multiple areas M within the specific area A to be estimated, the number of posts on social media related to each of the multiple areas M for a certain period after the disaster occurred, and a disaster scale value SI indicating the scale of the disaster in each of the multiple areas M. As a result, for example, based on the population of each of the multiple areas M and the number of posts on social media related to each of the multiple areas M for a certain period after the disaster occurred, the system calculates the ratio of the number of posts on social media related to each of the multiple areas M for a certain period after the disaster to the population of that area M, and extracts the area M with a particularly high ratio, thereby making it possible to estimate an area M that is likely to have suffered severe damage and a large number of people requiring rescue. Alternatively, for example, for each of multiple districts M, the ratio of the number of posts on social media about that district M during a certain period after the disaster occurs to the population of that district M can be calculated, and districts M with a large disaster scale value SI but a ratio that is too small can be extracted, thereby making it possible to estimate districts M that are likely to have suffered severe damage and have a large number of people requiring rescue. In this way, by using the number of posts on SNS as an indicator, it is possible to provide a severe damage area estimation system 1 that can accurately estimate, after a disaster occurs, the area M that is likely to suffer severe damage and have a large number of people requiring rescue.

[0040] When constructing the severely damaged area estimation system 1, it may be possible to estimate the severely damaged area M based solely on the number of posts on social media for each of multiple areas M within the specific area A to be estimated. However, in this case, if the number of posts on social media is large, it is not easy to determine whether this is due to a large population or the severity of the damage. In contrast, the severely damaged area estimation system 1 described above uses a combination of population distribution data 10, location-information-attached SNS data 11, and disaster scale data 12. As a result, the severely damaged area estimation system 1 is configured to be able to accurately estimate an area M that is considered to have suffered severe damage and have a large number of people requiring rescue.

[0041] In addition, the severely damaged area estimation unit 3 calculates the SNS post count multiplier for each of multiple areas M by dividing the number of SNS posts related to that area M by the population of that area M, and formulates, by regression analysis, the relationship (1) between the disaster scale value SI for that area M in the event of a disaster and a value based on the SNS post count multiplier (standardized SNS post count multiplier) in the event of a disaster in a specific area A, and for each of multiple areas M, multiplies the value W obtained by applying the disaster scale value SI for that area M to the relationship (1) by the population of that area M to calculate a standard value S for the number of SNS posts, which is the number of SNS posts that are expected to be made for that area M in the event of a disaster, and calculates the difference D between the number of SNS posts and the standard value S, and estimates the areas M that will suffer severe damage for each area M based on the difference D. According to the above configuration, the severely damaged area estimation unit 3 first calculates the SNS post count multiplier for each of the multiple areas M by dividing the number of posts on SNS related to that area M by the population of that area M. Based on this, the severely damaged area estimation unit 3 formulates a relational expression (1) between the disaster scale value SI for the specific area A and a value based on the SNS post count multiplier (normalized SNS post count multiplier) through regression analysis. The relational expression (1) thus obtained represents the relationship between the disaster scale value SI after the occurrence of the target disaster and the representative value of the value based on the SNS post count multiplier expected for the entire area M where the disaster scale value is the disaster scale value SI. Then, for each of the multiple areas M, the severely damaged area estimation unit 3 multiplies the value W obtained by applying the disaster scale value SI for that area M to the above relational expression (1), i.e., the value based on the SNS post count multiplier expected for that area M (normalized SNS post count multiplier) W, by the population of that area M. Since the SNS post count multiplier is a value obtained by dividing the number of posts on SNS by the population, by multiplying a value based on the SNS post count multiplier for the district M by the population of the district M, it is possible to obtain the reference value S for the number of SNS posts, which is the number of SNS posts expected to be made for the district M in the event of a disaster. Here, the difference D between the number of SNS posts actually made for each district M after the disaster occurs and the reference value S for the number of SNS posts calculated as described above indicates the degree to which the number of SNS posts actually made for the district M deviates from the expected number of posts for the district M. Therefore, the severely damaged district estimation unit 3 calculates this difference D for each district M and, by using this, can estimate the districts M that will suffer severe damage.

[0042] Furthermore, the severely damaged area estimation unit 3 estimates that the damage in each of the multiple areas M is severe if the value based on the difference D is greater than or equal to the upper threshold or less than or equal to the lower threshold. As already explained, when the number of SNS posts about a certain district M is higher than expected, it can be estimated that the district M is a district M where damage is severe and where there are many people needing rescue. Also, when the number of SNS posts about a certain district M is lower than expected, it can be estimated that the district M is a district where damage is severe and where there are many people needing rescue, for example, where there are many people trapped under collapsed buildings and in other situations where they are not allowed to move freely. Therefore, it is possible to estimate which districts M will suffer severe damage based on the difference D between the number of SNS posts and the standard value S for the number of SNS posts, which indicates how much the number of SNS posts actually made about the district M deviates from the expected number of posts about the district M. Based on this idea, with the above configuration, the difference D is calculated for each district M, and it is determined whether the value based on this difference D is equal to or greater than the upper threshold or equal to or less than the lower threshold. This makes it possible to estimate the district M that will suffer the most severe damage, both when the number of posts to SNS is higher or lower than expected, as described above.

[0043] In addition, the severely damaged area estimation unit 3 calculates a standardized SNS post count multiplier for each of the multiple areas M by dividing the SNS post count multiplier by the average value γ of the SNS post count multipliers for all of the areas M, and formulates a relationship (1) between the disaster scale value SI and the standardized SNS post count multiplier, using the standardized SNS post count multiplier as a value based on the SNS post count multiplier, and for each of the multiple areas M, calculates a standard value S for the number of SNS posts by multiplying the value W obtained by applying the disaster scale value SI for the area M to the relationship (1) and the average value γ by the population of the area M, and then calculates the difference D between the number of SNS posts and the standard value S. In addition, the severely damaged area estimation system 1 further includes a result display unit 4 that displays a map of the specific area A on the display device and displays the extent of damage in each area M by color-coding each area M. According to the above configuration, the severely damaged area estimation system 1 can be appropriately realized.

[0044] Furthermore, the above-described method for estimating severely damaged areas is a method for estimating areas M that are expected to suffer severe damage and have a large number of people requiring rescue after a disaster occurs, and includes a step (step S2) of estimating areas of severe damage, which estimates areas M that have a particularly large number of people requiring rescue and have suffered severe damage, based on the population of each of the areas M within the specific area A to be estimated, the number of posts on SNS regarding each of the areas M during a certain period after the disaster occurs, and a disaster scale value indicating the scale of the disaster in each of the areas M. According to the above configuration, as already explained with regard to the severe damage area estimation system 1, by using the number of posts on SNS as an indicator, it is possible to accurately estimate the area M that is likely to suffer severe damage and have a large number of people requiring rescue after a disaster occurs.

[0045] The severely damaged area estimation system 1 of the present invention is not limited to the above-described embodiment explained with reference to the drawings, and various other modifications are conceivable within the technical scope thereof. For example, in the severe damage area estimation system 1, the measured seismic intensity is used as the disaster scale value SI, but this is not limited to this. When the disaster is an earthquake, the disaster scale value SI can be calculated from the time history waveform, such as the maximum acceleration, maximum velocity, acceleration response spectrum value or velocity response spectrum value in a certain target period, cumulative absolute velocity, etc., for area M.

[0046] Furthermore, the disasters anticipated by the severe damage area estimation system 1 are not limited to earthquakes. The severely damaged area estimation system 1 can also be used, for example, after a tsunami occurs, to estimate an area M that is likely to suffer severe damage and have a large number of people needing rescue. In this case, the disaster scale value SI can be, for example, the water level or elevation of each area M. The severely damaged area estimation system 1 can also be used, for example, after a landslide occurs, to estimate an area M that is likely to have suffered severe damage and have a large number of people requiring rescue. In this case, the disaster scale value SI can be, for example, the distance or slope of each area M from a landslide risk area. The severe damage area estimation system 1 can also be used to estimate an area M where damage is severe and where there are many people in need of rescue, for example, after liquefaction has occurred. In this case, the disaster scale value SI can be, for example, the seismic intensity or ground physical properties of each area M. The severely damaged area estimation system 1 can also be used, for example, after a large-scale fire breaks out, to estimate an area M that is likely to be severely damaged and have many people needing rescue. In this case, the disaster scale value SI can be, for example, the distance from the source of the fire to each area M, the wind direction, etc. The severely damaged area estimation system 1 can also be used, for example, after a volcanic eruption, to estimate an area M that is likely to suffer severe damage and have many people in need of rescue. In this case, the disaster scale value SI can be, for example, the distance from the source of the eruption of each area M, wind direction, slope, etc. The severely damaged area estimation system 1 can also be used to estimate areas M that are likely to suffer severe damage and have many people in need of rescue after a heavy snowfall, for example. In this case, the disaster scale value SI can be, for example, the snow depth, snow volume, snowfall volume, etc. of each area M. The severely damaged area estimation system 1 can also be used to estimate an area M that is likely to suffer severe damage and have many people in need of rescue after a typhoon or strong wind occurs. In this case, the maximum wind speed, average wind speed, wind direction, etc. of each area M can be used as the disaster scale value SI. The severely damaged area estimation system 1 can also be used to estimate areas M that are likely to suffer severe damage and have many people in need of rescue after a flood or inundation occurs. In this case, the disaster scale value SI can be, for example, the water level, elevation, accumulated rainfall, rainfall intensity, etc. of each area M. The severely damaged area estimation system 1 can also be used to estimate an area M where damage is severe and where there are many people requiring rescue, for example, after a heatstroke incident. In this case, the heat index (WBGT: wet bulb globe temperature), temperature, humidity, etc. of each area M can be used as the disaster scale value SI.

[0047] In the above embodiment, the disaster is an earthquake, the disaster scale value SI is the measured seismic intensity, and the disaster scale value threshold β is set to a threshold value (4 in the example embodiment) at which the number of posts to SNS suddenly increases, but this is not limited to this. For example, the average seismic intensity value when the standardized SNS post count magnification is approximately 1 may be calculated, and this may be set as the disaster scale value threshold β. Furthermore, it goes without saying that the specific value of the disaster scale threshold β can be determined to be different from the value 4 exemplified in the embodiment, depending on the type of disaster and the index adopted as the disaster scale value SI.

[0048] Furthermore, the population of each district M used in the severely damaged district estimation system 1 does not necessarily represent the exact number of people living in that district M. For example, the usage rate of social media varies depending on age. While the usage rate of social media is high, for example, exceeding 60%, among people in their teens and twenties, it is low, for example, below 20%, among people in their fifties and sixties. Therefore, for example, in a district M where the proportion of people in their teens and twenties in the total population is high, the actual number of posts on social media may be greater than the reference value S for the number of posts on social media, due to the population composition of the district M. In this case, even if the actual number of posts on social media is greater than the reference value S for the number of posts on social media, it may not be possible to say that the damage in that district M is severe. Alternatively, for example, in a district M where the proportion of people in their fifties and sixties in the total population is high, the actual number of posts on social media may be smaller than the reference value S for the number of posts on social media, due to the population composition of the district M. In this case, even if the actual number of posts on SNS is less than the reference value S for the number of posts on SNS, it may not be possible to say that the damage to the area M is severe. To deal with situations like the one above, it is possible to obtain the population by age group for each district M, multiply the SNS usage rate for each age group by the sum of the results, and use the resulting value as the population of that district M. This may enable a more accurate calculation of the population that is able to use SNS. This reduces the variation in the vertical axis when generating a scatter diagram like the one shown in Figure 5, and allows for more accurate setting of the coefficient α and the disaster scale threshold β when formulating relational equation (1).

[0049] In addition to this, it is possible to select and discard the configurations given in the above embodiment and each modified example, or to change them to other configurations as appropriate. [Explanation of symbols]

[0050] 1 Severely damaged area estimation system 4 Result display section 2. Information Collection Department A Specific Area 3 Severely Damaged Area Estimation Department M Area

Claims

1. A severely damaged area estimation system that estimates areas where damage is severe and where there are many people who need rescue after a disaster occurs, A severely damaged area estimation system characterized by having a severely damaged area estimation unit that estimates, among multiple areas within a specific area to be estimated, areas with a particularly large number of people in need of rescue and where the damage is severe, based on the population of each of multiple areas within the specific area to be estimated, the number of posts on SNS regarding each of the multiple areas during a certain period after the disaster occurs, and a disaster scale value indicating the scale of the disaster in each of the multiple areas.

2. The Department of Estimated Severely Damaged Areas is For each of the plurality of districts, calculate an SNS post number magnification by dividing the number of posts to the SNS related to the district by the population of the district; Formulating a relational equation between the disaster scale value and the value based on the SNS post number magnification in the specific area when the disaster occurs by regression analysis; For each of the plurality of districts, calculate a reference value for the number of posts to the SNS, which is the number of posts that are expected to be made to the SNS for the district when the disaster occurs, by multiplying the value obtained by applying the disaster scale value for the district to the relational expression by the population of the district, and calculate the difference between the number of posts to the SNS and the reference value; For each of the districts, the districts where the damage is severe are estimated based on the difference.

2. The severely damaged area estimation system according to claim 1.

3. The severely damaged district estimation unit estimates that the damage in a district is severe when a value based on the difference is equal to or greater than an upper threshold or equal to or less than a lower threshold in each of the districts.

3. The severely damaged area estimation system according to claim 2.

Images