Landslide disaster prediction method and landslide disaster prediction device

Abstract

To provide a method for determining how high a landslide disaster occurring risk is.SOLUTION: A landslide disaster prediction method comprises: a past precipitation data preparation step of preparing spot past precipitation data including one or more past precipitation index values representing one or more precipitation indexes respectively and clock times when the values are measured, which have been observed in a spot past precipitation event that is a past precipitation event at a spot to be predicted; a prediction target precipitation index value preparation step of preparing one or more prediction target precipitation index values; an index value excess precipitation event extraction step of extracting, from the spot past precipitation data, index value excess precipitation events in which the past precipitation index values are the prediction target precipitation index values or more in all of the one or more precipitation indexes; and a most recent spot high precipitation event determination step of determining that the spot past precipitation event occurred at a time closest to a prediction target time from among the index value excess precipitation events is a most recent spot high precipitation event.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a landslide disaster prediction method and a landslide disaster prediction device for predicting the occurrence of landslides such as slope collapses and mudslides caused by rainfall. [Background technology]

[0002] Landslides, such as slope failures and mudslides, can cause serious damage to human lives. Therefore, it is important to predict the occurrence of landslides and accurately determine the timing of evacuation.

[0003] When heavy rainfall occurs, even for a short period of time, the probability of a landslide disaster occurring increases. In addition, when rainfall continues for a long period of time, the probability of a landslide disaster occurring increases even if the instantaneous rainfall intensity is not that strong. Therefore, when predicting landslide disasters, it is necessary to take into account both the short-term intensity of rainfall and the long-term total amount of rainfall.

[0004] Patent Document 1 describes a method for predicting the occurrence of landslides due to ongoing rainfall events using a short-term rainfall index value that indicates the intensity of short-term rainfall and a long-term rainfall index value that indicates the total amount of rainfall over a long period of time. In this method, a graph is created in advance in which data points (past data points 91) of short-term rainfall index values ​​and long-term rainfall index values ​​observed in past rainfall events are plotted on a graph (FIG. 18) in which one of the short-term rainfall index value and the long-term rainfall index value is on the X-axis and the other is on the Y-axis. In this graph, past data points 91 (which are referred to as boundary past data points 911) that form the outer edge of an area that includes all past data points 91 are connected in a stepped manner by straight lines parallel to the X-axis and Y-axis, which are referred to as boundary lines 92 (solid lines in FIG. 18).

[0005] Then, based on the data obtained up to the present time (time of prediction) in the ongoing rainfall event, a long-term rainfall index value and a short-term rainfall index value are calculated and plotted on the graph as a prediction time data point 93. Such plotting of prediction time data points 93 is continued successively during the rainfall event, and when this prediction time data point 93 is plotted outside the boundary line 92 (point 931 in FIG. 18), it means that rainfall never experienced before is occurring and particularly strict vigilance against landslides is required. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2015-232537 A [Non-patent literature]

[0007] [Non-Patent Document 1] Yasutaka Makihara and Masanobu Hirasawa, "Accuracy of the Tank Model in Predicting the Risk of Slope Failures," Meteorological Agency Research Bulletin, Vol. 45, No. 2, pp. 35-70, 1993. Summary of the Invention [Problem to be solved by the invention]

[0008] In the method described in Patent Document 1, when the prediction data point 93 is inside the boundary line 92, that is, when the current rainfall has not reached the level of rainfall that has never been experienced in the past, it is not possible to determine the extent to which the risk of a landslide has increased by comparing it with past rainfall.

[0009] The problem that the present invention aims to solve is to provide a landslide prediction method and a landslide prediction device that can provide an index for determining the extent to which the risk of a landslide has increased by comparing it with past rainfall. [Means for solving the problem]

[0010] The landslide disaster prediction method according to the present invention, which has been achieved to solve the above problems, comprises: a) a past rainfall data preparation step of preparing local past rainfall data including one or more past rainfall index values ​​each representing one or more rainfall indices observed in a local past rainfall event, which is a past rainfall event at a location to be predicted, and the time at which they were measured; b) preparing one or more forecast target rainfall index values, each of which represents one or more rainfall indexes obtained at a time of forecast target at the location; c) extracting an index value exceeding rainfall event from the local past rainfall data, in which the past rainfall index value is equal to or greater than the forecast target rainfall index value in one or more rainfall indices; d) determining a local past rainfall event that occurred closest to the time of the forecast target time among the index value exceeding rainfall events as a local high rainfall event; The present invention is characterized by having the following.

[0011] Here, a "rainfall event" refers to a series of events from the start of rainfall to the end of rainfall, or from the end of rainfall until a certain period of time (for example, 24 hours) has elapsed.

[0012] The "time" can be determined by the rainfall index if it is determined by the rainfall index. Alternatively, it may be the maximum instantaneous rainfall in the past rainfall event. Furthermore, it may be the start or end time of the past rainfall event. This "time" includes at least the year in which the past rainfall index value was obtained, and may also include units of the month, day, hour, and even minutes.

[0013] The "prediction target time" may be the current time when the prediction is performed, or may be a time after the current time, i.e., in the future (for example, several hours from the current time). When performing prediction for the future, the predicted rainfall index value is determined by adding the observed rainfall value up to the current time to the forecast rainfall value from the current time to the prediction target time.

[0014] In the landslide disaster prediction method according to the present invention, all past rainfall events (local past rainfall events) at the prediction target point (local) where one or more rainfall indices (all of them in the case of "multiple") are equal to or greater than the value at the time of prediction are extracted (the extracted events are "index value exceeding rainfall events"). Among the index value exceeding rainfall events, the index value exceeding rainfall event at the time closest to the time of prediction is determined as the most recent local high rainfall event. The most recent local high rainfall event indicates "the rainfall at the time of prediction at the local" since "which past rainfall event (since when)". As a rule of thumb, the larger the rainfall index in a rainfall event, the lower the occurrence frequency of the rainfall event. In other words, the larger the rainfall index, the longer the time interval between occurrence of the rainfall event. Therefore, the older the occurrence time of the most recent local high rainfall event determined by the method of the present invention, the greater the rainfall index at the time of prediction at the local is, indicating that its magnitude is unprecedented. This indicates that there is a high possibility of landslides occurring at the time of the prediction, and information on recent local heavy rainfall events determined by this method serves as an index for judging the degree to which the risk of landslides occurring has increased. In addition, among rainfall events of equal or greater magnitude than those at the time of the prediction, the most recent events are more vividly remembered in people's minds, and there is a high possibility that more numerous and detailed records remain. Therefore, the extraction of such recent local heavy rainfall events is of great help in dealing with rainfall events at the time of the prediction.

[0015] In addition, when there is no index value exceeding rainfall event throughout the entire period for which past rainfall data records exist, it is not possible to determine the most recent local high rainfall event according to the above definition, so for convenience it is advisable to treat the start time of recording of the past rainfall data as the occurrence time of the most recent local high rainfall event.

[0016] The "one or more rainfall indices" are preferably multiple rainfall indices. By using multiple rainfall indices, it is possible to make a judgment from a more multifaceted perspective than when a single rainfall index is used, and it is possible to reduce the occurrence of so-called oversights (when a landslide actually occurs despite the judgment that the possibility of the landslide occurring is low).

[0017] It is sufficient to use at least two types of rainfall indices, the first and second rainfall indices, as the plurality of rainfall indices, but N types of rainfall indices (N≧3, ie, three or more types) may be used.

[0018] For example, the effective rainfall defined as follows can be used as the rainfall index (see Patent Document 1). When hourly rainfall data is used, the effective rainfall at a certain time t is a function X(M, t) including a parameter M, X(M, t)=X(M, t-1)exp(α)+R(t) …(1) Here, X(M, t-1) is the effective rainfall one hour before time t (time (t-1)), and R(t) is the rainfall (hourly rainfall) for one hour between time (t-1) and t. Also, α is a negative coefficient called the reduction coefficient, which is expressed by using the parameter M. α = ln(0.5) / M It is expressed as: Equation (1) of effective rainfall X(M, t) means that the amount of rainfall R(t) that falls at a certain time t remaining in the soil decreases to half of R(t) after the passage of time M. Therefore, the parameter M is called the "half-life". Two effective rainfalls with different half-lives M can be used as the first rainfall index and the second rainfall index, respectively. When calculating the effective rainfall, an initial value X(M, 0) is set when time t is 0. The initial value X(M, 0) is usually set to 0, which is the theoretical minimum value.

[0019] When the effective rainfall is used as the multiple rainfall indices, multiple effective rainfalls with different half-lives M may be used, and further, the multiple effective rainfalls may be used in combination with other rainfall indices. Alternatively, an effective rainfall with a certain half-life M may be used in combination with one or more rainfall indices other than the effective rainfall. Also, one or more rainfall indices other than the effective rainfall may be used without using the effective rainfall. Examples of rainfall indices other than the effective rainfall include the soil rainfall index, the amount of water stored in each tank in the tank model (see Non-Patent Document 1) used to calculate the soil rainfall index, the 60-minute rainfall, the accumulated rainfall since the start of rainfall (hereinafter simply referred to as "accumulated rainfall"), and the rainfall duration.

[0020] In the landslide disaster prediction method according to the present invention, In the past rainfall data preparation step, past rainfall index values ​​obtained for N types of rainfall indexes, which are three or more types, are prepared as the one or more past rainfall index values; In the prediction target rainfall index value preparation step, the prediction target rainfall index value is prepared for each of the N types of rainfall indices; In the index value exceeding rainfall event extraction step, a plurality of combinations of rainfall indices each consisting of two or more and less than N of the N types of rainfall indices are prepared, and for each combination, an index value exceeding rainfall event is extracted in which the past rainfall index value is equal to or greater than the forecast target rainfall index value for all the rainfall indices in the combination; In the most recent local high rainfall event determination step, for each of the combinations, a local past rainfall event that occurred at the closest time to the prediction target time among the index value exceeding rainfall events is identified as a most recent local high rainfall event candidate, and the local past rainfall event that occurred at the oldest time among the most recent local high rainfall event candidates is determined as the most recent local high rainfall event. This may be done.

[0021] In addition, in each of the multiple combinations of rainfall indices, when there is no index value exceeding rainfall event throughout the entire period for which past rainfall data is recorded, for convenience, it is convenient to treat the start time of recording of the past rainfall data as the occurrence time of the most recent local high rainfall event candidate for that combination.

[0022] In the index value exceeding rainfall event extraction step, in addition to extracting an index value exceeding rainfall event for each of the combinations of the plurality of rainfall indices, an index value exceeding rainfall event may be extracted using a single rainfall index for one or more rainfall indices (not combined with other rainfall indices, and each of the rainfall indices when there are multiple rainfall indices). In that case, in the most recent local high rainfall event determination step, most recent local high rainfall event candidates are obtained from index value exceeding rainfall events for each of the combinations of the plurality of rainfall indices, and most recent local high rainfall event candidates are obtained from index value exceeding rainfall events for each of the single rainfall indices, and the most recent local high rainfall event candidate that occurred the earliest among the combinations of the most recent local high rainfall event candidates is determined to be the most recent local high rainfall event. Even when extracting index value exceeding rainfall events using a single rainfall index, if there are no index value exceeding rainfall events over the entire period for which past rainfall data records exist, it is convenient to treat the start time of recording the past rainfall data as the occurrence time of the most recent local high rainfall event candidate for that rainfall index.

[0023] As described above, there are many rainfall indices, and the indices and their combinations suitable for predicting the occurrence of landslides vary depending on the topography, geology, soil, vegetation, etc. at the prediction target point. It is difficult to select which of the many rainfall indices is suitable for predicting the occurrence of landslides at the target point, or which of the combinations of rainfall indices is suitable for predicting the occurrence of landslides at the target point. Therefore, past rainfall index values ​​and prediction target rainfall index values ​​are prepared for each of N rainfall indices (three or more), and index value excess rainfall events are extracted for each combination of multiple rainfall indices consisting of two or more and less than N of the N rainfall indices, and recent local high rainfall event candidates are identified. Furthermore, when index value excess rainfall events are extracted for one or more rainfall indices, recent local high rainfall event candidates are also identified for the single rainfall indices (if there are multiple rainfall indices, each of them). Then, the local past rainfall event that occurred at the earliest time among the most recent local high rainfall event candidates is determined as the most recent local high rainfall event. This makes it possible to determine the most recent local high rainfall event without considering the selection of the rainfall index or the combination thereof. Here, by determining the local past rainfall event that occurred at the earliest time among the most recent local high rainfall event candidates as the most recent local high rainfall event, even if any of the rainfall indexes or the combination thereof is optimal, the most recent local high rainfall event that is newer than the most recent local high rainfall event candidate obtained from the optimal rainfall index or the combination thereof, i.e., that indicates a lower risk of a landslide disaster, will not be determined as the most recent local high rainfall event, so that the possibility of overlooking the risk of a landslide disaster can be reduced.

[0024] Alternatively, in the index value exceeding rainfall event extraction step, from index value exceeding rainfall events extracted using a single rainfall index for each of a plurality of rainfall indices (without using a combination of rainfall indices), a most recent local high rainfall event candidate may be obtained for each single rainfall index, and the most recent local high rainfall event that occurred at the earliest time may be determined as the most recent local high rainfall event based only on the most recent local high rainfall event candidates. In this way, by obtaining the most recent local high rainfall event candidates using only each single rainfall index, the process of determining the most recent local high rainfall event candidates can be simplified. The handling when there is no index value exceeding rainfall event for a single rainfall index over the entire period in which past rainfall data records exist is the same as above.

[0025] Alternatively, the extraction of index value excess rainfall events and the identification of nearby local heavy rainfall event candidates may be performed for each of a combination of rainfall indices and one or more individual rainfall indices.

[0026] In the landslide disaster prediction method according to the present invention, a warning judgment time preparation step for preparing a warning judgment time, which is the end time of a local past rainfall event that caused a landslide disaster in the past at the location, the time when a landslide disaster occurred at a time other than the period of the local past rainfall event, or the time when an event that increases the risk of a landslide disaster occurred; a landslide disaster occurrence risk determination process for determining that the risk of a landslide disaster occurrence is high when the warning determination time is between the time when the most recent local heavy rainfall event occurred and the prediction target time; The configuration may include the following:

[0027] In this way, by determining that the risk of a landslide is high when the alert judgment time is between the time when the most recent local heavy rainfall event occurred and the time to be predicted, it is possible to more reliably recognize that the risk of a landslide is high. Note that in this specification, "the time between the time when the most recent local heavy rainfall event occurred and the time to be predicted" includes the same time as the time when the most recent local heavy rainfall event occurred.

[0028] In the case where a landslide disaster has occurred in the past due to a rainfall event at the prediction target location, the warning judgment time can be the end time of the local past rainfall event that caused the landslide disaster. On the other hand, in the case where a landslide disaster occurs at a time other than the period of the local past rainfall event, such as when a landslide disaster occurs after the end of the local past rainfall event, or when a landslide disaster occurs due to an event other than the local past rainfall event, the time when the landslide disaster occurred can be set as the warning judgment time. Furthermore, in the case where an event that increases the risk of a landslide disaster has occurred in the past even if it does not result in a landslide disaster, the time when the event occurred can be set as the warning judgment time. Examples of causes of such events include strong shaking due to an earthquake at the prediction target location. Such strong shaking loosens the ground, increasing the risk of a landslide disaster. Another example of a cause of an event that increases the risk of a landslide disaster is the occurrence of large-scale fallen trees due to wind damage. When fallen trees occur, the roots of trees that support the ground die and rainwater becomes more likely to seep into the ground, increasing the risk of slope collapse. In addition, when volcanic ash accumulates due to a volcanic eruption, the soil's infiltration capacity decreases, which increases flood flow and increases the risk of debris flow. Furthermore, artificial development activities such as tree cutting, road construction, and installation of solar panels, as well as changes in land use patterns that change the hydrological environment of slopes and watersheds, can also cause events that increase the risk of landslides.

[0029] The landslide disaster prediction device according to the present invention comprises: a) a local past rainfall data acquisition unit that acquires local past rainfall data including one or more past rainfall index values ​​each representing one or more rainfall indexes observed in a local past rainfall event that is a past rainfall event at a location to be predicted, and the time at which they were measured; b) a forecast target rainfall index value acquisition unit that acquires one or more forecast target rainfall index values, each of which represents one or more rainfall indexes obtained at a forecast target time at the location; c) an index value exceeding rainfall event extraction unit that extracts, from the local past rainfall data, an index value exceeding rainfall event in which the past rainfall index value is equal to or greater than the forecast target rainfall index value in one or all of a plurality of rainfall indices; d) a local high rainfall event determining unit that determines a local past rainfall event that occurred at a time closest to the forecast target time among the index value exceeding rainfall events as a local high rainfall event; The present invention is characterized by comprising:

[0030] In the landslide disaster prediction device, it is preferable that the "one or more rainfall indices" are multiple rainfall indices.

[0031] The landslide disaster prediction device is In the past rainfall data acquisition unit, past rainfall index values ​​obtained for N types of rainfall indexes, which are three or more types, are acquired as the one or more past rainfall index values; The prediction target rainfall index value acquisition unit acquires the prediction target rainfall index value for each of the N types of rainfall indices, In the index value exceeding rainfall event extraction unit, a plurality of combinations of rainfall indices each consisting of two or more and less than N of the N types of rainfall indices are prepared, and for each combination, an index value exceeding rainfall event is extracted in which the past rainfall index value is equal to or greater than the forecast target rainfall index value for all the rainfall indices in the combination; In the most recent local heavy rainfall event determination unit, for each of the combinations, a local past rainfall event that occurred at the closest time to the prediction target time among the index value exceeding rainfall events is identified as a most recent local heavy rainfall event candidate, and the local past rainfall event that occurred at the oldest time among the most recent local heavy rainfall event candidates is determined as the most recent local heavy rainfall event. It may have the following configuration.

[0032] The landslide disaster prediction device further comprises: a warning decision time acquisition unit that acquires a warning decision time, which is the end time of a local past rainfall event that caused a landslide disaster in the past at the location, the time when a landslide disaster occurred at a time other than the period of the local past rainfall event, or the time when an event that increases the risk of a landslide disaster occurred; a landslide disaster occurrence risk determination unit that determines that a risk of a landslide disaster occurrence is high when the alert determination time is between the time when the most recent local heavy rainfall event occurred and the prediction target time; The device may include:

[0033] In the present invention, the landslide disaster prediction device may be owned by an administrator of a local government or the like, and may be connected to a user terminal such as a PC (personal computer), smartphone, tablet, or the like owned by a user such as a resident via a network such as the Internet. The present invention also provides a program for use with such a user terminal. The user terminal program for a landslide disaster prediction system according to the present invention is a program for a user terminal that is used by connecting a computer to the landslide disaster prediction device via a network, A local high-rainfall event acquisition unit that acquires the local high-rainfall event from the local high-rainfall event determination unit via a network; a local heavy rainfall event notification unit that notifies a user of the user terminal of the local heavy rainfall event acquired by the local heavy rainfall event acquisition unit; It is intended to function as a

[0034] The most recent local heavy rainfall event notification unit can notify the user of the user terminal of the most recent local heavy rainfall event, for example, by displaying information about the most recent local heavy rainfall event on a display unit or by emitting a sound.

[0035] According to the user terminal program for the landslide disaster prediction system of the present invention, it is possible to accurately notify users of user terminals, such as residents, of the most recent local heavy rainfall event.

[0036] In the case where the landslide disaster prediction device has the warning judgment time acquisition unit and the landslide disaster occurrence risk judgment unit, the user terminal program for the landslide disaster prediction system according to the present invention can function as a landslide disaster occurrence risk notification unit that notifies predetermined information when the landslide disaster occurrence risk judgment unit judges that the risk of landslide disaster occurrence at the location is increasing. By making the user terminal function as a landslide disaster occurrence risk notification unit in this way, the risk of landslide disaster occurrence can be more reliably conveyed to the user of the user terminal. The predetermined information can be notified to the user of the user terminal, for example, by displaying the information on a display unit or emitting a voice.

[0037] In addition, when the program for a user terminal for a landslide disaster prediction system is made to function as a landslide disaster risk notification unit, it may also function as the most recent local high rainfall event acquisition unit and the most recent local high rainfall event notification unit, or the functions as the most recent local high rainfall event acquisition unit and the most recent local high rainfall event notification unit may be omitted (i.e., the most recent local high rainfall event may not be notified to the user of the user terminal), and only information to that effect may be notified to the user when the risk of landslide disaster occurrence is increasing.

[0038] The warning judgment time acquisition unit and the landslide risk determination unit may be provided in a user terminal together with the landslide prediction device or in place of the landslide prediction device. In that case, the user terminal program for the landslide prediction system according to the present invention further comprises: a warning judgment time input unit that allows a user to input a warning judgment time, which is the end time of a local past rainfall event that caused a landslide in the past at the location, the time when a landslide occurred outside the period of the local past rainfall event, or the time when an event that increases the risk of a landslide occurred; a user terminal-side landslide disaster occurrence risk determination unit that acquires a time when the most recent local heavy rainfall event occurred from the most recent local heavy rainfall event determination unit, and determines that a risk of a landslide disaster occurrence is high when the alert judgment time inputted into the alert judgment time input unit is a time between the time when the most recent local heavy rainfall event occurred and the prediction target time. Function as.

[0039] The alert judgment time input by the user through the alert judgment time input unit can be the same as the alert judgment time targeted in the landslide disaster prediction method having the alert judgment time preparation process and the landslide disaster occurrence risk judgment process.

[0040] The program for a user terminal for a landslide disaster prediction system according to the present invention further comprises: a prediction target location candidate information acquisition unit that acquires information on a plurality of prediction target location candidates; a prediction target location selection unit that allows a user to select one of the plurality of prediction target location candidates as the location; It may also function as

[0041] By allowing the user to select one of multiple candidate prediction locations as the location to be predicted, it is possible to accurately convey information regarding the risk of landslides at locations within the user's residential area or other locations of interest to the user. Effect of the Invention

[0042] The landslide disaster prediction method and device of the present invention can obtain the most recent local heavy rainfall event, which indicates how long it has been since the last rainfall and serves as an indicator for determining the degree to which the risk of a landslide disaster has increased by comparing it with past rainfall. [Brief description of the drawings]

[0043] [Figure 1] 1 is a schematic configuration diagram showing first and second embodiments of a landslide disaster prediction device according to the present invention; [Diagram 2] 1 is a flowchart showing a first embodiment of a landslide disaster prediction method according to the present invention. [Diagram 3] 1 is a graph showing the first and second past rainfall index values, the first and second forecast target rainfall index values, the index value exceedance time, and the occurrence time of the most recent local heavy rainfall event in the landslide disaster prediction method of the first embodiment. [Figure 4] A diagram to explain how to determine the date since rainfall. [Diagram 5] 5 is an example of a graph displayed on the display unit in the first embodiment. [Figure 6] FIG. 6 is a diagram for explaining information indicated by the graph in FIG. 5 . [Figure 7] 4 is a flowchart showing a second embodiment of a landslide disaster prediction method according to the present invention. [Figure 8] A diagram showing combinations of two selected from N rainfall indices. [Figure 9] FIG. 11 is a schematic configuration diagram showing a third embodiment of a landslide disaster prediction device according to the present invention. [Figure 10] 11 is a flowchart showing a third embodiment of a landslide disaster prediction method according to the present invention. [Figure 11] 13 shows examples of graphs displayed on the display unit in the third embodiment. [Figure 12] 13 shows another example of a graph displayed on the display unit in the third embodiment. [Figure 13] FIG. 13 is a schematic configuration diagram showing a user terminal program for a landslide disaster prediction device and a landslide disaster prediction system according to a fourth embodiment of the present invention. [Figure 14] 1 is a graph showing an example (lower part) of determining the occurrence time of the most recent local heavy rainfall event based on the actual measured value of 60-minute rainfall (upper part) in the first embodiment. [Figure 15]A graph showing an example (bottom row) of determining the occurrence time of the most recent local high rainfall event based on the forecast value of accumulated rainfall (top row) and the forecast value of 60-minute rainfall (middle row) obtained as of 6:00 p.m. on July 3, 2020 in the first embodiment. [Figure 16] A graph showing an example (bottom row) of determining the occurrence time of the most recent local high rainfall event based on the forecast value of accumulated rainfall (top row) and the forecast value of 60-minute rainfall (middle row) obtained as of 9:00 p.m. on July 3, 2020 in the first embodiment. [Figure 17] 13 is a graph showing an example (lower part) of determining the occurrence time of the most recent local heavy rainfall event based on the actual measured value of 60-minute rainfall (upper part) in the second embodiment. [Figure 18] FIG. 1 is a diagram for explaining a conventional landslide prediction method, showing data points based on past rainfall, a boundary line set based on the data points, and data points for currently ongoing rainfall. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0044] First to fourth embodiments of the landslide disaster prediction method and landslide disaster prediction device according to the present invention will be described with reference to Figs. 1 to 13. First, in the first embodiment, a method and device for determining a recent local high rainfall event that serves as an index for landslide disaster prediction using two rainfall indices, a first rainfall index and a second rainfall index, as the one or more rainfall indices will be described. In the second embodiment, a method and device for determining a recent local high rainfall event using three or more rainfall indices will be described. The recent local high rainfall event obtained by the method and device of the first and second embodiments is itself an index for determining the degree to which the risk of a landslide disaster has increased, as described above. In the third embodiment, a method and device having a configuration for performing an operation to determine the risk of a landslide disaster occurrence using the recent local high rainfall event obtained by the method and device of the first or second embodiment will be described. In the fourth embodiment, in addition to the landslide disaster prediction device, a program for a user terminal for a landslide disaster prediction system will also be described.

[0045] Before describing each embodiment, rainfall indices will be described. Rainfall indices include effective rainfall, soil rainfall index, 60-minute rainfall, accumulated rainfall, rainfall duration, and the like. Among these, for effective rainfall, multiple effective rainfalls with different half-lives may be used as multiple rainfall indices, or one effective rainfall obtained for one half-life may be used in combination with a rainfall index other than the effective rainfall. Among the rainfall indices exemplified here, the 60-minute rainfall indicates the short-term rainfall intensity, and the soil rainfall index, accumulated rainfall, and rainfall duration indicate the long-term rainfall total amount. The effective rainfall indicates the short-term rainfall intensity when the half-life is shortened, and indicates the long-term rainfall total amount when the half-life is extended.

[0046] (1) First embodiment As shown in FIG. 1, the landslide disaster prediction device 10 of the first embodiment includes a local past rainfall data acquisition unit 11, a prediction target rainfall index value acquisition unit 12, an index value exceeding rainfall event extraction unit 13, and a recent local high rainfall event determination unit 14. These units are embodied by software that executes the landslide disaster prediction method of the first embodiment, and hardware such as a central processing unit operated by the software. The functions of these units will be described later together with the explanation of the landslide disaster prediction method of the first embodiment. The landslide disaster prediction device 10 further includes an input unit 16 that is used when a user inputs predetermined data to the local past rainfall data acquisition unit 11 and the prediction target rainfall index value acquisition unit 12, a display unit 17 that displays information indicating the recent local high rainfall event obtained as a result of processing by the above units, and a storage unit 18 that stores various data.

[0047] Below, the operation of the landslide disaster prediction device 10 and the landslide disaster prediction method of the first embodiment will be explained using the flowchart of Fig. 2 while referring to the graph of Fig. 3 and the explanatory diagram of Fig. 4. Note that, although a graph is shown in Fig. 3 to facilitate understanding of the method of the first embodiment, creating such a graph is not essential in this embodiment.

[0048] First, the local past rainfall data acquisition unit 11 acquires the local past rainfall data consisting of a first past rainfall index value, which is a value representing a first rainfall index, a second past rainfall index value, which is a value representing a second rainfall index different from the first past rainfall index, and the time of the occurrence of the past rainfall event at which these values ​​were obtained, for each past rainfall event at the prediction target point, by the following method, and stores it in the storage unit 18 (step S1 in FIG. 2: past rainfall data preparation step). The local past rainfall data may be acquired by a user directly inputting the local past rainfall data using the input unit 16, by a user mounting a removable medium on which the local past rainfall data is recorded to a reading unit (not shown) of the landslide disaster prediction device 10 and inputting the data by performing a predetermined operation using the input unit 16, or by inputting the local past rainfall data stored in another server via a network. Alternatively, the local past rainfall data may be calculated by inputting the original data to the local past rainfall data acquisition unit 11, and the local past rainfall data may be calculated by the local past rainfall data acquisition unit 11 using the original data. For example, a 60-minute rainfall amount may be input as the original data, and the local past rainfall data such as the effective rainfall amount and the accumulated rainfall amount may be calculated based on the original data. The original data may be input directly using the input unit 16, input from a removable medium using the reading unit, input via a network, or the like.

[0049] The graph in FIG. 3 shows an example of data of the first past rainfall index value and the second past rainfall index value. In this graph, the horizontal axis indicates the first past rainfall index value and the vertical axis indicates the second past rainfall index value, and these values ​​are shown as gray dots (past data points 21). Each past data point 21 corresponds to one past time, but these past times are not shown in FIG. 3 except for some (times are shown for index value excess past data points 211, which are part of the past data points 21, but these will be described later). The thin dashed line in the graph is the boundary line 92 (see FIG. 18) used in the method of Patent Document 1. This boundary line 92 is shown only for reference and is not used in this embodiment. Points and lines other than the past data points 21 and the boundary line 92 in this graph will be described in the following explanation of the method of the first embodiment.

[0050] Next, the prediction target rainfall index value acquisition unit 12 acquires the first prediction target rainfall index value and the second prediction target rainfall index value at the prediction target point at the prediction target time (step S2: prediction target rainfall index value preparation step). The first prediction target rainfall index value and the second prediction target rainfall index value can be acquired using the same acquisition method as the first past rainfall index value and the second past rainfall index value. The prediction target time may be the present time or the future. When the prediction target time is the future, the first prediction target rainfall index value and the second prediction target rainfall index value are determined using the observed value of rainfall up to the present time plus the forecast value of rainfall from the present time to the prediction target time.

[0051] In Fig. 3, the first and second forecast target rainfall index values ​​are indicated by white circles as forecast target time data points 23. Fig. 3 also indicates the first and second forecast target rainfall index values ​​obtained during the rainfall event at the forecast target time and before the forecast target time as data points 230 during the current rainfall event before the forecast target time. In this example, the first and second forecast target rainfall index values ​​both increase as the rainfall event progresses. However, since the forecast target time data points 23 are inside the boundary line 92, the method of Patent Document 1 cannot determine the degree to which the risk of landslide occurrence has increased at the forecast target time by comparing it with past rainfall.

[0052] Next, the index value exceeding rainfall event extraction unit 13 extracts from the past rainfall data those in which the first past rainfall index value is equal to or greater than the first forecast target rainfall index value and the second past rainfall index value is equal to or greater than the second forecast target rainfall index value, and extracts the local past rainfall event from which the past rainfall data was acquired (measured) as an index value exceeding rainfall event (step S3: index value exceeding rainfall event extraction process).

[0053] 3, a first rainfall index is a first forecast target rainfall index value and a straight line parallel to the Y-axis is shown as a first rainfall index boundary line 221, and a second rainfall index is a second forecast target rainfall index value and a straight line parallel to the X-axis is shown as a second rainfall index boundary line 222. FIG. 3 further shows an index value exceeding region 22 in which the value of the first rainfall index is equal to or greater than the value on the first rainfall index boundary line 221 and the value of the second rainfall index is equal to or greater than the value on the second rainfall index boundary line 222. Among the past data points 21, a local past rainfall event corresponding to an index value exceeding past data point 211 existing within this index value exceeding region 22 corresponds to an index value exceeding rainfall event. In the example shown in Fig. 3, five past data points 21 exist within the index value exceedance region 22, and the times "1974 / 06 / 18 19:00", "2004 / 07 / 04 02:00", "2004 / 07 / 04 03:00", "2000 / 10 / 02 01:00", and "1995 / 07 / 20 22:00" are recorded for these five past data points 21 (index value exceedance past data points 211). For example, "1974 / 06 / 18 19:00" indicates an index value exceedance rainfall event that occurred at 19:00 (7:00 p.m.) on June 18, 1974. The two index value exceedance past data points 211, "2004 / 07 / 04 02:00" and "2004 / 07 / 04 03:00", were both obtained during the same local past rainfall event that occurred from July 3rd to July 5th, 2004.

[0054] Next, the most recent local heavy rainfall event determination unit 14 determines the local past rainfall event that occurred closest to the target time among the index value exceeding rainfall events extracted by the index value exceeding rainfall event extraction unit 13 as the most recent local heavy rainfall event (step S4: most recent local heavy rainfall event determination step). This most recent local heavy rainfall event indicates "the rainfall at the target time in the local area since which past rainfall event (since when)", or more specifically, "the rainfall that has not been experienced since which past rainfall event (since when)".

[0055] Of the times at which multiple index value exceedance rainfall events occurred shown in FIG. 3, the time closest to the time of prediction is "2004 / 07 / 04 03:00" (the time framed in the figure). Therefore, in the example of FIG. 3, the local past rainfall event corresponding to this "2004 / 07 / 04 03:00" corresponds to the most recent local high rainfall event. Note that in the example of FIG. 3, as described above, two index value exceedance past data points 211, "2004 / 07 / 04 02:00" and "2004 / 07 / 04 03:00", are obtained from the same local past rainfall event, but here, the occurrence time of the most recent local high rainfall event is identified using "2004 / 07 / 04 03:00", which is the time closest to the time of prediction.

[0056] Here, the most recent local high rainfall event will be further explained with reference to FIG. 4. If an index value exceeding rainfall event occurs at time B closer to the forecast time than time A, which is one of the times when the index value exceeding rainfall event occurs (left diagram of FIG. 4), rainfall equivalent to or exceeding the forecast target time occurred at time B, which is closer to the forecast time than time A, so the current rainfall cannot be said to be "rainfall since time A". On the other hand, if an index value exceeding rainfall event does not occur at a time closer to the forecast time than time C, which is one of the index value exceeding times (right diagram of FIG. 4), the current rainfall can be said to be "rainfall since time C". Therefore, in this embodiment, such a local past rainfall event that occurred at time C is considered to be the most recent local high rainfall event.

[0057] Next, the nearest local heavy rainfall event determination unit 14 notifies the user of the nearest local heavy rainfall event by displaying information indicating the determined nearest local heavy rainfall event on the display unit 17 (step S5). Through the above operations, the operation of the landslide disaster prediction device 10 and the landslide disaster prediction method of the first embodiment are completed.

[0058] The information displayed on the display unit 17 is typically the time when the most recent local heavy rainfall event occurred. Alternatively, the name given to each local past rainfall event may be stored in the storage unit 18 together with the local past rainfall data, and the name may be displayed on the display unit 17 together with the time, or only the name. Such names may include names named by the Japan Meteorological Agency such as "Heisei 29 (2017) Northern Kyushu Heavy Rains," "Typhoon No. 15 of 2015," and "Ise Bay Typhoon" (Typhoon No. 15 of 1959), and names used by media organizations such as "Western Japan Heavy Rains" (heavy rains that occurred in July 2018 and caused damage across western Japan).

[0059] 5 may be displayed as information on the display unit 17. In this graph, the horizontal axis indicates the prediction target time, and the vertical axis indicates the occurrence time of the most recent local heavy rainfall event, with the occurrence time moving upward as the occurrence time becomes older. The occurrence time of the most recent local heavy rainfall event is plotted while changing the prediction target time.

[0060] Using FIG. 6, the information shown in the graph displayed on the display unit 17 as in FIG. 5 will be described. A point plotted on this graph indicates that the current rainfall is rainfall that has not been experienced since the occurrence time of the most recent local heavy rainfall event indicated by that point, in other words, the largest amount of rainfall since that time. The higher the plotted point is, that is, the further back in time the vertical axis is, the higher the risk of landslides. Also, when the line connecting the plotted points changes over time from the lower left to the upper right, the risk of landslides tends to increase, and when the line changes over time from the upper left to the lower right, the risk of landslides is decreasing.

[0061] If an event that reduces resistance to the occurrence of slope collapse, debris flow, etc. occurs at the time of occurrence of the most recent local heavy rainfall event indicated by the plotted point (the time on the vertical axis corresponding to line 251 in Figure 6) (Case 1), the time of occurrence of the most recent local heavy rainfall event will be the same as or older than the time of occurrence of the past event during the forecast period from the time indicated by line 261 (time on the horizontal axis). This indicates that the risk of landslides occurring is particularly high from the time indicated by line 261 onwards.

[0062] Furthermore, if there are no index value exceeding rainfall events over the entire period for which past rainfall data records exist, i.e., if the time of occurrence of the most recent local heavy rainfall event coincides with the start time of recording of past rainfall data, this indicates that the rain is unlike anything experienced since rainfall data recording began, and it can be determined that the level of risk is particularly high.

[0063] On the other hand, if events that reduce resistance to slope collapse, debris flows, etc. occur only above the plotted point, i.e., at times older than the time of the most recent local high rainfall event indicated by that point (straight line 252 in Figure 6) (Case 2), it can be determined that the risk of landslides is relatively low.

[0064] According to the landslide disaster prediction device 10 and the landslide disaster prediction method of the first embodiment described above, the most recent local heavy rainfall event is determined, which indicates "since which past rainfall event (since when) it has rained." In other words, since the older the most recent local heavy rainfall event is, the higher the possibility of a landslide disaster occurring at the time of prediction is, the most recent local heavy rainfall event is an index for determining the degree to which the risk of a landslide disaster occurring has increased.

[0065] (2) Second embodiment The landslide disaster prediction device of the second embodiment has the same configuration as the landslide disaster prediction device 10 of the first embodiment, but the functions of the local past rainfall data acquisition unit 11, the prediction target rainfall index value acquisition unit 12, the index value excess rainfall event extraction unit 13, and the recent local heavy rainfall event determination unit 14 are different from those of the first embodiment. The functions of these units will be explained below together with the landslide disaster prediction method of the second embodiment.

[0066] Hereinafter, the landslide disaster prediction method of the second embodiment will be described with reference to the flowchart shown in Fig. 7 and Fig. 8. First, the local past rainfall data acquisition unit 11 acquires local past rainfall data consisting of the first past rainfall index value to the Nth past rainfall index value, which are values ​​representing the first rainfall index to the Nth rainfall index, which are N types of rainfall indexes (N is a natural number of 3 or more), for each past rainfall event at the prediction target point, and the time of occurrence of the past rainfall event at which these values ​​were obtained (step S11: past rainfall data preparation step). The method of acquiring the local past rainfall data is the same as in the first embodiment.

[0067] Next, the prediction target rainfall index value acquisition unit 12 acquires the first prediction target rainfall index value to the Nth prediction target rainfall index value, which are values ​​representing N types of rainfall indexes at the prediction target point at the prediction target time (step S12: prediction target rainfall index value preparation step). The method of acquiring these prediction target rainfall index values ​​is the same as in the first embodiment. Also, as in the first embodiment, the prediction target time may be the present time or a future time.

[0068] Next, the index value exceeding rainfall event extraction unit 13 selects two of the first rainfall index to the Nth rainfall index, the Kth rainfall index and the (K+L)th rainfall index (K is a natural number from 1 to (N-1), and L is a natural number from 1 to (NK). Therefore, (K+L) is a natural number from (K+1) to N). First, two rainfall indices, the first rainfall index and the second rainfall index, are selected by setting K=1 and L=1 (step S13).

[0069] Next, the index value exceeding rainfall event extraction unit 13 extracts from the past rainfall data an index value exceeding rainfall event, which is a local past rainfall event in which the Kth past rainfall index value is equal to or greater than the Kth prediction target rainfall index value and the (K+L)th past rainfall index value is equal to or greater than the (K+L)th prediction target rainfall index value (step S14: index value exceeding rainfall event extraction step). As described above, in the case of K=1 and L=1, an index value exceeding rainfall event in which the first past rainfall index value is equal to or greater than the first prediction target rainfall index value and the second past rainfall index value is equal to or greater than the second prediction target rainfall index value is extracted from the past rainfall data.

[0070] Next, the most recent local high rainfall event determination unit 14 identifies the time closest to the prediction target time among the index value exceedance times in the combination of the Kth rainfall index and the (K+L)th rainfall index extracted by the index value exceedance rainfall event extraction unit 13, and identifies the local past rainfall event that occurred at that time as the most recent local high rainfall event candidate (step S15). Note that, when the index value exceedance past data points 211 are obtained for multiple times within the same index value exceedance rainfall event, such as the index value exceedance past data points 211 of "2004 / 07 / 04 02:00" and "2004 / 07 / 04 03:00" shown in Fig. 3, the time closest to the prediction target time among those data is set as the occurrence time of the most recent local high rainfall event candidate, and the process of step S20 described later is performed.

[0071] After step S15, if L has not reached (NK) (No in step S16), 1 is added to the value of L (step S17), and the operations of steps S14 and S15 are performed. After step S15, if L has reached (NK) (Yes in step S16), the process proceeds to step S18.

[0072] In step S18, if K has not reached (N-1) (No in the same step), 1 is added to the value of K and the value of L is returned to 1 (step S19), and the operations of steps S14 and S15 (and step S17 if No in S16) are performed. In step S18, if K has reached (N-1), the process proceeds to step S20.

[0073] When step S20 is reached, two of the N rainfall indices are selected. N The most recent local heavy rainfall event candidates are obtained for C2 combinations (see FIG. 8). For example, when N=6, the most recent local heavy rainfall event candidates are obtained for 6C2=15 combinations. In step S20, the most recent local heavy rainfall event determination unit 14 determines the most recent local heavy rainfall event that occurred at the oldest time (farthest from the time to be predicted) among these most recent local heavy rainfall event candidates as the most recent local heavy rainfall event (the most recent local heavy rainfall event determination step).

[0074] The nearest local heavy rainfall event determination unit 14 further performs an operation to display information indicating the determined nearest local heavy rainfall event on the display unit 17, thereby notifying the user of the nearest local heavy rainfall event (step S21). The contents indicated by the information indicating the nearest local heavy rainfall event and the operation for displaying it are the same as those in the first embodiment. With the above operations, the operation of the landslide disaster prediction device 10 and the landslide disaster prediction method of the second embodiment are completed.

[0075] According to the second embodiment of the landslide disaster prediction method and device, a local high rainfall event candidate is identified for each of a plurality of combinations of rainfall indices consisting of two of three or more rainfall indices, and the local past rainfall event that occurred at the earliest time among the local high rainfall event candidates is determined as the local high rainfall event, so that the local high rainfall event can be determined without considering the selection of the combination of rainfall indices. Here, by determining the local past rainfall event that occurred at the earliest time among the local high rainfall event candidates as the local high rainfall event, even if any of the combinations of rainfall indices is optimal, a local high rainfall event that is newer than the local high rainfall event candidate obtained from the optimal combination of rainfall indices, i.e., that indicates a low risk of a landslide disaster, will not be determined as the local high rainfall event. Therefore, the possibility of overlooking the risk of a landslide disaster can be reduced.

[0076] (3) Third embodiment 9, the landslide disaster prediction device 30 of the third embodiment has, in addition to the components of the landslide disaster prediction device 10 of the first embodiment, a warning judgment time acquisition unit 31 and a landslide disaster occurrence risk determination unit 32. The functions of the warning judgment time acquisition unit 31 and the landslide disaster occurrence risk determination unit 32 will be explained in the following paragraphs together with the explanation of the landslide disaster prediction method of the third embodiment.

[0077] The landslide disaster prediction method of the third embodiment will be described with reference to the flowchart shown in Fig. 10. First, the nearest local heavy rainfall event is determined by the same operations as steps S1 to S4 in the landslide disaster prediction method of the first embodiment or steps S11 to S20 in the landslide disaster prediction method of the second embodiment.

[0078] The warning judgment time acquisition unit 31 acquires a warning judgment time (step S31: warning judgment time preparation step). Here, if a landslide disaster has occurred in the past due to a rainfall event at the prediction target point, the warning judgment time can be the end time of the rainfall event that caused the landslide disaster. On the other hand, if a landslide disaster has occurred in the past after the end of a rainfall event or if a landslide disaster has occurred due to an event other than a rainfall event, the disaster occurrence time can be the warning judgment time. Furthermore, if an event that increases the risk of a landslide disaster has occurred in the past even if it did not result in a landslide disaster, the occurrence time of the event can be the warning judgment time. Examples of causes that cause such events include strong shaking due to an earthquake and large-scale fallen trees. There may be only one warning judgment time, or there may be multiple warning judgment times. If the warning judgment time acquisition unit 31 acquires multiple warning judgment times, the user may select one of the multiple warning judgment times depending on the conditions of each slope or stream.

[0079] The method of obtaining the alert judgment time may be a method in which the user directly inputs the alert judgment time using the input unit 16, or a method in which the user inserts a removable medium on which the alert judgment time is recorded into a reading unit (not shown) of the landslide disaster prediction device 30 and then inputs the time by performing a specified operation using the input unit 16, or a method in which the alert judgment time stored in another server is input via a network.

[0080] Although step S31 is described here as being executed after step S4 or S20, in reality it may be executed at any timing before the next step S32 is executed.

[0081] Next, the landslide disaster occurrence risk determination unit 32 determines whether the warning judgment time is a time between the occurrence time of the most recent local heavy rainfall event and the predicted time (Step S32: landslide disaster occurrence risk determination process). Here, the "time between the occurrence time of the most recent local heavy rainfall event and the predicted time" includes the same time as the occurrence time of the most recent local heavy rainfall event. If the determination result is Yes, the landslide disaster occurrence risk determination unit 32 notifies information indicating that the risk of a landslide disaster occurrence is increasing (Step S33). Specific examples of the information will be described later. If the determination result is No, the operation at the predicted time is terminated as it is.

[0082] The notification in step S33 may be made by displaying characters or images, or by displaying a graph shown in FIG. 11. The graph in FIG. 11 is obtained by adding a landslide disaster occurrence risk judgment line 41, which is a straight line parallel to the X-axis indicating the time corresponding to the alert judgment time on the vertical axis of the graph shown in FIG. 5. This graph means that the risk of a landslide disaster occurrence is high at the time of prediction when a point or line 42 indicating the occurrence time of the most recent local high rainfall event is at the same level as or above the landslide disaster occurrence risk judgment line 41. On the other hand, as shown in FIG. 12, if the entire point or line 421 indicating the occurrence time of the most recent local high rainfall event is below the landslide disaster occurrence risk judgment line 41, it can be determined that the risk of a landslide disaster occurrence is relatively low. The landslide disaster occurrence risk judgment line 41 may be only one or may be multiple. When there are multiple landslide disaster occurrence risk judgment lines 41, the user may select one of the multiple landslide disaster occurrence risk judgment lines 41 depending on the conditions of each slope or stream.

[0083] According to the third embodiment of the landslide disaster prediction device 30 and the landslide disaster prediction method, when the warning judgment time is between the time of occurrence of the most recent local heavy rainfall event and the prediction target time, it is possible to more reliably recognize the increasing risk of a landslide disaster occurring.

[0084] (4) Fourth embodiment A fourth embodiment of the present invention will be described with reference to Fig. 13. In the fourth embodiment, an example of a landslide disaster prediction system is shown in which a landslide disaster prediction device 30 according to the third embodiment owned by a local government or the like and user terminals 81 such as PCs (personal computers), smartphones, tablets, etc. owned by residents or the like are connected via the Internet 80.

[0085] The landslide disaster prediction device 30 basically has the configuration described in the third embodiment, but each component of the landslide disaster prediction device 30 processes each of the multiple prediction target points. That is, for each of the multiple prediction target points, the local past rainfall data acquisition unit 11 acquires local past rainfall data, the prediction target rainfall index value acquisition unit 12 acquires the prediction target rainfall index value, the index value exceeding rainfall event extraction unit 13 extracts index value exceeding rainfall events, and the nearest local high rainfall event determination unit 14 determines the nearest local high rainfall event. Also, for each of the multiple prediction target points, the warning judgment time acquisition unit 31 acquires the warning judgment time, and the landslide disaster occurrence risk determination unit 32 determines whether the risk of landslide disaster occurrence is high.

[0086] A user terminal program for a landslide disaster prediction system (hereinafter, abbreviated as "user terminal program") 50 is installed in the user terminal 81. The user terminal program 50 is a program that causes the user terminal 81 to function as a prediction target point candidate information acquisition unit 51, a prediction target point selection unit 52, a recent local heavy rainfall event acquisition unit 53, a recent local heavy rainfall event notification unit 54, a landslide disaster occurrence risk notification unit 55, a warning judgment time input unit 56, and a user terminal side landslide disaster occurrence risk determination unit 57.

[0087] The prediction target point candidate information acquisition unit 51 acquires information (prediction target point candidate information) on a plurality of prediction target points stored in a memory unit (not shown) of the landslide disaster prediction device 30 via the Internet 80. The prediction target point selection unit 52 allows the user of the user terminal 81 to select a selected prediction target point to be a point to be predicted on the user terminal 81 from the plurality of prediction target point candidate information acquired by the prediction target point candidate information acquisition unit 51.

[0088] Here, the prediction target point candidate information may be, for example, an identification code such as row and column numbers assigned to each area after dividing a map into a plurality of areas in a mesh shape. The identification code may not be notified to the user (not displayed on the screen), and the prediction target point selection unit 52 may display the plurality of areas on the map, and the user may select an area using an input device such as a mouse or a touch panel. The prediction target point candidate information acquisition unit 51 may acquire the prediction target point candidate information from a recording unit (not shown) of the user terminal 81 instead of acquiring the prediction target point candidate information from the storage unit of the landslide disaster prediction device 30. In that case, when the user terminal program 50 is installed in the user terminal 81, the information may be recorded in the recording unit of the user terminal 81. Alternatively, the prediction target point candidate information may be periodically transmitted from the landslide disaster prediction device 30 to the user terminal 81 and stored in the storage unit of the user terminal 81.

[0089] The most recent local high rainfall event acquisition unit 53 acquires information on the most recent local high rainfall event at the selected prediction target point, i.e., the prediction target point selected by the user, from the most recent local high rainfall event determination unit 14 via the Internet 80. The most recent local high rainfall event notification unit 54 performs an operation for notifying the user of the information on the most recent local high rainfall event acquired by the most recent local high rainfall event acquisition unit 53. The information on the most recent local high rainfall event may be displayed, for example, as characters or figures on the screen of the user terminal 81, or may be notified by voice. The most recent local high rainfall event acquisition unit 53 may check the information on the most recent local high rainfall event at any time, and the most recent local high rainfall event notification unit 54 may notify only when the information is updated.

[0090] Furthermore, when the landslide disaster occurrence risk determination unit 32 determines that the risk of a landslide disaster occurring at the selected prediction target point is increasing, the landslide disaster occurrence risk notification unit 55 receives a predetermined signal from the landslide disaster occurrence risk determination unit 32, and upon receiving the signal, notifies the user of information indicating that the risk of a landslide disaster occurring is increasing via the user terminal 81. This notification of information may be displayed by characters, figures, etc. on the screen of the user terminal 81, as with the notification of information on the most recent local heavy rainfall event, or may be notified by voice.

[0091] The warning judgment time input unit 56 controls the input device and screen of the user terminal 81 so that the user inputs a warning judgment time, which is the end time of a local past rainfall event that caused a landslide in the past, the time when a landslide occurred at a time other than the period of the local past rainfall event, or the time when an event that increases the risk of a landslide occurred. If a landslide has occurred in the past near the user's home, the end time of the local past rainfall event that caused the landslide or the occurrence time of the landslide (if it occurred at a time other than the period of the local past rainfall event) can be used as the warning judgment time input by the user. In addition, if an event that increases the risk of a landslide has occurred near the user's home due to the decline of vegetation due to disease or insect damage, artificial development activities, or changes in land use that change the hydrological environment of slopes and watersheds, the occurrence time of the event can be used as the warning judgment time.

[0092] The user terminal-side landslide disaster occurrence risk determination unit 57 acquires the time when the most recent local heavy rainfall event occurred from the most recent local heavy rainfall event determination unit 14, and determines that the risk of a landslide disaster occurrence is increasing if the alert judgment time input to the alert judgment time input unit 56 is between the time when the most recent local heavy rainfall event occurred and the prediction target time. When making this determination, the user terminal-side landslide disaster occurrence risk determination unit 57 transmits a predetermined signal to the landslide disaster occurrence risk notification unit 55. Upon receiving the predetermined signal, the landslide disaster occurrence risk notification unit 55 performs an operation of notifying the user via the user terminal 81 of information indicating that the risk of a landslide disaster occurrence is increasing.

[0093] According to the landslide disaster prediction device 30 and the user terminal program 50 of the fourth embodiment, the user is notified of the nearest local heavy rainfall event at the selected prediction target point selected by the user, so that the nearest local heavy rainfall event at the prediction target point required by the user, such as the user's residence, can be accurately notified. In addition, since it has a function of notifying the user of information indicating that the risk of landslide disaster occurrence at the selected prediction target point is increasing, the increasing risk of landslide disaster occurrence can be more accurately conveyed to the user.

[0094] In addition, the user inputs the alert judgment time using the alert judgment time input unit 56, and the landslide disaster occurrence risk judgment unit 57 on the user terminal determines whether or not the risk of landslide disaster occurrence is increasing based on that alert judgment time, so that the user can be more accurately informed of the increasing risk of landslide disaster occurrence based on the conditions required at the location to be predicted. EXAMPLES

[0095] Below, using Figures 14 to 17, we will explain the results of a simulation in which the landslide prediction method of the present invention was executed using an example of a rainfall event that actually caused a landslide, with the time during the period in which the rainfall event occurred as the prediction target time.

[0096] [Example 1] In Example 1, a simulation was performed using a rainfall index (details described later) calculated from the 60-minute rainfall observed in Ashikita-machi, Ashikita-gun, Kumamoto Prefecture (hereinafter abbreviated as "Ashikita-machi") during the heavy rainfall that occurred on July 3-4, 2020 (July 2020 heavy rain). The July 2020 heavy rain caused a lot of damage due to landslides and floods, and in Ashikita-machi alone, there were 11 deaths, 1 missing person, 73 houses completely destroyed, 157 houses partially destroyed on a large scale, 934 houses partially destroyed, 183 houses partially destroyed, and 500 houses partially damaged, resulting in severe human and material damage.

[0097] Even prior to the heavy rains of July 2020, since 1976, when rainfall observation data for Ashikita Town began to be kept, past rainfall events involving human and / or property damage occurred in the area on the following dates: ·July 12, 1982 ·July 24, 1982 (Heavy rains of July 1982) ·July 6-13, 1997 ·July 5-10, 2005 ·July 12, 2012 (Heavy rains in northern Kyushu in July 2012) August 24-25, 2015 (Typhoon No. 15 of 2015) ·July 6-8, 2018 (July 2018 heavy rain = Western Japan heavy rain)

[0098] In this embodiment, for all local past rainfall events for which rainfall observation data exists (occurring since 1976), including past local rainfall events that involved human and / or property damage, past rainfall index values ​​(1st to 51st past rainfall index values: N=51) were calculated from the observed 60-minute rainfall values ​​using 51 different half-lives between 0.1 and 3000 hours. Here, the 51 half-lives were determined so that the amount of change in the logarithmic value would be constant. Then, using the method of the second embodiment, a combination of two mutually different past rainfall index values ​​from the 1st to 51st past rainfall index values ​​was calculated. 51 For each of the C2 = 1275 combinations, recent local heavy rainfall event candidates were obtained. In addition, recent local heavy rainfall event candidates were obtained for each of the 51 cases using only one of the 1st to 51st past rainfall index values. Then, of the combined 1275 + 51 = 1326 recent local heavy rainfall event candidates, the one that occurred at the oldest time (farthest from the time of the forecast target) was determined to be the recent local heavy rainfall event. In addition, the time at which each of the local past rainfall events that caused human and / or property damage ended was determined to be a candidate for the warning decision time.

[0099] The upper part of Figure 14 shows the 60-minute rainfall data observed in Ashikita Town at each time every 60 minutes between 00:00 on July 3, 2020 and 18:00 on July 4, 2020 (6:00 p.m.) (hereinafter, "July 2020" will be omitted and written as "3rd x o'clock" or "4th x o'clock"). Note that the 60-minute rainfall data is not limited to that shown in this figure, but has been obtained for the period since 1976. The 1st to 51st past rainfall index values ​​at each time are obtained from the 60-minute rainfall data for the period since 1976. The lower part of Figure 14 shows the results of determining the most recent local high rainfall event using the 1st to 51st past rainfall index values ​​by the above method, with each time every 60 minutes as the forecast target time. The most recent local high rainfall event is shown on the vertical axis of the graph in the lower part as the time (year, month, day, hour) at which the most recent local high rainfall event occurred.

[0100] From the graph at the bottom, the following can be said. First, for the forecast period up to 6:00 on the 3rd, the occurrence time of the most recent local high rainfall event is the time of the rainfall event that occurred immediately before, from June 29 to July 1, 2020. After that, when the forecast period is from 7:00 to 14:00 on the 3rd, the occurrence time of the most recent local high rainfall event is July 2019. Furthermore, when the forecast period is from 15:00 on the 3rd to 2:00 on the 4th, the occurrence time of the most recent local high rainfall event goes back to the heavy rains of July 2018 (Western Japan Heavy Rains), and the risk of landslides has increased slightly. During this time period, it can be said that the rainfall was the largest in the period after the heavy rains of July 2018. For this reason, the possibility of landslides occurring was judged to be particularly high in the areas and surrounding areas where landslides occurred in the heavy rains of July 2018, and in areas where events that increase the risk of landslides occurred in the period after the heavy rains of July 2018, compared to other locations.

[0101] The 60-minute rainfall was already relatively high at over 20 mm from 0:00 to 2:00 on the 4th, and reached a record high of about 70 mm from 2:00 to 3:00 on the 4th. As a result, at 3:00 on the 4th, the time of the most recent high rainfall event in the area rose sharply to 1976 (the time when the observation data in Ashikita Town began to be recorded). This means that the rainfall is on a scale never experienced since the observation data began to be recorded, and indicates that the possibility of a landslide disaster occurring at 3:00 on the 4th is rapidly increasing. In fact, landslides occurred just after 4:00 on the 4th, at around 5:00, 5:15, 5:40, and 7:00 on the same day. Therefore, if some measures had been taken at 3:00 on the 4th, when the time of the most recent high rainfall event in the area suddenly went back, it is believed that the human and material damage caused by the landslide disaster could have been reduced.

[0102] In the simulations so far, the most recent local high rainfall events were obtained using the actual measured values ​​of 60-minute rainfall up to that point at each forecast time. In the following, the forecast values ​​of 60-minute rainfall after the forecast execution time obtained at 18:00 on the 3rd, which is the time of forecast execution (different from the forecast time), were used to obtain the most recent local high rainfall events for each time after 19:00 on the 3rd as the forecast time. The results are shown in Figure 15. In the upper part of the figure, the forecast values ​​of the accumulated rainfall are shown in dark colors, and the actual measured values ​​are shown in light colors. In the middle part, the forecast values ​​of the 60-minute rainfall are shown in dark colors, and the actual measured values ​​are shown in light colors. In the lower part, the occurrence time of the most recent local high rainfall event obtained using the forecast values ​​of the 60-minute rainfall is shown together with the occurrence time of the most recent local high rainfall event obtained using the actual measured values ​​of the 60-minute rainfall (the same as the data in the lower part of Figure 14).

[0103] From the data in Figure 15, it can be seen that, based on the forecast values, at 18:00 on the 3rd, it can be predicted that the risk of landslides will increase in the early morning of the following day (7:00 on the 4th) because the time of the most recent local high rainfall event is rapidly retraced. If the risk of early morning landslides can be predicted in the evening of the previous day, measures such as evacuation orders can be taken early, and human and material damage can be reduced. Note that, because there is a difference between the 60-minute rainfall forecast value at 18:00 on the 3rd and the actual measured 60-minute rainfall value, there is a difference of about three hours between the time when the most recent local high rainfall event occurs rapidly retraced (7:00 on the 4th) and the time when the landslide actually began to occur (around 4:00 on the 4th). However, both of these times are sufficient to take measures such as evacuation when viewed from the time of prediction execution (18:00 on the 3rd), so there is no particular problem.

[0104] If a simulation similar to the data in Figure 15 is run based on the 60-minute rainfall forecast value as of 9 p.m. on the 3rd, the time period in which the most recent local high rainfall event occurred is rapidly pushed back to 6 a.m. on the 4th (Figure 16), which is closer to the time determined from the actual measurements.

[0105] [Example 2] In Example 2, a simulation was performed using the same method as in Example 1, using the 1st to 51st past rainfall index values ​​calculated based on the 60-minute rainfall observed in Maniwa City, Okayama Prefecture (hereinafter abbreviated as "Maniwa City") during the heavy rainfall that occurred from July 16 to July 21, 2006 (July 2006 heavy rainfall). This heavy rain caused many slope collapses mainly in the northern part of Okayama Prefecture, including Maniwa City. In this area, many wind-fallen tree damages occurred due to Typhoon No. 23, which struck in October 2004, about one year and nine months ago, and a major feature of the disaster caused by the heavy rainfall in July 2006 was that about 74% of the collapsed slopes were slopes that had been damaged by wind-fallen trees. In this embodiment, the 1st to 51st past rainfall index values ​​were calculated from the observed 60-minute rainfall values ​​for all past rainfall events in the area, including Typhoon No. 23 of 2004 and Typhoon No. 10 of 1998, which struck in October 1998 prior to that.

[0106] The upper part of FIG. 17 shows data on 60-minute rainfall observed in Maniwa City at each time every 60 minutes between 00:00 on July 17, 2006 and 12:00 on July 19, 2006 (12:00 p.m.) (hereinafter, "July 2006" will be omitted and described as "17th, xx o'clock", "18th, xx o'clock", or "19th, xx o'clock"). As in the first embodiment, the 60-minute rainfall data is actually acquired for the period after 1976. The 1st to 51st past rainfall index values ​​at each time are obtained from the 60-minute rainfall data for the period after 1976. The lower part of FIG. 17 shows the results of obtaining the most recent local heavy rainfall events using these effective rainfalls and setting each time every 60 minutes as the target time for prediction.

[0107] From the graph at the bottom, the following can be said. First, for the forecast time up to 6:00 on the 17th, the time of occurrence of the most recent local high rainfall event is the time of the local past rainfall event that occurred on July 14, 2006, just before the forecast time. After that, when the forecast time becomes 8:00 on the 17th, the time of occurrence of the most recent local high rainfall event dates back to October 2004 (when Typhoon No. 23 of 2004 struck), and when the forecast time becomes 23:00 on the 18th, the time of occurrence of the most recent local high rainfall event dates back to October 1998 (when Typhoon No. 10 of 1998 struck).

[0108] In Maniwa City, landslides occurred in two time periods during the period from July 17 to 19, 2006, as shown in FIG. 17, from around 13:00 on the 17th to around 0:00 on the 18th, and from around 23:30 on the 18th to around 5:00 on the 19th. In the former of these time periods, the time of the most recent local heavy rainfall event was backdated to October 2004 at 8:00 on the 17th, which is before the time period. Therefore, if it is analyzed in advance that the state is prone to landslides due to fallen trees caused by a typhoon that occurred in October 2004, the method of this embodiment can recognize that heavy rain has occurred since the typhoon at 8:00 on the 17th, and this can be used as an opportunity to take measures such as evacuation before the landslide occurs. In addition, for the latter landslide occurrence time period, the time of the most recent local heavy rainfall event going back can be used as an opportunity to take measures such as evacuation.

[0109] The present invention is not limited to the above-described embodiment and examples, and various modifications are possible.

[0110] For example, in the second embodiment, the operation of determining a near-local heavy rainfall event candidate using a combination of two of N types of rainfall indices is performed by performing all possible combinations, i.e. N We carried out the C2 method, but N It may be possible to execute the process using only a part of the C2 combinations.

[0111] In addition, in each of the above embodiments, two rainfall indices are used when determining one most recent local high rainfall event or most recent local high rainfall event candidate, but only one or three or more rainfall indices may be used to determine one most recent local high rainfall event or most recent local high rainfall event candidate. Specifically, when only one rainfall index is used, an index value exceeding rainfall event in which the past rainfall index value of the rainfall index exceeds the prediction target rainfall index value is extracted, and then, among the index value exceeding rainfall events, a local past rainfall event that occurred at the time closest to the prediction target time may be determined as the most recent local high rainfall event or identified as the most recent local high rainfall event candidate. In addition, when three or more rainfall indices are used, index value exceeding rainfall events in which the past rainfall index values ​​are equal to or greater than the forecast target rainfall index value for all of the first to Pth rainfall indices (P is a natural number equal to or greater than 3) are extracted, and the local past rainfall event that occurred at the time closest to the forecast target time among these index value exceeding rainfall events is determined to be the most recent local high rainfall event or identified as a candidate for the most recent local high rainfall event.

[0112] In the fourth embodiment, the recent local heavy rainfall event notification unit 54 and the landslide disaster occurrence risk notification unit 55 are provided, but either one of them may be omitted. If the recent local heavy rainfall event notification unit 54 is omitted, users such as residents will not be notified of the recent local heavy rainfall event, but the landslide disaster occurrence risk notification unit 55 will notify the user of the risk of landslide disaster occurrence, allowing the user to obtain an indicator for determining the necessity of evacuation, etc. Also, in the fourth embodiment, the alert judgment time input unit 56 and the user terminal side landslide disaster occurrence risk determination unit 57 may be omitted.

[0113] In addition, in the fourth embodiment, multiple prediction target points are prepared and the user of the user terminal 81 selects a specific prediction target point, but when providing pinpoint information to an area with a high risk of landslides, for example, only one prediction target point may be targeted and the function for selecting the prediction target point may be omitted.

[0114] In each of the above embodiments, a predetermined display is performed on the display unit 17 to notify the user of the determined most recent local heavy rainfall event, but in addition to or instead of that, a notification may be made by voice. Also, in the fourth embodiment, information indicating the most recent local heavy rainfall event is notified to the terminals of users such as residents via an Internet line, but the notification targets are not limited to residents, and may be news organizations, local governments (other local governments if the local government has a landslide disaster prediction device), or the country. [Explanation of symbols]

[0115] 10, 30...Landslide disaster prediction device 11…Local past rainfall data acquisition section 12…Prediction target rainfall index value acquisition section 13…Excessive rainfall event extraction section 14...Decision-making section for recent local high rainfall events 16...Input section 17…Display section 18...Storage section 21...Past data points 211…Past data point where index value exceeded 22…Exceeding index value 221…1st rainfall indicator boundary line 222…Second rainfall indicator boundary line 23…Data point for prediction 230…Data points before the current rainfall event 31…Warning judgment time acquisition unit 32…Landslide Risk Assessment Division 41…Landslide danger line 42, 421...Lines showing the occurrence time of the most recent local high rainfall event 50...User terminal program for landslide disaster prediction system (user terminal program) 51... Prediction target location candidate information acquisition unit 52...Prediction target location selection section 53…Recent local high rainfall event recorder 54…Recent local heavy rainfall event notification section 55…Landslide Risk Notification Division 56...Warning judgment time input section 57…User terminal side landslide risk assessment unit 80…Internet 81...User terminal 91…Past data point 911…Border past data point 92...Borderline 93…Prediction data point 931 - Forecast data points plotted outside the boundary

Claims

1. a) a past rainfall data preparation step of preparing local past rainfall data including one or more past rainfall index values, each of which is a value representing one or more rainfall indexes observed in a local past rainfall event that is a past rainfall event at a location to be predicted, and the time at which they were measured; b) preparing one or more forecast target rainfall index values, each of which represents one or more rainfall indexes obtained at a time of forecast target at the location; c) extracting an index value exceeding rainfall event from the local past rainfall data, in which the past rainfall index value is equal to or greater than the forecast target rainfall index value in all of the one or more rainfall indices; d) determining a local past rainfall event that occurred closest to the time of the forecast target time among the index value exceeding rainfall events as the local most recent high rainfall event; and, e) providing an index of the likelihood of a landslide disaster occurring based on the occurrence time of the most recent local heavy rainfall event; A landslide disaster prediction method comprising the steps of:

2. The method for predicting landslide disasters according to claim 1, characterized in that the one or more rainfall indices are multiple rainfall indices.

3. In the past rainfall data preparation step, past rainfall index values ​​obtained for N types of rainfall indexes, which are three or more types, are prepared as the one or more past rainfall index values; In the prediction target rainfall index value preparation step, the prediction target rainfall index value is prepared for each of the N types of rainfall indices; In the index value exceeding rainfall event extraction step, a plurality of combinations of rainfall indices consisting of two or more and less than N of the N types of rainfall indices are prepared, and for each combination, index value exceeding rainfall events in which the past rainfall index value is equal to or greater than the forecast target rainfall index value for all the rainfall indices in the combination are extracted; In the most recent local high rainfall event determination step, for each of the combinations, a local past rainfall event that occurred at the closest time to the prediction target time among the index value exceeding rainfall events is identified as a most recent local high rainfall event candidate, and the local past rainfall event that occurred at the oldest time among the most recent local high rainfall event candidates is determined as the most recent local high rainfall event. The method for predicting landslide disasters according to claim 1 .

4. a) a past rainfall data preparation step of preparing local past rainfall data including one or more past rainfall index values, each of which is a value representing one or more rainfall indexes observed in a local past rainfall event that is a past rainfall event at a location to be predicted, and the time at which they were measured; b) preparing one or more forecast target rainfall index values, each of which represents one or more rainfall indexes obtained at a time of forecast target at the location; c) extracting an index value exceeding rainfall event from the local past rainfall data, in which the past rainfall index value is equal to or greater than the forecast target rainfall index value in all of the one or more rainfall indices; d) determining a local past rainfall event that occurred closest to the time of the forecast target time among the index value exceeding rainfall events as the local most recent high rainfall event; and, e) a warning judgment time preparation step for preparing a warning judgment time, which is the end time of a local past rainfall event that caused a landslide disaster in the past at the said location, the time when a landslide disaster occurred at a time other than the period of the local past rainfall event, or the time when an event that increases the risk of a landslide disaster occurred; f) a landslide disaster occurrence risk determination step of determining that the risk of a landslide disaster occurrence is high when the warning determination time is between the time when the most recent local heavy rainfall event occurred and the prediction target time; A landslide disaster prediction method comprising the steps of:

5. a) a local past rainfall data acquisition unit that acquires local past rainfall data including one or more past rainfall index values, each of which represents one or more rainfall indexes observed in a local past rainfall event that is a past rainfall event at a location to be predicted, and the time at which they were measured; b) a forecast target rainfall index value acquisition unit that acquires one or more forecast target rainfall index values, each of which represents one or more rainfall indexes obtained at a forecast target time at the location; c) an index value exceeding rainfall event extraction unit that extracts an index value exceeding rainfall event in which the past rainfall index value is equal to or greater than the forecast target rainfall index value in all of the one or more rainfall indices from the local past rainfall data; d) a local high rainfall event determining unit that determines a local past rainfall event that occurred closest to the time of the forecast target time among the index value exceeding rainfall events as a local high rainfall event; e) an index providing unit that provides the age of the occurrence time of the most recent local heavy rainfall event as an index of the likelihood of a landslide disaster occurring; A landslide disaster prediction device comprising:

6. The landslide disaster prediction device according to claim 5, characterized in that the one or more rainfall indices are multiple rainfall indices.

7. In the past rainfall data acquisition unit, the past rainfall index values ​​obtained for N types of rainfall indices, which are three or more types, are acquired as the one or more past rainfall index values; The prediction target rainfall index value acquisition unit acquires the prediction target rainfall index value for each of the N types of rainfall indices, In the index value exceeding rainfall event extraction unit, a plurality of combinations of rainfall indices each consisting of two or more and less than N of the N types of rainfall indices are prepared, and for each combination, an index value exceeding rainfall event is extracted in which the past rainfall index value is equal to or greater than the forecast target rainfall index value for all the rainfall indices in the combination; In the most recent local heavy rainfall event determination unit, for each of the combinations, a local past rainfall event that occurred at the closest time to the prediction target time among the index value exceeding rainfall events is identified as a most recent local heavy rainfall event candidate, and the local past rainfall event that occurred at the oldest time among the most recent local heavy rainfall event candidates is determined as the most recent local heavy rainfall event. The landslide disaster prediction device according to claim 5 .

8. a) a local past rainfall data acquisition unit that acquires local past rainfall data including one or more past rainfall index values, each of which represents one or more rainfall index values ​​observed in a local past rainfall event that is a past rainfall event at a location to be predicted, and the time at which they were measured; b) a forecast target rainfall index value acquisition unit that acquires one or more forecast target rainfall index values, each of which represents one or more rainfall indexes obtained at a forecast target time at the location; c) an index value exceeding rainfall event extraction unit that extracts an index value exceeding rainfall event in which the past rainfall index value is equal to or greater than the forecast target rainfall index value in all of the one or more rainfall indices from the local past rainfall data; d) a local high rainfall event determining unit that determines a local past rainfall event that occurred closest to the time of the forecast target time among the index value exceeding rainfall events as a local high rainfall event; e) a warning judgment time acquisition unit that acquires a warning judgment time, which is the end time of a local past rainfall event that caused a landslide disaster in the past at the said location, the time when a landslide disaster occurred at a time other than the period of the local past rainfall event, or the time when an event that increases the risk of a landslide disaster occurred; f) a landslide disaster occurrence risk determination unit that determines that the risk of a landslide disaster occurrence is high when the alert determination time is between the time when the most recent local heavy rainfall event occurred and the prediction target time; and A landslide disaster prediction device comprising:

9. A program for causing a computer, which is a user terminal connected to the landslide disaster prediction device according to any one of claims 5 to 7, to function, the computer being: A most recent local heavy rainfall event acquisition unit that acquires the most recent local heavy rainfall event from the index providing unit via a network; a local heavy rainfall event notification unit that notifies a user of the user terminal of the local heavy rainfall event acquired by the local heavy rainfall event acquisition unit; A program for a user terminal for a landslide disaster prediction system, which causes the user terminal to function as a

10. A program for causing a computer, which is a user terminal connected to the landslide disaster prediction device according to claim 8 via a network, to function, A program for a user terminal for a landslide disaster prediction system, characterized in that it causes the computer to function as a landslide disaster occurrence risk notification unit that notifies specified information when the landslide disaster occurrence risk determination unit determines that the risk of landslide disaster occurrence at the location is increasing.

11. Furthermore, the computer a warning judgment time input unit that allows a user to input a warning judgment time, which is the end time of a local past rainfall event that caused a landslide in the past at the location, the time when a landslide occurred outside the period of the local past rainfall event, or the time when an event that increases the risk of a landslide occurred; a user terminal-side landslide disaster occurrence risk determination unit that acquires a time when the most recent local heavy rainfall event occurred from the most recent local heavy rainfall event determination unit, and determines that a risk of a landslide disaster occurrence is high when the alert judgment time inputted into the alert judgment time input unit is a time between the time when the most recent local heavy rainfall event occurred and the prediction target time.

11. The program for a user terminal for a landslide disaster prediction system according to claim 9 or 10, characterized in that the program causes the user terminal to function as a

12. Furthermore, the computer a prediction target location candidate information acquisition unit that acquires information on a plurality of prediction target location candidates; a prediction target location selection unit that allows a user to select one of the plurality of prediction target location candidates as the location; The program for a user terminal for a landslide disaster prediction system according to any one of claims 9 to 11, characterized in that the program functions as a user terminal for a landslide disaster prediction system according to any one of claims 9 to 11.

Images