Flood risk assessment method and flood risk assessment system

Abstract

To provide a flood damage risk evaluation method and a flood damage risk evaluation system capable of quantitatively evaluating the risk of flood damage using the Laplacian, which is an index of ground irregularity.SOLUTION: A flood damage risk evaluation method that uses a computer to derive a discriminant for determining the presence or absence of flood damage, comprising: calculating, from the Laplacian L, which is an index of ground irregularity in a sample area including areas where flood damage has occurred in the past, a converted value L' of the Laplacian using (Equation 3); and deriving the above-mentioned discriminant by performing an analysis using the presence or absence of flood damage based on past flood damage data in the sample area as an objective variable and a predetermined index of the sample area including the converted value L' of the Laplacian as an explanatory variable.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a flood risk assessment method and a flood risk assessment system.

Background Art

[0002] In recent years, due to concentrated heavy rainfall in urban areas, many human and material damages have occurred, and the need for flood risk assessment has been increasing. In response, local governments and the like have created and published hazard maps for floods independently. On the other hand, companies with bases (factories, branch offices, etc.) scattered throughout the country may compare risks when implementing flood countermeasures to determine priorities. However, since conditions such as damage assumptions vary from local government to local government, it has been necessary to individually investigate the risks, which has been time-consuming. Therefore, methods for evaluating risks with the same indicators in different regions have been proposed. For example, in Patent Document 1, the unevenness of the ground is evaluated using the Laplacian, which is an index of the unevenness of the ground that can be calculated using nationwide unified elevation data. Specifically, in order to reflect the influence of local depressions or depressions extending over a wide area of the ground, a certain area is divided into multiple types of mesh sizes, and the Laplacian is calculated for each divided mesh. Then, the part that is repeatedly shown as a depression in the Laplacians of multiple types of meshes is evaluated as having a lower elevation than the surroundings and a high risk of flood occurrence.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, while evaluation results using the Laplacian to assess the degree of unevenness of the ground, as in Patent Document 1, can serve as an indicator of the risk of flooding, it is difficult to quantitatively evaluate the risk of flooding based on these evaluation results. To quantitatively assess the risk of flooding, it is advisable to derive a discriminant equation (for example, a relational equation for calculating the probability of flooding) that determines whether or not a flood will occur. However, the variability of the Laplacian in the selected sample area may be large when deriving the discriminant equation. In such cases, even though the Laplacian is an indicator that has a significant impact on flood occurrence, it becomes difficult to incorporate the Laplacian into the discriminant equation.

[0005] This invention has been made in view of the above problems, and its objective is to provide a flood risk assessment method and a flood risk assessment system that can quantitatively evaluate the degree of flood risk using the Laplacian, which is an index of ground unevenness. [Means for solving the problem]

[0006] The main invention for achieving the above objective is a flood risk assessment method that uses a computer to derive a discriminant formula for determining whether or not a flood will occur, and calculates a transformed value L' of the Laplacian from the Laplacian L, which is an index of the unevenness of the ground in a sample area including areas where floods have occurred in the past, using the following formula. TIFF0007877991000001.tif26170 This flood risk assessment method is characterized by deriving the discriminant formula by performing an analysis using the presence or absence of flood occurrence based on past flood data of the sample area as the dependent variable, and a predetermined index of the sample area including the transformed value L' of the Laplacian as the independent variable. Other features of the present invention will be made clearer by description in this specification and the accompanying drawings. [Effects of the Invention]

[0007] According to the present invention, a flood risk assessment method and a flood risk assessment system can be provided that can quantitatively evaluate the degree of flood risk using the Laplacian, which is an index of ground irregularities. [Brief explanation of the drawing]

[0008] [Figure 1] This is a block diagram of the flood risk assessment system. [Figure 2] This is a flowchart illustrating the flood risk assessment method of the first embodiment. [Figure 3] This figure shows sample divisions S1a to S12y and intermediate sample divisions S1 to S12, in which the sample region SA is divided by a predetermined mesh size. [Figure 4] This figure shows a database used to derive relational equations. [Figure 5] This diagram explains how to calculate the Laplacian L. [Figure 6] This is an explanatory diagram of the regression coefficient β and its significance probability P-value according to the evaluation method of the first embodiment. [Figure 7] Figure 7A shows an example of an area RA to be evaluated, and Figure 7B shows a database for calculating the probability p of flood occurrence in the area RA to be evaluated. [Figure 8] This diagram explains the regression coefficient β and its significance probability P-value according to the evaluation method of the comparative example. [Figure 9] Figure 9A is a graph showing the breakdown of the Laplacian L for the sample region SA, and Figure 9B is a graph showing the breakdown of the transformed value L' of the Laplacian for the sample region SA. [Figure 10] This diagram illustrates how the Laplacian L can be calculated using two different mesh sizes. [Figure 11] This is a flowchart illustrating the flood risk assessment method of the second embodiment. [Figure 12] This is an explanatory diagram of the regression coefficient β and its significance probability P-value according to the evaluation method of the second embodiment. [Modes for carrying out the invention]

[0009] This specification and the accompanying drawings make it clear at least the following: (Aspect 1) A flood hazard assessment method for deriving a discriminant for determining the presence or absence of floods using a computer, comprising calculating a transformed value L' of the Laplacian from the Laplacian L, which is an index of the unevenness of the ground in a sample area including locations where floods have occurred in the past, using the following formula: TIFF0007877991000002.tif26170Performing an analysis with the presence or absence of flood occurrence based on the past flood data of the sample area as the target variable and a predetermined index of the sample area including the transformed value L' of the Laplacian as the explanatory variable to derive the discriminant. The flood hazard assessment method is characterized by this.

[0010] According to Aspect 1, even when the variation of the Laplacian L is large, the variation can be suppressed with the transformed value L' of the Laplacian, and the coefficient of the transformed value L' of the Laplacian in the discriminant can be made a significant coefficient. Also, the transformed value L' of the Laplacian approximately reproduces the unevenness of the ground represented by the Laplacian L. Therefore, a discriminant can be derived that takes into account the Laplacian L, which greatly affects flood occurrence, and the flood hazard can be evaluated more accurately and quantitatively. Also, by using a computer, the discriminant can be derived easily and in a short time.

[0011] (Aspect 2) Calculating the transformed value L' of the Laplacian for each of the N types of divisions obtained by dividing the sample area into N types of mesh sizes, and using the transformed value L' of the Laplacian as the explanatory variable for all N types of these mesh sizes. The flood hazard assessment method according to Aspect 1 is characterized by this.

[0012] According to Aspect 2, by using the transformed values L' of the Laplacian of a plurality of mesh sizes as explanatory variables, a discriminant can be derived that takes into account local unevenness and unevenness over a wide range. Also, by using the transformed value L' of the Laplacian as the explanatory variable for all N types of mesh sizes, the process of confirming the significance probability of the coefficient when the Laplacian L of each mesh size is used as the explanatory variable becomes unnecessary. Therefore, the discriminant can be derived easily and in a short time.

[0013] (Aspect 3) A flood hazard degree evaluation method for deriving a discriminant formula for determining the presence or absence of flood disasters using a computer. The presence or absence of flood disasters based on past flood disaster data in a sample area including locations where flood disasters occurred in the past is used as the objective variable, and a predetermined index of the sample area including the Laplacian L, which is an index of the unevenness of the ground, is used as the explanatory variable. By performing analysis, a first step of obtaining the coefficient of the Laplacian L and its significance probability is carried out. When the significance probability is less than a predetermined threshold value, the Laplacian L is used as the explanatory variable. When the significance probability is greater than or equal to the predetermined threshold value, a converted value L' of the Laplacian calculated using the following formula is used as the explanatory variable. TIFF0007877991000003.tif26170 A second step of deriving the discriminant formula by performing analysis using the presence or absence of flood disasters in the sample area as the objective variable and the predetermined index of the sample area including an index related to the Laplacian L as the explanatory variable. The flood hazard degree evaluation method is characterized by having this.

[0014] According to Aspect 3, even when the variation of the Laplacian L is large, the variation can be suppressed with the converted value L' of the Laplacian, and the coefficient of the converted value L' of the Laplacian in the discriminant formula can be made a significant coefficient. Therefore, a discriminant formula that takes into account the Laplacian that greatly affects the occurrence of flood disasters can be derived. On the other hand, when the variation of the Laplacian L is small, since the Laplacian L itself that directly represents the unevenness of the ground is used as the explanatory variable, a more accurate discriminant formula can be derived. Therefore, the flood hazard degree can be evaluated more accurately and quantitatively. Also, by using a computer, the discriminant formula can be derived easily and in a short time.

[0015] (Aspect 4) The flood risk assessment method according to Embodiment 3, characterized in that, in the first step, the coefficient of the Laplacian L and its significance probability are determined for each of N mesh sizes that divide the sample area, and in the second step, the Laplacian L is used as the explanatory variable for mesh sizes in which the significance probability is less than the predetermined threshold, and the transformed value L' of the Laplacian is used as the explanatory variable for mesh sizes in which the significance probability is equal to or greater than the predetermined threshold.

[0016] According to Embodiment 4, by using indicators related to the Laplacian of multiple mesh sizes as explanatory variables, it is possible to derive a discriminant formula that takes into account both local and widespread irregularities. Furthermore, an explanatory variable related to the Laplacian (the transformed value of the Laplacian L' or the Laplacian L) is determined for each mesh size. Therefore, the Laplacian L, which directly represents the irregularities of the ground, can be used as an explanatory variable as much as possible, and a more accurate discriminant formula can be derived.

[0017] (Appendix 5) A flood risk assessment method according to any one of embodiments 1 to 4, characterized in that it determines whether or not flood damage occurs in the area to be evaluated based on the predetermined index in the area to be evaluated and the discriminant formula.

[0018] According to embodiment 5, the risk of flooding can be evaluated more accurately and quantitatively by a discriminant formula that takes into account the Laplacian L, which greatly influences the occurrence of flooding.

[0019] (Aspect 6) The flood risk assessment method according to any one of embodiments 1 to 5, characterized in that the predetermined indicators are the Laplacian L index, elevation, and mean S-wave velocity.

[0020] According to embodiment 6, the risk of flooding due to differences in elevation is taken into account, allowing for the deriving of a more accurate discriminant. Furthermore, the average S-wave velocity, which indirectly represents the topography, is also taken into account, allowing for the deriving of a more accurate discriminant. In addition, since the number of predetermined indicators (explanatory variables) is relatively small, the discriminant can be derived quickly and easily.

[0021] (Aspect 7) A storage unit stores data about a sample area that includes locations where floods have occurred in the past. Using the Laplacian L, which is an index of the unevenness of the ground in the sample area, a converted value L' of the Laplacian is calculated using the following formula. A flood risk assessment system characterized by having a calculation processing unit that derives a discriminant expression for determining whether or not a flood has occurred by performing an analysis with the presence or absence of a flood based on past flood data of the sample area as the dependent variable, and a predetermined index of the sample area including the transformed value L' of the Laplacian as the independent variable.

[0022] According to embodiment 7, even when the variation in the Laplacian L is large, the transformed value L' of the Laplacian can suppress the variation, and the coefficient of the transformed value L' of the Laplacian in the discriminant equation can be made a significant coefficient. Furthermore, the transformed value L' of the Laplacian roughly reproduces the unevenness of the ground represented by the Laplacian L. Therefore, a discriminant equation that takes into account the Laplacian L, which greatly influences the occurrence of floods, can be derived, and the risk of floods can be evaluated more accurately and quantitatively. In addition, the discriminant equation can be derived quickly and easily by the calculation processing unit.

[0023] The following describes an embodiment of the flood risk assessment method according to the present invention, with reference to the drawings. ===First Embodiment=== Figure 1 is a block diagram of the flood risk assessment system. Figure 2 is a flowchart illustrating the flood risk assessment method of the first embodiment. Figure 3 is a diagram showing sample divisions S1a to S12y and intermediate sample divisions S1 to S12, in which the sample area SA is divided by a predetermined mesh size. Figure 4 is a diagram showing a database for deriving relational equations. Figure 5 is a diagram illustrating the calculation method of the Laplacian L. Figure 6 is an explanatory diagram of the regression coefficient β and its significance probability P value according to the evaluation method of the first embodiment. Figure 7A is a diagram showing an example of the area to be evaluated RA, and Figure 7B is a diagram showing a database for calculating the probability of flood occurrence p in the area to be evaluated RA.

[0024] In the flood risk assessment method of the first embodiment, a computer 5 is used to derive a relational expression (a discriminant expression for determining whether or not a flood occurs) for calculating the probability p of a flood occurring in a certain area (area to be evaluated). Specifically, the probability of flooding p is the probability of inundation occurring. Furthermore, using the probability of flooding p as the dependent variable and several indicators assumed to be factors in flooding as independent variables, a relationship is derived using logistic regression analysis. The relationship derived from logistic regression analysis is (Equation 1) below.

[0025] TIFF0007877991000005.tif20170

[0026] Then, the probability of flooding in the evaluation area RA is calculated based on the derived relation (the equation obtained by substituting the regression coefficient β into Equation 1) and the explanatory variables of the evaluation area RA. Specifically, similar to the sample area SA (Figure 3) used when deriving the relation, the evaluation area RA (Figure 7A) is divided into meshes of a predetermined size, and the probability of flooding p is calculated for each of the divided evaluation areas RA1, RA2, etc. In this embodiment, the evaluation area RA is divided into 50m meshes, and the probability of flooding p is calculated for each 50m x 50m evaluation area RA1, RA2, etc.

[0027] Furthermore, flooding is not limited to water damage. For example, one could use an area containing locations where landslides have occurred in the past as the sample area, and derive a relational equation to calculate the probability p of water damage (landslide) occurring, using the probability of past landslides as the dependent variable.

[0028] Computer 5 corresponds to a flood risk assessment system and can be exemplified by a system equipped with a memory unit 1, an arithmetic processing unit 2 (e.g., CPU), an input unit 3 (e.g., mouse and keyboard), a display unit 4 (e.g., display), etc., as shown in Figure 1. Data for deriving relational equations can be input to and stored in the memory unit 1 via the input unit 3. The arithmetic processing unit 2 can perform the following processes, such as deriving relational equations, according to the program stored in the memory unit 1. Furthermore, computer 5 (flood risk assessment system) may be a dedicated device for performing flood risk assessment, or it may be a general-purpose computer device.

[0029] First, the evaluator or computer 5, which assesses the risk of flooding, determines the sample area SA to be an existing area that includes a flood-affected location Q where flooding has occurred in the past, as shown in the flow chart of Figure 2 (S01). Next, the evaluator or computer 5 collects past flood data and data related to explanatory variables for the sample area SA (S02).

[0030] For past flood data, data is acquired for each "sample section" created by dividing the sample area SA into a predetermined mesh size (in this case, a 50m mesh). Explanatory variables are obtained for each 50m sample segment, or for each "intermediate sample segment" created by dividing the sample area SA with a mesh larger than 50m (in this case, a 250m mesh).

[0031] The following explanation will primarily use, as shown in Figure 3, 12 sample intermediate divisions S1 to S12, obtained by dividing the sample area SA with 250m meshes, and 25 samples each (S1a to S1y, S2a to S2y…S12a to S12y), obtained by dividing each sample intermediate division S1 to S12 with 50m meshes. In Figure 3, the flooding locations Q are shown with shading. Also, in Figure 3, only a portion of the sample area SA is divided, and the display of some meshes is omitted, but the entire sample area SA is divided with meshes (50m mesh and 250m mesh).

[0032] Furthermore, for determining the sample area SA and collecting past flood data, for example, land history survey data published nationwide by the Ministry of Land, Infrastructure, Transport and Tourism (Land Classification Basic Survey (Land History Survey) by the Land Information Division, National Land Policy Bureau, Ministry of Land, Infrastructure, Transport and Tourism https: / / nlftp.mlit.go.jp / kokjo / inspect / landclassification / land / land_history_2011 / pdf_landform_03.html), and the Tokyo Metropolitan Government Bureau of Construction's "Past Flood Records ~Inundation Record Map~" https: / / www.kensetsu.metro.tokyo.lg.jp / jigyo / river / suishin / suigai_kiroku / kako.html can be used.

[0033] The evaluator or computer 5 calculates the probability of flood occurrence p (whether or not a flood occurs) for each sample division S1a to S12y (each 50m mesh) based on past flood data as exemplified above. Then, as shown in Figure 4, the probability of flood occurrence p is stored in the storage unit 1, associated with each sample division S1a to S12y.

[0034] For example, as shown in Figure 3, sample section S1a does not include any areas that have been flooded in the past, so the flooding probability p for sample section S1a is set to "0" and stored in the storage unit 1 in association with the location information of sample section S1a. Also, sample section S1d has a wider area that has been flooded in the past. For this reason, the flooding probability p for sample section S1d is set to "1" and stored in the storage unit 1 in association with the location information of sample section S1d.

[0035] The probability of flooding occurring p in each sample category S1a to S12y can be calculated by dividing the sample category into "flooded" (1) and "not flooded" (0) based on the proportion of the area occupied by areas that have been flooded in the past relative to the total area of ​​that sample category. The threshold can be set appropriately based on the proportion of the area occupied by flooded areas, such as 30%, 50%, or 80%. In this case, setting a smaller threshold will result in a relationship that calculates a more conservative probability of flooding. Here, we define "flooded" (1) as occurring when the proportion of the area occupied by flooded areas is 50% or more, and "not flooded" (0) as occurring when it is less than 50%.

[0036] The explanatory variables are the various indicators that are assumed to be factors in flooding. In this embodiment, the following four indicators are designated as explanatory variables 1 to 4.

[0037] Explanatory variable 1 is the elevation data for each sample segment S1a to S12y, which is divided by a 50m-sized mesh. Elevation data for each sample category S1a to S12y can be obtained from sources such as the Geospatial Information Authority of Japan's Digital Elevation Model (https: / / fgd.gsi.go.jp / download / ref_dem.html) and the National Land Numerical Information Elevation / Slope 5th-order Mesh Data (https: / / nlftp.mlit.go.jp / ksj / gml / datalist / KsjTmplt-G04-d.html).

[0038] Explanatory variable 2 is the converted Laplacian value L' for each sample division S1a to S12y, which is divided by a 50m mesh. Explanatory variable 3 is the converted Laplacian value L' for each intermediate sample division S1 to S12, which is divided by a 250m mesh. As a preliminary step before calculating the converted Laplacian value L', computer 5 uses the calculation processing unit 2 to calculate the Laplacian L for each sample division S1a to S12y (50m mesh) and the Laplacian L for each intermediate sample division S1 to S12 (250m mesh). The Laplacian L is an index of ground unevenness (a value that quantifies and represents the variation in unevenness).

[0039] Specifically, the Laplacian L for a 50m mesh can be calculated using the following (Equation 2) based on elevation data for each sample division S1a to S12y (for example, data for explanatory variable 1). Similarly, the Laplacian L for a 250m mesh can also be calculated based on elevation data for each intermediate sample division S1 to S12. For elevation data, for example, the elevation data published nationwide by the Ministry of Land, Infrastructure, Transport and Tourism, as mentioned above, can be used.

[0040] TIFF0007877991000006.tif18170dx: Mesh spacing in the horizontal direction (east-west direction) dy: Mesh spacing in the vertical direction (north-south direction)

[0041] As shown in Figure 5, in (Equation 2), the elevation data of the sample section or intermediate sample section St for which the Laplacian L(m,n) is calculated is denoted as u(m,n). Then, the elevation data of the four sample sections or intermediate sample sections St1 to St4 adjacent to the target sample section or intermediate sample section St in the horizontal (east-west) and vertical (north-south) directions are denoted as u(m-1,n), u(m+1,n), u(m,n-1), and u(m,n+1).

[0042] If the Laplacian L(m,n) calculated by equation (2) is positive (L>0), it indicates a concave topography; if it is zero (L=0), it indicates a flat topography; and if it is negative (L<0), it indicates a convex topography. Furthermore, the larger the absolute value of the Laplacian L(m,n), the greater the unevenness compared to the surrounding ground. Therefore, the degree of unevenness in each sample division S1a to S12y or intermediate sample division S1 to S12 can be represented based on the Laplacian L(m,n) of that division.

[0043] Next, the arithmetic processing unit 2 of the computer 5 calculates the converted value L' of the Laplacian from the Laplacian L for each sample division S1a to S12y and the Laplacian L for each intermediate sample division S1 to S12 using the following equation (Equation 3) (S03 in Figure 2). For example, the converted value L' of the Laplacian can be calculated using a spreadsheet application stored in the memory unit 1. The transformed value L' of the Laplacian is the transformed value based on the common logarithm of the Laplacian L. Specifically, if the Laplacian L is a positive value (L>0), the transformed value L' is the common logarithm of the Laplacian L. If the Laplacian L is zero (L=0), the transformed value L' remains zero. If the Laplacian L is a negative value (L<0), the transformed value L' is the value obtained by taking the common logarithm of the absolute value of the Laplacian L and adding a minus sign to it.

[0044] TIFF0007877991000007.tif26170...(Formula 3)

[0045] Then, as shown in Figure 4, the transformed Laplacian value L' for each sample division S1a to S12y (each 50m mesh) is used as explanatory variable 2, and the transformed Laplacian value L' for each intermediate sample division S1 to S12 is used as explanatory variable 3.

[0046] Explanatory variable 4 is data on the ground conditions for each sample intermediate division S1 to S12 (every 250m mesh), specifically the average S-wave velocity data down to 30m underground for each sample intermediate division S1 to S12. For example, average S-wave velocity data down to 30m underground estimated from the microtopography of the National Research Institute for Earth Science and Disaster Resilience (NIED) J-SHIS surface ground (https: / / www.j-shis.bosai.go.jp / map / ) can be used.

[0047] The evaluator or computer 5 stores the data for explanatory variables 1 to 4 for each sample division S1a to S12y in the memory unit 1, associating them with each sample division S1a to S12y. In this way, a database is created to find the relational formula for calculating the probability of flood occurrence p (S04). Note that for explanatory variables 1 to 4 that can be obtained for each 250m mesh (in this case, explanatory variables 3 and 4), the same data is associated with the 25 sample divisions S1a to S12y belonging to the same intermediate sample division S1 to S12.

[0048] Next, the arithmetic processing unit 2 of the computer 5 performs logistic regression analysis based on the database (Figure 4) stored in the memory unit 1, using the probability of flood occurrence p for each sample category S1a to S1y as the dependent variable and explanatory variables 1 to 4, to derive a relational equation for calculating the probability of flood occurrence p (S05). For example, the relational equation obtained by logistic regression analysis can be derived using a spreadsheet application stored in the memory unit 1.

[0049] As shown in Figure 6, logistic regression analysis yields the intercept β0 of the relation and the regression coefficients β1 to β4 corresponding to each of the explanatory variables 1 to 4. By fitting these regression coefficients β0 to β4 into (Equation 1), a relation is obtained for calculating the probability of flood occurrence p in the evaluation area RA. Note that the regression coefficients β1 to β4 indicate the degree to which the dependent variable (probability of flood occurrence p) changes when the corresponding explanatory variables 1 to 4 change.

[0050] Then, the arithmetic processing unit 2 of computer 5 can calculate the probability p of flood occurrence in the area RA based on the derived relation (the equation obtained by substituting the value of β into equation 1) and explanatory variables 1 to 4 of the area RA to be evaluated (elevation data for 50m mesh, the transformed value L' of the Laplacian for 50m mesh and 250m mesh, and the average S-wave velocity data for 250m mesh) (S06).

[0051] Specifically, as shown in Figure 7A, data for explanatory variables 1 to 4 are collected for each evaluation target division RA1, RA2, etc., which is divided into meshes of a predetermined size (the same 50m mesh as the sample area SA). A database (Figure 7B) is created by associating this data with each evaluation target division RA1, RA2, etc., and stored in the storage unit 1. Then, the calculation processing unit 2 substitutes explanatory variables 1 to 4 of the evaluation target area RA into a relational expression, and calculates the probability of flood occurrence p for each evaluation target division RA1, RA2, etc.

[0052] Figure 8 is an explanatory diagram of the regression coefficient β and its significance probability P-value according to the evaluation method of the comparative example. Figure 9A is a graph showing the breakdown of the Laplacian L of the sample region SA, and Figure 9B is a graph showing the breakdown of the transformed value L' of the Laplacian of the sample region SA. In the first embodiment, the transformed value L' of the Laplacian is used as the explanatory variable when deriving the relation. Here, as a comparative example, we will explain the case in which the value of the Laplacian L itself is used as the explanatory variable when deriving the relation. Except for explanatory variables 2 and 3 related to the Laplacian L, the method is the same as in the first embodiment, and logistic regression analysis was performed based on the same sample region SA data. As a result, the obtained regression coefficients β1 to β4 and their significance probability P values ​​are shown in Figure 8. In the results in Figure 8, the significance probability P value of explanatory variable 2 (Laplacian L of the 50m mesh) was a high 0.401.

[0053] The significance probability p-value is the probability that the calculated value of β would occur due to data variability, assuming that the regression coefficient β is zero (irrelevant). In other words, a small significance probability p-value, almost zero, indicates that the probability of obtaining the calculated value of β (for example, -0.031 for explanatory variable 1, β1) is almost zero when β is assumed to be zero. Therefore, the calculated value of β (for example, β1 = -0.031) is significant and can be said to be a reliable (explanatory) result. Conversely, when the significance probability p-value is large, such as 0.401 (=40.1%) for the regression coefficient β2 of explanatory variable 2 in the comparative example (for example, more than 5%), it indicates that there is a 40.1% probability that β2 would be -13.205 even when β2 is assumed to be zero, which is unreliable. Therefore, the data for explanatory variable 2 in the comparative example has large variability, and the regression coefficient β2 obtained from such data is not significant and is an unreliable coefficient.

[0054] In fact, the data for explanatory variable 2 in sample area SA, which was used to derive the relational equation (results in Figure 8), specifically the breakdown of the Laplacian L for each sample division S1a to S12y (every 50m mesh) belonging to sample area SA, is shown in the bar graph and cumulative line graph in Figure 9A. The horizontal axis shows the value of Laplacian L, and the vertical axis shows the number of occurrences of the corresponding Laplacian L. As shown in Figure 9A, although more than half of the Laplacian L values ​​were 0.0005 or less, there were Laplacian L values ​​up to 0.08. This also indicates that the variability of Laplacian L for each 50m mesh used in sample area SA was large, resulting in a large significance probability P-value for the regression coefficient β2 corresponding to explanatory variable 2 (resulting in a regression coefficient β2 that lacks reliability).

[0055] Thus, depending on the sample area SA selected to calculate the relationship, the variability of the Laplacian L for each mesh (in this case, every 50m mesh) may be large. As a result, the significance probability P-value of the regression coefficient β for the Laplacian L becomes large. However, if a regression coefficient β with a large significance probability P-value is substituted into the relationship, the reliability (accuracy) of the flood occurrence probability p-value calculated from the relationship decreases. On the other hand, the Laplacian (the degree of unevenness of the ground) is an indicator that has a large influence on the occurrence of floods (inundation). Therefore, if the regression coefficient β for the Laplacian L is excluded because of its large significance probability P-value, the Laplacian indicator is not taken into account, and the reliability of the flood occurrence probability p-value calculated from the relationship decreases.

[0056] Therefore, in the first embodiment, the following flood risk assessment method is performed using the computer 5. First, the Laplacian L of the sample region SA, which includes areas where floods have occurred in the past, is used to calculate the Laplacian transformed value L' with respect to the common logarithm of the Laplacian L using the aforementioned (Equation 3). Then, using the probability of flood occurrence p (whether or not a flood occurs) based on past flood data of the sample region SA as the dependent variable, and a predetermined index of the sample region SA, including the transformed value L' of the Laplacian, as the independent variable, an analysis (in this case, logistic regression analysis) is performed to derive a relational equation for calculating the probability of flood occurrence p (the equation obtained by substituting the regression coefficient β into Equation 1, which is a discriminant equation for determining whether or not a flood occurs). The predetermined indicators are those assumed to be factors contributing to flood damage. In this embodiment, the predetermined indicators are the elevation data for 50m mesh, the converted value L' of the Laplacian for 50m mesh and 250m mesh, and the average S-wave velocity data for 250m mesh.

[0057] According to the flood risk assessment method described above, even when the Laplacian L has a large variation, as shown in Figure 9A, the transformed value L' of the Laplacian L is expressed as the common logarithm (exponential) of the Laplacian L, thus reducing the variation of the transformed value L'. In the graph in Figure 9B, the horizontal axis shows the value of the transformed value L' of the Laplacian, and the range from the minimum value to the maximum value (0 to 0.7) of the transformed value L' is smaller than that of the Laplacian L in Figure 9A. This also shows that the variation of the transformed value L' of the Laplacian is reduced.

[0058] Furthermore, in the calculation of the Laplacian's transformed value L' (Equation 3), if the Laplacian L is a positive value, the transformed value L' will also be a positive value, indicating that both represent a concave topography. If the Laplacian L is zero, the transformed value L' will also be zero, indicating that both represent a flat surface. If the Laplacian L is a negative value, the transformed value L' will also be a negative value, indicating that both represent a convex topography. Also, if the absolute value of the Laplacian L is large, the absolute value of the transformed value L' will also be large, thus roughly reproducing the relationship between the degree of concavity and convexity. Therefore, similar to the Laplacian L, the transformed value L' of the Laplacian can also be used to evaluate the degree of convexity for each sample division S1a to S12y or intermediate sample division S1 to S12.

[0059] Therefore, even when the Laplacian L of the selected sample region SA is highly variable, logistic regression analysis can be performed using the transformed value L' of the Laplacian as the explanatory variable, and the regression coefficients β2 and β3 related to the Laplacian L can be made significant coefficients (statistically meaningful coefficients). Thus, it is possible to derive a relationship that takes into account the Laplacian L, which has a significant impact on flood occurrence.

[0060] Then, based on the predetermined indicators (explanatory variables 1-4) in the evaluation area RA and the derived relational equation (the equation obtained by substituting the value of β into Equation 1), the probability of flood occurrence p in the evaluation area RA is calculated. In this way, the flood risk of the evaluation area RA can be quantitatively evaluated.

[0061] Furthermore, since the relationship is derived based on past cases (flood locations Q and explanatory variables in an actual sample area SA), a more accurate probability of flood occurrence p can be calculated based on the relationship. In addition, for explanatory variables 1 to 4, which are assumed to be factors in flooding, nationwide standardized data published by the Geospatial Information Authority of Japan, the National Research Institute for Earth Science and Disaster Resilience (J-SHIS), etc., can be used. Therefore, a relationship that is less prone to regional variability and can be evaluated using the same indicators nationwide, and the probability of flood occurrence p based on that relationship, can be obtained.

[0062] Furthermore, in this embodiment, since the relational expression (discriminant) is derived by logistic regression analysis, the range of the probability of flood occurrence p does not become -∞ to ∞ as in the result of simple regression analysis, and the probability of flood occurrence can be determined in the range of 0 to 1. Therefore, it is possible to quantitatively evaluate the probability of flood occurrence. Also, when "1" is set to indicate that a flood has occurred and "0" is set to indicate that a flood has not occurred, a value (threshold) within the range of 0 to 1 is set as appropriate. When the calculated probability p is greater than or equal to that value, it can be predicted that there is a risk of flood occurrence, and when the calculated probability p is less than that value, it can be predicted that there is little risk of flood occurrence, and thus it is possible to determine whether or not a flood will occur.

[0063] Furthermore, the relational equation is derived using a computer 5. Specifically, a flood risk assessment system is used that includes a storage unit 1 for storing data related to the sample region SA, and a calculation processing unit 2 for calculating the transformed value L' of the Laplacian and deriving the relational equation by logistic regression analysis using a predetermined index of the sample region SA containing the transformed value L' of the Laplacian as an explanatory variable. In this way, the relational equation can be derived easily and quickly.

[0064] Furthermore, in the logistic regression analysis results shown in Figure 8, the significance probability P-values ​​for all regression coefficients β are close to zero. However, this is not the only case; it is also possible that the explanatory variables other than the transformed value L' of the Laplacian L (in this case, elevation and mean S-wave velocity) may have large variability, resulting in a large significance probability P-value for the regression coefficients β. In that case, explanatory variables with a large significance probability P-value (e.g., 5% or more) may be excluded, and the relational equation may be derived. Alternatively, for explanatory variables with a large significance probability P-value, the common logarithm of that explanatory variable (see Equation 3) may be used as the transformed value, similar to the transformed value L' of the Laplacian, and logistic regression analysis may be performed again to derive the relational equation (regression coefficient β).

[0065] Figure 10 illustrates the Laplacian L obtained using two different mesh sizes. For example, as shown in Figure 10, when the Laplacian L of region X1, which is divided into a relatively large 250m mesh, is calculated, the Laplacian L of region X1 becomes positive (L>0), and region X1 is represented as a depression. Since region X1 has depressions over a wide area, this evaluation is generally correct. However, the elevation difference between local depressions A and flat areas B within region X1 is not reflected in the evaluation. Because the Laplacian L is calculated based on the average elevation data of a region divided into a mesh of a predetermined size, elevation differences (undulations) within the same mesh are not detected.

[0066] On the other hand, when calculating the Laplacian L for regions X11 to X15, which are created by dividing region X1 with a relatively small 50m mesh, the Laplacian L for region X12 is positive (L>0), representing a local depression A. However, the Laplacian L for region X13 is negative (L<0), representing a convex area, and the Laplacian L for region X14 is zero, representing a flat area. Therefore, the fact that regions X13 and X14 are part of a depression within region X1 is not reflected. Since the Laplacian L is calculated based on the elevation difference between adjacent meshes, if the mesh size is small, elevation differences (undulations) over a wide area larger than the mesh cannot be detected.

[0067] Therefore, in the first embodiment, the Laplacian conversion value L' is calculated for each of the N divisions obtained by dividing the sample region SA into N different mesh sizes. Here, the sample region SA is divided into two mesh sizes: 50m and 250m. The Laplacian conversion value L' for each sample division S1a to S12y, divided by the 50m mesh, is used as explanatory variable 2. The Laplacian conversion value L' for each intermediate sample division S1 to S12, divided by the 250m mesh, is used as explanatory variable 3.

[0068] By doing so, localized irregularities can be taken into account using explanatory variable 2 with a small mesh (50m mesh), and widespread irregularities can be taken into account using explanatory variable 3 with a large mesh (250m mesh). Therefore, a more accurate relational equation based on the irregularities of the ground can be derived, and based on this relational equation, the probability of flood occurrence p (flood risk) can be evaluated with greater accuracy.

[0069] Furthermore, in the first embodiment, the Laplacian transformation value L' is used as the explanatory variable for all N mesh sizes. For example, the significance probability P-value of the regression coefficient β3 corresponding to explanatory variable 3 (250m mesh) in the comparative example in Figure 8 is almost zero. Therefore, the regression coefficient β3 corresponding to the Laplacian L of the 250m mesh is a significant coefficient, and the probability of flood occurrence p can be calculated accurately by substituting it into the relational equation. However, in the first embodiment, regardless of the significance probability P-value of the regression coefficient β with the Laplacian L as the explanatory variable, the transformed value L' of the Laplacian is used as the explanatory variable for both the Laplacian L of the 50m mesh and the Laplacian L of the 250m mesh.

[0070] By doing so, it becomes unnecessary to perform logistic regression analysis with the Laplacian L as the explanatory variable and check whether the regression coefficient β corresponding to each mesh size is significant, as in the comparative example. Therefore, the relationship can be derived more easily and in less time. Furthermore, as mentioned above, even with the transformed value L' of the Laplacian, the unevenness of the ground is represented in the same way as the Laplacian L, so it is possible to derive a relationship that takes into account the Laplacian (unevenness of the ground), which has a significant impact on the occurrence of floods.

[0071] While 50m and 250m meshes were used as examples, the sample area SA may be divided using meshes of other sizes. However, it is desirable that the length of each side of the mesh be 5m or more. This allows for the use of highly consistent data published by the Ministry of Land, Infrastructure, Transport and Tourism, etc. Furthermore, compared to cases where the mesh is too fine, the data is less likely to become complicated, and the variation in the Laplacian L is also suppressed, making it easier to accurately represent the unevenness of the ground. Furthermore, it is preferable that the mesh size of one mesh be at least twice the mesh size of the other. This makes it easier to represent various irregularities of different scales. Furthermore, the mesh size for elevation (50m mesh in this case) and the mesh size for average S-wave velocity (250m mesh in this case) are not particularly limited.

[0072] Alternatively, the sample region SA may be divided into three or more mesh sizes, and the transformed Laplacian value L' of each mesh size may be used as an explanatory variable. This allows for the deriving of a relational equation that takes into account various irregularities of different scales. However, if the number of explanatory variables increases too much, the calculation of the relational equation and the probability of flood occurrence p becomes complicated.

[0073] Alternatively, the mesh size used to divide the sample region SA may be limited to one, and the explanatory variable for the transformed Laplacian L' may be limited to one. In this case as well, the transformed Laplacian L' for that single mesh size is used as the explanatory variable, regardless of the variability of the Laplacian L for that single mesh size (the significance probability of the explanatory variable).

[0074] Furthermore, it is desirable that the predetermined indicators used as explanatory variables (indicators assumed to be factors in flood damage) be an indicator related to the Laplacian L (the transformed value of the Laplacian L' in the first embodiment), elevation, and mean S-wave velocity.

[0075] For example, in regions X1 and X2 shown in Figure 10, the ground surface is the same, so the Laplacian L (and its transformed value L') is also the same. However, region X1, which is at a lower elevation than region X2, is considered to have a higher risk of flooding. Therefore, by using elevation as an explanatory variable in addition to the transformed value L' of the Laplacian, we can derive a more accurate relational expression for calculating the probability p of flooding.

[0076] Furthermore, the average S-wave velocity is an indicator influenced by the hardness of the ground in that area, and indirectly represents the topography of that area (e.g., mountains, plains, alluvial fans, etc.). Therefore, the average S-wave velocity is also thought to influence the risk of flooding. Thus, by using the average S-wave velocity as an explanatory variable in addition to the Laplacian transformation value L', it is possible to derive a more accurate relational expression for calculating the probability of flooding p.

[0077] As described above, in the first embodiment, four indicators considered to have a significant impact on flood occurrence are used as explanatory variables: elevation of the 50m mesh, the transformed value L' of the Laplacian for the 50m and 250m meshes, and the average S-wave velocity of the 250m mesh. In this way, it is desirable not to increase the number of explanatory variables unnecessarily, and the relationship can be derived more easily and quickly compared to cases where five or more indicators are used as explanatory variables. Furthermore, when calculating the probability of flood occurrence p in the evaluation area RA, processing such as data collection for explanatory variables can be made easier.

[0078] However, the analysis is not limited to the above; it is sufficient that the Laplacian transformed value L' is included as at least one explanatory variable. For example, altitude and mean S-wave velocity do not need to be included as explanatory variables, and five or more indicators, including the four indicators mentioned above, may be used as explanatory variables. Examples of other indicators (explanatory variables) include the following:

[0079] For example, in the above example, the elevation for each sample division S1a to S12y (every 50m mesh) is used as an explanatory variable, but the elevation for each intermediate sample division S1 to S12 (every 250m mesh) could also be used as an explanatory variable. Alternatively, slope data for each intermediate sample division S1 to S12 (every 250m mesh) may be used as an explanatory variable. Slope data can be obtained from sources such as the National Land Numerical Information Elevation and Slope 5th-order Mesh Data (https: / / nlftp.mlit.go.jp / ksj / gml / datalist / KsjTmplt-G04-d.html). Furthermore, for example, the slope data of sample divisions S1a to S12y adjacent to the target sample divisions S1a to S12y (for example, adjacent to the upstream or downstream side) may be used as explanatory variables. In addition, the slope data of the intermediate sample divisions S1 to S12 to which the adjacent sample divisions S1a to S12y belong may also be used as explanatory variables.

[0080] ===Second Embodiment=== Figure 11 is a flowchart illustrating the flood risk assessment method of the second embodiment. Figure 12 is an explanatory diagram of the regression coefficient β and its significance probability P value according to the assessment method of the second embodiment. In the first embodiment, regardless of the significance probability P value of the regression coefficient β of the Laplacian L, the transformed value L' of the Laplacian is used as the explanatory variable for all meshes (50m meshes and 250m meshes). In contrast, in the second embodiment, depending on the significance probability P-value of the regression coefficient β of the Laplacian L, either the Laplacian L itself is used as the explanatory variable, or the transformed value L' of the Laplacian is used as the explanatory variable.

[0081] As shown in the flow chart of Figure 11, the process is the same as in the first embodiment up to the point where the evaluator or computer 5 determines a sample area SA that includes locations where flooding has occurred in the past (flooded areas Q) (S11), and collects past flood data and data related to explanatory variables for the sample area SA (S12).

[0082] Subsequently, in the second embodiment, the probability of flood occurrence p based on past flood data of the sample area SA is used as the dependent variable, and predetermined indicators of the sample area SA, including the Laplacian L, are used as independent variables. Logistic regression analysis is then performed to calculate the regression coefficient β (coefficient) of the Laplacian L and its significance probability P value (S13, corresponding to the first step). In other words, similar to the comparative example described in the first embodiment (Figure 8), logistic regression analysis is performed using the Laplacian L of the 50m mesh and 250m mesh as they are, as independent variables. As a result, the regression coefficients β1 to β4 for each independent variable 1 to 4 and their significance probability P values ​​are obtained.

[0083] Next, the evaluator or computer 5 checks whether any of the regression coefficients β2 and β3 of the Laplacian L have a significance probability P value greater than or equal to a predetermined threshold (S14). Below, 5% is used as an example of the predetermined threshold, but the value of the threshold is not particularly limited. If the significance probability P-value for the regression coefficient β is less than 5%, the Laplacian L is used as the explanatory variable. If the significance probability for the regression coefficient β is 5% or greater, the transformed value L' of the Laplacian is calculated using (Equation 3) (S15), and the transformed value L' of the Laplacian is used as the explanatory variable. Then, using the probability of flooding in the sample area SA (whether or not flooding occurs) as the dependent variable, and a predetermined index of the sample area SA, which includes an index related to the Laplacian L (at least one of the Laplacian L and the transformed value L' of the Laplacian), as the independent variable, logistic regression analysis is performed to derive the regression coefficient β and the relational equation (the equation obtained by substituting β into Equation 1, which is a discriminant equation for determining whether or not flooding occurs) (S14~S16, corresponding to the second step).

[0084] By doing so, even when the variability of the Laplacian L in the sample region SA selected to derive the relational equation is large, a significant regression coefficient β can be obtained by using the transformed value L' of the Laplacian as the explanatory variable. Therefore, a relational equation that takes into account the Laplacian L (ground topography) can be derived without excluding the regression coefficient β related to the Laplacian L. On the other hand, when the variability of the Laplacian L in the sample region SA is small, the Laplacian L itself can be used as the explanatory variable. Therefore, a more accurate relational equation can be derived based on the Laplacian L, which directly represents the ground topography, compared to the transformed value L' of the Laplacian. As a result, the probability p of flood occurrence in the evaluation area RA can be calculated with high accuracy based on the relational equation, and the flood risk can be quantitatively evaluated.

[0085] Furthermore, similar to the first embodiment, nationwide standardized data can be used as the dependent and independent variables. Therefore, regional variations are less likely to occur, and a relational expression and the probability of flood occurrence p can be derived that can be evaluated using the same indicator nationwide. In addition, by using the computer 5 (that is, by using a flood risk assessment system equipped with a calculation processing unit 2 that executes the first and second steps), the relational expression and the probability of flood occurrence p can be derived easily and quickly.

[0086] Furthermore, similar to the first embodiment, it is desirable to use an index relating to the Laplacian L of N types (here, two types: 50m mesh and 250m mesh) as an explanatory variable. By doing so, it is possible to derive a relational expression that takes into account local and widespread irregularities, and to determine the probability of flood occurrence p with greater accuracy. In this case, in S13 (first step), as shown in Figure 8, the regression coefficient β of the Laplacian L and its significance probability P value are determined for each of the N types (here, two types) of mesh sizes that divide the sample area SA.

[0087] Then, in steps S14-S16 (second step), for mesh sizes with a significance probability of less than 5%, the Laplacian L is used as the explanatory variable, and for mesh sizes with a significance probability of 5% or more, the transformed value L' of the Laplacian is used as the explanatory variable. For example, in the results shown in Figure 8, the significance probability P-value of the regression coefficient β2 corresponding to the Laplacian L for the 50m mesh is 5% or greater (0.401 = 40.1%). The significance probability P-value of the regression coefficient β3 corresponding to the Laplacian L for the 250m mesh is less than 5% (almost 0). In this case, as shown in Figure 12, for the 50m mesh, the transformed value L' of the Laplacian is used as explanatory variable 2, and for the 250m mesh, the Laplacian L itself is used as explanatory variable 3.

[0088] Furthermore, as shown in Figure 8, if there is a regression coefficient β with a significance probability P-value of 5% or more in S14, the computer 5 performs a new logistic regression analysis using predetermined indicators (for example, four indicators: elevation of the 50m mesh, the transformed value L' of the Laplacian for the 50m mesh, the Laplacian L for the 250m mesh, and the average S-wave velocity of the 250m mesh) as explanatory variables (S16). In this way, regression coefficients β1 to β4 are calculated and a relational equation (equation obtained by substituting β into equation 1) is derived.

[0089] On the other hand, if there is no regression coefficient β with a significance probability P-value of 5% or more in S14, the Laplacian L is used as the explanatory variable for all N mesh sizes. Therefore, computer 5 can derive the relationship by directly applying the regression coefficient β obtained in S13 (first step) without performing a new logistic regression analysis. This shortens the process of deriving the relationship. Furthermore, if there is no regression coefficient β with a significance probability P-value of less than 5% in S14, the transformed Laplacian value L' is used as the explanatory variable for all N mesh sizes.

[0090] As described above, it is desirable to decide for each mesh size whether to use the Laplacian L as the explanatory variable or the transformed value L' of the Laplacian as the explanatory variable. By using the transformed value L' of the Laplacian as the explanatory variable, regression coefficients β for mesh sizes with large variability in the Laplacian L are not excluded. For mesh sizes with small variability in the Laplacian L, by using the Laplacian L as the explanatory variable, a more accurate relationship can be derived based on the Laplacian L which directly represents the unevenness of the ground.

[0091] While the example uses indicators related to the Laplacian L of 50m and 250m meshes as explanatory variables, the number and size of meshes are not particularly limited. For example, there may be one mesh size, or there may be three or more. In addition to the indicator related to the Laplacian L, the elevation of a 50m mesh and the average S-wave velocity of a 250m mesh were given as examples of other specified indicators, but the number and types of specified indicators are not particularly limited.

[0092] ===Other Embodiments=== The embodiments described above are for the purpose of facilitating understanding of the present invention and are not intended to limit its interpretation. The present invention can be modified and improved without departing from its spirit, and it goes without saying that equivalents thereof are included. In particular, the embodiments described below are also included in the present invention.

[0093] In the above embodiment, examples were given of using data such as land history survey data, elevation / slope 5th-order mesh data from the National Land Numerical Information System, and J-SHIS surface ground data from the National Research Institute for Earth Science and Disaster Resilience as data used for past flood data and explanatory variables. However, the embodiment is not limited to these; other data may be used as long as it is, for example, nationwide standardized and reasonably reliable.

[0094] Furthermore, in the above embodiment, logistic regression analysis was used to derive a discriminant (relational expression) for determining whether or not flood damage occurred, but the analysis method is not limited to this. For example, the discriminant may be derived based on other analysis methods such as linear discriminant analysis or quadratic discriminant analysis. In that case as well, by using the transformed value L' of the Laplacian, it is possible to derive a highly accurate discriminant that takes the Laplacian L into account.

[0095] Furthermore, computer 5 may be equipped with machine learning functions, such as deep learning, as a prediction means for deriving a discriminant equation (predictive model). In this case, when new input data (dependent variable, independent variable) is acquired, the machine learning function can update the discriminant equation to take that new input data into account, thereby enabling more accurate prediction of whether or not floods will occur. [Explanation of symbols]

[0096] 1 storage section, 2. Processing Unit, 3. Input section, 4 Display section, 5. Computer (flood risk assessment system), Q: Locations where flooding occurred, SA sample area, S1~S12 Sample Intermediate Section, S1a~S12y Sample classification, RA Evaluation Areas

Claims

1. A flood risk assessment method that uses a computer to derive a discriminant formula for determining whether or not a flood will occur, From the Laplacian L, which is an index of the ground surface irregularities in a sample area including areas where floods have occurred in the past, the transformed value L' of the Laplacian is calculated using the following formula. The dependent variable is whether or not a flood occurred based on past flood data in the aforementioned sample area. Using a predetermined index of the sample region including the Laplacian transformation value L' as an explanatory variable, A flood risk assessment method characterized by deriving the aforementioned discriminant formula by performing an analysis.

2. A flood risk assessment method according to claim 1, The sample region is divided into N different mesh sizes, and the Laplacian transformation value L' is calculated for each of the N divisions. A flood risk assessment method characterized in that the Laplacian transformation value L' is used as the explanatory variable for all of the N types of mesh sizes.

3. A flood risk assessment method that uses a computer to derive a discriminant formula for determining whether or not a flood will occur, The dependent variable is whether or not a flood occurred in a sample area that includes locations where floods have occurred in the past, based on past flood data. Using a predetermined index of the sample area, including the Laplacian L which is an index of the unevenness of the ground, as an explanatory variable, The first step involves performing an analysis to determine the coefficient of the Laplacian L and its significance probability, If the significance probability is less than a predetermined threshold, the Laplacian L is used as the explanatory variable. If the significance probability is equal to or greater than the predetermined threshold, the transformed value L' of the Laplacian calculated using the following formula will be used as the explanatory variable. The dependent variable is whether or not the aforementioned flood occurred in the sample area. Using the predetermined index of the sample region, which includes the index related to the Laplacian L, as an explanatory variable, The second step involves deriving the discriminant formula by performing an analysis, A method for assessing flood risk, characterized by having the following features.

4. A flood risk assessment method according to claim 3, In the first step described above, for each of the N mesh sizes that divide the sample region, the coefficient of the Laplacian L and its significance probability are determined. In the second step described above, For the mesh size for which the significance probability is less than the predetermined threshold, the Laplacian L is used as the explanatory variable. A flood risk assessment method characterized in that, for the mesh size in which the significance probability is equal to or greater than the predetermined threshold, the transformed value L' of the Laplacian is used as the explanatory variable.

5. A flood risk assessment method according to any one of claims 1 to 4, A flood risk assessment method characterized by determining whether or not flood damage occurs in the area to be evaluated based on the predetermined index in the area to be evaluated and the discriminant formula.

6. A flood risk assessment method according to any one of claims 1 to 4, A flood risk assessment method characterized in that the predetermined indicators are the Laplacian L index, elevation, and mean S-wave velocity.

7. A storage unit that stores data about a sample area including locations where floods have occurred in the past, From the Laplacian L, which is an index of the unevenness of the ground in the aforementioned sample area, the transformed value L' of the Laplacian is calculated using the following formula. The dependent variable is whether or not a flood occurred based on past flood data in the aforementioned sample area. Using a predetermined index of the sample region including the Laplacian transformation value L' as an explanatory variable, A processing unit that performs analysis to derive a discriminant formula for determining whether or not a flood has occurred, A flood risk assessment system characterized by having the following features.

Images