Method for correction of rainfall data, system for correction of rainfall data, program for correction of rainfall data, and correction device

Abstract

To realize a method for providing rainfall data with high accuracy in real time.SOLUTION: The method for correction of rainfall data includes: a first acquisition step of acquiring satellite rainfall data; a second acquisition step of acquiring ground rainfall data that is either configured such that mutually different accumulation times are set for a plurality of observation points or have non-uniform densities of the observation points; and a correction step of correcting the satellite rainfall data so as to minimize an objective function including a ground rainfall consistency term.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a rainfall data correction method, a rainfall data correction system, a rainfall data correction program, and a correction device. [Background technology]

[0002] In developing countries or remote islands, there are cases where rainfall cannot be estimated appropriately due to a lack of ground observation equipment such as weather radar or rain gauges. To address this issue, JAXA (Japan Aerospace Exploration Agency) developed the Global Satellite Mapping of Precipitation (GSMaP), which combines data from multiple satellites. GSMaP is used to estimate rainfall in many countries, particularly developing countries (see, for example, Non-Patent Document 1 below). [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] "Performance study of GSMaP satellite rainfall product in central Vietnam", [online], [Retrieved March 5, 2024], Internet<https: / / doi.org / 10.11520 / jshwr.27.0_100034> Summary of the Invention [Problem to be solved by the invention]

[0004] However, GSMaP is rainfall data estimated based on satellite observations (e.g., cloud positions identified from satellites), and therefore, the accuracy of GSMaP-based rainfall data may be lower than that of rainfall data observed on the ground.

[0005] There is also a method to correct GSMaP rainfall data using the global surface rain gauge map provided by the NOAA (National Oceanic and Atmospheric Administration) CPC (Climate Prediction Center). However, there is a three-day delay before the NOAA CPC rainfall data is created, so it is not possible to correct GSMaP rainfall data in real time to predict rainfall.

[0006] One aspect of the present invention has been made in consideration of the above problems, and its object is to provide real-time, highly accurate rainfall data. [Means for solving the problem]

[0007] In order to solve the above problem, the rainfall data correction method according to aspect 1 of the present invention includes a first acquisition step of acquiring satellite rainfall data, which is rainfall data estimated based on satellite observation; a second acquisition step of acquiring ground rainfall data, which is rainfall data observed on the ground, in which different accumulation times are set for multiple observation points or the density of observation points for the ground rainfall data is uneven; and a correction step of correcting the satellite rainfall data to the corrected rainfall data so as to minimize an objective function including a ground rainfall matching term that represents the error between the ground rainfall data and the corrected rainfall data.

[0008] In the rainfall data correction method according to aspect 2 of the present invention, in the above-mentioned aspect 1, the ground rainfall matching term may include an error between the ground rainfall data and the corrected rainfall data for the accumulated time for the ground rainfall data whose accumulated time is longer than the unit time of the satellite rainfall data.

[0009] In a rainfall data correction method according to aspect 3 of the present invention, in aspect 1 or 2 above, the corrected rainfall data may be obtained by multiplying the satellite rainfall data by an intermediate variable, the objective function may be a function with the intermediate variable as a variable, and in the correction step, the objective function may be minimized under a non-negative constraint that the intermediate variable is greater than or equal to 0.

[0010] In the rainfall data correction method according to aspect 4 of the present invention, in any of aspects 1 to 3 above, the objective function further includes a satellite rainfall matching term representing the error between the satellite rainfall data and the corrected rainfall data, and in the correction step, the ground rainfall matching term may be calculated for a location and time where the ground rainfall data exists, and the satellite rainfall matching term may be calculated for a location and time where the ground rainfall data does not exist but the satellite rainfall data exists.

[0011] In the rainfall data correction method according to aspect 5 of the present invention, in any of aspects 1 to 4 above, the objective function may further include a spatial direction regularization term that evaluates the continuity of two adjacent corrected rainfall data in the spatial direction.

[0012] In the rainfall data correction method according to aspect 6 of the present invention, in any of aspects 1 to 5 above, the objective function may further include a time direction regularization term that evaluates the continuity of two adjacent corrected rainfall data in the time direction.

[0013] In the rainfall data correction method according to a seventh aspect of the present invention, in any one of the first to sixth aspects, different weighting coefficients may be set for the ground rainfall matching term for a plurality of observation points.

[0014] In a rainfall data correction method according to aspect 8 of the present invention, in aspect 7 above, the weighting coefficient of an observation point in a first region may be greater than the weighting coefficient of an observation point in a second region having a lower density of observation points than the first region.

[0015] In order to solve the above problem, the rainfall data correction system according to aspect 9 of the present invention comprises an acquisition unit that acquires satellite rainfall data, which is rainfall data estimated based on satellite observation, and ground rainfall data, which is rainfall data observed on the ground, including rainfall data for which different accumulation times are set for multiple observation points, or where the density of observation points for the ground rainfall data is uneven; a correction unit that corrects the satellite rainfall data to the corrected rainfall data so as to minimize an objective function including a ground rainfall matching term that represents the error between the ground rainfall data and the corrected rainfall data; and an output control unit that outputs the corrected rainfall data to the outside.

[0016] A rainfall data correction program according to a tenth aspect of the present invention is a rainfall data correction program for causing a computer to function as the rainfall data correction system according to the ninth aspect, and causes the computer to function as the correction unit.

[0017] In order to solve the above problem, a correction device according to aspect 11 of the present invention is a correction device used in a rainfall data correction system, which acquires satellite rainfall data, which is rainfall data estimated based on satellite observation, and ground rainfall data, which is rainfall data observed on the ground, and which includes rainfall data for which different accumulation times are set for multiple observation points, or where the density of observation points for the ground rainfall data is uneven, and corrects the satellite rainfall data to the corrected rainfall data so as to minimize an objective function including a ground rainfall matching term that represents the error between the ground rainfall data and the corrected rainfall data. [Effects of the Invention]

[0018] According to one aspect of the present invention, it is possible to provide accurate rainfall data in real time. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is a block diagram showing an example of a configuration of a main part of a rainfall data correction system according to an embodiment of the present invention. [Figure 2]FIG. 2 is a schematic diagram showing an example of distribution of ground rainfall data. [Figure 3] 4 is a flowchart showing a flow of processing executed by the rainfall data correction system. [Figure 4] This is a precipitation map shown by ground rainfall data. [Figure 5] This is an example of GSMaP. [Figure 6] This is an example of a comparative precipitation map created by correcting GSMaP data using a global surface rain gauge map provided by NOAA CPC. [Figure 7] 1 is an example of a precipitation map according to the present embodiment, created by correcting GSMaP data using ground rainfall data. DETAILED DESCRIPTION OF THE INVENTION

[0020] [Embodiment 1] (1. Overview of the rainfall data correction system) 1 is a block diagram showing an example of the configuration of the main parts of a rainfall data correction system 10 according to this embodiment. First, the rainfall data correction system 10 that executes the rainfall data correction method according to this embodiment will be described with reference to FIG.

[0021] 1, the rainfall data correction system 10 communicates with an external system 30 via an external network 20. The external system 30 includes a satellite rainfall data management server 31 and a ground rainfall data management server 32.

[0022] The satellite rainfall data management server 31 is a server that manages rainfall data estimated based on satellite observation (satellite rainfall data). The satellite rainfall data is, for example, rainfall data based on GSMaP managed by JAXA (hereinafter referred to as GSMaP data). The GSMaP data has a spatial resolution of 10 km and a time resolution of 1 hour. In this embodiment, an example will be described in which the satellite rainfall data is GSMaP data.

[0023] The ground rainfall data management server 32 is a server that manages rainfall data observed on the ground (ground rainfall data). The ground rainfall data is a combination of rainfall amounts observed by ground observation equipment (weather radar, rain gauges, etc.) installed in various locations. Details of the ground rainfall data will be described later with reference to FIG. 2.

[0024] The rainfall data correction system 10 includes a communication unit 11 (acquisition unit), a correction unit 12, and an output control unit 13. The rainfall data correction system 10 is, for example, a device (such as a PC) installed with a program that executes a rainfall data correction method described below.

[0025] The communication unit 11 transmits and receives data to and from the external system 30 via the external network 20. Specifically, the communication unit 11 acquires satellite rainfall data from the satellite rainfall data management server 31 and acquires ground rainfall data from the ground rainfall data management server 32.

[0026] The correction unit 12 corrects the satellite rainfall data to corrected rainfall data using the ground rainfall data. Details of the rainfall data correction method performed by the correction unit 12 will be described later with reference to FIG.

[0027] The output control unit 13 outputs to the outside the corrected rainfall data derived by the correction unit 12. The output control unit 13 causes a display device having a display unit, for example, to display the corrected rainfall data.

[0028] The correction unit 12 may exist as a correction device independent of the device including the communication unit 11 and the output control unit 13. The correction unit 12 may also be included in the external system 30. Specifically, a computer server included in the external system 30 may correct the satellite rainfall data to corrected rainfall data using ground rainfall data, or the ground rainfall data management server 32 may be configured integrally with the correction unit 12. In this case, the communication unit 11 acquires the corrected rainfall data from the computer server.

[0029] (2. Ground rainfall data) FIG. 2 is a schematic diagram showing an example of the distribution of ground rainfall data. As described above, ground rainfall data is a combination of rainfall amounts observed by ground observation facilities installed in various locations. The rainfall data correction system 10 regards the rainfall amounts observed by ground observation facilities in grid-like divisions (pixels) as rainfall data at positions corresponding to each division. For example, in FIG. 2, the rainfall data at position [p, q] is the rainfall amount observed by the ground observation facility in division P1, and the rainfall data at position [p+1, q] is the rainfall amount observed by the ground observation facility in division P2. There is no ground rainfall data at locations where there are no ground observation facilities.

[0030] The width of the above-mentioned section, which is the spatial resolution of the ground rainfall data, is, for example, 10 km, which is the spatial resolution of the GSMaP data. That is, the rainfall data correction system 10 acquires the ground rainfall observed at position [i, j] in association with the estimated rainfall at position [i, j] of the GSMaP data.

[0031] As shown in Figure 2, the time resolution of ground rainfall data measured at ground observation facilities in various locations may differ from one another. In other words, different accumulation times may be set for multiple observation points. For example, in Figure 2, the ground observation facility in section P1 measures rainfall in 3-hour intervals, while the ground observation facility in section P2 measures rainfall in 6-hour intervals. In other words, the unit of rainfall data at position [p, q] is [mm / 3h], and the unit of rainfall data at position [p+1, q] is [mm / 6h].

[0032] Also, as shown in Figure 2, the number of ground observation facilities may vary depending on the region. In other words, the density of observation points for ground rainfall data may be uneven. For example, in Figure 2, the density of observation points is high in region R1, but low in region R2.

[0033] The correction unit 12 of the rainfall data correction system 10 corrects the satellite rainfall data using ground rainfall data in which different accumulation times are set for multiple observation points or the density of observation points is uneven, as described above. This makes it possible to provide real-time, highly accurate rainfall data using data observed in real time by ground observation facilities.

[0034] (3. Example of rainfall data correction system operation) 3 is a flowchart showing the flow of processing executed by the rainfall data correction system 10. Hereinafter, the rainfall data correction method using the above-mentioned ground rainfall data will be described with reference to FIG.

[0035] First, in S1, the communication unit 11 acquires satellite rainfall data from the satellite rainfall data management server 31 via the external network 20 (first acquisition step). For example, the communication unit 11 may acquire GSMaP data from a server managed by JAXA every hour, which is the time resolution of the GSMaP data.

[0036] Next, in S2, the communication unit 11 acquires ground rainfall data from the ground rainfall data management server 32 via the external network 20 (second acquisition step). The communication unit 11 acquires the rainfall observed by ground observation facilities located in the observation target area (hereinafter referred to as the target area) and the positions [i, j] of the ground observation facilities. The communication unit 11 acquires ground rainfall data from the servers managed by the ground observation facilities for each time resolution of the ground rainfall data measured by the ground observation facilities in each location. For example, in the example shown in FIG. 2, the communication unit 11 acquires rainfall data at position [p, q] every three hours and rainfall data at position [p+1, q] every six hours.

[0037] Next, in S3, the correction unit 12 corrects the satellite rainfall data to corrected rainfall data using the ground rainfall data (correction step). The correction unit 12 extracts satellite rainfall data within the target area from the acquired satellite rainfall data, and corrects the satellite rainfall data within the target area using the ground rainfall data. Specifically, the correction unit 12 derives the corrected rainfall data by solving an optimization problem that minimizes an objective function including a ground rainfall matching term that represents the error between the ground rainfall data and the corrected rainfall data.

[0038] Next, in S4, the output control unit 13 outputs the corrected rainfall data derived by the correction unit 12 to the outside. The output control unit 13 causes a display device including a display unit, for example, to display the corrected rainfall data.

[0039] <Objective function> The optimization problem for minimizing the above-mentioned objective function is expressed, for example, by the following equation (1).

number

[0040] x t [i,j] is the hourly satellite rainfall data [mm / h] at time t and position [i,j]. t [i,j] is the hourly corrected rainfall data [mm / h] at time t and position [i,j]. R t [i,j] is the accumulated time T corresponding to position [i,j] at time t and position [i,j]. i,j Ground rainfall data per [mm / T i,j h].

[0041] In this optimization problem, satellite rainfall data for a total of L hours at times t, t-1, t-2, ... t-L+1 (time t-L+1 going back in time from the current time t) is simultaneously corrected.

[0042] In addition, in equation (1), at a certain time t-τ, the position [i,j]∈S t-τ Ground rainfall data R t [i,j] is given, i.e., S t is the ground rainfall data R at time t. t [i,j] is the set of positions [i,j] where [i,j] exists.

[0043] In addition, the corrected rainfall data a t [i,j] is the satellite rainfall data x t [i,j] is an intermediate variable α t [i,j] multiplied by (a t [i,j]=α t [i,j]x t [i,j]), and the objective function is the intermediate variable α t In this optimization problem, we represent the intermediate variable α for all times t-τ (τ=0,1,…,L-1) and all positions [i,j] so that the objective function is minimized. t [i,j] are determined simultaneously.

[0044] In this optimization problem, the intermediate variable α t A non-negative constraint that [i,j] is 0 or greater may be imposed. This allows the corrected rainfall data a t This can prevent negative rainfall from occurring.

[0045] The first term of the objective function (w 1、t [i,j]) is the satellite rainfall data x t [i,j] and corrected rainfall data a t The satellite rainfall matching term represents the error with [i,j]. The satellite rainfall matching term is an intermediate variable α t [i,j] should be as close to 1 as possible (i.e., the corrected rainfall data a t [i,j] is the satellite rainfall data x t[i,j]).

[0046] The second term of the objective function (w 2、t [i,j]) is the ground rainfall data R t [i,j] and corrected rainfall data a t The ground rainfall matching term represents the error with [i,j]. The ground rainfall matching term is calculated over the accumulation time T i,j For ground rainfall data whose unit time is longer than the unit time (1 hour) of satellite rainfall data, i,j The ground rainfall matching term is calculated for the position [i,j]∈S t-τ Accumulation time T i,j Minutes of corrected rainfall data a t [i,j](=α t [i,j]x t [i,j]) as much as possible from the ground rainfall data R t It contributes to getting closer to [i,j].

[0047] The third term of the objective function (w 3、t [i,j] and w 4、t The term including [i,j]) is a spatial regularization term that evaluates the continuity of two adjacent corrected rainfall data in the spatial direction. In detail, the spatial regularization term is an intermediate variable α t [i,j], α t The difference between [i+1,j] and the intermediate variable α between two adjacent corrected rainfall data in the longitude direction t [i,j], α t The spatial direction regularization term is the difference between the corrected rainfall data a t Contributes to smoothing in the spatial direction of [i,j].

[0048] The fourth term of the objective function (w 5、t The term including [i,j]) is a time-direction regularization term that evaluates the continuity of two adjacent corrected rainfall data in the time direction. In detail, the fourth term of the objective function is the intermediate variable α t [i,j], α t+1The time direction regularization term represents the difference between [i,j] and the corrected rainfall data a t Contributes to smoothing in the time direction of [i,j].

[0049] w in the objective function 1、t [i,j]~w 5、t [i,j] are weighting coefficients that determine the magnitude of the contribution of the first, second, and third terms of the objective function, the latitude component of the third term, the longitude component of the third term, and the fourth term, respectively.

[0050] The correction unit 12 calculates the satellite rainfall matching term for positions and times where satellite rainfall data exists. The correction unit 12 calculates the ground rainfall matching term for positions and times where ground rainfall data exists. The correction unit 12 calculates the satellite rainfall matching term without calculating the ground rainfall matching term for positions and times where ground rainfall data does not exist but satellite rainfall data exists. That is, for a position [i, j] and time t where ground rainfall data exists, w 1、t [i,j]~w 5、t [i,j] are all set to be greater than 0. On the other hand, for the position [i,j] and time t where there is no ground rainfall data but there is satellite rainfall data, the weighting coefficient w 2、t [i,j] is set to 0, and the other weighting factors are set to be greater than 0.

[0051] Weighting factor w 1、t [i,j]~w 5、t [i,j] may be a fixed value or a variable value with respect to the position [i,j] and the time t. That is, the weighting coefficient w 1、t [i,j]~w 5、t [i, j] may be set in advance or may be adjusted by the correction unit 12 in accordance with data acquired by the communication unit 11.

[0052] For example, the weighting coefficient w of the ground rainfall matching term 2、t[i,j] may be different for multiple observation points (i.e., depending on the position [i,j]). Specifically, the weighting coefficient for an observation point in a first region (e.g., region R1 in Figure 2) may be larger than the weighting coefficient for an observation point in a second region (e.g., region R2 in Figure 2) where the density of observation points is lower than that of the first region. This increases the contribution of the ground rainfall matching term in regions where the density of observation points is higher and the reliability is higher. This improves the accuracy of the corrected rainfall data.

[0053] In addition, the weighting coefficient w of the spatial direction regularization term 3、t [i,j] may be set according to the topography of the location [i,j]. For example, the weighting coefficient in mountainous areas, coastal areas, and other places where rainfall fluctuates rapidly may be set smaller than the weighting coefficient in basins.

[0054] Also, the weighting factor w 1、t [i,j]~w 5、t [i, j] may be set to 0 depending on the specificity of the data acquired by the communication unit 11. For example, when the ground rainfall data at the position [p, q] at the time t1 indicates an outlier, the weight coefficient w of the ground rainfall consistency term is 2、t1 The value of [p,q] may be set to 0.

[0055] By dealing with the optimization problem of minimizing the above objective function, the following effects are achieved. That is, the rainfall data correction system 10 minimizes the objective function including the ground rainfall matching term, thereby achieving the spatial and temporal distribution of ground rainfall data in the target area (i.e., S t-τ and T i,j ) regardless of the actual situation, corrected rainfall data can be derived.

[0056] Furthermore, the rainfall data correction system 10 can derive corrected rainfall data for all positions [i, j] by minimizing an objective function that includes a spatial direction regularization term, even if ground rainfall data for all positions [i, j] does not exist.

[0057] Similarly, the rainfall data correction system 10 can derive corrected rainfall data for all times t by minimizing an objective function including a time direction regularization term, even if ground rainfall data for all times t does not exist.

[0058] In addition, the corrected rainfall data is a t [i,j]=α t [i,j]x t Since it is given in the form of [i,j], the movement of rain clouds in the satellite rainfall data can be carried over to the corrected rainfall data. t This can prevent fixed rain clouds from appearing.

[0059] The objective function for deriving the corrected rainfall data is not limited to that shown in equation (1). For example, in the spatial regularization term and the time regularization term, α t [i,j] is α t [i,j]x t [i,j] may be substituted. The reason for taking the difference of the intermediate variables in the third and fourth terms of the objective function above is to reduce the amount of calculation required for the optimization algorithm described below. Also, in each term of the objective function, the square of the difference value may be substituted with the absolute value of the difference value.

[0060] Techniques for solving the above optimization problem include, for example, FISTA (Fast Iterative Shrinkage Thresholding Algorithm) and ADMM (Alternating Direction Method of Multipliers). Below, we will explain the optimization techniques using FISTA and ADMM.

[0061] Hereinafter, the size of the target area is I × J. The intermediate variable at each time is α t [1,1],α t [1,2],…,α t [1,J],α t [2,1],α t [2,2],…,α t [2,J],…,αt [I,1], α t [I,2], …, α t become [I,J]. All intermediate variables α t-τ [i,j] (τ = 0, 1, …, L - 1, i = 1, 2, …, I, j = 1, 2, …, J) are collectively represented by the following vertical vector.

Number

[0062] Since the dimension of the vertical vector α is IJL, it is expressed as follows:

Number

[0063] <Optimization Method Using FISTA> The optimization problem shown in Equation (1) can be expressed as follows using vectors and matrices.

Number

Number

Number

[0064] As a result, α (k) converges to the optimal solution (y (k) also converges to the optimal solution).

[0065] <Optimization Method Using ADMM> The optimization problem shown in Equation (1) can be expressed as follows by newly adding auxiliary variables z1, z2, z3, z4, z5.

Number

Number

Number

[0066] This allows α (k) converges to the optimal solution (ξ1 (k) , ξ2 (k) , ξ3 (k) , ξ4 (k) , ξ5 (k) converges to γ ​​times the optimal solution of the dual problem).

number

number

[0067] Unlike FISTA, ADMM can also handle regularization terms with non-differentiable points such as the following:

number

number

number

[0068] (4. Precipitation Map) FIG. 4 is an example of a precipitation map shown by ground rainfall data. FIG. 5 is an example of GSMaP. FIG. 6 is an example of a precipitation map according to a comparative example, created by correcting GSMaP data using a global ground rain gauge map provided by NOAA CPC. FIG. 7 is an example of a precipitation map according to this embodiment, created by correcting GSMaP data using ground rainfall data. Here, FIGS. 4 to 7 are precipitation maps for the same region and the same time. Also, FIGS. 4 to 7 show rainfall data obtained by accumulating rainfall over 24 hours.

[0069] Comparing Figures 4 and 5, we can see that in areas R3 and R4, for example, the GSMaP data differs from the ground rainfall data actually measured on the ground. This indicates that the accuracy of GSMaP rainfall data may be low because it is estimated based on satellite observations.

[0070] Furthermore, comparing Figures 4 and 6, for example, in area R3, the contours of the comparative adjusted rainfall data are blurred compared to the ground rainfall data actually measured on the ground. This indicates that the accuracy and resolution are slightly lower than the raw data because the GSMaP data is adjusted using spatially smoothed NOAA CPC rainfall data rather than raw data (ground rainfall data actually measured on the ground). Furthermore, because there is a three-day delay before the NOAA CPC rainfall data is generated, the comparative adjusted rainfall data is not provided in real time.

[0071] 4 and 7, it can be seen that the corrected rainfall data according to this embodiment reflects the ground rainfall data, even in areas R3 and R4. That is, the corrected rainfall data according to this embodiment corrects the GSMaP data using raw data, and therefore can provide accurate rainfall data in real time.

[0072] (5. Action and Effects) The rainfall data correction method described above can be used to provide disaster prevention information to densely populated areas vulnerable to meteorological disasters. This effect will also contribute to the achievement of Goal 11 of the United Nations' Sustainable Development Goals (SDGs), such as "Make cities and human settlements inclusive, safe, resilient and sustainable."

[0073] [Software implementation example] The functions of the rainfall data correction system 10 (hereinafter referred to as the "device") can be realized by a program that causes a computer to function as the device, and a program that causes a computer to function as each control block of the device (particularly each part included in the correction unit 12).

[0074] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The control device and storage device execute the program, thereby realizing the functions described in each of the above embodiments.

[0075] The program may be non-transitory and may be recorded on one or more computer-readable recording media. The recording media may or may not be included in the device. In the latter case, the program may be supplied to the device via any wired or wireless transmission medium.

[0076] Furthermore, some or all of the functions of the control blocks can be realized by logic circuits. For example, an integrated circuit in which a logic circuit that functions as each of the control blocks is formed is also included in the scope of the present invention. In addition, the functions of the control blocks can also be realized by, for example, a quantum computer.

[0077] Furthermore, each process described in each of the above embodiments may be executed by AI (Artificial Intelligence). In this case, the AI ​​may run on the control device or on another device (for example, an edge computer or a cloud server).

[0078] [Additional Notes] Each unit in the rainfall data correction system 10 may be provided on a separate server. That is, the acquisition unit that acquires satellite rainfall data and ground rainfall data, the correction unit that corrects the satellite rainfall data, and the output control unit that outputs the corrected rainfall data may be provided on separate servers.

[0079] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Explanation of symbols]

[0080] 10 Rainfall data correction system 11. Communication Department (Acquisition Department) 12 Correction unit 13 Output control section

Claims

1. a first acquisition step of acquiring satellite rainfall data, which is rainfall data estimated based on satellite observation; a second acquisition step of acquiring ground rainfall data, which is rainfall data observed on the ground, wherein different accumulation times are set for a plurality of observation points, or the density of the observation points for the ground rainfall data is non-uniform; and a correction step of correcting the satellite rainfall data to the corrected rainfall data so as to minimize an objective function including a ground rainfall matching term that represents an error between the ground rainfall data and the corrected rainfall data.

2. 2. The rainfall data correction method according to claim 1, wherein the ground rainfall matching term includes an error between the ground rainfall data and corrected rainfall data for an accumulated time longer than the unit time of the satellite rainfall data.

3. the corrected rainfall data is obtained by multiplying the satellite rainfall data by an intermediate variable; the objective function is a function that uses the intermediate variables as variables, The rainfall data correction method according to claim 1 , wherein the correction step minimizes the objective function under a non-negative constraint that the intermediate variable is equal to or greater than 0.

4. The objective function further includes a satellite rainfall matching term that represents an error between the satellite rainfall data and the corrected rainfall data; In the correction step, For the location and time for which the ground rainfall data exists, the ground rainfall matching term is calculated; The rainfall data correction method according to claim 1 , wherein the satellite rainfall matching term is calculated for a position and time for which the ground rainfall data is not available but the satellite rainfall data is available.

5. The rainfall data correcting method according to claim 1 , wherein the objective function further includes a spatial direction regularization term that evaluates continuity between two pieces of corrected rainfall data adjacent in a spatial direction.

6. The rainfall data correcting method according to claim 1 , wherein the objective function further includes a time direction regularization term that evaluates continuity between two corrected rainfall data adjacent in the time direction.

7. The rainfall data correction method according to claim 1 , wherein different weighting coefficients are set for the ground rainfall matching term for a plurality of observation points.

8. The rainfall data correction method according to claim 7 , wherein the weighting coefficient of an observation point in a first region is greater than the weighting coefficient of an observation point in a second region where the density of observation points is lower than that of the first region.

9. an acquisition unit that acquires satellite rainfall data, which is rainfall data estimated based on satellite observation, and ground rainfall data, which is rainfall data observed on the ground, including rainfall data for which different accumulation times are set for multiple observation points, or the density of the observation points for the ground rainfall data is uneven; a correction unit that corrects the satellite rainfall data to the corrected rainfall data so as to minimize an objective function including a ground rainfall matching term that represents an error between the ground rainfall data and the corrected rainfall data; an output control unit that outputs the corrected rainfall data to an external device.

10. A rainfall data correction program for causing a computer to function as the rainfall data correction system according to claim 9, the rainfall data correction program causing a computer to function as the correction unit.

11. A correction device used in a rainfall data correction system, By acquiring satellite rainfall data, which is rainfall data estimated based on satellite observation, and ground rainfall data, which is rainfall data observed on the ground, including rainfall data with different accumulation times set for multiple observation points, or where the density of observation points for the ground rainfall data is uneven, a correction device that corrects the satellite rainfall data to the corrected rainfall data so as to minimize an objective function including a ground rainfall matching term that represents an error between the ground rainfall data and the corrected rainfall data;

Images