JP2000187082A - Rainfall prediction correction method, its device and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically correct predicted rainfall strength corresponding to unaccountable rainfall region change (generation, disappearance, development and attenuation). SOLUTION: A generation/disappearance judgment part 16 judges generation and disappearance based on each time sequential rainfall data in the grid stored in a rainfall data memory 12 for each grid and rainfall intensity data of the grid in correspondent relation with the grid calculated for each advective vector stored in advective vector memory 14. A generation/disappearance correction parameter operation part 17 determines correction parameters (correction coefficient and correction constant) with a specific method about the grids which are judged as generation or disappearance by the generation/disappearance judgment part 16. A rainfall movement prediction generation/disappearance correction operation part 18 corrects rainfall prediction operation results (predicted rainfall intensity) obtained by a rainfall movement prediction operation part 15 with a correction parameter obtained with the generation/ disappearance correction parameter operation part 17.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rainfall prediction correction method, an apparatus therefor, and a recording medium.

[0002]

2. Description of the Related Art In recent years, various methods have been proposed or put into practical use for short-term rainfall prediction in which local and short-term rainfall is predicted using a rainfall radar or the like. Representative methods generally known include a rain area tracking method and a method based on an advection equation (hereinafter referred to as an advection model method).

First, the rain area tracking method will be briefly described.

In the rain area tracking method, first, a rainfall observation area is divided into a plurality of grids, and continuous time-series rainfall data observed for each grid is obtained. The data input here is data on the amount of rainfall (rainfall intensity) per unit time for each grid, and is observed by rainfall observation means such as a digital radar rain gauge.

Next, the obtained time-series rainfall data is binarized according to, for example, a predetermined threshold value of the amount of rainfall to obtain a time-series rainfall area pattern. Next, when the rain area pattern at a certain time is moved in each direction and each distance,
Calculate the cross-correlation coefficient with the rain area pattern at a later time.
The direction and distance of the movement of the rain area pattern when the calculated value of the cross-correlation coefficient indicates the maximum value is determined by the movement vector of the rain area between the observation times of the two rain area patterns used in the calculation. I reckon. Then, assuming that the rain area will move according to the movement vector in the future, the acquired time-series rainfall data is moved based on the movement vector to generate movement prediction data of future rainfall.

On the other hand, in the advection model method, first, time-series rainfall data is acquired in the same manner as in the rain area tracking method, and the parameters of the advection equation are calculated by linear least square estimation using this data. This will be described in detail below.

First, let the rainfall intensity at the time t at the observation grid point at the position (x, y) be z (x, y, t), and the space-time change of z is expressed by the following equation (1) (advection equation). Shall be.

∂z / ∂t + u (∂z / ∂x) + v (∂z / ∂y) = w (1) (u, v) represents an advection vector, and w represents a developmental weakening term.
Here, u = α ₁ x + α ₂ y + α ₃ v = α ₄ x + α ₅ y + α ₆ (2) It is assumed that w = α ₇ x + α ₈ y + α _9, and the advection model parameters α ₁ ,.
α ₉ using the time series rainfall distribution data obtained above,
It is obtained by linear least square estimation described below.

First, each of M × N grid points (x _i , x _i ,
_{y j), i = 1,} ..., M, j = 1, ..., for N, to the difference of the advection equation at time t _k. Here, the time difference width is assumed to be Δt, and z (x _i ,
When y _j , t _k ) is represented as z _ijk , (∂z / ∂t) _ijk = (z _{ij (k + 1)} -z _ijk ) / Δt (∂z / ∂x) _ijk = (z _{(i + 1) jk−} z _{(i−1) jk} ) / 2Δx (3) (∂z / ∂y) _ijk = (zi _{(j + 1) k −} zi _{(j−1) k} ) / 2Δy Can be approximated. The data of z up to time t _{k + 1} is a known value input in the previous process. By substituting the above equation (3) into the equation (1), the following equation is obtained.

(Α ₁ x _i + α ₂ y _j + α ₃ ) (z _{(i + 1) jk} −z _{(i-1) jk} ) / 2Δx + (α ₄ x _i + α ₅ y _j + α ₆ ) ((z _{i (j + 1) k -z} i (j-1) k) / 2Δy - (α 7 x i + α 8 y j + α 9) = - (z ij (k + 1) -z ijk) / Δt-d _ij (4) Here, d _ij is introduced to represent the residual, and this value should be originally zero if the assumptions so far are correct. Α ₁ ,…, α ₉ are determined so that the sum of squares of the differences is minimized.

[0011]

(Equation 1)

Α obtained by partially differentiating α ₁ ,..., Α _9
1, ..., it may be solved a system of linear equations and unknowns α _9.

[0013] advection model parameters alpha ₁ according to the above procedures, ..., calculating the alpha _9, then the parameter alpha ₁ of advection equation, ..., an alpha ₉ by substituting the above equation (2), each observation The advection vector (u, v) and the development decay term w for each grid are calculated.

Finally, based on the time series rain data obtained at the first time (at the latest time), the advection vector (u, v) calculated as described above, and the development decay term w, each observation grid is obtained. The future rainfall intensity is calculated, and the calculated data is combined to create prediction data of future rain area movement.

In the advection model method, for example,
"Development of Real-time Rainfall and Flow Rate Prediction System Using Digital Radar Data" (Takuma Takao, Grants-in-Aid for Scientific Research (1987) (1)) Research Report (Showa 6)
March 3))).

[0016]

In such a short-term rainfall prediction, a change in the rainfall area which cannot be explained by the above-described prediction method may occur due to a physical factor such as a topographical factor. . For example, there are sudden local rainfall occurrence phenomenon and rainfall area disappearance phenomenon. In addition, there is a local development / decay phenomenon that is difficult to explain by movement prediction using an advection vector.

Conventionally, with respect to such unexplained changes in the rainfall area, a person extracts the cause, analyzes and analyzes the change, and performs a correction calculation based on his / her experience / intuition. The predicted rainfall intensity was corrected by determining various parameters. For this reason, it is unreliable, and a person who is well versed in rainfall prediction has to perform troublesome work. Further, with regard to the above-mentioned rainfall generation / extinction phenomena, in particular, it is impossible to express rainfall generation by the advection model method, and sudden rainfall disappearance cannot be expressed even for rainfall extinction (rainfall extinction due to reduction can be expressed).

An object of the present invention is a method of correcting a predicted rainfall value in a short-time rainfall prediction, and automatically corrects a predicted rainfall intensity corresponding to an unexplained change in a rainfall area (occurrence, disappearance, development, attenuation). It is an object of the present invention to provide a rainfall prediction / correction method, a device and a storage medium which can be performed efficiently.

[0019]

According to a first rainfall prediction and correction method of the present invention, an advection vector is identified using rainfall intensity data adjacent in time series, and a future advancing grid is determined based on the advection vector. A method for correcting the prediction result in a rainfall intensity prediction method for predicting rainfall intensity, wherein each rainfall intensity data stored in a time series of the determination target grid,
Based on the rainfall intensity data of a certain grid associated with each advection vector corresponding to each rainfall intensity data of the determination target grid, it is determined whether the target grid is a rainfall occurrence area or a rain disappearance area, For a grid determined to be a rainfall occurrence area or a rainfall annihilation area, a correction parameter is calculated by a predetermined method according to the determination result, and the calculated rainfall area is used as a rainfall occurrence area or a rainfall annihilation area. The prediction result of the rainfall intensity is corrected for each of the grids determined as above.

According to the first rainfall prediction correction method, it is possible to automatically correct the predicted rainfall intensity corresponding to the unexplainable rainfall area change phenomenon of occurrence / disappearance, without the need for manual labor. It is possible to perform highly accurate rainfall prediction corresponding to the unique phenomenon.

A second rainfall prediction correction method according to the present invention is as follows.
A method of identifying the advection vector using the rainfall intensity data adjacent in time series, and correcting the prediction result in the rainfall intensity prediction method of predicting the rainfall intensity of each future grid based on the advection vector, Based on the rainfall intensity data accumulated in time series of the determination target grid and the rainfall intensity data of a certain grid related by each advection vector corresponding to each rainfall intensity data of the determination target grid, Determine whether the grid is a rainfall development zone or a rain attenuation zone, for the grid determined to be a rain development zone or a rain attenuation zone, calculate a correction parameter by a predetermined method according to the determination result, Using the calculated correction parameters, the prediction result of the rainfall intensity is corrected for each of the grids determined to be in the rainfall development area or the rain attenuation area.

According to the second rainfall prediction correction method, it is possible to automatically correct the predicted rainfall intensity corresponding to the unexplainable rainfall area change phenomenon of development and attenuation, without the need for manual labor. It is possible to perform highly accurate rainfall prediction corresponding to the unique phenomenon.

[0023]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a functional block diagram showing an example of the configuration of the rainfall prediction operation device according to the first embodiment.

First, the observation grid points (grids) will be described. As shown in FIG. 3, for example, a certain observation region in the sky includes a plurality of grids (for example, 2.5 km ×
(2.5 km mesh). In this method, the horizontal direction of the observation area is represented by xy coordinates, and the plane is divided into grids at substantially equal intervals. Then, for example, when the position of a certain grid is i-th in the x direction and j-th in the y direction with reference to the lower left in the drawing of the observation region, the grid (i, j) is represented by the coordinates (i, j). Call. In the description of the present embodiment, as shown in the figure, “movement source grid point”, which will be described later, is referred to as a grid point (i ′, j ′).
The “destination grid point” is expressed as a grid point (i, j).

In FIG. 1, the rainfall data input unit 11
As shown in FIG. 3 and the like, rainfall intensity data obtained at regular time intervals (Δt described later) is input to each grid obtained by dividing the observation area in the sky as shown in FIG. 3 and the like (for example, observation data by a rainfall radar from a meteorological service support center or the like). Is transmitted at regular time intervals), and passes the input rainfall intensity data to the rainfall data storage unit 12 and the advection vector identification calculation unit 13.

The rainfall data storage unit 12 stores and accumulates rainfall intensity data input from the rainfall data input unit 11.
This means that, for example, continuous time-series rainfall intensity data is stored for each grid.

The advection vector identification calculation unit 13 receives the current time (for example, t ₀ ) input from the rainfall data input unit 11.
Of the advection vector (representing the moving speed vector of the rainfall area for each grid point) based on the rainfall intensity data of the above and the rainfall intensity data of the previous time (t ₀ −Δt) stored in the rainfall data storage unit 12. Perform an identification operation.

The advection vector storage unit 14 stores the advection vector calculated by the advection vector identification calculation unit 13.
accumulate.

The rainfall movement prediction calculation unit 15 calculates the advection vector (this time) calculated by the advection vector identification calculation unit 13.
The rainfall prediction calculation is performed based on the rainfall intensity data (this time) input from the rainfall data input unit 11.

Regarding the rainfall data input section 11 to rainfall movement prediction calculation section 15, substantially the same functions are described in detail in a prior application filed by the present applicant (Japanese Patent Application No. 9-265679). Therefore, a detailed description will not be given here.

In FIG. 1, the following are the main features of the rainfall prediction calculation device of the present embodiment, that is, the occurrence / disappearance determination unit 16, the occurrence / disappearance correction parameter calculation unit 17, and the rainfall movement prediction occurrence / deletion correction calculation. The unit 18 will be described in detail.

First, the occurrence / disappearance determination unit 16 will be described with reference to FIG.

The occurrence / disappearance determination unit 16 determines the following for each grid based on the time-series rainfall intensity data stored in the rainfall data storage unit 12 and the advection vector stored in the advection vector storage unit 14. Is determined by the method described in (1). This occurrence / disappearance is a kind of the unexplained phenomenon described above.

First, for example, a time zone for which occurrence / disappearance is to be determined (for example, several hours during which rainfall conditions are uniform)
Is set by an operator or the like, for example. This time period is, for example, (start time t = t ₀ −NΔt) to (end time t = t ₀ ), and the end time t ₀ and the data input sampling period (constant time interval) Δ from the rainfall radar.
It is set using t and N (the number of sampled data in the time zone; the number of data observed in the time zone). The above-mentioned time zone is defined as a time zone in which the rainfall condition is uniform. In the present invention, even if the entire rainfall condition is uniform, occurrence, development, attenuation, and disappearance in a part thereof cannot be explained. It is shown as an example to clarify that it corresponds to the occurrence of a peculiar phenomenon, and the rainfall state does not necessarily have to be uniform.

The occurrence / disappearance determining unit 16 first selects one grid to be determined (hereinafter, referred to as a target grid) (step S1), and this grid is stored in the rainfall data storage unit 12. Each rainfall intensity data is extracted, and each advection vector corresponding to each rainfall intensity data is extracted from the advection vector storage unit 14 (step S2).

FIG. 2 is a diagram showing the relationship between rainfall intensity data and advection vectors.

FIG. 3 is a conceptual diagram of a model in which an observation region is divided into a plurality of grids, and is a diagram showing a positional relationship between rainfall intensities associated with advection vectors.

FIG. 2 shows that each rainfall intensity data is associated with each advection vector. For example, the advection vector at time t = t ₀ −2Δt is expressed as
Rain intensity data of t ₀ -2Δt and time t = t ₀ -3Δ
t rainfall intensity data. This is
It shows the relationship between each advection vector and the rainfall intensity data used at the time of identifying the advection vector, and does not mean a time-series relationship in a certain grid. Normally, if the rain area is moving, the two rainfall intensity data are different grid data.
Therefore, the relationship between (two) rainfall intensity data items associated with such a certain advection vector means a correspondence relationship between a “source grid point” and a “destination grid point” described later. Also.

For example, in FIG. 3, the grid point (i, j) shown in FIG.
j ′) is associated with the advection vector. This is because the (two) rainfall intensity data associated with the advection vector is at time t = t ₀ − (n−
1) Rain intensity data at grid point (i, j) at Δt (n; any integer from 1 to N) and time t = t ₀ −nΔ
This means that the data is the rainfall intensity data of the grid point (i ′, j ′) at t.

If the advection vector is accurate, the rain state of the grid point (i ′, j ′) and the grid point (i, j,
The rain condition of j) should be almost the same (although it may fluctuate somewhat), but in reality, an unexplained phenomenon occurs, for example, one of which has rainfall and the other has no rainfall. ing. The first embodiment provides a correction method for coping with such a phenomenon.

From the above description, to explain the processing in step S2, the grid to be determined is set to the grid point (i,
j), and the respective time-series rainfall intensity data and the advection vectors respectively corresponding to the rainfall intensity data are extracted.

Next, regarding each of these advection vectors,
Another related rainfall intensity data is extracted. For example,
Using the examples of FIGS. 2 and 3 described above,
Time t = t₀-2t corresponding to rainfall intensity data
The advection vector is calculated at time t = t ₀Advection vector of -2Δt
And another rainfall intensity corresponding to this advection vector
Time t = t as degree data₀The above lattice point of -3Δt
The rainfall intensity data of (i ′, j ′) is extracted. this is,
In other words, the judgment target grid is fixed as the destination grid.
And each time t = t₀− (N−1) Δt (n; 1 to N)
For each of the integers of
The rainfall intensity data of the child (i ', j') is shown in FIG.
Using the correspondence indicated by the advection vector
And extracting from the rainfall data storage unit 12.
(Step S3).

Based on the various data extracted from the rainfall data storage unit 12 and the advection vector storage unit 14 as described above, the occurrence / disappearance determination unit 16 first determines the first
, The number N1 of data that satisfies each of the third to third conditions,
Find N2 and N3.

That is, in the above data (n; 1 to N), regarding the rainfall intensity Z (i, j, t), Z (i, j, t0− (n−1) Δt) ≠ 0 and Z (I ′, j ′, t0−nΔt) = 0 Data that satisfies the first condition of
Let it be 1. In other words, the number N1 of data in which the “source grid point” is “no rainfall” and the “destination grid point” is “rainfall” is obtained (step S4).

In the data of (n; 1 to N), regarding the rainfall intensity Z (i, j, t), Z (i, j, t0− (n−1) Δt) = 0 and Z (i ', J', t0−nΔt) ≠ 0, and detects data satisfying the second condition.
Let it be 2. In other words, the number N2 of data in which the “source grid point” is “with rain” and the “destination grid point” is “without rain” is obtained (step S5).

In the data of (n; 1 to N), regarding the rainfall intensity Z (i, j, t), Z (i, j, t0− (n−1) Δt) ≠ 0 and Z (i ', J', t0−nΔt) ≠ 0, and detects the data satisfying the third condition.
3 (step S6). In other words, the number N3 of data in which both the “source grid point” and the “destination grid point” are “with rain” is obtained.

Next, N1, N2, N
3 is used to calculate A = N1 / (N1 + N2 + N3) B = N2 / (N1 + N2 + N3) (step S7).

Then, C = AB (step S
8), α is set as a threshold (α ≧ 0), and if C> α (step S9, YES), it is determined that the area is a rainfall occurrence area (step S10).

If C <-α (step S11), it is determined that the area is within the rainfall disappearance area (step S12).

As described above, the occurrence / disappearance determination unit 16
Each rainfall intensity data (observation data) stored in time series with respect to the target grid and rainfall intensity data associated with the advection vector corresponding to each of the rainfall intensity data (a grid (corresponding to the target grid ( Rainfall intensity data Δt before the “grid point of the movement source”) and
By detecting how often (probability) the phenomenon considered to be “occurring” or “disappearing” has occurred in relation to the target grid, if any of the phenomena tends to occur more than a certain degree. If it is conceivable, the position of the target grid is determined to be a rainfall occurrence area or a rainfall disappearance area. In other words, the position of the target grid is located at a position where a change in the rainfall area (a sudden local rainfall generation phenomenon, a rainfall disappearance phenomenon, or the like) that cannot be explained due to geographical or topographical factors is likely to occur. (The target grid has a characteristic of “occurrence” or a characteristic of “disappearance”).

For the threshold α, for example, a person who is involved in rainfall prediction is set to a value that is considered appropriate for making the above determination in consideration of these situations. The threshold value is determined not only by α but also, for example, α ′ (≧ 0). If C> α, it is determined that the area is a rainfall occurrence area, and if C <−α ′, the area is a rainfall extinction area. You may.

Then, the above determination is sequentially performed for all the lattices (step S13).

As described above, the occurrence / disappearance determination unit 16
Determines, for each grid, whether it is a rainfall occurrence area or a rainfall extinction area, and passes the determination result to the generation / extinction correction parameter calculation unit 17.

The generation / destruction correction parameter calculation unit 17
In accordance with the determination result received from the occurrence / disappearance determination unit 16, for each grid, for example, the processing shown in FIG. 5 is executed to determine correction parameters (correction coefficients and correction constants).

That is, when the occurrence / disappearance determination unit 16 determines that the grid is a rainfall occurrence area (step S21, YES), the N1 data that satisfies the first condition is subjected to the rainfall. Strength Z (i, j, t
0− (n−1) Δt) are sequentially added, and an average rainfall intensity is calculated by dividing the result of the addition (sum of rainfall intensity for data satisfying the first condition) by N1 ( Step S22). Then, the calculated average rainfall intensity is used as a correction constant (step S23).

On the other hand, when the occurrence / extinction determining section 16 determines that the grid is a rainfall disappearance area (step S24), the correction constant is set to 0 (step S2).
5).

In the first embodiment, since only occurrence / disappearance is observed, the correction coefficient is set to "0".

As described above, when the occurrence / extinction correction parameter calculation section 17 determines that the area is a rainfall occurrence area, the correction coefficient = 0 and the correction constant is equal to the average rainfall intensity. In this case, the correction coefficient = 0 and the correction constant = 0 are determined as the correction parameters.

The rainfall movement prediction occurrence / destruction correction calculation unit 18
Is calculated by the following formula based on the correction parameter calculated by the occurrence / destruction correction parameter calculation unit 17 and the rainfall prediction calculation result (predicted rainfall intensity) calculated by the rainfall movement prediction calculation unit 15. Correct rainfall intensity.

(Corrected rainfall intensity) = (correction coefficient) × (predicted rainfall intensity) + (correction constant) That is, for a grid determined to be a rainfall extinction area, the predicted rainfall intensity is forcibly set to ' Correct to 0 '(no rainfall). On the other hand, for a grid determined to be a rainfall occurrence area, an average value of the rainfall intensity when a rainfall occurrence phenomenon is considered to have occurred in the past in the past is used as a predicted value of the rainfall intensity of the grid.

As described above, in the first embodiment, the correction method corresponding to the occurrence / disappearance phenomenon, which is a kind of unexplainable phenomenon, has been described. However, in the second embodiment described below, another explanation will be given. Provide a correction method corresponding to a peculiar development / decay phenomenon which is an impossible phenomenon.

FIG. 6 is a functional block diagram of a rainfall prediction operation device having a rainfall prediction correction method according to the second embodiment.

In the configuration shown in FIG. 1, the rainfall data input unit 11, the rainfall data storage unit 12, the advection vector identification calculation unit 13, the advection vector storage unit 14, and the rainfall movement prediction calculation unit 15 are shown in FIG. The configuration is substantially the same as that shown in the drawings, and the same reference numerals are given, and detailed description thereof will be omitted.

The development / decay judging section 21 performs, for each grid, based on the time-series rainfall intensity data stored in the rainfall data storage section 12 and the advection vector stored in the advection vector storage section 14, for example. By executing the processing shown in FIG.
Judgment of development / decay.

First, the occurrence / disappearance determination unit 16
In a similar manner to the case of, for example, a time zone to be subjected to the development / decay determination (for example, several hours in which the rainfall condition is substantially uniform) is set in advance, for example. Then, a determination process described below is performed for each target grid (step S3).
1).

First, each rainfall intensity data of the target grid within the set time zone is stored in the rainfall data storage unit 12.
The advection vector data corresponding to each rainfall intensity data is further extracted from the advection vector storage unit 14 (step S32). Then (using the correspondence indicated by the advection vector as shown in FIG. 2 above),
Another rainfall intensity data corresponding to each of these advection vectors is extracted from the rainfall data storage unit 12. In other words, the determination target grid is fixed as the movement destination grid, and each time t = t ₀ − (n−1) Δt (n; 1) in the above time zone.
(The arbitrary integer from to N) to extract the rainfall intensity data of the corresponding movement source grid (step S3)
3).

Next, data satisfying the third condition is detected from the extracted data (step S3).
4). Then, for each of the detected data, a ratio (rainfall intensity ratio R) of the rainfall intensity between the source and destination grids is calculated. Then, an average rain intensity ratio is calculated by adding the calculated rain intensity ratios R and dividing the addition result (sum of the rain intensity ratios R for data satisfying the third condition) by N3. That is, the average rain intensity ratio based on the target grid (the destination grid (i, j)) is obtained by the following equation (step S35).

[0069]

(Equation 2)

(Where m is data that satisfies the third condition) Then, when β (β> 1) and γ (1 ≧ γ> 0) are set as threshold values, and the average rainfall intensity ratio> β (step S36,
YES), it is determined that the rainfall intensity tends to develop at the position of the target grid point (step S37).

If the average rainfall intensity ratio <γ (step S38, YES), it is determined that the rainfall intensity tends to attenuate at the position of the target grid point (step S3).
9).

That is, first, in the conventionally known advection model method, it is basically considered that the rain state of the source grid shifts to the destination grid (rain intensity ratio = 1). Although based on concepts, rainfall conditions usually vary somewhat in practice. However, even if it changes, it is generally considered that the value is close to '1', that is, there is no large change. On the other hand, in the local development / decay phenomena to be corrected in the second embodiment, it is considered that relatively large fluctuations occur, and the fluctuation rate is high on average in the past with respect to the target grid. Humans or the like set appropriate values that can be regarded as the thresholds β and γ. If it is determined that the value exceeds the range indicated by the two threshold values β and γ based on the rainfall intensity ratio = 1, the position of the target grid cannot be explained due to geographical or topographical factors. In a location where a large change in the rainfall area (unique development and attenuation phenomena) is likely to occur (the tendency to develop (characteristics) or the tendency to attenuate (characteristics)
There is).

Then, the above determination is sequentially performed for all the lattices (step S40).

FIG. 8 is a flowchart for explaining an example of the processing in the development / decay correction parameter calculation section 22.

The development / decay correction parameter calculation unit 22
For the grids determined to have “development” or “decay” in accordance with the determination result received from the development / decay determination unit 21 (step S41, YES), the correction parameters (correction coefficient, correction constant, ) Is determined (step S42).

Correction coefficient = average rainfall intensity ratio Correction constant = 0 Rainfall movement prediction development / attenuation correction calculation unit 23
Based on the correction parameter determined by the attenuation correction parameter calculator 22 and the rainfall prediction calculation result (predicted rainfall intensity) obtained by the rainfall movement prediction calculator 15, the predicted rainfall intensity is corrected by the following formula. .

(Corrected rainfall intensity) = (correction coefficient) × (predicted rainfall intensity) + (correction constant) As described above, in the rainfall prediction correction method of the second embodiment, the movement prediction using the advection vector will be described. Correction can be performed by estimating a grid that tends to cause difficult unique development / decay phenomena.

In the above description of the embodiment, the configuration of the rainfall prediction operation device has been described by dividing it into the configuration of the first embodiment and the configuration of the second embodiment. However, the present invention is not limited to this. Alternatively, both may be mixed. In this case, for example, in the occurrence / disappearance determination unit 16, step S1 in FIG.
After the processing of Step S6, the information relating to the data that satisfies the third condition is passed to the development / decay determination unit 21, and the development / decay determination unit 21 receives the information and performs the processing from Step S35 onward in FIG. May be executed.

The rainfall prediction correction method according to the present invention is realized on an information processing device such as a workstation. For example, a program in which at least processing for realizing the above-described determination and correction method is described in an arbitrary programming language is stored in a storage unit or a portable storage medium described later, and the program is executed as necessary. It is realized by reading and executing.

FIG. 9 is a configuration block diagram showing an example of an information processing device for realizing the rainfall prediction operation device of the present embodiment.

In the figure, the information processing device 30 has a CP
U31, storage unit 32 (including portable storage medium 32a), memory 33, input / output unit 34, and the like.

The CPU 31 is a central processing unit that controls the entire information processing device 30.

The storage unit 32 is a storage device such as an HDD in which at least a program for implementing the above-described rainfall prediction and correction method according to the present invention is stored. Or storage unit 3
Reference numeral 2 may be a combination of a portable storage medium 32a and its drive / reading device (for example, a floppy disk (FD) and a floppy disk drive (FDD)). The portable storage medium 32a includes a CD-ROM, a memory card, a DVD, an MO, and the like in addition to the FD.

The memory 33 temporarily stores a program stored in the storage unit 32 and stores the program in the CPU.
And a RAM to be executed by the CPU 31. The memory 33
There is also a storage area for temporarily storing various data generated while the CPU 31 is executing the program.

The input / output unit 34 is for inputting, for example, observation data by the rainfall radar sent from the meteorological service support center or the like, and outputting the corrected predicted values and the like to the outside if necessary. It is an input / output interface.

Although not shown, an output unit such as a display and a printer, and an input unit such as a keyboard and a mouse may be provided. For example, input of a set value and an instruction of a correction process are performed from an input unit, and a rainfall prediction result after correction is graphically displayed on the output unit.

The present invention is not limited to the above-described rainfall prediction correction method and its apparatus (the above-mentioned information processing apparatus and the like), but a computer-readable recording medium (storage medium) storing the above-mentioned program. The form of itself is also included.

The “recording medium” is, for example, the storage unit 32
(Portable storage medium 32a). Alternatively, a storage unit (not shown) that stores the above program in an external device is also included. That is, it is only necessary that the program can be read and executed, and thus the location (medium) for storing / storing the program is not particularly limited. For example, even if the program is stored in an external device, the program is stored in a storage unit (not shown) in the external device via a communication interface, a public line, a dedicated line, or the like. If loaded into the information processing device 30 and executed, the rainfall prediction correction method and device according to the present invention can be realized.

[0089]

As described above in detail, according to the rainfall prediction correction method and the apparatus thereof according to the present invention, the occurrence and occurrence of unusual phenomena which cannot be explained by the conventional advection model method.
By estimating a grid that is considered to have characteristics that cause annihilation / development / attenuation phenomena, and correcting the rainfall prediction result on such a grid according to the characteristics, it is possible to contribute to improvement of prediction accuracy.

[Brief description of the drawings]

FIG. 1 is a functional block diagram illustrating an example of a configuration of a rainfall prediction operation device according to a first embodiment.

FIG. 2 is a diagram showing a relationship between rainfall intensity data and advection vectors.

FIG. 3 is a diagram illustrating a positional relationship between rainfall intensities associated with advection vectors in a model conceptual diagram in which an observation region is divided into a plurality of grids.

FIG. 4 is a flowchart illustrating an example of a process performed by an occurrence / disappearance determination unit illustrated in FIG. 1;

FIG. 5 is a flowchart illustrating an example of a process performed by an occurrence / destruction correction parameter calculation unit illustrated in FIG. 1;

FIG. 6 is a functional block diagram illustrating an example of a configuration of a rainfall prediction operation device according to a second embodiment.

FIG. 7 is a flowchart illustrating an example of processing of a development / decay determination unit illustrated in FIG. 6;

FIG. 8 is a flowchart illustrating an example of a process of a development / decay correction parameter calculation unit illustrated in FIG. 6;

FIG. 9 is a configuration block diagram illustrating an example of an information processing device that realizes the rainfall prediction operation device of the present embodiment.

[Explanation of symbols]

 Reference Signs List 11 rainfall data input unit 12 rainfall data storage unit 13 advection vector identification calculation unit 14 advection vector storage unit 15 rainfall movement prediction calculation unit 16 generation / disappearance determination unit 17 generation / disappearance correction parameter calculation unit 18 rainfall movement prediction generation / disappearance correction calculation Unit 21 Development / Attenuation Determination Unit 22 Development / Attenuation Correction Parameter Operation Unit 23 Rainfall Movement Prediction Development / Attenuation Correction Operation Unit 30 Information Processing Device 31 CPU 32 Storage Unit 32a Portable Storage Medium 33 Memory 34 Input / Output Unit

 ──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Takashi Aoki 1-1, Tanabe-Nitta, Kawasaki-ku, Kawasaki-shi, Kanagawa Prefecture Inside Fuji Electric Co., Ltd. (72) Inventor Akihiro Miyamoto 1-Fuji-cho, Hino-shi, Tokyo Within FFC

Claims (6)

[Claims] An advection vector is identified using rainfall intensity data adjacent in time series, and the prediction result in a rainfall intensity prediction method for predicting future rainfall intensity of each grid based on the advection vector is corrected. A method, wherein each of the rainfall intensity data accumulated in time series of the determination target grid and the rainfall intensity data of a certain grid related by each advection vector corresponding to each rainfall intensity data of the determination target grid. It is determined whether the target grid is a rain occurrence area or a rain extinction area based on the target grid, and the grid determined to be a rain generation area or a rain extinction area is corrected by a predetermined method according to the determination result. Parameters, and using the calculated correction parameters, supplement the prediction result of the rainfall intensity for each of the grids determined to be the rainfall occurrence area or the rainfall extinction area. A rainfall prediction correction method characterized by correcting. 2. A method of determining whether the target grid is a rainfall occurrence area or a rainfall annihilation area includes determining the target grid as a destination grid, a grid associated with the advection vector as a source grid, and Based on the probability that the rain state of the source grid is no rain and the rain state of the destination grid is rain, and the probability that the rain state of the source grid is rain and the rain state of the destination grid is no rain. The rainfall prediction correction method according to claim 1, wherein the determination is performed. 3. An advection vector is identified using rainfall intensity data adjacent in time series, and the prediction result in a rainfall intensity prediction method for predicting future rainfall intensity of each grid based on the advection vector is corrected. A method, wherein each of the rainfall intensity data accumulated in time series of the determination target grid and the rainfall intensity data of a certain grid related by each advection vector corresponding to each rainfall intensity data of the determination target grid. Based on the determination, it is determined whether the target grid is a rain development zone or a rain attenuation zone, and the grid determined to be a rain development zone or a rain attenuation zone is corrected by a predetermined method according to the determination result. Parameters, and using the calculated correction parameters, for the grids determined to be in the rainfall development area or the rain attenuation area, supplement the prediction result of the rainfall intensity for each grid. A rainfall prediction correction method characterized by correcting. 4. A determination as to whether the target grid is a rainfall development zone or a rain attenuation zone is performed by setting the target grid as a destination grid, a grid associated with the advection vector as a source grid, and If the variation rate of the rainfall intensity obtained as the ratio between the rainfall intensity of the source grid and the rainfall intensity of the destination grid exceeds a preset range, it is determined that the target grid is a rainfall development area or a rain attenuation area. The rainfall prediction correction method according to claim 3, wherein: 5. An advection vector for each grid is identified based on rainfall intensity data obtained at predetermined intervals for each grid obtained by dividing the observation region, and a future rainfall intensity of each grid is determined based on the advection vector. In the predicting device, for each grid, from the accumulated rainfall intensity and time-series data of the advection vector, to detect the actual rainfall state change state between the related grids associated with each advection vector, A characteristic determining means for determining whether or not there is any characteristic of the occurrence or disappearance or development or attenuation of rain in each of the grids based on the detection result; and the occurrence or disappearance or development or attenuation by the characteristic determination means. Parameter calculation for calculating a parameter by a predetermined calculation method according to the characteristic for a grid determined to have any of the characteristics And stage, using the parameters calculated by the parameter calculating means, and correcting means for correcting the rainfall intensity prediction of the corresponding lattice, rainfall prediction apparatus characterized by having a. 6. When used in a computer, for each grid obtained by dividing the observation area, the accumulated rainfall intensity and the time series data of the advection vector are used to determine the inter-related grids associated with each advection vector. A function of detecting a change state of an actual rainfall intensity state, and determining, based on the detection result, whether or not each of the grids has any characteristic of generation or extinction or development or attenuation of rainfall; A function of calculating a parameter by a predetermined calculation method according to the characteristic for a grid determined to have any characteristic of disappearance or development or attenuation, and using the calculated parameter to A computer-readable recording medium storing a program for realizing: a function of correcting a rainfall intensity prediction of a grid;