JP6709941B2 - Precipitation forecasting apparatus and precipitation forecasting method - Google Patents

Description

The present invention relates to a precipitation prediction device and a precipitation prediction method for predicting precipitation.

Conventionally, by preparing a predicted rainfall amount by a kinematic method, a predicted rainfall amount by a physical method, and a different weight function for each predicted rainfall amount, and selecting and using the weight function, A precipitation amount prediction method for accurately predicting precipitation from 10 hours to 10 hours is disclosed (see Patent Document 1).

Japanese Patent No. 3851641

Since the precipitation prediction method of Patent Document 1 mainly targets a phenomenon in which the spatial scale is larger than the cumulonimbus and the time scale is long, the spatiotemporal fluctuation of the rapidly developing cumulonimbus cannot be captured. Moreover, the precipitation amount prediction method of Patent Document 1 is for predicting the amount of precipitation, and cannot provide the probability of occurrence of precipitation.

The present invention provides a precipitation prediction device and a precipitation prediction method capable of predicting the probability of occurrence of heavy rain in addition to the prediction of precipitation for rapidly developing cumulonimbus, as compared with the prior art. To aim.

The precipitation prediction apparatus according to the present invention,
An observation data acquisition unit that acquires at least precipitation observation data;
A numerical value that divides the observation area into grids and uses the observation data acquired from the observation data acquisition unit to calculate the temporal change of the precipitation situation by computer using physical laws and predict the future precipitation situation for each grid A prediction section,
The observation area is divided into grids, the movement of past precipitation areas and the distribution of present precipitation are obtained using the observation data acquired from the observation data acquisition unit, and the distribution of precipitation after a short time is predicted for each grid. With the nowcast prediction unit,
For the numerical prediction data obtained from the numerical prediction unit and the Nowcast prediction data obtained from the Nowcast prediction unit, the number of misalignment grids indicating a predetermined area is determined, and the correct answer rate is calculated for each number of misalignment grids. Correct rate calculation part
In consideration of the accuracy rate obtained by the accuracy rate calculation unit, the optimal synthesis coefficient is estimated in real time, and a synthesis unit that synthesizes the latest numerical prediction data and nowcast prediction data using the synthesis coefficient,
It is characterized by including.

・・・（１）
ただし、
O(n)(i,j) は、格子(i,j)での観測値のFraction、
M(n)(i,j) は、格子(i,j)での予測値のFraction、
Fractionは、n×n領域に対して降水量が予め定めた所定の値以上となる格子の割合、
nは、Fractionを計算する正方形の１辺の格子数、
Nは、 評価領域の全格子数、
Σ_iΣ_jは、評価領域内の和
である。 In the precipitation prediction device according to the present invention,
The correctness rate calculation unit is characterized by calculating the correctness rate for each misalignment grid number using the following equation (1).

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the lattice where the precipitation amount is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area.

The precipitation prediction apparatus according to the present invention,
It is characterized by further comprising a position shift lattice number input unit capable of inputting the position shift lattice number.

In the precipitation prediction device according to the present invention,
The synthesizing unit synthesizes the latest numerical prediction data and nowcast prediction data by using the following equation (2).
R(n,t)=C(n,t)×R1(n,t)+(1−C(n,t))×R2(n,t) (2)
However,
n is the number of grids on one side of the square for calculating Fraction,
t is time,
R1(n,t) is nowcast prediction data after t hours with the number of grids n on one side,
R2(n,t) is the numerical prediction data after t hours when the number of grids on one side is n,
C(n,t) is a composite coefficient (a function of n and t, after t time) of the nowcast prediction data R1(n,t) with respect to the numerical prediction data R2(n,t) at the number n of grids on one side. 0≦C(n,t)≦1),
Is.

The precipitation prediction method according to the present invention is
Acquiring at least precipitation observation data;
Dividing the observation region into a grid, using the observation data, a step of predicting numerical prediction data indicating future precipitation conditions for each grid by calculating a time change of the precipitation condition by a computer using a physical law,
The observation area is divided into grids, the movement of the past precipitation area and the distribution of the present precipitation are obtained using the observation data, and nowcast prediction data showing the distribution of precipitation after a short time is predicted for each grid. Steps,
With respect to the numerical prediction data and the nowcast prediction data, determining the number of misalignment grids indicating a predetermined area, and calculating a correct answer rate for each number of misalignment grids,
Considering the correct answer rate, estimating the optimal synthesis coefficient in real time, and synthesizing the latest numerical prediction data and nowcast prediction data using the synthesis coefficient,
It is characterized by having.

・・・（１）
ただし、
O(n)(i,j) は、格子(i,j)での観測値のFraction、
M(n)(i,j) は、格子(i,j)での予測値のFraction、
Fractionは、n×n領域に対して降水量が予め定めた所定の値以上となる格子の割合、
nは、Fractionを計算する正方形の１辺の格子数、
Nは、 評価領域の全格子数、
Σ_iΣ_jは、評価領域内の和
である。 In the precipitation prediction method according to the present invention,
The step of calculating the correct answer rate for each misaligned grid number is characterized by using the following equation (1).

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the lattice where the precipitation amount is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area.

The precipitation prediction method according to the present invention is
The method further comprises the step of inputting the number of misaligned grids.

In the precipitation prediction method according to the present invention,
The step of synthesizing the latest numerical prediction data and nowcast prediction data is characterized by using the following equation (2).
R(n,t)=C(n,t)×R1(n,t)+(1−C(n,t))×R2(n,t) (2)
However,
n is the number of grids on one side of the square for calculating Fraction,
t is time,
R1(n,t) is nowcast prediction data after t hours with the number of grids n on one side,
R2(n,t) is the numerical prediction data after t hours when the number of grids on one side is n,
C(n,t) is a composite coefficient (a function of n and t, after t time) of the nowcast prediction data R1(n,t) with respect to the numerical prediction data R2(n,t) with the number n of grids on one side. 0≦C(n,t)≦1),
Is.

According to such a precipitation predicting apparatus and a precipitation predicting method, it is possible to predict the probability of occurrence of heavy rain in addition to predicting the amount of precipitation for a rapidly developing cumulonimbus cloud.

The system block diagram of the precipitation prediction apparatus of this embodiment is shown. The ratio of the predicted correct answer rate by numerical prediction and the predicted correct answer rate by Nowcast prediction to the predicted time is shown. A comparison of observed data and predicted data for each grid is shown. The relationship between the accuracy rate and the prediction time is shown. The flowchart of the precipitation prediction method of this embodiment is shown. The precipitation prediction data obtained by the precipitation prediction apparatus and precipitation prediction method of this embodiment are shown.

An embodiment according to the present invention will be described with reference to the drawings.

FIG. 1 shows a system block diagram of the precipitation prediction apparatus of this embodiment.

The precipitation prediction device 10 includes an observation data acquisition unit 11, a numerical prediction unit 12, a nowcast prediction unit 13, a correct answer rate calculation unit 14, and a synthesis unit 16.

The observation data acquisition unit 11 acquires observation data from ground observations, meteorological satellites, radars, etc., from meteorological organizations, space agencies, etc., in countries around the world. Since the observation is performed at various places and times, the observation data is acquired from each acquisition source at every predetermined time. Since some observation data have low accuracy and cannot be used due to human error or equipment failure, these data are excluded. The acquired observation data is output to the numerical prediction unit 12 and the nowcast prediction unit 13.

The numerical prediction unit 12 is a unit that divides the observation region into a grid and uses the observation data acquired from the observation data acquisition unit 11 to make a numerical prediction for each grid. Numerical forecasting predicts future precipitation by calculating the temporal change of precipitation using a physical law by a computer.

The nowcast prediction unit 13 divides the observation region into a grid pattern, uses the observation data acquired from the observation data acquisition unit 11, and uses the movement of the precipitation region in the past and the distribution of the present precipitation, This is the part that predicts the distribution of precipitation for each grid.

・・・（１）
ただし、
O(n)(i,j) は、格子(i,j)での観測値のFraction、
M(n)(i,j) は、格子(i,j)での予測値のFraction、
Fractionは、n×n領域に対して降水量が予め定めた所定の値以上となる格子の割合、
nは、Fractionを計算する正方形の１辺の格子数、
Nは、 評価領域の全格子数、
Σ_iΣ_jは、評価領域内の和
である。 The correct answer rate calculation unit 14 allows the positional deviation between the numerical prediction data obtained from the numerical prediction unit 12 and the nowcast prediction data obtained from the nowcast prediction unit 13 while allowing the positional deviation lattice number (allowing the positional deviation). The accuracy rate for each spatial scale) is calculated using the following equation (1).

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the lattice where the precipitation amount is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area.

The misalignment grid number input unit 15 appropriately inputs the necessary misalignment grid number. Increasing the number n of misaligned grids increases the accuracy of predicting the precipitation occurrence probability. However, the forecast that it will rain anywhere in a wide area is accurate but impractical. On the other hand, when the number of misalignment grids n is reduced, the accuracy of prediction of the probability of precipitation will decrease. In other words, it is difficult to pinpoint the forecast that it will rain tens of minutes after this point. Therefore, it is preferable that the number n of misaligned grids be appropriately determined according to the user's request.

The synthesizing unit 16 estimates the optimal synthesis coefficient in real time in consideration of the accuracy rate calculated by the accuracy rate calculating unit 14, and uses the synthesis coefficient to calculate the numerical prediction data and the nowcast prediction data by the following equation ( 2) is used for synthesis.
R(n,t)=C(n,t)×R1(n,t)+(1−C(n,t))×R2(n,t) (2)
However,
n is the number of grids on one side of the square for calculating Fraction,
t is time,
R1(n,t) is nowcast prediction data after t hours with the number of grids n on one side,
R2(n,t) is the numerical prediction data after t hours when the number of grids on one side is n,
C(n,t) is a composite coefficient (a function of n and t, after t time) of the nowcast prediction data R1(n,t) with respect to the numerical prediction data R2(n,t) at the number n of grids on one side. 0≦C(n,t)≦1),
Is.

The synthesis coefficient is determined in consideration of the correct answer rates of the numerical prediction data and the nowcast prediction data. The correct answer rate of the prediction data is calculated in real time by comparing with the true value in the past hour.

FIG. 2 shows the ratio of the predicted correct answer rate by numerical prediction to the predicted correct answer rate by nowcast prediction with respect to the predicted time.

As shown in FIG. 2, in the numerical prediction, the accuracy rate is low when the prediction time is short, and the accuracy rate is high when the prediction time is long. In the Nowcast prediction, the accuracy rate is high when the prediction time is short, and the accuracy rate is low when the prediction time is long. Therefore, when the prediction time is short, it is preferable to obtain a synthesis coefficient that emphasizes nowcast prediction and when the prediction time is long, emphasizes numerical prediction.

Further, when predicting localized heavy rainfall, prediction with high resolution with a grid of 1 km or less is required. When the prediction is performed with a high grid resolution, the accuracy of the prediction data for each grid is generally low. That is, the prediction with high grid resolution has a large error.

FIG. 3 shows a comparison between observed data and predicted data for each grid. FIG. 3A shows observed data, and FIG. 3B shows predicted data. The grid in the shaded area indicates the position of the amount of precipitation above a predetermined amount.

In the example shown in FIG. 3, when a grid scale of n=1 that does not allow positional deviation is selected, the prediction data is determined to be different from the observation data. However, if a grid scale of n=5 that allows a positional deviation of 5 grids is selected, the prediction data and the observation data both have 6 positions of the precipitation amount equal to or more than a predetermined amount. To be judged.

・・・（１）
ただし、
O(n)(i,j) は、格子(i,j)での観測値のFraction、
M(n)(i,j) は、格子(i,j)での予測値のFraction、
Fractionは、n×n領域に対して降水量が予め定めた所定の値以上となる格子の割合、
nは、Fractionを計算する正方形の１辺の格子数、
Nは、 評価領域の全格子数、
Σ_iΣ_jは、評価領域内の和
である。 In the present embodiment, the correct answer rate calculation unit 14 uses the following formula (1) to improve the prediction accuracy even at a high grid resolution by allowing a certain amount of positional deviation and maintaining the prediction accuracy. Then, the synthesizing unit 16 synthesizes the numerical prediction data and the nowcast prediction data using the result of the correct answer rate calculating unit 14.

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the lattice where the precipitation amount is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area.

Expression (1) shows the correct answer rate of Fraction when the number n of positional deviation lattices at the coordinate (i, j) to be evaluated is allowed. Fraction has O and M. FractionO is the precipitation occurrence probability when the positional deviation is allowed for the actually observed rain. Fraction M is a precipitation occurrence probability when the positional deviation is allowed with respect to the predicted rain.

The formula (1) is obtained for each position shift lattice number with respect to the numerical prediction data and the nowcast prediction data obtained in the past, and the correct answer rate of the prediction can be calculated. The formula (1) is calculated for each predicted lead time. Then, the synthesis coefficient is determined from the ratio of the correct answer rate for each lead time of the numerical prediction and the nowcast prediction. Therefore, the synthesis coefficient is determined by each position shift lattice number and each lead time.

That is, the correct answer rate calculation unit 14 calculates, for each lead time, a precipitation amount distribution and a precipitation occurrence probability distribution in which the numerical prediction data and the nowcast prediction data are allowed by the number n of misaligned grids. Then, the synthesizing unit 16 synthesizes the latest numerical prediction data and the nowcast prediction data based on the synthesis coefficient for the prediction for each lead time and the number n of misaligned grids, and the precipitation distribution and the precipitation occurrence probability distribution. Ask for.

As described above, the precipitation prediction device 10 of the present embodiment can obtain the precipitation occurrence probability distribution in addition to the precipitation prediction distribution.

Next, the relationship between the misaligned grid number n and the probability of occurrence of precipitation will be described.

Increasing the number of misaligned grids n increases the accuracy of prediction of precipitation occurrence probability. However, the forecast that it will rain anywhere in a wide area is accurate but impractical. On the other hand, when the number of misalignment grids n is reduced, the accuracy of prediction of the probability of precipitation will decrease. In other words, it is difficult to pinpoint the forecast that it will rain tens of minutes after this point. Therefore, it is preferable that the number n of misaligned grids be determined according to the user's request.

FIG. 4 shows the relationship between the accuracy rate and the prediction time. FIG. 4(a) shows the case where 50 mm of precipitation is predicted for one hour, and FIG. 4(b) shows the case where 20 mm of precipitation is predicted for one hour. L in the graph shown in FIG. 4 indicates a positional deviation allowable scale (km).

For example, when it is determined that the correct answer rate of prediction is 50%, it is preferable to use the data above the allowable line shown in FIG.

Next, a control flow of the precipitation prediction device 10 of the present embodiment will be described.

FIG. 5 shows a flowchart of the precipitation prediction method of this embodiment.

First, in step 1, past actual precipitation data is read (ST1). Subsequently, the past numerical prediction data and nowcast prediction data are read (ST2).

Next, in step 3, the number n of misaligned grids is determined (ST3). The number n of misaligned grids can be appropriately determined by the user.

Next, in step 4, the accuracy rate shown in the equation (1) is calculated (ST4). Then, in step 5, the composite coefficient of the numerical prediction data and the nowcast prediction data is calculated (ST5). At the time of calculating the correct answer rate and the synthesis coefficient, the precipitation data, the prediction data, the number n of misaligned grids, and the like obtained in steps 1 to 3 are used.

Next, in step 6, the latest numerical prediction data and nowcast prediction data are read (ST6). Subsequently, in step 7, the precipitation amount prediction and the precipitation probability prediction for each of the number of positional deviation grids in the latest numerical value prediction data and nowcast prediction data read in step 6 are obtained (ST7).

Next, in step 8, the prediction based on the numerical prediction data and the prediction based on the nowcast prediction data are combined (ST8). When synthesizing, the synthesizing coefficient calculated in step 5 is used to obtain from the equation (2). Further, it is preferable to consider the allowable line and the like shown in FIG.

Finally, in step 9, a composite prediction file that predicts the precipitation amount and precipitation probability for each misaligned grid number is output (ST9).

FIG. 6 shows precipitation prediction data obtained by the precipitation prediction apparatus and precipitation prediction method of this embodiment. FIG. 6A shows rainfall prediction data, and FIG. 6B shows precipitation occurrence probability data.

As shown in FIGS. 6A and 6B, the precipitation prediction apparatus and the precipitation prediction method according to the present embodiment can obtain precipitation occurrence probability data in addition to the precipitation amount prediction data. Further, in the precipitation occurrence probability data, it is possible to appropriately obtain the probability that more than a predetermined amount of rain will fall in a predetermined area by selecting the position shift grid number n. For example, in the example shown in FIG. 6B, the probability that 50 mm or more of rain will fall on a 5 km square around the periphery is obtained.

As described above, the precipitation prediction apparatus 10 of the present embodiment uses at least the observation data acquisition unit 11 that acquires observation data of precipitation and the observation region obtained by dividing the observation region into a grid shape and using the observation data acquired from the observation data acquisition unit. Numerical prediction unit 12 that predicts the future rainfall situation for each grid by calculating the temporal change of precipitation situation using the law, and observation data acquired from observation data acquisition unit 11 by dividing the observation region into a grid Using the, the nowcast prediction unit 13 that obtains the movement of the past precipitation area and the distribution of the present precipitation and predicts the distribution of the precipitation after a short time for each grid, and the numerical prediction data obtained from the numerical prediction unit 12. With respect to the nowcast prediction data obtained from the nowcast prediction unit 13, a correct answer rate calculation unit 14 that determines a positional deviation lattice number n indicating a predetermined area and calculates a correct answer rate for each positional deviation lattice number n; A synthesis unit 16 that estimates the optimal synthesis coefficient in real time in consideration of the correct answer rate obtained by the rate calculation unit 14 and synthesizes the latest numerical prediction data and nowcast prediction data using the synthesis coefficient is provided. .. Therefore, it is possible to predict the probability of heavy rainfall in addition to the prediction of precipitation for rapidly developing cumulonimbus clouds.

・・・（１）
ただし、
O(n)(i,j) は、格子(i,j)での観測値のFraction、
M(n)(i,j) は、格子(i,j)での予測値のFraction、
Fractionは、n×n領域に対して降水量が予め定めた所定の値以上となる格子の割合、
nは、Fractionを計算する正方形の１辺の格子数、
Nは、 評価領域の全格子数、
Σ_iΣ_jは、評価領域内の和
である。 Further, in the precipitation prediction device 10 of the present embodiment, the correct answer rate calculation unit 14 uses the following equation (1) to calculate the correct answer rate for each misalignment grid number n. Therefore, the prediction accuracy can be improved even with high resolution.

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the lattice where the precipitation amount is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area.

Further, the precipitation prediction device 10 of the present embodiment includes a position shift grid number input unit 15 capable of inputting the position shift grid number n. Therefore, it is possible to obtain the prediction data that meets the user's request.

In addition, in the precipitation prediction device 10 of the present embodiment, the synthesis unit 16 synthesizes the latest numerical prediction data and nowcast prediction data by using the following formula (2). Therefore, it is possible to obtain more accurate combined data.
R(n,t)=C(n,t)×R1(n,t)+(1−C(n,t))×R2(n,t) (2)
However,
n is the number of grids on one side of the square for calculating Fraction,
t is time,
R1(n,t) is nowcast prediction data after t hours with the number of grids n on one side,
R2(n,t) is the numerical prediction data after t hours when the number of grids on one side is n,
C(n,t) is a composite coefficient (a function of n and t, after t time) of the nowcast prediction data R1(n,t) with respect to the numerical prediction data R2(n,t) at the number n of grids on one side. 0≦C(n,t)≦1),
Is.

Further, the precipitation prediction method of the present embodiment includes at least a step of acquiring observation data of precipitation, dividing the observation region into a grid, and using the observation data, a temporal change of the precipitation situation by a computer using a physical law. The step of calculating and predicting the numerical prediction data indicating the future precipitation situation for each grid, dividing the observation area into grids, and using the observation data, obtaining the movement of the past precipitation area and the distribution of the present precipitation , A step of predicting the nowcast prediction data showing the distribution of precipitation after a short time for each grid, and determining the position shift grid number indicating a predetermined area for the numerical prediction data and the nowcast prediction data A step of calculating a correct answer rate for each number n of grids, a step of estimating an optimum synthesis coefficient in real time in consideration of the correct answer rate, and a step of synthesizing the latest numerical prediction data and nowcast prediction data using the synthesis coefficient And have. Therefore, it is possible to predict the probability of heavy rainfall in addition to the prediction of precipitation for rapidly developing cumulonimbus clouds.

・・・（１）
ただし、
O(n)(i,j) は、格子(i,j)での観測値のFraction、
M(n)(i,j) は、格子(i,j)での予測値のFraction、
Fractionは、n×n領域に対して降水量が予め定めた所定の値以上となる格子の割合、
nは、Fractionを計算する正方形の１辺の格子数、
Nは、 評価領域の全格子数、
Σ_iΣ_jは、評価領域内の和
である。 Further, in the precipitation prediction method of the present embodiment, the following equation (1) is used in the step of calculating the correct answer rate for each of the misaligned grid numbers n. Therefore, the prediction accuracy can be improved even with high resolution.

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the lattice where the precipitation amount is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area.

In addition, the precipitation prediction method of the present embodiment has a step of inputting the number n of positional deviation grids. Therefore, it is possible to obtain the prediction data that meets the user's request.

Further, in the precipitation prediction method of the present embodiment, the step of synthesizing the latest numerical prediction data and nowcast prediction data uses the following equation (2). Therefore, it is possible to obtain more accurate combined data.
R(n,t)=C(n,t)×R1(n,t)+(1−C(n,t))×R2(n,t) (2)
However,
n is the number of grids on one side of the square for calculating Fraction,
t is time,
R1(n,t) is nowcast prediction data after t hours with the number of grids n on one side,
R2(n,t) is the numerical prediction data after t hours when the number of grids on one side is n,
C(n,t) is a composite coefficient (a function of n and t, after t time) of the nowcast prediction data R1(n,t) with respect to the numerical prediction data R2(n,t) with the number n of grids on one side. 0≦C(n,t)≦1),
Is.

The present invention is not limited to this embodiment. That is, in the description of the embodiments, many specific details are included for the purpose of illustration, but those skilled in the art may make various variations and modifications to these details.

10...Precipitation prediction device 11...Observation data acquisition unit 12...Numerical prediction unit 13...Nowcast prediction unit 14...Correct rate calculation unit 15...Position shift grid number input unit 16...Synthesis unit

Claims (8)

An observation data acquisition unit that acquires at least precipitation observation data;
A numerical value that divides the observation area into grids and uses the observation data acquired from the observation data acquisition unit to calculate the temporal change of the precipitation situation by computer using physical laws and predict the future precipitation situation for each grid A prediction section,
The observation area is divided into grids, the movement of past precipitation areas and the distribution of present precipitation are obtained using the observation data acquired from the observation data acquisition unit, and the distribution of precipitation after a short time is predicted for each grid. With the nowcast prediction unit,
For the numerical prediction data obtained from the numerical prediction unit and the Nowcast prediction data obtained from the Nowcast prediction unit, the number of misalignment grids indicating a predetermined area is determined, and the correct answer rate is calculated for each number of misalignment grids. Correct rate calculation part
In consideration of the correct answer rate obtained by the correct answer rate calculation section, the optimal synthesis coefficient is estimated in real time, and a synthesis section for synthesizing the latest numerical prediction data and nowcast prediction data using the synthesis coefficient,
A precipitation prediction device comprising: The precipitation prediction device according to claim 1, wherein the correct answer rate calculation unit calculates a correct answer rate for each misalignment grid number using the following equation (1).

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the grid where the amount of precipitation is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area. The precipitation prediction apparatus according to claim 1 or 2, further comprising a position shift grid number input unit capable of inputting the position shift grid number. The precipitation prediction apparatus according to claim 1, wherein the composition unit composes the latest numerical prediction data and nowcast prediction data using the following formula (2). ..
R(n,t)=C(n,t)×R1(n,t)+(1−C(n,t))×R2(n,t) (2)
However,
n is the number of grids on one side of the square for calculating Fraction,
t is time,
R1(n,t) is nowcast prediction data after t hours with the number of grids n on one side,
R2(n,t) is the numerical prediction data after t hours when the number of grids on one side is n,
C(n,t) is a composite coefficient (a function of n and t, after t time) of the nowcast prediction data R1(n,t) with respect to the numerical prediction data R2(n,t) at the number n of grids on one side. 0≦C(n,t)≦1),
Is. Acquiring at least precipitation observation data;
Dividing the observation region into a grid, using the observation data, a step of predicting numerical prediction data indicating future precipitation conditions for each grid by calculating a time change of the precipitation condition by a computer using a physical law,
The observation area is divided into grids, the movement of the past precipitation area and the distribution of the present precipitation are obtained using the observation data, and nowcast prediction data showing the distribution of precipitation after a short time is predicted for each grid. Steps,
With respect to the numerical prediction data and the nowcast prediction data, determining the number of misalignment grids indicating a predetermined area, and calculating a correct answer rate for each number of misalignment grids,
Considering the correct answer rate, estimating the optimal synthesis coefficient in real time, and synthesizing the latest numerical prediction data and nowcast prediction data using the synthesis coefficient,
A method for predicting precipitation, comprising: The method of calculating a correct answer rate for each misaligned grid number uses the following formula (1), and the precipitation prediction method according to claim 4.

...(1)
However,
O(n)(i,j) is the Fraction of the observed value at the lattice (i,j),
M(n)(i,j) is the Fraction of the predicted value at the grid (i,j),
Fraction is the ratio of the lattice where the precipitation amount is greater than or equal to a predetermined value for the n×n region,
n is the number of grids on one side of the square for calculating Fraction,
N is the total number of grids in the evaluation area,
Σ _i Σ _j is the sum within the evaluation area. The precipitation prediction method according to claim 4 or 5, further comprising the step of inputting the number of misaligned grids. The precipitation prediction method according to claim 5, wherein the step of synthesizing the latest numerical prediction data and the nowcast prediction data uses the following formula (2).
R(n,t)=C(n,t)×R1(n,t)+(1−C(n,t))×R2(n,t) (2)
However,
n is the number of grids on one side of the square for calculating Fraction,
t is time,
R1(n,t) is nowcast prediction data after t hours with the number of grids n on one side,
R2(n,t) is the numerical prediction data after t hours when the number of grids on one side is n,
C(n,t) is a composite coefficient (a function of n and t, after t time) of the nowcast prediction data R1(n,t) with respect to the numerical prediction data R2(n,t) at the number n of grids on one side. 0≦C(n,t)≦1),
Is.

Images