JP6949332B2 - Lightning risk judgment device - Google Patents

Description

The present invention relates to a lightning risk determination device for determining the risk of lightning.

Conventionally, a technique for determining the occurrence of lightning using a weather radar and an outside air temperature has been disclosed (see Patent Document 1). The technique described in Patent Document 1 determines the freezing altitude based on the outside air temperature, generates a lightning icon when the reflectance for altitudes above the freezing altitude is larger than the lightning threshold, and sets the freezing altitude and a predetermined distance value. Generates a hail icon when the altitude of the sum of is greater than the hail threshold.

Japanese Unexamined Patent Publication No. 2011-128150

Numerical Weather Prediction Division, "Basic Knowledge of Numerical Weather Prediction and Latest Numerical Weather Prediction System", Japan Meteorological Agency Forecast Department, 2012 Numerical Weather Prediction Training Textbook, November 2012 Shingo Shimizu, Tsuyoshi Maesaka, "Analysis Method of Multiple Doppler Radar Data Using Variant Method for Estimating Three-dimensional Wind Speed Field", National Research Institute for Earth Science and Disaster Prevention Research Report No. 70, January 2007 (http: http: //dil-opac.bosai.go.jp/publication/nied_report/PDF/70/70shimizu.pdf) TAKEHARU KOUKETSU, 8 outsiders, "A Hydrometeor Classification Method for X-Band Polarimetric Rader: Construction and Validation Focusing on Solid Hydrometeors under Moist Environments", American Meteorological Society, JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY, VOLUME 32, pp2052-2074, Nov 2015 (https://journals.ametsoc.org/doi/abs/10.1175/JTECH-D-14-00124.1) HYPERLINK "https://journals.ametsoc.org/author/Hauser%2C+Dani%C3%A8le" Daniele Hauser HYPERLINK "https://journals.ametsoc.org/author/Amayenc%2C+Paul" Paul Amayenc, " Retrieval of Cloud Water and Water Vapor Contents from Doppler Radar Data in a Tropical Squall Line ”, American Meteorological Society, JOURNAL OF ATMOSPHERIC SCIENCES, VOL 43, No.8, pp823-838, 15 APRIL, 1986 (https://journals. ametsoc.org/doi/abs/10.1175/1520-0469%281986%29043%3C0823%3AROCWAW%3E2.0.CO%3B2) Lawrence D. Carey and Steven A. Rutledge, "The Relationship between Precipitation and Lightning in Tropical Island Convection: A C-Band Polarimetric Radar Study", American Meteorological Society, MONTHLY WEATHER REVIEW, VOL 128, pp2687-2710, AUGUST 2000 (https) //journals.ametsoc.org/doi/full/10.1175/1520-0493%282000%29128%3C2687%3ATRBPAL%3E2.0.CO%3B2) Gregory N. Seroka, Richard E. Orville, Courtney Schumacher, "Radar Nowcasting of Total Lightning over the Kennedy Space Center", American Meteorological Society, WEATHER AND FORECASTING, VOL 27, pp189-204, FEBRUARY 2012 (https://journals. ametsoc.org/doi/pdf/10.1175/WAF-D-11-00035.1)

However, the technique described in Patent Document 1 does not predict hail and lightning in relation to each other, and therefore the prediction accuracy is not good.

An object of the present invention is to provide a lightning risk determination device capable of accurately predicting the risk of lightning as compared with the prior art.

The lightning risk determination device according to the present invention is
Meteorological information acquisition department that detects cumulonimbus clouds,
A three-dimensional data creation unit that creates three-dimensional data including current data representing the current cumulonimbus cloud status and past data representing the past cumulonimbus cloud status using objective analysis technology from the meteorological data acquired by the meteorological information acquisition unit. When,
Nowcast prediction data that predicts the movement of the turbulent cloud from the current state data and the past data and represents the situation of the turbulent cloud in the future, and a numerical value created by performing weather prediction using the cloud resolution numerical model with the current state data as the initial value. A movement prediction unit that creates future prediction data including prediction data,
A state threshold input unit that defines a predetermined threshold value for a state change,
A state change calculation unit that identifies a cumulonimbus cloud whose state change is equal to or greater than the threshold value from the current state data, the past data, and the future prediction data, and calculates the state change in the cumulonimbus cloud.
A risk calculation unit that calculates lightning risk data representing the risk of lightning occurrence from the state change calculated by the state change calculation unit, and a risk calculation unit.
Regarding the future prediction data, a blending unit that blends the results of the nowcast prediction data and the numerical prediction data, and
A lightning information creation unit that creates position information of a high-risk place from the lightning risk data calculated by the risk calculation unit and the blending unit, and a lightning information creation unit.
It is characterized by having.

The lightning risk determination device according to the present invention is
It is characterized by including an observation data input unit that inputs statistical information created by learning processing using at least one of the three-dimensional data, the lightning risk data, and the observation data to the risk calculation unit. do.

The lightning risk determination device according to the present invention is
The state change calculation unit calculates the volume change of the ascending current in the cumulonimbus cloud, and calculates the volume change.
The state threshold input unit is characterized in that it defines a threshold value of the volume of the ascending current to be compared with the value calculated by the state change calculation unit.

The lightning risk determination device according to the present invention is
The state change calculation unit calculates the volume change of the hail in the cumulonimbus cloud.
The state threshold input unit is characterized in that it defines a threshold value for the volume of hail to be compared with a value calculated by the state change calculation unit.

According to such a lightning risk determination device, it is possible to accurately predict the risk of lightning.

Shows the mechanism of lightning in cumulonimbus clouds. The system block of the lightning risk determination device of this embodiment is shown. The image of cumulonimbus cloud detection of this embodiment is shown. The three-dimensional data creation part of this embodiment is shown. The flow of analysis of the three-dimensional data creation unit of this embodiment is shown. The data at each time of the three-dimensional data creation unit of this embodiment is shown. The control flowchart of the three-dimensional data creation part of this embodiment is shown. It shows the volume of hail at each altitude with respect to the time elapsed since the cumulonimbus cloud was generated. 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 to the predicted time and the predicted correct answer rate by nowcast prediction is shown. A comparison of observation data and prediction data for each grid is shown. The relationship between the correct answer rate and the predicted time is shown. The flowchart of the precipitation prediction method of this embodiment is shown. The precipitation prediction data obtained by the precipitation prediction device and the precipitation prediction method of the present embodiment is shown. The system block of the cumulonimbus cloud prediction system of this embodiment is shown. The flowchart of the cumulonimbus cloud prediction system of this embodiment is shown.

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

FIG. 1 shows the mechanism of lightning in a cumulonimbus cloud.

Lightning comes from cumulonimbus clouds. Cumulonimbus clouds are formed by the strong updraft lifting the air in the lower layer and the water vapor in the air becoming water droplets in the sky. At altitudes below freezing, not only raindrops but also ice grains such as hail and ice crystals are formed. Ice particles grow by colliding with surrounding water droplets called supercooled water droplets in the ascending current. Eventually, the ice particles begin to fall when they become too large to be supported by the updraft.

During this ascent and fall, the ice particles collide with each other, causing charge transfer between the large and small particles. The sign of the electric charge charged by each ice grain is determined by the mass of water contained in the atmosphere per unit volume called the amount of cloud water and the ambient air temperature. When there is an appropriate amount of cloud water, large ice particles are charged with a negative charge and small ice particles are charged with a positive charge when the temperature is lower than -10 ° C. When such ice particles continue to collide with each other, a large amount of electric charge is stored in the cumulonimbus cloud.

Air is an insulator that does not conduct electricity, but when the potential difference exceeds 3 million V per meter, a phenomenon called dielectric breakdown occurs, and electric discharge that allows electricity to pass through the air begins, causing lightning. Lightning includes lightning strikes and cloud discharges. Lightning strikes are a phenomenon in which electricity flows between cumulonimbus clouds and the ground, and cloud discharge is a phenomenon in which electricity flows within cumulonimbus clouds or between different cumulonimbus clouds.

The lightning risk determination device 10 of the present embodiment determines the volume change of the hail and the airflow in the cumulonimbus cloud, and determines that it is dangerous when a rapid increase is observed.

FIG. 2 shows a system block of the lightning risk determination device 10 of the present embodiment. FIG. 3 shows an image of cumulonimbus cloud detection according to this embodiment.

The lightning risk determination device 10 includes a weather information acquisition unit 11 that detects a turbulent cloud, a three-dimensional data creation unit 12 that creates three-dimensional data, a movement prediction unit 13 that predicts the movement of the turbulent cloud, and a state change in the turbulent cloud. The state change calculation unit 14 that calculates the above, the state threshold input unit 15 that defines and inputs the threshold value to be compared with the value calculated by the state change calculation unit 14, and the state change calculated by the state change calculation unit 14 The risk calculation unit 16 for calculating the above, the observation data input unit 17 for inputting statistical information created by learning and processing three-dimensional data, and the movement prediction unit 13 obtained from nowcast and numerical prediction, respectively. The blending unit 18 that blends the results calculated by the risk calculation unit 16 from the future prediction data of the above, and the current data calculated by the risk calculation unit 16 and the future prediction data blended by the blending unit 18 It includes a lightning information creating unit 19 that creates position information, a moving direction, and a moving speed.

The meteorological information acquisition unit 11 of the present embodiment acquires observation data obtained from a working single polarization radar or the like of the Japan Meteorological Agency or the like. The observation data may be data representing the elevation angle of the polar coordinate system.

The three-dimensional data creation unit 12 uses the observation data acquired by one or more meteorological information acquisition units 11 as an input for the meteorological field objective analysis, and uses the meteorological field objective analysis technology to obtain elevation angle data of a plurality of polar coordinate systems in Cartesian coordinates. Convert to grid-like 3D data in the system.

In the three-dimensional data creation unit 12 of the present embodiment, data may be created by using a technique for restoring a three-dimensional cumulonimbus cloud by objective analysis. The three-dimensional data creation unit 12 creates current status data indicating the current status of cumulonimbus clouds, and saves the status of past cumulonimbus clouds as past data. Here, a technique for restoring a three-dimensional cumulonimbus cloud by objective analysis will be described.

FIG. 4 shows the three-dimensional data creation unit 12 of the present embodiment. FIG. 5 shows the flow of the three-dimensional data creation unit 12 objective analysis of the present embodiment. The horizontal axis of FIG. 5 represents the flow of time. FIG. 6 shows the data at each time of the three-dimensional data creation unit 12 of the present embodiment.

The three-dimensional data creation unit 12 includes an observation data acquisition unit 121, an initial value creation unit 122, a forecast value creation unit 123, a restart file creation unit 124, an estimation value creation unit 125, a provisional analysis unit 126, and the like. The analysis unit 127 is provided.

The observation data acquisition unit 121 acquires observation data such as ground observations, meteorological satellites, and radars from meteorological organizations or space organizations around the world. Since observations are made at various places and times, observation data is acquired from each acquisition destination at predetermined time intervals. In addition, it is preferable that the observation data acquisition unit 121 also acquires background field data such as grid point data of the global model or meso model of the Japan Meteorological Agency. Some observation data are not available due to low accuracy due to human error or equipment failure, so these data are excluded. The acquired observation data is output to the initial value creation unit 122.

The initial value creation unit 122 creates an initial value at the initial time from the background field data and the observation data acquired by the observation data acquisition unit 121. As shown in FIG. 6, the observation data cannot be used as it is because the distribution is non-uniform. Therefore, the initial value creation unit 122 creates an initial value in a spatially and temporally uniform format.

The forecast value creating unit 123 continuously creates the forecast value of the meteorological field at predetermined time intervals from the initial value created by the initial value creating unit 122. The forecast values are created in a spatially and temporally uniform format like the initial values. Here, the meteorological field refers to atmospheric pressure, temperature, humidity, wind direction, wind speed, precipitation, snow depth, snowfall depth, sunshine duration, solar radiation, clouds, visibility, atmospheric phenomena, and the like.

The restart file creation unit 124 creates a restart file to be used for this analysis. The restart file is one before the forecast variable (wind, temperature, atmospheric pressure, water substance, etc.), the time change term of the forecast variable, the boundary value, and the restart time at the predetermined restart time t before the analysis time. It is created by the forecast variable and the like at the time step time t−Δt of (see Non-Patent Document 1).

The estimated value creating unit 125 creates an estimated value at the analysis time in which the observation data acquired by the observation data acquisition unit 121 is input at predetermined time intervals after the initial time. Estimates are created in a spatially and temporally uniform format, similar to forecast values.

The provisional analysis unit 126 calculates the analysis increment from the new observation data acquired by the observation data acquisition unit 121 and the estimated value created by the estimated value creating unit 125. The analysis increment is a value that corrects the error between the estimated value and the new observation value. Analysis increments are created in a spatially and temporally uniform format, similar to estimates. In the present embodiment, it is preferable to perform the analysis by using the three-dimensional variational method.

Here, the three-dimensional variational method will be briefly described. In the three-dimensional variational method, the following equations (1) and (2) are used.
[Number 1]
J = 1/2 (xx ^b ) ^T B ^-1 (xx ^b ) + 1/2 [H (x) -y] ^T R ^-1 [H (x) -y] (1)
∇ x J = B ^-1 (xx ^b ) + H ^T R ^-1 [H (x) -y] (2)
However,
B ^-1 is the inverse of the background error covariance matrix,
R ^-1 is the inverse of the observation error covariance matrix,
y is the observed value,
x ^b is the estimated value,
H (x) is a function (observation operator) that finds the quantity equivalent to the observation from the analysis value.
H ^T is the transposed matrix of the matrix H when the observation operator corresponds to the linear transformation of the analysis values.

Equation (1) is called a cost function, and is a quantification of the error between the observed value and the estimated value of the numerical model in a quadratic form. Equation (2) is called a gradient vector of the cost function, which is obtained by differentiating the cost function of the scalar quantity with the estimated value of the vector quantity to be obtained from now on, and is a vector quantity having the same dimension as the estimated value.

The gradient vector provides quantitative information on how to modify the estimate to reduce the error. The inverse matrix of the background error covariance matrix and the inverse matrix of the observation error covariance matrix are known and stored in advance. That is, the three-dimensional variational method is a problem of exploring an unknown value x that minimizes the cost function as an analysis value by using a gradient vector.

The analysis unit 127 traces back to the restart time, smoothes the analysis increment calculated by the provisional analysis unit 126 at the analysis time, and continuously corrects the restart forecast value from the data at the restart time. Output the analysis value. As a correction method, it is preferable to use an IAU (Incremental Analysis Update) filter or the like. After the main analysis, the forecast value creating unit 13 continuously creates the forecast value with the new main analysis value as the initial value. That is, the calculation can be restarted as soon as the next observed value is acquired.

Here, the IAU filter will be described. The IAU filter uses the following equation (3).
[Number 2]
∂A / ∂t = Adv.A + Diff.A + Src.A + β (t ^{--t 0} ) * Inc.A (3)

The time evolution of numerical prediction is generally determined by the advection term (Adv), diffusion term (Diff), and occurrence / extinction term (Src) as shown in Eq. (3). In the IAU filter, β (t ^{--t 0} ) * Inc. A is added to change the way of time evolution and improve the prediction.

β (t --t ⁰ ) is given as a function of the time difference between a certain time t and the time t ^{0 when the assimilation took place.} It becomes maximum when t ^{matches t 0} , and approaches 0 as the time difference increases. ^{Inc.A is xx b} obtained by Eqs. (1) and (2) at a certain time t ⁰ .

FIG. 7 shows a control flowchart of the three-dimensional data creation unit 12 of the present embodiment.

First, in step 1, the background field data and the observation data acquired from the Japan Meteorological Agency by the observation data acquisition unit 121 are assimilated by the initial value creation unit 122 to create the initial value (ST1). The background field data may be acquired when the initial value is created.

Subsequently, in step 2, the forecast value creating unit 123 continuously creates the forecast value of the meteorological field from the initial value (ST2).

Subsequently, in step 3, the restart file creation unit 124 creates a restart file (ST3). The restart file is created at a predetermined restart time before the analysis time.

In step 4, it is determined whether or not there is new observation data (ST4). Since new observation data is input at the analysis time, usually no new observation data is input until the analysis time. If there is new observation data, the process proceeds to step 5. If there is no new observation data, the process returns to step 4.

In step 5, the estimated value creating unit 125 creates the forecast value of the analysis time as the estimated value (ST5).

Subsequently, in step 6, the provisional analysis unit 126 calculates the analysis increment from the estimated value and the new observation data (ST6).

Next, in step 7, the analysis unit 127 holds the analysis increment and goes back to the restart time (ST7).

Next, in step 8, the analysis unit 127 performs the main analysis from the restart file and the analysis increment, and outputs the main analysis value (ST8). In this analysis, it is preferable to use the smoothed analysis increment to continuously correct the restart forecast value from the data at the restart time.

Next, in step 9, it is determined whether or not the analysis period has ended (ST9). When the analysis period ends, the control ends.

When the analysis period ends, in step 10, the present analysis value is set as a new initial value, the analysis increment is deleted, and the process returns to step 2 (ST10). That is, the objective analysis device of the present embodiment repeats the objective analysis until the analysis period ends.

The movement prediction unit 13 predicts the movement of cumulonimbus clouds using the current state data and past data created by the three-dimensional data creation unit 12.

As an example, the movement vector may be calculated from the past data and the current data by pattern matching by the cross-correlation number method or the like, and output as future prediction data. By using the movement prediction unit 13, it is possible to accurately predict future lightning. The prediction data obtained in this way is called nowcast prediction data.

Further, the prediction data may be created by using the current state data as the initial value and performing the weather prediction by the cloud resolution numerical model. The prediction data obtained in this way is called numerical prediction data. As the cloud resolution numerical model, for example, CRESS (Cloud Resolving Storm Simulator) or the like may be used.

FIG. 8 shows the volume of hail at each altitude with respect to the time elapsed since the cumulonimbus cloud was generated.

The state change calculation unit 14 calculates the change in the state of the turbulent cloud from the current state data and the past data created by the three-dimensional data creation unit 12 and the future prediction data created by the movement prediction unit 13. Indicators of state change include a predetermined time volume of the ascending current, a predetermined time volume of the region determined to be hail, a mass per unit volume of hail, a vertically integrated hail mass or echo. At least one of the apex altitudes is sufficient.

For example, the calculated state change is set to be equal to or higher than a predetermined threshold value input from the state threshold value input unit 15. The state change calculation unit 14 identifies cumulonimbus clouds that are considered to be the same from the three-dimensional data from the past to the present, and calculates the state change. To identify cumulonimbus clouds that appear to be the same, existing cumulonimbus cloud automatic tracking technology may be used.

Here, the variables that evaluate the state change will be described. FIG. 8 shows the volume of hail as an example of these state changes.

To determine the volume of the updraft, combine the Doppler velocities observed by multiple radars and estimate the three components of the wind. The three components are east-west wind, north-south wind, and vertical wind, and in the coordinate system where the vertical upward direction is positive, the positive vertical wind is called the updraft. The total volume of the grid grid in which an ascending flow above the threshold value is detected is defined as the "increasing flow volume". For a method of estimating the three components of wind, refer to Non-Patent Document 2.

For the volume of the region determined to be hail, for example, the total volume of the grid grid classified as hail may be obtained by using a method for discriminating precipitation particles using a dual polarization radar. Non-Patent Document 3 may be referred to for a method for discriminating precipitation particles using a dual polarization radar.

The mass per unit volume of hail is, for example, per unit volume from measured values such as reflection intensity and polarization parameters in a grid grid classified as hail using a method of discriminating precipitation particles using a dual polarization radar. It is the mass per unit volume in each grid grid calculated by using the method of estimating the mass of. For the method of estimating the mass per unit volume of hail, refer to Non-Patent Documents 4 and 5.

The vertically integrated mass of precipitation particles is the mass of precipitation particles in each grid grid using a method of estimating the mass of precipitation particles (rain, haze, snow) per unit volume from measured values such as reflection intensity and polarization parameters. The mass is calculated, and the mass of each grid is integrated in the vertical direction (see Non-Patent Document 6).

The echo peak altitude is the highest altitude reached on the isosurface of the radar reflection intensity in the cumulonimbus cloud.

The isosurface echo intensity is the reflection intensity on the isosurface of a certain temperature by creating an isosurface of the temperature from the three-dimensional distribution of the temperature. Specifically, the temperature at which the hail is negatively charged at about −10 ° C. may be selected, and the reflection intensity at an altitude of −10 ° C. may be extracted.

The risk calculation unit 16 includes the volume change rate of the ascending current calculated by the state change calculation unit 14, the volume change rate of the hail, the mass change rate of the hail, the change rate of the mass of the vertically integrated precipitation particles, and the echo peak altitude change rate. Or, using the isothermal surface echo intensity change rate and the statistical information input from the observation data input unit 17, a cumulonimbus cloud having a change rate of a predetermined value or more is determined to have a high risk of lightning. The risk determination is modeled by statistical information.

The observation data input unit 17 is a risk calculation unit for statistical information created by learning processing using at least one of three-dimensional data, lightning risk data, and observation data such as an LMA (Lightning Mapping Array) sensor. Enter in 16. By performing the learning process, it becomes possible to predict the risk of lightning more accurately. The observation data input unit 17 does not have to be used.

The modeled lightning risk data is output as a horizontal distribution map. The horizontal distribution map is output as a current horizontal distribution map and a future horizontal distribution map using the current data and past data created by the three-dimensional data creation unit 12 and the future prediction data created by the movement prediction unit 13, respectively. NS.

Of these, regarding the future prediction data, the results of the nowcast prediction data and the numerical prediction data created by the movement prediction unit are blended in the blending unit 18, respectively. Here, the blending prediction will be described.

FIG. 9 shows a system block diagram of the blending unit 18 of the present embodiment.

The blending unit 18 includes an observation data acquisition unit 181, a numerical value prediction unit 182, a nowcast prediction unit 183, a correct answer rate calculation unit 184, a position shift grid number input unit 185, and a synthesis unit 186.

The observation data acquisition unit 181 acquires observation data such as ground observations, meteorological satellites, and radars from meteorological institutions or space agencies around the world. Since observations are made at various places and times, observation data is acquired from each acquisition destination at predetermined time intervals. Some observation data are not available due to low accuracy due to human error or equipment failure, so these data are excluded. The acquired observation data is output to the numerical prediction unit 182 and the nowcast prediction unit 183.

The numerical weather prediction unit 182 is a part that divides the observation area into a grid pattern and performs numerical weather prediction for each grid using the observation data acquired from the observation data acquisition unit 181. Numerical weather prediction predicts future precipitation by calculating the time change of precipitation with a computer using the laws of physics.

The nowcast prediction unit 183 divides the observation area into a grid pattern, and uses the observation data acquired from the observation data acquisition unit 181 to use the movement of the past precipitation area and the distribution of the current precipitation, after a short period of time. This is the part that predicts the distribution of precipitation for each grid.

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 184 allows the number of misalignment grids (permissible misalignment) with respect to the numerical prediction data obtained from the numerical prediction unit 182 and the nowcast prediction data obtained from the nowcast prediction unit 183. The correct answer rate for each spatial scale) is calculated using the following equation (4).

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

The position shift grid number input unit 185 appropriately inputs the required position shift grid number. Increasing the number of misaligned lattices n increases the accuracy of predicting the probability of precipitation. However, although the forecast that it will rain anywhere within a wide range is correct, it is not practical. On the contrary, if the number of misaligned lattices n is reduced, the accuracy of predicting the probability of precipitation will decrease. That is, it is difficult to make a pinpoint forecast that it will rain at this point after tens of minutes. Therefore, it is preferable that the number of misaligned lattices n is appropriately determined according to the request of the user.

The synthesis unit 186 estimates the optimum synthesis coefficient in real time in consideration of the correct answer rate obtained by the correct answer rate calculation unit 184, and uses the synthesis coefficient to calculate the numerical prediction data and the nowcast prediction data by the following formula ( Synthesize using 5).
R (n, t) = C (n, t) x R1 (n, t) + (1-C (n, t)) x R2 (n, t) (5)
However,
n is the number of grids on one side of the square for which Fraction is calculated,
t is time,
R1 (n, t) is the nowcast prediction data after t time with the number of lattices n on one side.
R2 (n, t) is the numerical prediction data after t time with the number of lattices n on one side.
C (n, t) is the composite coefficient (function of n and t, of n and t) of the nowcast prediction data R1 (n, t) after t time with respect to the numerical prediction data R2 (n, t) with the number of lattices n on one side. 0 ≤ C (n, t) ≤ 1),
Is.

The composite coefficient is determined in consideration of the accuracy rate of the numerical prediction data and the nowcast prediction data. The correct answer rate of the predicted data is calculated in real time by comparing it with the true value of the past one hour.

FIG. 10 shows the ratio of the predicted correct answer rate by numerical prediction to the predicted time and the predicted correct answer rate by nowcast prediction.

As shown in FIG. 10, in the numerical prediction, when the prediction time is short, the correct answer rate is low, and when the prediction time is long, the correct answer rate is high. In nowcast prediction, the correct answer rate is high when the prediction time is short, and the correct answer rate is low when the prediction time is long. Therefore, it is preferable to obtain a composite coefficient that emphasizes nowcast prediction when the prediction time is short and emphasizes numerical prediction when the prediction time is long.

In addition, when predicting localized heavy rainfall, it is required to predict with high resolution with a grid of 1 km or less. When forecasting with high grid resolution, the accuracy of the forecast data for each grid is generally low. That is, the prediction with high lattice resolution has a large error.

FIG. 11 shows a comparison between the observed data and the predicted data for each grid. FIG. 11 (a) shows observation data, and FIG. 11 (b) shows prediction data. The grid in the shaded area indicates the position of precipitation above a predetermined amount.

In the example shown in FIG. 11, if a grid scale of n = 1 that does not allow positional deviation is selected, the predicted data is determined to be different from the observed data. However, if a grid scale of n = 5 that allows 5 grids of positional deviation is selected, the predicted data and the observed data both have the same result as the observed data because there are 6 positions of precipitation above a predetermined amount. It is judged.

In the present embodiment, the accuracy rate calculation unit 184 can tolerate a certain degree of positional deviation by using the equation (4), and can maintain the prediction accuracy even at a high grid resolution. Then, the synthesis unit 186 synthesizes the numerical prediction data and the nowcast prediction data by using the result of the correct answer rate calculation unit 184.

Equation (4) shows the correct answer rate of Fraction when the number of positional deviation lattices n at the coordinates (i, j) to be evaluated is allowed. Fraction has O and M. FractionO is the probability of precipitation when the position shift is allowed with respect to the actually observed rain. Fraction M is the probability of precipitation when the position shift is allowed with respect to the predicted rain.

Equation (4) 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. Further, the equation (4) is calculated for each predicted lead time. Then, the composite 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 the number of misaligned lattices and each lead time.

That is, the correct answer rate calculation unit 184 calculates the precipitation distribution and the precipitation probability distribution that allow the numerical prediction data and the nowcast prediction data by the number of positional deviation lattices n for each lead time. Then, the synthesis unit 186 synthesizes the latest numerical prediction data and the nowcast prediction data based on the synthesis coefficient with respect to the prediction for each lead time and the number of position shift lattices n, and the precipitation distribution and the precipitation occurrence probability distribution. Ask for.

As described above, the blending unit 18 of the present embodiment can obtain the precipitation probability distribution in addition to the precipitation prediction distribution.

FIG. 12 shows the relationship between the correct answer rate and the predicted time. FIG. 12 (a) shows a case where a precipitation of 50 mm is predicted in one hour, and FIG. 12 (b) shows a case where a precipitation of 20 mm is predicted in one hour. L in the graph shown in FIG. 12 indicates a position shift allowable scale (km).

For example, if it is determined that a prediction accuracy rate of 50% is required, it is preferable to use data above the permissible line shown in FIG.

Next, the control flow of the blending unit 18 of the present embodiment will be described.

FIG. 13 shows a flowchart of the precipitation prediction method of the present embodiment.

First, in step 11, the past actual precipitation data is read (ST11). Subsequently, in step 12, the past numerical prediction data and the nowcast prediction data are read (ST12).

Next, in step 13, the number of misaligned lattices n is determined (ST13). The number of misaligned lattices n can be appropriately determined by the user.

Next, in step 14, the correct answer rate shown in the equation (4) is calculated (ST14). Subsequently, in step 15, the composite coefficient of the numerical prediction data and the nowcast prediction data is calculated (ST15). When calculating the correct answer rate and the composite coefficient, the precipitation data, the prediction data, the position shift lattice number n, etc. obtained in steps 11 to 13 are used.

Next, in step 16, the latest numerical prediction data and nowcast prediction data are read (ST16). Subsequently, in step 17, precipitation prediction and precipitation probability prediction for each position shift grid number in the latest numerical prediction data and nowcast prediction data read in step 16 are obtained (ST17).

Next, in step 18, the prediction based on the numerical prediction data and the prediction based on the nowcast prediction data are combined (ST18). At the time of synthesis, it is obtained from the equation (5) using the synthesis coefficient calculated in step 15. Further, it is preferable to consider the allowable line and the like shown in FIG.

Finally, in step 19, a composite prediction file that predicts the amount of precipitation and the probability of precipitation for each number of misaligned lattices is output (ST19).

FIG. 14 shows precipitation prediction data obtained by the blending unit 18 of the present embodiment. FIG. 14 (a) shows rainfall prediction data, and FIG. 14 (b) shows precipitation probability data.

As shown in FIGS. 14A and 14B, the blending unit 18 of the present embodiment can obtain precipitation probability data in addition to precipitation prediction data. Further, in the precipitation probability data, by selecting the position shift lattice number n, it is possible to appropriately obtain the probability that a predetermined amount or more of rain will fall in a predetermined region. For example, in the example shown in FIG. 14B, the probability that it will rain 50 mm or more in a 5 km square around the circumference is obtained.

The lightning information creation unit 19 predicts the three-dimensional distribution using the time-series lightning risk data calculated by the risk calculation unit 16 and the blending unit 18 and the movement vector calculated by the movement prediction unit 13. After that, the position of the current cumulonimbus cloud with high lightning risk, and the future movement direction and speed of the cumulonimbus cloud are calculated.

As described above, according to the lightning risk determination device 1 of the present embodiment, it is possible to accurately predict the lightning risk.

FIG. 15 shows a system block of the cumulonimbus cloud prediction system of the present embodiment.

The turbulence cloud prediction system 21 receives from the lightning risk determination device 10, the receiver information input unit 24 for inputting receiver information, and the turbulence cloud information and receiver information input unit 24 calculated by the lightning risk determination device 10. It includes a turbulent cloud / receiver relationship calculation unit 25 that calculates each relationship from the receiver information input by the person, and an output unit 26 that outputs the result of the calculation by the turbulent cloud / receiver relationship calculation unit 25.

The receiver information input unit 24 is for the receiver to input his / her own information in advance. For example, the receiver information input unit 24 may use a mobile terminal or the like. Recipient's position information 24a that informs the position where the receiver wants to know whether or not it is dangerous, receiver's danger setting information 24b that informs the danger level and distance set by the receiver, and the position deviation set by the receiver. The receiver's misalignment tolerance information 24c that informs the permissible range is input.

The location information 24a of the receiver may be a place where the receiver currently exists, a place where the receiver will move, a place where the receiver wants to know, or the like. The location may be specified from latitude / longitude information such as GPS. The recipient chooses at least one of these locations.

The receiver's risk setting information 24b is information on the degree of risk set by the receiver. For example, the receiver identifies at least one of the phenomena of concern from rain, wind, lightning, and hail, and selects the risk of that phenomenon for each level. The degree of danger is the hourly rainfall or cumulative rainfall in the case of rain, the wind speed in the case of wind, the activity of thunder nowcast set by the Japan Meteorological Agency in the case of lightning, the existence or fall confirmation in the sky in the case of hail, etc. At least two levels such as caution and caution should be set with reference to. The recipient chooses at least one of these levels.

The position shift allowance information 24c of the receiver is a range in which the position shift set by the receiver can be tolerated. For example, the receiver may adjust the position from no misalignment to 20 km. The misalignment distance may be set continuously or stepwise. Since the size of cumulonimbus clouds is about 10 km, it is preferable to set the maximum value to twice that.

The cumulonimbus / receiver relation calculation unit 25 converts the coordinate system of the cumulonimbus cloud into the coordinate system of the receiver, and calculates the risk level of the position set by the receiver in chronological order for each phenomenon. Cumulonimbus clouds constantly change in size and move. In addition, the receiver may change the phenomenon, position, etc. for which he / she wants to know the degree of danger every hour. Therefore, the relationship between the cumulonimbus cloud and the receiver is calculated by matching the coordinate system.

The output unit 26 outputs the result calculated by the cumulonimbus / receiver relation calculation unit 5. The output unit 26 tells whether the location set by the receiver in the receiver information input unit 24 is a dangerous position or not, and whether the time set by the receiver in the receiver information input unit 24 is a dangerous time. Danger time information 26b notifying whether or not, danger type information 26c notifying whether the type such as rain, wind, thunder or hail set by the receiver in the receiver information input unit 24 is dangerous, and the receiver is the receiver. At least one of the danger level information 26d and the like that informs which level of the danger level set by the information input unit 24 is output.

The receiver information input unit 24 and the output unit 26 may be a personal computer, a mobile terminal, or the like. The receiver can input the information of the receiver and the information to be known from a personal computer or a mobile terminal or the like, and can see the information about the turbulent cloud after the calculation on the mobile terminal or the like.

FIG. 16 shows a flowchart of the cumulonimbus cloud prediction system of the present embodiment.

First, in step 21, the lightning risk determination device 10 calculates and outputs information on dangerous lightning (ST21).

Next, in step 22, the receiver information input unit 24 acquires the receiver information (ST22). The acquired receiver information may be the receiver's position information 24a, the receiver's danger setting information 24b, the receiver's position shift tolerance information 24c, and the like.

Next, in step 23, the cumulonimbus / receiver relationship calculation unit 25 calculates the relationship between the cumulonimbus cloud and the receiver (ST23). The relationship between the cumulonimbus cloud and the receiver may be obtained by converting the coordinate system of the cumulonimbus cloud into the coordinate system of the receiver and calculating the risk level of the position set by the receiver in chronological order for each phenomenon.

Next, in step 24, the output unit 25 outputs the relationship between the cumulonimbus cloud and the receiver (ST24). The output unit 26 may output at least one of the danger position information 26a, the danger time information 26b, the danger type information 26c, the danger level information 26d, and the like.

As described above, according to the cumulonimbus cloud prediction system 1, the lightning risk determination device 10 can accurately predict the risk of lightning, and it is possible to accurately inform the receiver of the cumulonimbus cloud information.

As described above, the lightning risk determination device 10 of the present embodiment represents the current state of the turbulent cloud by using the meteorological information acquisition unit 11 for detecting the turbulent cloud and the meteorological data acquired by the meteorological information acquisition unit 11 by using an objective analysis technique. The three-dimensional data creation unit 12 that creates three-dimensional data including the current data and the past data showing the situation of the past turbulent clouds, and the nowcast prediction that predicts the movement of the turbulent clouds from the current data and the past data and shows the situation of the future turbulent clouds. Defines a movement prediction unit 13 that creates future prediction data including numerical prediction data created by performing weather prediction using a cloud resolution numerical model with data and current data as initial values, and a predetermined threshold value for state changes. The risk of lightning occurrence is calculated from the state threshold input unit, the state change calculation unit 14 that calculates the state change in the turbulent cloud from the current state data, past data, and future prediction data, and the state change calculated by the state change calculation unit 14. The risk calculation unit 16 that calculates the lightning risk data to be represented, the blending unit 18 that blends the results of the nowcast prediction data and the numerical prediction data with respect to the future prediction data, and the lightning calculated by the risk calculation unit 16 and the blending unit 18. It is provided with a lightning information creation unit 19 that creates location information of a high-risk place from risk data. Therefore, it is possible to accurately predict the risk of lightning.

Further, the lightning risk determination device 10 of the present embodiment sends statistical information created by learning processing using at least one of three-dimensional data, lightning risk data, and observation data to the risk calculation unit 16. The observation data input unit 17 for input is provided. Therefore, it is possible to predict the risk of lightning more accurately.

Further, in the lightning risk determination device 10 of the present embodiment, the state change calculation unit 14 calculates the volume change of the ascending current in the cumulonimbus cloud, and the state threshold input unit 16 is the value calculated by the state change calculation unit. Define a threshold for the volume of the ascending current to be compared. Therefore, it is possible to predict the risk of lightning more accurately.

Further, in the lightning risk determination device 10 of the present embodiment, the state change calculation unit 14 calculates the volume change of the hail in the cumulonimbus cloud, and the state threshold input unit 16 compares it with the value calculated by the state change calculation unit. Defines the threshold for the volume of hail. Therefore, it is possible to predict the risk of lightning more accurately.

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

1 ... Storm cloud prediction system 2 ... Storm cloud detection unit 3 ... Storm cloud information calculation unit 4 ... Recipient information input unit 5 ... Storm cloud receiver-related calculation unit 6 ... Output unit 10 ... Lightning risk determination device 11 ... Meteorological information acquisition unit 12 ... Three-dimensional data creation unit 13 ... Movement prediction unit 14 ... State change calculation unit 15 ... State threshold input unit 16 ... Risk calculation unit 17 ... Observation data input unit 18 ... Blending unit 19 ... Lightning information creation unit

Claims (4)

Meteorological information acquisition department that detects cumulonimbus clouds,
A three-dimensional data creation unit that creates three-dimensional data including current data representing the current cumulonimbus cloud status and past data representing the past cumulonimbus cloud status using objective analysis technology from the meteorological data acquired by the meteorological information acquisition unit. When,
A state threshold input unit that defines a predetermined threshold value for a state change,
Nowcast prediction data that predicts the movement of the turbulent cloud from the current state data and the past data and represents the situation of the turbulent cloud in the future, and a numerical value created by performing weather prediction using the cloud resolution numerical model with the current state data as the initial value. A movement prediction unit that creates future prediction data including prediction data,
A state change calculation unit that identifies a cumulonimbus cloud whose state change is equal to or greater than the threshold value from the current state data, the past data, and the future prediction data, and calculates the state change in the cumulonimbus cloud.
A risk calculation unit that calculates lightning risk data representing the risk of lightning occurrence from the state change calculated by the state change calculation unit, and a risk calculation unit.
A blending unit that blends the results of the nowcast prediction data and the numerical prediction data with respect to the future prediction data.
A lightning information creation unit that creates position information of a high-risk place from the lightning risk data calculated by the risk calculation unit and the blending unit, and a lightning information creation unit.
A lightning risk determination device characterized by being equipped with. It is characterized by including an observation data input unit that inputs statistical information created by learning processing using at least one of the three-dimensional data, the lightning risk data, and the observation data to the risk calculation unit. The lightning risk determination device according to claim 1. The state change calculation unit calculates the volume change of the ascending current in the cumulonimbus cloud, and calculates the volume change.
The lightning risk determination device according to claim 1 or 2, wherein the state threshold value input unit defines a threshold value of an ascending current volume to be compared with a value calculated by the state change calculation unit. The state change calculation unit calculates the volume change of the hail in the cumulonimbus cloud.
The lightning risk determination device according to claim 1 or 2, wherein the state threshold value input unit defines a threshold value for the volume of hail to be compared with a value calculated by the state change calculation unit.

Images