JP7256487B2 - Weather forecasting device, weather forecasting method, and program - Google Patents

Description

TECHNICAL FIELD Embodiments of the present invention relate to a weather forecasting device, a weather forecasting method, and a program.

Conventionally, there has been disclosed a system for calculating a predicted value indicating the possibility of occurrence of turbulence using a plurality of indices related to the occurrence of turbulence. In this system, a coefficient determined for each grid is used to weight a plurality of indices associated with each grid to calculate the possibility of occurrence of turbulence in the grid. However, the probability of occurrence of turbulence needs to take into account the interrelationship between indicators and factors that cause turbulence. there were.

Also, conventionally, when providing turbulence prediction information, absolute prediction values were handled. However, since weather prediction includes uncertainty, it is not possible to judge how reliable the calculated predicted value is. Compared to other meteorological phenomena, turbulence has a low average occurrence probability, but once it occurs, for example, when an aircraft encounters turbulence, it can have a large impact on human lives and aircraft. Regarding the occurrence of turbulence, it is considered that more useful information can be obtained by treating how likely it is to occur, that is, by treating it as a probability, rather than whether or not it will occur.

In addition, there is a method using a Bayesian network as a technique for predicting the probability of meteorological phenomena in general. However, the above prediction technique does not disclose specific processing of a technique for predicting the occurrence of turbulence.

JP 2009-192262 A JP 2008-134145 A JP 2009-052976 A

A problem to be solved by the present invention is to provide a weather prediction device, a weather prediction method, and a program capable of predicting the occurrence of turbulence more accurately.

A weather prediction device according to an embodiment has an index derivation unit and a probability derivation unit. The index deriving unit calculates, based on the input meteorological conditions, a first index indicating the likelihood of turbulence occurring, a second index indicating the degree of change in the potential temperature direction in the altitude direction, and the strength of updraft or downdraft. At least two indicators are derived from the third indicator that indicates the degree of difficulty. The probability derivation unit classifies at least two indices derived based on future weather conditions by the index derivation unit based on criteria for each index, and derives a probability of occurrence of turbulence based on the pattern of the classification result. .

The figure which shows the structure of the weather prediction system 1. FIG. The figure which shows the functional structure of the weather prediction apparatus 20. FIG. 4 is a flowchart showing the flow of processing executed by the weather prediction device 20; 4 is a diagram showing an example of a determination map 42; FIG. The figure which shows an example of the Bayesian network in this embodiment. FIG. 4 is a diagram showing an example of an image displayed on a display unit 36;

Hereinafter, a weather prediction device, a weather prediction method, and a program according to embodiments will be described with reference to the drawings.

FIG. 1 is a diagram showing the configuration of a weather forecast system 1. As shown in FIG. A weather forecasting system 1 including a weather forecasting device 20 includes an information providing device 10 and a weather forecasting device 20 . The information providing device 10 and the weather prediction device 20 communicate via a network NW such as a LAN (Local Area Network), a WAN (Wide Area Network), or the Internet.

The information providing device 10 acquires three-dimensional observation data of atmospheric conditions from a radar device (weather radar) or other device, and transmits the acquired observation data to the weather prediction device 20 . Observation data is, for example, a polar coordinate radar echo GPV (Grid Point Value). A polar coordinate radar echo GPV is a reflection intensity (echo) for each three-dimensional grid acquired by a radar device. A three-dimensional grid is a divided area divided by a predetermined width in three directions with a certain point as the center. In the case of the polar coordinate radar echo GPV, there are divided regions divided by a predetermined width from a predetermined observation position (for example, a radar device) in the distance direction, azimuth direction, and elevation angle (φ) direction. Observation data includes, for example, physical quantities such as precipitation, temperature, atmospheric pressure, wind direction, and wind speed. Observation data may be directly transmitted to the weather prediction device 20 from the radar device.

The information providing device 10 also transmits forecast data to the weather forecasting device 20 . The prediction data is, for example, a numerical forecast model GPV derived from a meso scale model (MSM), a global numerical forecast model (GSM), or the like. This prediction data may include, in addition to the types of data that can be acquired from the observed data, data of other types derived from the data.

FIG. 2 is a diagram showing the functional configuration of the weather forecasting device 20. As shown in FIG. The weather forecasting device 20 includes, for example, a communication interface 22, a data storage unit 24, a numerical weather prediction calculation unit 26, a forecast data storage unit (prediction result output unit) 28, an index derivation unit 30, and a probability derivation unit 32. , an information generation unit 34 , a display unit 36 , a map generation unit (criterion setting unit) 38 , and a storage unit 40 . The data storage unit 24, the numerical weather prediction calculation unit 26, the forecast data storage unit 28, the index derivation unit 30, the probability derivation unit 32, and the information generation unit 34 are provided by a processor such as a CPU (Central Processing Unit). It may be realized by executing a program stored in the storage unit 40 . Further, all or part of these functional units are realized by hardware such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), and FPGA (Field-Programmable Gate Array), and these functional units may have a circuit configuration for realizing the function of Also, these functional units may be realized by cooperation of software and hardware.

The storage unit 40 is realized by, for example, nonvolatile storage media such as ROM (Read Only Memory), flash memory, and HDD (Hard Disk Drive), and volatile storage media such as RAM (Random Access Memory) and registers. be done. The storage unit 40 stores programs executed by the processor, information such as a determination map 42 and a conditional probability map 43, which will be described later.

The data storage unit 24 acquires through the communication interface 22 the observation data and prediction data that are the basis of the weather prediction transmitted from the information providing device 10 . The numerical weather prediction calculator 26 predicts the future (predetermined time) weather conditions based on the observation data, forecast data, etc. acquired by the data storage unit 24 and a predetermined prediction model. The predetermined forecast model is CReSS (Cloud Resolving Storm Simulator), WRF (the Weather Research & Forecasting model), or the like.

The numerical weather forecast calculation unit 26 uses the forecast data stored in the data storage unit 24 as initial values, assimilates the observation data into the forecast model, and obtains consistency with the forecast data, thereby obtaining a more subdivided three-dimensional grid. Make weather forecasts. A finer three-dimensional grid is a three-dimensional grid that is finer than the three-dimensional grid of the size targeted by the meso-scale numerical forecast model, the global numerical forecast model, and the like. In addition, the numerical weather forecast calculation unit 26 can derive more detailed information than the information targeted by the meso-scale numerical forecast model, the global numerical forecast model, and the like.

The prediction data storage unit 28 stores information about future weather conditions predicted by the numerical weather prediction calculation unit 26 .

Based on the future meteorological conditions output from the forecast data storage unit 28, the index deriving unit 30 calculates the Richardson number (first index) indicating the likelihood of occurrence of turbulence, the temperature indicating the degree of change in the altitude direction of the potential temperature At least two indices are derived: the gradient (second index) and the vertical flow (third index) indicating the strength of the updraft or downdraft. Also, the index derivation unit 30 derives at least two indexes based on the past weather conditions acquired in the data storage unit 24 .

The probability derivation unit 32 classifies the derived indices based on predetermined criteria, and derives the probability of occurrence of turbulence based on the pattern of the classification results. Details of the processing of the index derivation unit 30 and the probability derivation unit 32 will be described later.

The information generation unit 34 generates information to be presented to the user based on the probability of occurrence of turbulence derived by the probability derivation unit 32, and causes the display unit 36 to display the generated information. The display unit 36 includes, for example, a display device such as an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence) display.

The map generation unit 38 classifies at least two indices derived by the index derivation unit 30 based on past weather conditions based on the criteria for each index. Then, based on the conditional probability, a criterion for the classification result pattern is set. The map generator 38 generates the determination map 42 based on conditional probabilities, for example. Also, the map generator 38 updates the determination criteria or the determination map 42 .

FIG. 3 is a flow chart showing the flow of processing executed by the weather prediction device 20. As shown in FIG. The following process derives the probability of occurrence of turbulence for each three-dimensional grid. In the following description, it is assumed that the index deriving unit 30 derives three indices, a first index, a second index and a third index.

First, the data storage unit 24 acquires the observation data and prediction data that are the basis of the weather prediction transmitted from the information providing device 10 (step S100). Next, the numerical weather prediction calculation unit 26 predicts future weather conditions based on the observation data and prediction data for each grid (step S102).

Next, the index deriving unit 30 derives the Richardson number (step S104). The Richardson number is an index that indicates the likelihood of occurrence of turbulence, and is derived from the following formula (1). In the numerator of formula (1), the temperature gradient (details will be described later) indicates thermodynamic stability, and the denominator vertical shear (vertical wind shear) indicates kinematic instability. In the formula, "Ri" is Richardson's number, "g" is the acceleration of gravity, "θ" is the potential temperature, "θ ₀ " is the average potential temperature in the range of the altitude direction (z), "u" is the east-west wind speed, " v” is the north-south wind speed. Turbulence tends to occur when the Richardson number is small, that is, when kinematic instability outweighs thermodynamic stability.

Next, the index deriving unit 30 derives the potential temperature gradient (step S106). The potential temperature gradient is the rate of change in the altitude direction of the potential temperature (temperature when a certain air mass moves to the ground surface), and indicates thermodynamic stability. The rate of change is the rate of change of the potential temperature in the altitude (plus or minus z) direction of the grid of interest. For example, it is a value obtained by adding the degree of change of the potential temperature in the minus z direction and the plus z direction with respect to the potential temperature of the target grid and dividing by two. Turbulence tends to occur when the potential temperature gradient is small, or when the potential temperature gradient is large but the vertical shear is larger than that and the Richardson number is small.

Next, the index deriving unit 30 derives a vertical current (step S108). The vertical flow is an index (the absolute value of the vertical component of the wind direction on the grid) that indicates the strength of the upflow or downflow. Vertical flow is a factor that can also trigger the generation of turbulence, and the greater the absolute value, the greater the shaking when encountering it.

Next, the probability deriving unit 32 derives the probability of occurrence of turbulence based on the index derived in steps S104 to S108 and the determination map 42 (step S110). For example, the probability derivation unit 32 classifies each of the first index to the third index as “large” when larger than the reference value set for each index, and as “small” when smaller than the reference value set for each index. By applying the classification result to the determination map 42, the probability of occurrence of turbulence is derived.

FIG. 4 is a diagram showing an example of the determination map 42. As shown in FIG. For example, 8 cases (Case 1 to Case 8) are generated by combining a large or small Ri number, a large or small potential temperature gradient, and a large or small vertical flow. "Case 1" has the lowest probability of occurrence of turbulence, followed by "Case 2", . . . "Case 8".

In this embodiment, each index is described as being classified into two values of "large" or "small", but may be classified into three or more values. In this case, the determination map 42 is formed with 9 or more Cases, each of which is given a probability.

An example of a method for generating the determination map 42 will now be described. The determination map 42 is generated using a Bayesian network technique that is formed by the causal relationship between the occurrence of turbulence and the index derived by the index derivation unit 30 . A Bayesian network is a method of aggregating probabilistic interactions for a chain of uncertain events. FIG. 5 is a diagram showing an example of a Bayesian network in this embodiment. The illustrated example shows that the probability of occurrence of turbulence (1) has a causal relationship with each weather element (Richardson number (2), potential temperature gradient (3), and vertical flow (4)). Conditional probability maps 43A to 43C are generated for each causal weather element for the probability of occurrence of turbulence (1). For example, based on past weather conditions, the probabilities (conditional probabilities) of large and small numbers of Ri in cases where turbulence has occurred and cases where turbulence has not occurred are aggregated. For potential temperature gradients and vertical currents, the probabilities of magnitude of meteorological elements are similarly aggregated for cases where turbulence occurs and those where turbulence does not occur.

In order to generate the conditional probability map 43, the map generation unit 38 acquires weather information obtained at predetermined sampling intervals during a predetermined period. Then, the map generating unit 38 processes the acquired weather information for each sampling interval and for each grid, and if it is larger than the reference value set for each index, it is “large”, and if it is smaller than it, it is “small”. ”.

The above-mentioned first index to third index are evidence nodes (child nodes) that can be determined from information on future weather conditions predicted by the numerical weather prediction calculation unit 26 at the determination time, and determination results of the respective evidence nodes. , a probability value defined in the conditional probability map 43 is selected. Bayes' theorem is then used to back-estimate from the probability values of each node to determine the probability of occurrence of turbulence.

In the Bayesian network configuration, for example, eight cases can be combined depending on the size of the child node, and the probability of occurrence of turbulence in eight stages corresponding to each case is derived. By considering the combination of such child nodes, it is possible to express the interrelationship between indices and the cause of turbulence. For example, when the Richardson number is small and the potential gradient and vertical flow are large, the turbulence associated with the Kelvin-Helmholtz wave, which is generally known as one of the causes of turbulence, is likely to occur. there is Thus, the determination map 42 of FIG. 4 described above is generated.

Note that the map generation unit 38 may update the created conditional probability map 43 and determination map 42 on a daily basis based on the weather information acquired by the data storage unit 24 . As a result, the latest information can be reflected in the determination map 42, so that the probability of occurrence of turbulence can be derived with higher accuracy.

Further, in the above example, the probability of occurrence of turbulence is derived based on three indices, but the probability of occurrence of turbulence may be derived based on two indices. For example, a probability that turbulence will occur may be given to the combination of the Richardson number Ri and the potential temperature gradient.

Returning to the description of FIG. Next, the information generation unit 34 generates an image including information indicating the possibility of occurrence of turbulence based on the result derived in step S110, and causes the display unit 36 to display the generated image (step S112). . Specifically, the information generator 34 divides the probability (posterior probability) of each Case by the probability (posterior probability) of the Case with the lowest probability (posterior probability). As a result, it is possible to derive relative risk information indicating how many times turbulence is more likely to occur than the safest case, which serves as a reference, and display the derived information on the display unit 36 . This completes the processing of one routine in this flow chart.

In addition, risk information may be derived based on cases other than the case with the lowest probability (posterior probability). If there is a specific weather condition, the corresponding case may be used as the standard.

FIG. 6 is a diagram showing an example of an image displayed on the display unit 36. As shown in FIG. The illustrated example shows the risk of occurrence of turbulence in a cross section cut out at a predetermined altitude at a certain time. For example, the information generating unit 34 generates images by giving colors and shadings to each case, or dividing them by patterns, etc., and displays the generated images on the display unit 36 or display units of other devices.

As mentioned above, by deriving the probability of occurrence of turbulence using the Bayesian network method, it is possible to consider the uncertainty contained in weather forecasts, and also consider the combination of large and small indices placed in child nodes. By doing so, it is possible to express the interrelationship between the indicators and the factors that cause turbulence, so that the reliability of the risk level information can be improved. In addition, by deriving the degree of risk relative to a certain standard Case, it is easy to grasp routes and time periods with a lower degree of risk. By providing this information to aviation-related parties, it is expected that it will contribute to the safer and more optimal operation of aircraft and the efficiency of operational decisions.

According to the embodiment described above, the probability derivation unit 32 derives the possibility of occurrence of turbulence based on the index derived by the index derivation unit 30 and the determination map 42 generated by the map generation unit 38. Thus, the occurrence of turbulence can be predicted more accurately.

According to at least one embodiment described above, based on the input weather conditions, the first index indicating the likelihood of occurrence of turbulence, the second index indicating the degree of change in the altitude direction of the potential temperature, and the an index derivation unit 30 for deriving at least two of the third indexes indicating the strength of the airflow or downdraft; By having the probability derivation unit 32 for classifying based on the criteria of and deriving the probability of occurrence of turbulence based on the pattern of the classification result, the occurrence of turbulence can be predicted more accurately.

While several embodiments of the invention have been described, these embodiments have been presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and modifications can be made without departing from the scope of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, as well as the scope of the invention described in the claims and equivalents thereof.

1 weather forecasting system 10 information providing device 20 weather forecasting device 22 communication interface 24 data storage unit 26 numerical weather forecast calculation unit 28 prediction data storage unit 30 index derivation unit 32 Probability derivation unit 34 Information generation unit 36 Display unit 38 Map generation unit 40 Storage unit 42 Decision map 43 (43A to 43C) Conditional probability map

Claims (6)

Based on the input meteorological conditions, the first index indicates the likelihood of occurrence of turbulence, the second index indicates the degree of change in the altitude direction of potential temperature, and the third index indicates the strength of updraft or downdraft. an index derivation unit for deriving a plurality of indices including
Each of the plurality of indices derived based on future weather conditions by the index deriving unit is classified based on criteria for each index, and turbulence is detected based on the pattern of the classification result of the plurality of indices and criteria. a probability derivation unit that derives the probability of occurrence;
Obtaining a conditional probability that turbulence will occur for each of the plurality of indicators having a causal relationship with the occurrence of the turbulence for each of the patterns of classification results obtained by classifying each of the plurality of indicators based on the criteria for each indicator; A criterion setting unit that sets the criterion for the pattern of the classification result based on the conditional probability, and is a Bayesian network method formed by a causal relationship between the occurrence of the turbulence and the plurality of indices. A criterion setting unit that determines the conditional probability that turbulence will occur using
A weather forecasting device. wherein the first index is the Richardson number;
The weather prediction device according to claim 1. a prediction result output unit that acquires information about weather, predicts future weather conditions based on the acquired information, and outputs the predicted future weather conditions to the index derivation unit;
The weather prediction device according to claim 1 or 2. An information generation unit that generates an image containing information indicating the probability of occurrence of turbulence derived by the probability derivation unit and displays the generated information on a display unit, further comprising:
The weather prediction device according to any one of claims 1 to 3. the computer
Based on the input meteorological conditions, the first index indicates the likelihood of occurrence of turbulence, the second index indicates the degree of change in the altitude direction of potential temperature, and the third index indicates the strength of updraft or downdraft. an index derivation process for deriving a plurality of indices including
Each of the plurality of indices derived based on future weather conditions by the index deriving process is classified based on criteria for each index, and turbulence is detected based on the pattern of the classification results of the plurality of indices and criteria. A probability derivation process for deriving the probability of occurrence;
Obtaining a conditional probability that turbulence will occur for each of the plurality of indicators having a causal relationship with the occurrence of the turbulence for each of the patterns of classification results obtained by classifying each of the plurality of indicators based on the criteria for each indicator; A setting process for setting the criterion for the pattern of the classification result based on the conditional probability, using a Bayesian network technique formed by the causal relationship between the occurrence of the turbulence and the plurality of indicators. A setting process for obtaining the conditional probability that turbulence will occur at
A weather forecasting method that performs to the computer,
Based on the input meteorological conditions, the first index indicates the likelihood of occurrence of turbulence, the second index indicates the degree of change in the altitude direction of potential temperature, and the third index indicates the strength of updraft or downdraft. an index derivation process for deriving a plurality of indices including
Each of the plurality of indices derived based on future weather conditions by the index deriving process is classified based on criteria for each index, and turbulence is detected based on the pattern of classification results of the plurality of indices and criteria. A probability derivation process for deriving the probability of occurrence;
Obtaining a conditional probability that turbulence will occur for each of the plurality of indicators having a causal relationship with the occurrence of the turbulence for each of the patterns of classification results obtained by classifying each of the plurality of indicators based on the criteria for each indicator; A setting process for setting the criterion for the pattern of the classification result based on the conditional probability, using a Bayesian network technique formed by the causal relationship between the occurrence of the turbulence and the plurality of indicators. A setting process for obtaining the conditional probability that turbulence will occur at
program to run.

Images