Control device, control method, and control program
Patent Information
- Application Number
- PCT/JP2025/011444
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-03-24
- Publication Date
- 2026-10-01
Smart Images
Figure JP2025011444_01102026_PF_FP_ABST
Abstract
Description
Control device, control method, and control program
[0001] The present disclosure relates to a control device, a control method, and a control program.
[0002] In the field of weather and oceanographic prediction, the process of generating initial values for prediction simulation of observation values using observation data (data assimilation) is known as a technology that complementarily connects observation and prediction simulation based on numerical prediction models (hereinafter referred to as prediction sim) (see Non-Patent Documents 1 and 2).
[0003] There is also a known technology that executes prediction sim and estimates the rate of change in the predicted value with respect to changes in the initial value as sensitivity (see Non-Patent Document 3).
[0004] Japan Meteorological Agency, "Chapter 3 Data Assimilation", [online], [retrieved February 28, Reiwa 7], <URL: https: / / www.jma.go.jp / jma / kishou / books / nwptext / 45 / 1_chapter3.pdf>Miyoshi, et al., "Data Assimilation in Meteorology", [online], [retrieved February 28, Reiwa 7], <URL: https: / / www.metsoc.jp / tenki / pdf / 2007 / 2007_04_0015.pdf>Meteorological Society of Japan, "Forecast Sensitivity Analysis", [online], [retrieved February 28, Reiwa 7], <URL: https: / / www.metsoc.jp / tenki / pdf / 2012 / 2012_11_1039.pdf>
[0005] The technologies of Non-Patent Documents 1 and 2 have the following problems. To improve the prediction accuracy of prediction sim, it is desirable to sufficiently and accurately prepare observation data for generating initial values. However, the amount of observation data in sea areas is insufficient relative to the number of grids on the sea surface, which is the calculation unit of prediction sim.
[0006] The technology described in Non-Patent Document 3 has the following problem: Since sensitivity to initial values is not synonymous with sensitivity to observed data, in order to see the effect of observed data on prediction simulations, it is necessary to run the prediction simulation again using observed data in the highly sensitive region and evaluate the quality of the prediction results themselves. In other words, the prediction simulation must be run many times, which is computationally expensive.
[0007] This disclosure is made in view of the above circumstances, and the purpose of this disclosure is to provide a technology that enables the efficient acquisition of observational data effective for prediction.
[0008] A control device according to one aspect of the present disclosure includes: a construction unit that constructs a weather estimation model for estimating weather observation values using weather observation data and calculates statistical information related to the observation accuracy of the weather estimation model; a derivation unit that derives observation position candidates to improve the observation accuracy based on the statistical information; and a control unit that determines an observation position from among the observation position candidates based on weather forecast information and sets the observation position in an observation instrument.
[0009] A control method in one aspect of the present disclosure is a control method performed by a control device, comprising: constructing a weather estimation model for estimating weather observation values using weather observation data; calculating statistical information related to the observation accuracy of the weather estimation model; deriving candidate observation locations to improve the observation accuracy based on the statistical information; determining an observation location from among the candidate observation locations based on weather forecast information; and setting the observation location in the observation equipment.
[0010] A control program in one aspect of the present disclosure causes a computer to perform the following processes: construct a weather estimation model for estimating weather observation values using weather observation data; calculate statistical information related to the observation accuracy of the weather estimation model; derive candidate observation locations to improve the observation accuracy based on the statistical information; and determine an observation location from among the candidate observation locations based on weather forecast information and set the observation location in an observation instrument.
[0011] According to this disclosure, observational data that is effective for prediction can be efficiently acquired.
[0012] Figure 1 shows an example of the overall system configuration. Figure 2 is a flowchart showing an example of the overall system operation. Figure 3 is a reference diagram used when explaining the overall system operation. Figure 4 shows an example of the control device configuration. Figure 5 is a sequence diagram showing an example of the overall system operation. Figure 6 is a reference diagram used when explaining a specific example. Figure 7 shows an example of observation frequency information. Figure 8 shows an example of constructing a sea surface wind model. Figure 9 shows an example of constructing a sea surface wind model. Figure 10 shows an example of the number of observations and error rate per distance. Figure 11 shows an example of candidate observation position information. Figure 12 shows an example of moving observation equipment. Figure 13 shows an example of the control device hardware configuration.
[0013] Embodiments of this disclosure will be described below with reference to the drawings. In the drawings, the same parts are denoted by the same reference numerals and their descriptions are omitted.
[0014] This disclosure aims to efficiently acquire observational data that is effective for prediction.
[0015] To achieve this objective, this disclosure constructs a meteorological estimation model for estimating meteorological observation values using meteorological observation data, and sets observation positions on observation equipment to improve the observation accuracy based on statistical information related to the observation accuracy of the meteorological estimation model.
[0016] For example, observation equipment can be moved to locations with low observation frequency and used for observations. This allows for the efficient acquisition of observational data that is effective for prediction. As a result, the accuracy of weather estimation models can be improved, and the accuracy of initial values for prediction simulations can be improved. Furthermore, the computational cost of prediction simulations is reduced, and the accuracy of the prediction simulations themselves is also improved.
[0017] Figure 1 shows an example of the overall configuration of the system according to this embodiment.
[0018] The system comprises a control device 1 that constructs a weather estimation model using weather observation data, a prediction device 2 that generates initial values for a prediction simulation using the observation values estimated by the weather estimation model and executes a prediction simulation based on these initial values, a client terminal 3 that instructs the start or end of observation, and an observation device 4 that performs observations.
[0019] Figure 2 is a flowchart showing an example of the overall operation of the system. Figure 3 is a reference diagram used to explain the overall operation of the system.
[0020] The control device 1 constructs a weather estimation model for estimating weather conditions for initial value generation using existing observation data observed by the observation instrument 4, and calculates statistical information related to the accuracy of the weather estimation model during its construction (step S1).
[0021] Next, in order to improve the accuracy of the weather estimation model, the control device 1 derives observation requirements to be observed by the observation instrument 4 (for example, candidate observation locations, physical parameters) based on the statistical information calculated in step S1 (for example, frequency distribution of observation locations, error between observation data from the observation instrument 4 and observation values estimated by the weather estimation model) (step S2).
[0022] Next, the control device 1 determines the observation location based on the observation requirements derived in step S2 and external weather forecast information, and sets the determined observation requirements (e.g., observation location, physical parameters) in the observation instrument 4 (step S3).
[0023] For example, the control device 1 determines an observation location from among multiple candidate observation locations based on weather forecast information obtained from the Japan Meteorological Agency (e.g., typhoon track forecast, typhoon intensity forecast). For example, the control device 1 determines the observation location with the highest priority from among multiple candidate observation locations based on the position and speed of the observation equipment 4 in operation.
[0024] Next, the observation instrument 4 moves to the observation position set by the control device 1, performs observations, and transmits the observed data to the control device 1 (step S4).
[0025] Next, the control device 1 updates the weather estimation model by adding the observation data from the destination transmitted from the observation instrument 4 (observation data based on the observation requirements derived in step S2) to the existing observation data, and transmits the observation values estimated by the updated weather estimation model to the prediction device 2 (step S5).
[0026] Finally, the prediction device 2 synthesizes the observed values transmitted from the control device 1 and the analyzed values output from the prediction sim to generate initial values, and then executes the prediction sim using the generated initial values (step S6).
[0027] The control device 1 repeatedly executes steps S2 to S6. When the error between the observation data from the observation instrument 4 and the observed values estimated by the weather estimation model reaches a threshold, the control device 1 may determine that the construction of the weather estimation model is complete and finalize the weather estimation model.
[0028] The prediction simulation and initial value generation performed by prediction device 2 are based on existing technologies. The prediction simulation is a process that uses a numerical weather prediction model to simulate the prediction result at a predicted time (T+n) from initial values at a predetermined time T. Initial value generation is the process called "data assimilation" mentioned at the beginning. In other words, it is a process that generates initial values for prediction at time T by combining the observed data at time T with the analyzed values at time T output from the prediction simulation.
[0029] Conventionally, initial values were generated using observation data from observation instrument 4, as shown by the dashed arrow in Figure 3. In contrast, this embodiment generates initial values using observation values estimated by a weather estimation model with improved accuracy. This allows for the efficient acquisition of observation data that is effective for prediction.
[0030] Figure 4 shows an example of the configuration of the control device 1.
[0031] The control device 1 is connected to the forecasting device 2, the client terminal 3, and the observation equipment 4 via a wireless / wired network N. The control device 1 is also connected to a server of a weather forecasting agency (not shown) via the network N to acquire external weather forecast information, etc.
[0032] The control device 1 comprises a processing unit 10 and a storage unit 20. The processing unit 10 and the storage unit 20 are mutually communicative. The processing unit 10 comprises a construction unit 101, an output unit 102, and a control unit 103. The storage unit 20 comprises a first storage unit 201, a second storage unit 202, and a third storage unit 203.
[0033] The construction unit 101 has the function of executing step S1 in Figure 2.
[0034] For example, the construction unit 101 has the function of reading observation data from the observation instrument 4 from the first storage unit 201 and constructing a weather estimation model for estimating desired weather observation values using the read observation data.
[0035] For example, the construction unit 101 has a function to calculate statistical information related to the observation accuracy of the weather estimation model. The construction unit 101 also has a function to store various parameters necessary for constructing the weather estimation model and the statistical information as weather estimation model information in the second storage unit 202.
[0036] The construction unit 101 has the function of executing step S5 in Figure 2.
[0037] For example, the construction unit 101 has the function of updating the weather estimation model by adding the observation data after the position change transmitted from the observation instrument 4 to the existing observation data, and transmitting the observed values of the updated weather estimation model to the prediction device 2 as observed values for generating initial values for the prediction sim of the observed target.
[0038] The derivation unit 102 has the function of executing step S2 in Figure 2.
[0039] For example, the derivation unit 102 reads statistical information from the second storage unit 202 and has the function of deriving observation locations (for example, observation locations with insufficient observation data) that improve the observation accuracy in the weather estimation model, based on the observation locations and errors in the observation data when constructing the weather estimation model. The derivation unit 102 also has the function of storing the derived observation locations as candidates for observation locations in the third storage unit 203.
[0040] The control unit 103 has the function of executing step S3 in Figure 2.
[0041] For example, the control unit 103 reads out candidate observation locations from the third storage unit 203, and further reads out the current position of the observation instrument 4, the observation path of the observation instrument 4, and the estimated speed of the observation instrument 4. Based on the read out candidate observation locations and external weather forecast information, the control unit 103 has a function to store the determined observation location in the third storage unit 203.
[0042] For example, the control unit 103 has a function of moving the observation device 4 to the determined observation position by setting the determined observation position in the observation device 4.
[0043] The first storage unit 201 has, for example, a function of storing observation data obtained by the observation device 4.
[0044] The second storage unit 202 has, for example, a function of storing weather estimation model information.
[0045] The third storage unit 203 has, for example, a function of storing observation position candidates, the current position of the observation device 4, the observation path of the observation device 4, the estimated speed of the observation device 4, and the observation position.
[0046] FIG. 5 is a sequence diagram showing an example of the overall operation of the system.
[0047] The client terminal 3 transmits a start signal for starting observation to the observation device 4 (step S101), and transmits the start signal to the control device 1 (step S102).
[0048] The observation device 4 determines whether it is necessary to continue observation (step S103).
[0049] When it is determined that observation needs to be continued, the observation device 4 performs observation at the current observation position (step S104), and transmits the obtained observation data to the control device 1 (step S105). Thereafter, the first storage unit 201 stores the observation data from the observation device 4 (step S106).
[0050] When it is determined that observation does not need to be continued, the observation device 4 notifies the client terminal 3 that no observation will be performed (step S107). Thereafter, the process is terminated.
[0051] After step S106, the construction unit 101 determines whether it is necessary to construct a weather estimation model (step S108).
[0052] If it is determined that a weather estimation model needs to be constructed, the construction unit 101 reads observation data from the first storage unit 201 (step S109) and reads various parameters necessary for constructing the weather estimation model from the second storage unit 202 (step S110). The construction unit 101 uses the read observation data and various parameters to construct a weather estimation model for estimating desired weather observation values, calculates statistical information related to the observation accuracy of the weather estimation model (step S111), and transmits the calculated statistical information to the second storage unit 202 (step S112). Subsequently, the second storage unit 202 stores the statistical information (step S113).
[0053] If it is determined that the weather estimation model will not be constructed, the construction unit 101 notifies the client terminal 3 that the weather estimation model will not be constructed (step S114). After that, the process is terminated.
[0054] After step S113, the derivation unit 102 reads statistical information from the second storage unit 202 (step S115), and based on the read statistical information, derives one or more observation locations where observation data is missing, such as the observation location and the error in the observation data when constructing the weather estimation model (step S116), and stores these observation locations as observation location candidates in the third storage unit 203 (step S117). Subsequently, the third storage unit 203 stores the observation location candidates (step S118).
[0055] The control unit 103 reads out candidate observation locations from the third storage unit 203, and further reads out the current position of the observation instrument 4, the observation path of the observation instrument 4, and the estimated speed of the observation instrument 4 (step S119). Based on the read out candidate observation locations and external weather forecast information, the control unit 103 determines the observation location (step S120).
[0056] The control unit 103 stores the observation position in the third storage unit 203 (step S121). Subsequently, the third storage unit 203 stores the observation position (step S122). The control unit 103 sets the observation position to the observation instrument 4 (step S123).
[0057] The observation device 4 moves to the observation position (step S124) and returns to step S103. If it is determined that observation should be continued, it performs observations at the new observation position (step S104') and transmits the observed data to the control device 1 (step S105').
[0058] Subsequently, the construction unit 101 updates the weather estimation model by adding the observation data after the position change to the existing observation data (step S111'), and transmits the observation values estimated by the updated weather estimation model to the prediction device 2 as observation values for initial value generation (not shown in Figure 5).
[0059] Next, I will explain a specific example.
[0060] The prediction target is typhoons in ocean areas. The initial physical parameter used in the typhoon prediction simulation is the wind (hereinafter referred to as sea surface wind), which represents the typhoon intensity. For sea surface wind, a weather estimation model that utilizes observational data is referred to as the sea surface wind model.
[0061] The sea surface wind under a typhoon has a distinctive positional relationship with respect to the typhoon's direction of movement. Therefore, it is important to collect sea surface wind speeds for each quadrant in the typhoon coordinate system shown in Figure 6, where the typhoon's center is the origin and the direction of the typhoon's movement is the positive y-axis of the x-y plane.
[0062] To manage wind speeds for multiple typhoons using a unified standard, the observation location is defined as the distance from the typhoon's center O to the actual observation location P, divided by the radius R of the typhoon's maximum wind speed. The wind speed is defined as the actual wind speed V divided by the typhoon's maximum wind speed.
[0063] For example, suppose the center position O of typhoon n is (LatT_n, LonT_n), the maximum wind speed is Vmax_n, and the radius of the maximum wind speed is Rmax_n. The wind speed observation position P is (LatO_n, LonO_n), and the wind speed value is Vo_n. In this case, the normalized observation position (LatSO_n, LonSO_n) is calculated by equation (1). The normalized wind speed Vos_n is calculated by equation (2).
[0064]
[0065]
[0066] Standardized sea winds are generally converted to an altitude of 10 meters (References 1-3).
[0067] Reference 1: WG Large, et al., “Observations and Simulations of Upper-Ocean Response to Wind Events during the Ocean Storms Experiment,” Journal of Physical Oceanography, 25, November 1995, pp. 2831-2852.
[0068] Reference 2: SE Zedler, et al., “Analyses and simulations of the upper ocean's response to Hurricane Felix at the Bermuda Testbed Mooring site: 13-23 August 1995”, Journal of Geophysical Research, Vol. 107, No. C12, 3232, doi:10.1029 / 2001JC000969, 2002, pp. 25-1-25-9.
[0069] Reference 3: WG Large and others, “Open Ocean Momentum Flux Measurements in Moderate to Strong Winds”, Journal of Physical Oceanography, 11, 1981, p.324-p.336.
[0070] First, the construction unit 101 determines the observation frequency for each quadrant. Figure 7 shows an image of the observation frequencies for quadrants Q1 to S4.
[0071] The quadrant divisions can be four (Q1-Q4), or even eight (further subdivided), or other numbers besides four. Also, since the observation position is normalized by the radius R of the maximum wind speed, the maximum wind speed is on a circle with radius 1, and the tendency of the inflow angle of the sea surface wind differs inside and outside this circle. Therefore, the area can be divided into the inside of the maximum wind speed circle (Q1_in) and the outside of the maximum wind speed circle (Q1_out).
[0072] Furthermore, the distance from the center O can be fixed by setting a threshold value (e.g., 10) in the normalized coordinate system. Alternatively, it can be dynamically set to be within the circle of the typhoon's strong wind area (normalized by the radius of the typhoon's maximum wind speed), taking into account the typhoon's projected path.
[0073] Next, the construction unit 101 constructs a sea surface wind model using standardized observation data. Examples of sea surface wind model construction are shown in Figures 8 and 9.
[0074] In a normalized coordinate system, if r* is the distance from the center O to the observation position, the normalized wind speed Vos_n increases from the center O to around r* = 1, as shown in Figure 8, and then decreases beyond that point. Also, as shown in Figure 9, the angle of incidence (inflow angle_n) of the sea surface wind tends to be negative when it is smaller than r* and positive when it is larger.
[0075] When modeling, the entire model can be approximated by a polynomial, or approximation curves can be calculated separately at the boundary of r*=1. Also, since r* has infinitely many values, intervals can be defined and set as r*_a1, r*_a2, r*_a3, ... For example, the distance reached by the approximate movement speed of observation instrument 4 can be used as the boundary.
[0076] Next, the construction unit 101 calculates the error rate (error information) with respect to the sea surface wind model for each quadrant Q and each distance r*. Figure 10 shows an image of the error information for each distance r* in quadrant Q1.
[0077] In the case of the sea surface wind model with a curve as shown in Figure 8, the error rate is calculated using equation (3). In the case of the sea surface wind model with a straight line as shown in Figure 9, the error rate is calculated using equation (4).
[0078]
[0079]
[0080] Error information may be calculated using RMSE (Root Mean Square Error) or standard deviation. Correlation coefficients, confidence intervals, and prediction intervals may also be used to evaluate the magnitude of the error.
[0081] Next, the derivation unit 102 calculates the priority of observation position candidates based on the observation frequency for each quadrant (Figure 7) and error information from when the sea surface wind model was constructed (Figure 10), and derives the observation position candidates.
[0082] To improve the accuracy of wind speed estimation using the sea surface wind model, priority observation locations are derived by selecting areas where the overall number of wind speeds (number of observations) is insufficient, the distance to typhoons is insufficient during model construction, and the distance r* with a large error with the sea surface wind model. For example, locations with a small "number of observations during sea surface wind model construction" or a small "1 - error rate" in Figure 10 are designated as high-priority observation locations.
[0083] If the number of wind speeds (observations) is sufficient, then lower priority locations can be derived using the reverse approach. For example, the derivation can be performed under conditions such as sufficient overall observations and observations during the construction of the sea surface wind model, a higher number of observations in a given quadrant compared to other quadrants, and small errors during the construction of the sea surface wind model.
[0084] Priority can be expressed as either a priority number or a score value.
[0085] When determining priority in stages, first select the quadrant with the fewest "number of observations" in Figure 7, and then select the distance r* with the fewest "number of observations when constructing the sea surface wind model" or "1 - error rate" in Figure 10. Alternatively, you may select the distance r* with a large "error rate" instead of "1 - error rate".
[0086] Alternatively, the entire system could be evaluated with a score of S, and priority could be determined based on that score.
[0087]
[0088] The locations with lower priority can be obtained by extracting the lower priority items derived above.
[0089] Figure 11 shows an image of the candidate observation locations. The candidate observation locations will be updated whenever wind speed (observation data) is obtained. However, the construction of the sea surface wind model will be considered complete when the error information reaches a certain level. For example, this would be when the error rate is 5% or less across the entire area.
[0090] Next, the output unit 102 acquires typhoon forecast information published by the Japan Meteorological Agency and other organizations. This typhoon forecast information includes, for example, a track forecast (where the typhoon's center will move) and an intensity forecast (for example, central pressure, maximum wind speed, and radius of maximum wind speed).
[0091] Finally, the control unit 103 arbitrarily combines candidate observation locations, location information of observation instrument 4, observation path of observation instrument 4, estimated speed of observation instrument 4, and typhoon forecast information, determines the observation location using the combined information, and transmits the said observation location to observation instrument 4.
[0092] If there are multiple observation instruments 4, the destination is determined based on the position and movement speed of each observation instrument 4, prioritizing those with the highest priority for reliable observation. If the movement speed is insufficient and the instrument cannot reach the target location, it is excluded from the list of potential observation locations. Figure 12 shows an image of the movement of the observation instruments.
[0093] As described above, according to this embodiment, (1) a weather estimation model is constructed for estimating weather observation values using weather observation data, and statistical information related to the observation accuracy of the weather estimation model is calculated, (2) candidate observation locations that improve the observation accuracy are derived based on the statistical information, and (3) an observation location is determined from the candidate observation locations based on weather forecast information, and the observation location is set in the observation equipment, so that observation data that is effective for prediction can be acquired effectively and efficiently.
[0094] This disclosure is not limited to the embodiments described above. Numerous modifications are possible within the scope of the gist of this disclosure. For example, it can be applied to observing atmospheric temperature and humidity using a drone.
[0095] The control device 1 of this embodiment described above can be realized using a general-purpose computer system, for example, as shown in Figure 13, which includes a CPU 901, a memory 902, a storage 903, a communication device 904, an input device 905, and an output device 906. The memory 902 and the storage 903 are storage devices. In this computer system, each function of the control device 1 is realized when the CPU 901 executes a predetermined program loaded onto the memory 902.
[0096] The control device 1 may be implemented on a single computer. The control device 1 may be implemented on multiple computers. The control device 1 may also be a virtual machine implemented on a computer. The program for the control device 1 can be stored on a computer-readable recording medium such as an HDD, SSD, USB memory, CD, or DVD. A computer-readable recording medium is, for example, a non-transitory recording medium. The program for the control device 1 can also be distributed via a communication network.
[0097] 1 Control device 2 Prediction device 3 Client terminal 4 Observation equipment 10 Processing unit 20 Memory unit 101 Construction unit 102 Output unit 103 Control unit 201 First memory unit 202 Second memory unit 203 Third memory unit 901 CPU 902 Memory 903 Storage 904 Communication device 905 Input device 906 Output device
Claims
1. A control device comprising: a construction unit that constructs a weather estimation model for estimating weather observation values using weather observation data and calculates statistical information related to the observation accuracy of the weather estimation model; a derivation unit that derives candidate observation locations to improve the observation accuracy based on the statistical information; and a control unit that determines an observation location from among the candidate observation locations based on weather forecast information and sets the observation location in an observation instrument.
2. The control device according to claim 1, wherein the statistical information comprises a frequency distribution of observation locations and an error between the observation data from the observation instrument and the observed value estimated by the weather estimation model, the derivation unit calculates the priority of the observation location candidates based on the frequency distribution and the error, and the control unit determines the observation location candidate with the highest priority as the observation location based on the position and movement speed of the observation instrument.
3. A control method performed by a control device, comprising: constructing a weather estimation model for estimating weather observation values using weather observation data; calculating statistical information related to the observation accuracy of the weather estimation model; deriving candidate observation locations to improve the observation accuracy based on the statistical information; determining an observation location from among the candidate observation locations based on weather forecast information; and setting the observation location in the observation equipment.
4. A control program that causes a computer to execute the following processes: construct a weather estimation model for estimating weather observation values using weather observation data, and calculate statistical information related to the observation accuracy of the weather estimation model; derive candidate observation locations that improve the observation accuracy based on the statistical information; and determine an observation location from among the candidate observation locations based on weather forecast information, and set the observation location in the observation equipment.