JP7386136B2 - Cloud height measurement device, measurement point determination method, and cloud type determination method - Google Patents

Description

The present invention relates to a technique for realizing cloud height measurement over the entire sky by correcting the output of a low-accuracy cloud height estimation method using local cloud height values obtained from a small number of cloud height measurement devices.

Weather information is generally obtained from video data captured by meteorological satellites and weather forecasts based on general meteorology, and is disclosed not only within Japan but throughout the world. Some types of weather information are acquired using unique measurement methods, and the cost of measuring the information is high. However, this weather information is widely used by businesses and individuals, and has a wide variety of uses. The weather information that is generally disclosed in this way covers a wide area, and corresponds to an area of several kilometers square at most. Therefore, it is effective for predicting weather phenomena on a regional basis.

On the other hand, there are many cases where local weather information for a specific location is required. Taking aircraft operation management as an example, by obtaining pinpoint information about clouds and airflow on the aircraft's path, safe and efficient operations can be achieved. In particular, it is not hard to imagine that it is extremely important to observe weather information over airfields and around runways in order to ensure safety during takeoff and landing of aircraft. Airfields are generally equipped with a wide variety of observation equipment, and are constantly being visually observed by observers. One of these observation items is the altitude from the ground to the cloud level (hereinafter referred to as cloud height), which plays an important role in determining whether aircraft can take off or land. Currently, laser distance measuring devices such as ceilometers are widely used as a standard means of measuring cloud height.

A ceilometer is an optical device that uses a laser. Specifically, we irradiate a powerful laser beam, detect the optical signals scattered and reflected at the cloud base (here we use clouds as an example, but the target is not particularly limited), and detect the cloud based on the arrival time and signal strength. This is a device that can measure height with an accuracy of several meters. However, ceilometers can only measure a narrow range directly above the sensor. Therefore, it is not suitable for measuring the entire sky, including directly above and around airports and runways. Although it is possible to expand the measurement range by installing multiple ceilometers, it is not practical because ceilometers are generally expensive.

Image processing is an example of a cloud height measuring means that does not use such an optical device. For example, there are methods such as stereo cameras that measure the distance to an object based on the parallax between two cameras, and methods that utilize AI to directly estimate the distance from two-dimensional images. Its use is progressing. Although images have the advantage of being able to measure the entire sky, clouds have less color information and texture than general objects, so there is a problem in that the distance measurement accuracy of the image processing technology described above decreases.

Therefore, there is a need for technology that can accurately measure cloud height information for the entire sky at any point. Patent Document 1 discloses a technique of calculating cloud temperature information using an infrared radiation thermometer and measuring cloud base height through predetermined calculation processing. Additionally, Patent Document 2 discloses a technique for measuring cloud height, cloud movement speed, and wind speed by shooting a stereo camera composed of two wide-angle cameras toward the zenith.

Japanese Patent Application Publication No. 2004-170350 JP2019-60754A

The technology described in Patent Document 1 corrects the effects of water vapor and carbon dioxide in the atmosphere by using an optical filter that transmits a specific wavelength band when photographing the sky using an infrared radiation thermometer. Accurate temperature distribution can be obtained. Cloud height can be estimated using calculations that take into account the temperature gradient of the temperature lapse rate in the troposphere, the dry adiabatic lapse rate, and the wet adiabatic lapse rate for the obtained temperature. However, although the above-mentioned calculation is effective in the international standard atmosphere, it is not robust against changes in atmospheric conditions, and therefore various coefficients must be obtained in advance, resulting in an increase in the number of observation items.

In the technology described in Patent Document 2, a stereo camera is configured with two cameras whose installation locations have known altitude, longitude, and latitude, and the azimuth and zenith angle of the subject are determined from images taken by each camera at the same time. read out. The process assumes the altitude of the subject, calculates the longitude and latitude based on the assumed altitude, and then assumes a new altitude until the values match for each camera or the difference falls within a preset tolerance. Cloud height can be measured repeatedly. However, it is generally difficult to accurately obtain the positions and installations of cameras. Furthermore, for purposes of measuring cloud bases, the distance between the two cameras constituting the stereo camera (referred to as the baseline length) needs to be approximately 50 meters to 300 meters, which limits the means for calibration. In addition, since optical sensors in general, not just wide-angle cameras, are industrial products, there are individual differences, and even if the housing is installed horizontally, the optical axis does not necessarily go perpendicular to the ground, so the zenith position in the image Separate calibration is required for accurate determination. Although Patent Document 2 discloses a method of executing based on this calibration process, there is a problem that stars cannot be observed depending on weather conditions or seasons, making it impossible to measure cloud height.

The present invention uses local but highly accurate cloud height information acquired from an optical device such as a ceilometer to correct low-accuracy wide-ranging cloud height information acquired from images. The purpose is to measure height with high precision.

The all-sky cloud height measurement device according to the present invention includes a local cloud height measurement unit that receives data from a laser and measures the heights of a plurality of cloud bases in a local range, and an all-sky image data from a sensor. a low-accuracy cloud height estimator that receives the image data and estimates the relative heights of a plurality of cloud bases based on the image data; and a cloud height correction unit that corrects the plurality of cloud bases estimated by the low-precision cloud height estimation unit using the heights of the plurality of cloud bases measured by the local cloud height measurement unit. It is configured as an all-sky cloud height measuring device.

According to the present invention, it is possible to measure cloud height over the entire sky with high precision.

FIG. 2 is a diagram showing an example of the configuration of a total sky cloud height measuring device. FIG. 2 is a diagram for explaining the equipment configuration of a total sky cloud height measuring device. FIG. 3 is a diagram for explaining input data of a low-precision cloud height estimator. It is a figure showing an example of low precision cloud height information. FIG. 3 is a diagram for explaining the nature of low-precision cloud height information. FIG. 2 is a diagram for explaining a measurement method using an all-sky cloud height measuring device. It is a figure for demonstrating the processing flow by which a cloud height correction part determines the position of a measurement point. FIG. 3 is a diagram for explaining a method of setting measurement point candidates. FIG. 7 is a diagram illustrating a specific example in which a cloud base correction unit determines a measurement point. It is a figure showing an example of recorded local cloud height and low-precision cloud height. FIG. 3 is a diagram for explaining a specific example of correcting a cloud height value. It is a figure which shows the example of the case where the installation position of a ceilometer and a camera is installed in a preferable position. FIG. 2 is a diagram for explaining a method of calibrating a total sky cloud height measuring device when there are no clouds. It is a figure showing an example of recorded local cloud height and low-precision cloud height.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a mode for carrying out the present invention (hereinafter referred to as "this embodiment") will be described in detail with reference to the drawings as appropriate. In this embodiment, an example will be described in which the laser distance measuring device is a ceilometer, but the device is not particularly limited as long as it is a device capable of measuring cloud height, such as a radar echo or an ultrasonic distance measuring device. In addition, in this embodiment, an example will be described in which the sensor is a camera that can photograph the entire sky, but it may also be a fisheye camera that can photograph the entire sky at once, or it can photograph the entire sky in parts. It is also possible to photograph only a portion of the entire sky that requires cloud height measurement, and there is no particular limitation as long as it is a means that can measure clouds.

Further, the following description and drawings are examples for explaining the present invention, and are omitted and simplified as appropriate for clarity of explanation. The present invention can also be implemented in various other forms. Unless specifically limited, each component may be singular or plural.

The position, size, shape, range, etc. of each component shown in the drawings may not represent the actual position, size, shape, range, etc. in order to facilitate understanding of the invention. Therefore, the present invention is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings.

In the following description, various information may be described using expressions such as "table" and "list," but various information may be expressed using data structures other than these. "XX table", "XX list", etc. are sometimes referred to as "XX information" to indicate that they do not depend on the data structure. When describing identification information, when expressions such as "identification information", "identifier", "name", "ID", and "number" are used, these expressions can be replaced with each other.

When there are multiple components having the same or similar functions, the same reference numerals may be given different suffixes for explanation. However, if there is no need to distinguish between these multiple components, the subscripts may be omitted in the description.

In addition, in the following explanation, processing performed by executing a program may be explained, but the program is executed by a processor (for example, a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit)). The processor may be the main body of the processing in order to perform the processing using appropriate storage resources (for example, memory) and/or interface devices (for example, communication ports). Similarly, the subject of processing performed by executing a program may be a controller, device, system, computer, or node having a processor. The main body of the processing performed by executing the program may be an arithmetic unit, and may include a dedicated circuit (for example, FPGA (Field-Programmable Gate Array) or ASIC (Application Specific Integrated Circuit)) that performs specific processing. .

A program may be installed on a device, such as a computer, from a program source. The program source may be, for example, a program distribution server or a computer-readable storage medium. When the program source is a program distribution server, the program distribution server includes a processor and a storage resource for storing the program to be distributed, and the processor of the program distribution server may distribute the program to be distributed to other computers. Furthermore, in the following description, two or more programs may be realized as one program, or one program may be realized as two or more programs.

An example of the configuration of the all-sky cloud height measuring device will be explained using FIGS. 1 to 11.

FIG. 1 shows an example of the configuration of a total sky cloud height measuring device 1. As shown in FIG. The all-sky cloud height measuring device 1 uses local but highly accurate cloud height information acquired by a laser ranging device to correct the low-accuracy all-sky cloud height information acquired from image data. This is a measurement device that can measure highly accurate cloud height information for the entire sky. To give an overview of each function shown in FIG. 1, the local cloud height measuring unit 2 has a function of receiving information from a laser distance measuring device such as a ceilometer and measuring cloud height in a local range. be. The low-precision cloud height estimation unit 3 has a function of receiving image data of the entire sky from a sensor and estimating at least the relative cloud height of each cloud layer based on the image data. The cloud height correction unit 4 assumes the measurement location of the local cloud height measurement unit 2 in the output of the low precision cloud height estimation unit 3, calculates the measurement result at the assumed measurement location, and calculates the measurement result at the assumed measurement location. This function corrects the cloud height information of the low-precision cloud height estimator 3 by extracting measurement points where the similarity between the output and the temporal change is high.

The total sky cloud height measuring device 1 includes a processing device such as a CPU, which is a control unit, and storage such as a ROM (Read Only Memory), a RAM (Random Access Memory), an HDD (Hard Disk Drive), and an SSD (Solid State Drive). As the hardware, a general communicable computer can be used as the hardware, which has a storage medium or a storage device as a section, and a communication device as a communication section such as an NIC (Network Interface Card). Each function of the total sky cloud height measuring device 1 is realized by, for example, a CPU reading a program from a ROM, reading/writing from/to a RAM, and executing processing.

The program may be provided, for example, by being read from a USB (Universal Serial Bus) memory, which is an example of an external storage medium, or by being downloaded from another computer via a network. Furthermore, in the following, each function of the total sky cloud height measuring device 1 is installed in a general computer as hardware, but all or part of these functions can be installed in one or more computers such as a cloud. Similar functions may be realized by being distributed and communicating with each other.

First, an example of the use of the all-sky cloud height device 1 will be explained using FIG. 2. In FIG. 2, a ceilometer 2_A that transmits local cloud height information to the local cloud height measurement unit 2 and a camera 3_A that transmits image data to the low-precision cloud height estimation unit 3 are directed toward the sky that is the measurement target. There is a cloud C_1 in the sky and a cloud C_2 above it, and there are measurement points P_1 and P_2 between the measurement point of the ceilometer 2_A and the measurement line 2_B indicating the direction. Here, a case where the ceilometer 2_A can measure two or more cloud layers is shown as an example, but there is no particular limitation as long as it can measure at least one cloud layer. In addition, for convenience, the measurement range of the ceilometer 2_A is shown as a line, and the point where it intersects with the cloud P is explained as a point, but the measurement line 2_B is not limited to a line, and can have a horizontal range such as a conical shape. may be used, and is not particularly limited. In that case, the measurement point P may be a local area rather than a point, and in that case, the average value, maximum value, etc. of cloud height values within the area may be determined as a representative value.

In this way, the local cloud height measurement unit 2 may arbitrarily select a region for measuring the heights of a plurality of cloud bases, and measure the height of the cloud base in the selected region. In addition, the local cloud height measurement unit 2 determines two or more measurement points and calculates the regression line of the measurement points, thereby determining the position of the measurement line at the plurality of cloud bases estimated by the low-precision cloud height estimation unit 3. may be determined.

An example of camera image 3_B acquired from camera 3_A will be described using FIG. 3. From the viewpoint of the camera 3_A, the cloud C_2 is located behind the cloud C_1, so naturally the hidden part of the cloud C_2 is not visible in the camera image 3_B. Although the measurement line 2_B is not actually shown in the camera image 3_B, it is shown for convenience of explanation. In this case, the measurement results of ceilometer 2_A output cloud heights for two layers, the lower of which is the cloud height of cloud C_1 observed at measurement point P_1, and the other value is the cloud height observed at measurement point P_2. This is the cloud height of cloud C_2. Hereinafter, the local cloud height measuring section 2, the low precision cloud height estimating section 3, and the cloud height correcting section 4 will be explained in detail.

The local cloud height measurement unit 2 receives cloud height information from the ceilometer 2_A, and outputs the cloud height of at least one cloud layer or cloud mass when clouds are present in the measurement range. The ceilometer 2_A irradiates a laser pulse, and calculates the altitude to the cloud from the time and intensity until a detector built in the ceilometer 2_A detects the laser light backscattered by the cloud as a reflected signal. The number of measurable cloud layers, maximum altitude, and minimum altitude vary depending on product specifications such as Ceilometer 2_A's laser intensity and detector sensitivity. In cases such as the examples shown in FIGS. 1 and 2, the local cloud height measurement unit 2 preferably accepts binary values indicating the altitudes of measurement point P_1 and measurement point P_2. It is sufficient to accept the cloud height of C_1. The output of the ceilometer 2_A may actually be an intensity distribution of laser light or a probability distribution of cloud height, and is not particularly limited as long as it can specify the cloud light at the measurement point P by calculation.

The low-accuracy cloud height estimation unit 3 will be explained using FIG. 4. The low-accuracy cloud height estimation unit 3 has a function of receiving data obtained by photographing the entire sky from the camera 3_A, and outputting low-accuracy cloud height information 3_C. The low-precision cloud height information 3_C is data as shown in FIG. 4, and in this example, cloud C_1 is arranged so as to partially overlap cloud C_2 on a clear day. At this time, the low-precision cloud height information 3_C is information from which at least the area of cloud C and their relative cloud heights can be obtained. In the example of FIG. 4, the regions of cloud C_1, cloud C_2, and clear weather are provided with low-accuracy cloud height information, but the altitude of regions other than cloud C does not necessarily have to be obtainable. Specifically, the cloud height of cloud C_1 is cloud height A, and the cloud height of cloud C_2 is cloud height B. The respective values of cloud height A and cloud height B do not necessarily have to match the actual cloud height, and it is sufficient as long as at least relative cloud height can be obtained. In this case, it is sufficient if the relationship of cloud height A<cloud height B can be obtained. The low-precision cloud height information 3_C does not have to be in the image format shown in FIG. 4; for example, it may be an array that stores pairs of the pixel position of the camera image 3_B and the cloud height value, and the cloud height value is a relative value. Since it is sufficient that the height can be determined, normalized values may be stored. In this way, the low-accuracy cloud height estimating unit 3 holds the relative height relationships among a plurality of cloud bases as data in a format that can be determined.

The low-precision cloud height information 3_C can be obtained, for example, by image processing technology using deep learning. Specifically, a true value cloud height image is generated in which an all-sky camera image 3_B, the region of cloud C reflected in the camera image 3_B, and the actual cloud height are embedded. From the true value cloud height image, the position of cloud C in the image and the true value of its cloud height can be obtained. By using this true value cloud height image as training data, designing and learning a neural network that takes camera image 3_B as input and outputs a depth image with estimated depth for each pixel, it is possible to estimate the cloud height of camera image 3_B. Therefore, low precision cloud height information 3_C can be obtained. Alternatively, the low-precision cloud height information 3_C can be obtained from thermography using an infrared radiation thermometer by utilizing the characteristic that the temperature varies depending on the altitude of cloud C, specifically, the higher the cloud height, the lower the temperature. At this time, when using a general infrared radiation thermometer, it may not be possible to accurately measure the temperature of cloud C due to the influence of water vapor, carbon dioxide, and dust in the atmosphere, but as mentioned earlier, low-precision cloud height information 3_C Since it is sufficient to obtain the relative cloud height, there is no need to take into account the reduction in measurement accuracy due to external factors in the atmosphere. Alternatively, a device such as a stereo camera that can directly measure the distance to the measurement target may be used. In this way, the low-accuracy cloud height estimation unit 3 estimates the relative height of the cloud base from the temperature distribution obtained by measuring light of a predetermined wavelength (for example, infrared rays and far infrared rays), and calculates the height. The height of the cloud can be taken as the height of the clouds.

In general, the maximum measurement distance of a stereo camera is determined mainly by the baseline length of the monocular cameras that make up the stereo camera, and when measuring the distance of a distant object such as cloud C, the baseline length should be set from 50 meters. It should be set at about 300 meters. Stereo cameras generally have a property that the distance measurement performance deteriorates as the measurement target is farther away, so it is difficult to accurately measure cloud height, but it is possible to meet the requirements of low precision cloud height information 3_C. can. In this way, the low-accuracy cloud height estimating unit 3 performs calculations suitable for the device that measures input data, and the low-accuracy cloud height information 3_C only needs to be able to obtain at least the relative cloud height of the cloud C. The acquisition means is not particularly limited.

The cloud height correction unit 4 will be explained in detail using FIGS. 5 to 11. In order to correct the low-precision cloud height information 3_C using the cloud height output by the local cloud height measurement unit 2, it is necessary to specify the position corresponding to the measurement point P of the ceilometer 2_A in the low-precision cloud height information 3_C. There is. The necessity of specifying the position of the measurement point P will be explained using FIG. In FIG. 5, the positions of the measurement line 2_B, measurement point P_1, and measurement point P_2 are illustrated assuming three patterns with respect to the low-precision cloud height information 3_C. As mentioned above, the measurement line 2_B is generally invisible, so the position of the measurement point P is unknown. Moreover, since accurate cloud height cannot be obtained from the low-precision cloud height information 3_C, it is also difficult to estimate the approximate position of the measurement point P, especially when a plurality of clouds C exist.

For example, in the case of measurement line 2_B', measurement point P_1' exists somewhere on measurement line 2_B'. Since it is unknown whether the measurement point P_2' exists based on the low-precision cloud height information 3_C, if the cloud height is obtained in binary form from the local cloud height measurement unit 2, its existence is recognized. However, since cloud C may exist further above cloud C_2, it cannot be determined that the cloud height is that of cloud C_2. Similarly, in the case of measurement line 2_B'', measurement point P_1'' and measurement point P_2'' exist somewhere on measurement line 2_B'', and in case of measurement line 2_B'', measurement point P_1' '' and measurement point P_2''' exist somewhere on measurement line 2_B'''.

In order to specify the actual position of the actual measurement line 2_B, a parameter representing the relative positional relationship between the camera 3_A and the ceilometer 2_A is required. Hereinafter, this parameter will be referred to as an external parameter. Common methods for estimating external parameters include a method of calculating by measuring the installation position and orientation of the camera 3_A and ceilometer 2_A, and a method of calculating by measuring the installation position and orientation of the camera 3_A and ceilometer 2_A. There is a method of calculating external parameters by measuring the object (plate) with a camera 3_A and a ceilometer 2_A, and estimating the relative position and orientation to the object to be photographed from each. However, since the output of the ceilometer 2_A is a discrete cloud height or a format in which no correspondence with image data can be obtained, it is difficult to apply the conventional calibration method. Hereinafter, a method for specifying the position of the measurement point P on the low-precision cloud height information 3_C will be explained.

FIG. 6 shows an outline of time-series measurements using the all-sky cloud height measuring device 1. First, at time t1, only cloud C_2 is observed above ceilometer 2_A. Therefore, the ceilometer 2_A measures the cloud height of the cloud C_2, and the local cloud height measurement section 2 receives it and transmits it to the cloud height correction section 4. At time t2, cloud C_1 moves to the right, intersects measurement line 2_B of ceilometer 2_A for the first time, and forms measurement point P_1 on the outline of cloud C_1. Furthermore, following time t1, cloud C_2 intersects measurement line 2_B at measurement point P_2. Therefore, at this point, there are two cloud height values received from the local cloud height measuring section 2. At time t3, cloud C_1 and cloud C_2 continue to intersect measurement line 2_B, so the local cloud height measuring unit 2 outputs the cloud height in binary, and since the same cloud C is being measured, each cloud There is little change in high prices.

On the other hand, in the low-precision cloud height information 3_C acquired at the same time, the actual position of the measurement point P_2 cannot be confirmed because it is hidden behind the cloud C_1. At time t4, cloud C_1 and cloud C_2 further move to the right and do not intersect with measurement line 2_B, so local cloud height measuring section 2 does not output a cloud height value. On the other hand, since the position of the measurement line 2_B and the actual cloud height are unknown in the low-precision cloud height information 3_C acquired at the same time, it is unknown whether the measurement point P exists.

FIG. 7 shows a processing flow in which the cloud height correction unit 4 identifies the position of the measurement point P in the low precision cloud height information 3_C. The cloud height correction unit 4 sets the measurement point candidate 4_A on the low-precision cloud height information 3_C in step S401. In step S402, the cloud height correction unit 4 records the low-accuracy cloud height at each measurement point candidate 4_A by referring to the low-accuracy cloud height information 3_C. In step S403, the cloud height correction unit 4 determines whether the number of cloud height recordings is less than the set number of times, and if it is determined that the number of cloud height recordings is less than the set number of times (step S403; no), the process returns to step 401 and the subsequent processing is performed.

On the other hand, if the cloud height correction unit 4 determines that the number of cloud height recordings is not less than the set number of times (step S403; yes), in step S404, the cloud height correction unit 4 uses the local cloud height information 2_C received from the local cloud height measurement unit 2 and each The correlation with the low-precision cloud height at the measurement point candidate 4_C is calculated. The cloud height correction unit 4 determines the point with the strongest correlation among the measurement point candidates 4_C as the measurement point P in step S404. That is, the cloud height correction unit 4 determines that the correlation between the height of the cloud base measured by the local cloud height measurement unit 2 and the relative height of the cloud base estimated by the low precision cloud height estimation unit 3 is the highest. In addition to strong points, positions that satisfy a predetermined relationship such as a strong correlation of a certain degree or more are determined as measurement points. Hereinafter, steps S401 to S405 will be explained in detail.

Step S401 will be explained using FIG. 8. In step S401, the cloud height correction unit 4 sets a measurement point candidate 4_A assuming the position of the measurement point P in order to determine a measurement point P whose position is unknown based on the low precision cloud height information 3_C. FIG. 8 shows an example in which 144 measurement point candidates 4_A are set for the low-precision cloud height information 3_C acquired at a certain time t, and each measurement point candidate 4_A is arranged at equal intervals. In this embodiment, for convenience of explanation, an example is shown in which 144 measurement point candidates 4_A are set, but in reality, at least one or more measurement point candidates may be set, and they do not need to be arranged at equal intervals as shown in FIG. Preferably, the cloud height correction unit 4 sets one measurement point candidate 4_A for each information unit (for example, a pixel if the form is an image) of the low-precision cloud height information 3_C, thereby determining the measurement point P with higher precision. can be determined.

Furthermore, if the relative position of the ceilometer 2_A and camera 3_A is clear and the position where the measurement line 2_B can exist can be limited on the low-precision cloud height information 3_C, the cloud height correction unit 4 determines that the measurement line 2_B is The measurement point candidates 4_A may be set only in areas where they can exist, and in this case, the amount of calculation for determining the measurement points P can be reduced. In this way, the cloud height correction unit 4 estimates the position of the measurement point from the relative positional relationship between the sensor such as the camera 3_A and the laser such as the ceilometer 2_A, and limits the candidate points that are candidates for the measurement point. do.

The process of step S402 will be explained using FIG. 8. When the cloud height correction unit 4 acquires the low-accuracy cloud height information 3_C as shown in FIG. For reference, the cloud height of measurement point candidate 4_A_1 at time t is recorded as cloud height A. Furthermore, since the measurement point candidate 4_A_2 is located above the cloud C_2, the cloud height correction unit 4 records the cloud height of the measurement point candidate 4_A_2 at time t as cloud height B. The cloud height correction unit 4 performs the above processing for all measurement point candidates 4_A set in step S401. Then, the cloud height correction unit 4 simultaneously records the local cloud height 2_C acquired from the local cloud height measurement unit 2 as the local cloud height 2_C at time t.

In step S403, the cloud height correction unit 4 determines that the cloud height information recorded in step S402 has reached a preset number of recordings, and if the number of recordings has not been reached, repeats steps S401 and S402. . The number of recordings used for determination may be at least one time, and is determined so that the number of data for local cloud height 2_C and measurement point candidate 4_C, which will be described later, is suitable for the correlation calculation method.

Step S404 will be explained using FIGS. 9 and 10. In step S404, in order to determine the measurement point P, the cloud height correction unit 4 determines the measurement point candidate 4_C that has the closest cloud height to the local cloud height 2_C recorded up to the previous step. Calculate the correlation of changes in direction. FIG. 9 shows excerpts of scenes at times t5, t6, and t7 when measurements were taken using the all-sky cloud height measurement device 1, and shows representative points among the measurement point candidates 4_A preset in step S401. Three measurement point candidates 4_A_1-4_A_3 are illustrated as points. The upper part of Fig. 9 shows a side view of the total sky cloud height measuring device 1 and the clouds C_1 and C_2 to be measured at each time, and the position of the actual measurement point P, and the lower part shows the position of the actual measurement point P at each time. The low-precision cloud height information 3_C received from the low-precision cloud height estimation unit 3 and the position of the measurement point candidate 4_A are shown.

FIG. 10 shows a graph in which the recorded local cloud height 2_C and the low-precision cloud height information of measurement point candidates 4_A_1, 4_A_2, and 4_A_3 are plotted in chronological order, focusing on cloud C_1. Note that at time t6, both cloud C_1 and cloud C_2 and measurement line 2_B intersect at measurement point P_1 and measurement point P_2, respectively, so although the local cloud height 2_C is recorded as a binary value at time t6, here For convenience of explanation, only one value of the lowest cloud height is shown. The actual cloud heights of cloud C_1 and cloud C_2 are cloud height C and cloud height D, respectively, and the illustration assumes that the cloud base is flat, but the same method can be applied even if the cloud base is not actually flat. Applicable.

When cloud heights as shown in Fig. 10 are recorded, the local cloud height 2_C and the low-precision cloud height information 3_C at each measurement point candidate 4_A can be regarded as time series data, and the correlation can be calculated. . For example, if a correlation coefficient is used as an index indicating the relationship between two data, it is possible to find that there is a strong positive correlation between measurement point candidate 4_A_1 and local cloud height 2_C because the cloud height value is decreasing at the same timing (time t6). Recognize. In step S405, the cloud height correction unit 4 refers to the correlation value calculated in step S404, and determines the measurement point candidate 4_A having the strongest correlation as the measurement point P_1. By performing the same process for measurement point P_2 at another time t, it is possible to determine the position of measurement point P_2 in low-precision cloud height information 3_C.

For example, in the example shown in FIGS. 9 and 10, the cloud C_1 moves directly below the measurement point P_2 between time t6 and time t7. At that time, although the point on the cloud C_2 corresponding to the measurement point P_2 cannot be visually recognized from the low-precision cloud height information 3_C, since the position of the measurement point P_2 and the local cloud height 2_C are clear, the local cloud height As long as cloud height D is recorded at height 2_C, it can be predicted that cloud C_2 exists above cloud C_1. By continuing the measurement further, if there is a time when cloud height D is no longer recorded in local cloud height 2_C, it is known that a gap or outline of cloud C_2 exists on measurement line 2_B at that time, so the camera With 2_A, it is possible to measure the approximate shape of the cloud C hidden in the lower cloud C, which cannot normally be confirmed.

Through the above steps, the position of the measurement point P in the low-precision cloud height information 3_C can be specified. Furthermore, when the positions of two or more measurement points P are specified in the low-precision cloud height information 3_C, a straight line connecting those measurement points P may be used as the measurement line 2_B. If the position of the measurement line 2_B can be acquired on the low-precision cloud height information 3_C, the amount of calculation is reduced by setting the measurement point candidate 4_A on the measurement line 2_B or an area including its surroundings in step S401. At the same time, the estimation accuracy of the measurement point P can be expected to improve.

A method by which the cloud height correction unit 4 corrects the low precision cloud height information 3_C will be explained. Through steps S401 to S405, the cloud height correction unit 4 can acquire the position of the measurement point P in the low-precision cloud height information 3_C and the local cloud height 2_C at that position. Since the low-precision cloud height information 3_C can obtain relative cloud height as described above, it is possible to specify an area having the same cloud height as a certain point. In the example shown in FIG. 9, it can be seen that the actual cloud height value of measurement point candidate 4_A_1 is cloud height C at time t6, so cloud height correction unit 4 corresponds to cloud height A in the same way as measurement point candidate 4_A_1. By replacing all regions with cloud height C, low precision cloud height information 3_C can be corrected to local cloud height 2_C. Cloud C_2 can be similarly corrected by correcting the cloud height value from cloud height B to cloud height D.

Through the above processing, it is possible to correct cloud heights with high accuracy even in areas that do not intersect with the measurement line 2_B, and it is possible to measure cloud heights of the entire sky with high accuracy. In the example taken here, there are only two clouds C, and in each case, the measurement line 2_B and part of cloud C intersect, but there are other clouds, and low-precision cloud height information 3_C is the same as cloud C_1 or cloud C_2. If you have , you can correct the cloud height in the same way. In addition, if the low-precision cloud height information 3_C has a cloud height value that can be linearly supplemented to the actual cloud height value, such as a temperature distribution obtained by thermography or a parallax distribution and a parallax image obtained by a stereo camera, the low-precision cloud height information 3_C Even if there is a region of 3_C where local cloud height 2_C cannot be obtained, it can be corrected. Specifically, for example, if it is clear that cloud height A is 50% of cloud height B, and only the area of cloud height A is corrected by local cloud height 2_C, then the area of cloud height B is By correcting the cloud height value of cloud height A/0.5, it is possible to measure cloud light over the entire sky. Alternatively, the output of the cloud height correction unit 4 may be a conditional expression for the cloud height value, and an example thereof will be explained using FIG. 11.

It is assumed that there are clouds C_1 and C_2 in the entire sky, and only cloud C_1 intersects with measurement line 2_B, and does not intersect with cloud C_2 at any time before or after. The low-precision cloud height information 3_C is a value in which only the relationship of cloud height E<cloud height F is known, and the cloud height cannot be linearly interpolated, and may be a label value. In this case, even if the cloud height correction unit 4 corrects the cloud height value of the cloud C_1 from the cloud height E to the cloud height G from the ceilometer 2_A to the measurement point P using the local cloud height 2_C using the method described above, Since the cloud C_2 does not intersect with the measurement line 2_B, the cloud height value remains unknown even if the above-described method is applied. Therefore, the cloud height correction unit 4 can set the corrected cloud height value of the cloud C_2 to a conditional expression that the cloud height is equal to or greater than the cloud height G based on the relational expression of cloud height E<cloud height F. In this case, since the cloud height value of cloud C_2 can be limited, even if the actual cloud height value is unknown, it can be provided as useful whole-sky cloud height information. If cloud C exists in addition to cloud C_1 and cloud C_2, correction can be made using a more limited cloud height value conditional expression.

A configuration example of the total sky cloud height measuring device 1 in Example 2 will be explained using FIG. 12. In this embodiment, a method will be described in which the cloud height correction unit 4 corrects the low-precision cloud height information 3_C without setting the measurement point candidate 4_A by adjusting the installation positions of the ceilometer 2_A and the camera 3_A. Below, the position of the measurement point used by the all-sky cloud height measurement device 1 is determined by arranging the sensor such as the camera 3_A and the laser such as the ceilometer 2_A so that their respective measurement ranges coincide. It has been decided.

FIG. 12 shows an example in which the installation positions of the ceilometer 2_A and the camera 3_A are adjusted so that the measurement line 2_B of the ceilometer 2_A and the camera 3_A are installed at a specific position, for example, perpendicular to the center of the image. In such a case, in the low-precision cloud height information 3_C, the measurement line 2_B is not a line but a point, and the measurement point P is limited to the same point regardless of the cloud height. Therefore, the cloud base correction unit 4 can determine the position of the measurement point P without processing steps S401 to S405, and therefore the amount of calculation can be reduced. Alternatively, by adjusting the installation positions of ceilometer 2_A and camera 3_A so that the optical axes of measurement line 2_B and camera 3_A are as close as possible, preferably parallel, measurement point 2_B can be set to It can be a short line. Therefore, the cloud height correction unit 4 can easily determine the position of the measurement point P, as described above. Alternatively, by adjusting the installation position so that the relative position and orientation of the ceilometer 2_A and camera 3_A are known in advance, for example, low-accuracy clouds can be The position of the measurement line 2_B on the high information 3_C can be specified.

In (Equation 1), (u, v) are coordinates on the camera image 3_B or low-precision cloud height information 3_C, and fx, fy, cu, cv are camera parameters that are generally called internal parameters. 3_A, where (fx, fy) are the focal lengths per pixel in the x-axis and y-axis directions, and (cu, cv) are the coordinates of the intersection of the image plane and the optical axis. (x, y, z) are coordinates in the world coordinate system on the measurement line 2_B.

In this example, a method of identifying cloud types using the all-sky cloud height measuring device described in Examples 1 and 2 will be described. In the following, cloud types and amounts of clouds in the whole sky are identified based on the height of the cloud low measured by the all-sky cloud height measurement device 1 and the cloud types classified by a predetermined classification method using meteorological information. The method for determining the type is exemplified.

Clouds C are subdivided into classes called cloud types based on their shape and arrangement. The most common cloud types are called 10 types of clouds, specifically 10 types: stratocumulus, stratus, cumulus, altocumulus, altocumulus, stratostratus, cirrus, cirrocumulus, cirrostratus, and cumulonimbus. In this example, a case will be explained in which 10 types of cloud types are identified using the total sky cloud height measuring device 1, but there are a wide variety of methods for classifying cloud types, and of course, the classification method to which this example is applied is limited to 10 types. It is not limited to the seed cloud shape.

For example, when specifying 10 types of cloud shapes using a general image recognition method using an image of cloud C as input, cloud types with high similarity in appearance will be confused. For example, cirrocumulus clouds and altocumulus clouds have a high degree of similarity in that they have the same appearance as cloud fragments clustered together. Therefore, in general image recognition methods, cloud types are distinguished based on appearance, and cirrocumulus clouds and altocumulus clouds may be confused. Generally, cirrocumulus clouds occur between altitudes of 5,000 meters to 15,000 meters, while altocumulus clouds occur between altitudes of 2,000 meters to 7,000 meters. Since cloud height can be measured by the all-sky cloud height measuring device 1, cirrocumulus clouds and altocumulus clouds, which have a high degree of similarity in appearance, can be classified with high accuracy. Note that this embodiment is also effective for cloud types that are distinguished by cloud height. Naturally, it is also possible to correct cloud type determination results other than those obtained by general image processing, for example, it can also be used to correct visual observation results, and the means for cloud type determination in the first stage is not particularly limited.

Additionally, some types of clouds can be classified by referring to general weather information. Below, we will explain how to classify cloud types using publicly available weather information.

For example, weather radar is an example of publicly available weather information. Meteorological radar emits microwaves from an observation device and uses the intensity of the reflected waves, radar echoes, to observe the distribution of precipitation particles within a radius of several hundred kilometers. By referring to the distribution of precipitation particles measured by a weather radar, it is possible to determine the presence or absence of precipitation clouds. The measurement resolution of a weather radar is generally several kilometers square, and preferably it is equivalent to this resolution, but if the granularity is even finer and the installation location of the total sky cloud height measurement device 1 is clear, low-precision cloud By identifying the direction in the high information 3_C and the direction in the weather radar, it is possible to easily determine whether the cloud C measured by the all-sky cloud height measurement device 1 is a rainfall type. Classification becomes easier when the numbers are different. Furthermore, by directly observing meteorological phenomena such as clouds and fog using not only weather radar but also satellite photographs, and inputting the content and location of these meteorological phenomena, we can classify cloud types with higher accuracy. is possible.

Additionally, lightning observation data is made public as weather information. Lightning observation data is obtained by a lightning monitoring system receiving radio waves generated by lightning and recording the time and location of lightning occurrence. (often observed as lightning strikes) and ground discharges that occur between thunderclouds and the ground (observed as lightning strikes). Similar to the weather radar, by identifying the direction in the low-precision cloud height information 3_C and the direction of the lightning observation data, it is easy to determine whether there is a discharge in the cloud C measured by the total sky cloud height measuring device 1. be able to. For example, when a tower-shaped cumulus cloud is observed, by implementing this embodiment, the cloud type can be classified as a cumulonimbus cloud if discharge is confirmed, and as a tower-shaped cumulus cloud if no discharge is confirmed. Naturally, this embodiment is not limited to the classification of cumulus clouds and tower-like cumulus clouds, but can be widely applied to cases where cloud types can be easily classified based on the presence or absence of electric discharge, and is not particularly limited.

As an application example of this embodiment, the cloud type and cloud amount for each cloud layer of the entire sky are measured from the cloud height of the entire sky obtained by the all-sky cloud height measuring device 1 and the cloud types classified by the above-mentioned means. I can do it. Furthermore, when clouds C having different cloud heights are superimposed on the same plane parallel to the earth's surface, the proportion of the area occupied by clouds C in the entire sky can be used as the total cloud amount.

Further, by correcting the low-precision cloud height information 3_C in advance to a value close to the actual cloud height using general weather information, the performance of the total sky cloud height measuring device 1 can be improved. For example, humidity can be calculated from the difference between atmospheric temperature and dew point temperature. Using humidity, it can be seen that the atmospheric conditions are such that cloud C is likely to form at altitudes where the temperature is below the dew point temperature, and if cloud C is confirmed on low-precision cloud height information 3_C near that altitude, The cloud height of the cloud C may be corrected using the altitude calculated based on the humidity. In that case, since the cloud height of the low-precision cloud height information 3_C has a value close to the local cloud height 2_C, it is expected that the accuracy of the final corrected total sky cloud height information will improve.

Regarding the all-sky cloud height measuring device 1, if there is no cloud C in the sky or if there is a cloud C in the sky but the cloud C does not intersect with the measurement line 2_B at any time t, steps S401 to S405 are performed. Even if this is carried out, the position of the measurement point P of the low-precision cloud height information 3_C cannot be determined. Even in such a case, a method of determining the position of the measurement point P of the low-precision cloud height information 3_C or the measurement line 2_B by using a jig whose height is known is shown in FIGS. 13 and 14. explain. Hereinafter, the measurement points used in the all-sky cloud height measuring device are set by installing an object that can be measured by a laser such as a ceilometer 2_A and has a predetermined shape in the measurement range of the laser and a sensor such as a camera 3_A. For example, it is possible to determine the position of the measurement point in the absence of clouds.

A method for determining the position of the measurement point P when there is no suitable cloud C in the sky will be explained using FIG. 13. The board E_1 has a slit E_2 open therein, and preferably the dimensions of each and the position of the slit E_2 are known. Further, the board E_1 may be made of any material such as resin or metal, and is not particularly limited as long as the ceilometer 2_A can measure its altitude. The board E_1 used in this example is an example of the most basic shape, and the board E_1 preferably has a slit E_2, and other localized parts such as a recess with a known depth, a protrusion with a known height, etc. It is not particularly limited as long as it has a shape that can measure changes in the distance from the cloud height 2_C to the ceilometer 2_A and the board E_1. Furthermore, the board E_1 may be, for example, spherical, and does not necessarily have to be plate-shaped, and its shape is not particularly limited.

In the example shown in FIG. 13, times t9, t10, and t11 are set when the board E_1 is moved at a constant speed from the left to the right of the angle of view of the camera 3_A while maintaining the distance H between the camera 3_A and the ceilometer 2_A. Excerpts and illustrations are provided. The board E_1 is moved to pass above the ceilometer 2_A so as to intersect with the measurement line 2_B. At time t9, the measurement line 2_B intersects with the board E_1, so the measurement point P appears, and at the same time, the altitude H is recorded as the cloud height in the local cloud height 2_C. At time t10, the board E_1 further moves to the right and the measurement line 2_B passes through the slit E_2, so there is no measurement point P and no cloud height is recorded in the local cloud height 2_C. At time t11, board E_1 further moves to the right, and measurement line 2_B intersects board E_1 again.

At this time, it is sufficient that the low-precision cloud height information 3_C has at least information that allows the area of the board E_1 to be determined. It can be seen that at times when the altitude H is not recorded in the local cloud height 2_C, the measurement line 2_B passes outside the board E_1 or through the slit E_2. Furthermore, since the moving direction of the board E_1 is clear, by observing the timing of the rise and fall of the altitude recorded in the local cloud height 2_C in chronological order, we can determine the time t when the measurement line 2_B passes through the slit E_2. Easily identified.

A specific example will be described in detail using FIG. 14. As mentioned above, the local cloud height 2_C when the board E_1 is moved is the altitude H recorded from time t9 to time t10, and from time t10 to time t11, the measurement line 2_B passes through the slit E_2, so the altitude H is Not recorded. After time t11, the altitude H is recorded again. By performing steps S401 to S405 in the same manner as in the first embodiment, the measurement point candidate 4_A located at the edge of the slit E_2 can be determined as the measurement point P at time t10, or the measurement point candidates can be narrowed down.

In the example shown in FIG. 13, since the slit E_2 has a width in the y-axis direction, the measurement point P is limited to an area corresponding to the width of the slit E_2, and its position cannot be uniquely determined. Therefore, by rotating the board E_1 by 90 degrees and similarly moving it from above to below the low-precision cloud height information 3_C, it is possible to further limit the measurement point candidates 4_A, and finally determine the position of the measurement point P. can. Furthermore, by installing the board E_1 at different altitudes and performing the above steps, it is possible to determine multiple measurement points P at different altitudes, and if two or more measurement points P are determined, they can be connected. The line may be the measurement line 2_B.

By using the board E_1 according to the above procedure, even when suitable clouds C do not exist above the all-sky cloud height measuring device 1, or in environments where the sky cannot be measured such as indoors, the low-precision cloud height information 3_C can be used. The position of the measurement point P can be determined.

1 All-sky cloud height measurement device 2 Local cloud height measurement unit 3 Low-precision cloud height estimation unit 4 Cloud height correction unit 2_A Ceilometer 3_A Camera

Claims (10)

a local cloud height measurement unit that receives data from a laser and measures the heights of multiple cloud bases in a local range;
a low-accuracy cloud height estimator that receives all-sky image data from a sensor and estimates the relative heights of a plurality of cloud bases based on the image data;
Receiving the intersection points of the measurement range by the laser and the plurality of cloud bases, and using the intersection points and the heights of the plurality of cloud bases measured by the local cloud height measurement section, the low-precision cloud height estimating section estimates. a cloud height correction section that corrects multiple cloud bases;
An all-sky cloud height measuring device characterized by comprising: The local cloud height measurement unit arbitrarily selects a region for measuring the height of the plurality of cloud bases, and measures the height of the cloud base in the selected region.
2. The total sky cloud height measuring device according to claim 1. The low-accuracy cloud height estimator retains the relative height relationship of the plurality of cloud bases as data in a distinguishable format.
2. The total sky cloud height measuring device according to claim 1. The cloud height correction unit is configured to calculate a correlation with respect to a time change between the height of the cloud base measured by the local cloud height measurement unit and the relative height of the cloud base estimated by the low precision cloud height estimation unit. Determine the position that satisfies the relationship as the measurement point,
2. The total sky cloud height measuring device according to claim 1. The local cloud height measurement unit determines two or more measurement points based on the intensity of a reflected signal obtained from a laser pulse irradiated by the laser and the time taken to detect the reflected signal, and determining the position of a measurement line indicating the measurement location and direction of the laser in the plurality of cloud bases estimated by the cloud height estimation unit;
5. The total sky cloud height measuring device according to claim 4 . The cloud height correction unit estimates the position of the measurement point from the relative positional relationship between the sensor and the laser, and limits candidate points that are candidates for the measurement point.
5. The total sky cloud height measuring device according to claim 4 . The low-accuracy cloud height estimation unit estimates the relative height of the cloud base from a temperature distribution obtained by measuring light of a predetermined wavelength.
2. The total sky cloud height measuring device according to claim 1. A measurement point used in the all-sky cloud height measurement device according to claim 1 is set by installing an object that can be measured by the laser and has a predetermined shape in the measurement range of the laser and the sensor, determining the position of the measurement point in the absence of clouds;
A measurement point determination method characterized by the following. The position of the measurement point used in the all-sky cloud height measuring device according to claim 1 is determined by arranging the sensor and the laser so that their respective measurement ranges coincide.
A measurement point determination method characterized by the following. Identifying the type of cloud in the whole sky based on the height of the cloud low measured by the whole sky cloud height measuring device according to claim 1 and the cloud type classified by a predetermined classification method using meteorological information. A cloud type determination method characterized by:

Images