WO2020209704A1

WO2020209704A1 - Method for generating high-resolution marine data, and device using same

Info

Publication number: WO2020209704A1
Application number: PCT/KR2020/095061
Authority: WO
Inventors: 임효혁
Original assignee: 임효혁
Priority date: 2019-04-10
Filing date: 2020-04-07
Publication date: 2020-10-15

Abstract

The present invention relates to a method for generating high-resolution marine data and a marine data generation device using same, and while a machine learning-based marine resource data prediction model for predicting the relationship between marine weather learning data, marine physical learning data, and marine resource learning data is generated using learning data including the marine resource learning data, the marine weather learning data, and the marine physical learning data about each specific point in the ocean according to a specific latitude and a specific longitude, the marine data generation device acquires marine weather test data and marine physical test data about each lattice point for sectioning a certain area in the ocean by latitude and longitude, inputs, into the marine resource data prediction model, the marine weather test data and marine physical test data about each lattice point, so that the marine resource data prediction model predicts and outputs marine resource test data at each lattice point corresponding to the marine weather test data and marine physical test data about each lattice point, and generates marine data about the certain area in the ocean by referring to the marine resource test data at each lattice point.

Description

Method for generating high-resolution ocean data and apparatus using same

The present invention relates to a method for generating high-resolution ocean data and an apparatus using the same.

With the enactment of the Act on Ocean Space Planning and Management on April 18, 2019, the need for systematic and integrated management of all marine spaces is increasing.

However, as can be seen from FIG. 1, since only marine resource information in the observed marine area constructed mainly for fixed line and peak observation exists, a wide range of blank areas inevitably occurs, and reliable in unobserved marine areas. Since marine resource information did not exist, difficulties existed in establishing systematic policies for the entire ocean space.

In addition, conventionally, a method of predicting marine resource information for an unobserved ocean region is applied based on ocean observation information built mainly for fixed line and peak observation, but a method by linear interpolation exists between two points. Because the marine resource information is predicted by a simple proportional equation derived from the observed values only at two points, without considering the effects of various variables, it is possible to produce predicted data that is far from the actual data. There was a problem that it was difficult to obtain accurate information on unobserved marine areas.

For example, if mackerel is not caught in an area above 25 degrees Celsius, a three-dimensional grid space including sea level and the space below it is assumed, the observed water temperature of grid A is 20 degrees Celsius, and the observed water temperature of grid C is It is assumed that the water temperature is 30 degrees Celsius, and the unobserved water temperature of grid B, located between grid A and grid C, is 25 degrees Celsius, and the water temperature from grid B to grid C exceeds 25 degrees Celsius.

In the above situation, although the catch of mackerel from grid B to grid C is actually 0 due to the water temperature, according to conventional linear interpolation, the catch of mackerel from grid A to grid C decreases linearly. There is a problem of predicting data that is far from the data.

or. According to the conventional linear interpolation, the water temperature of the D grid, which is an intermediate point between the A grid and the C grid, was predicted to be 25 degrees Celsius, and only the level of expecting mackerel to be caught between the A grid and the D grid was limited.

Therefore, there is a need for an improvement method to solve the above problems.

An object of the present invention is to solve all of the above-described problems.

In addition, the present invention grasps the relationship between marine meteorological data and marine physical data based on information actually observed at a wide range and marine resource data observed at a limited point, Another purpose is to produce resource data.

In addition, the present invention provides integrated ocean data for the entire ocean space by integrating ocean resource data for an actually observed point and ocean resource data for a blank area where observation is restricted with ocean meteorological data and ocean physical data. It has another purpose.

In order to achieve the object of the present invention as described above, and to realize the characteristic effects of the present invention described later, the characteristic configuration of the present invention is as follows.

According to an aspect of the present invention, in a method for generating high-resolution ocean data, (a) ocean resource data for learning, ocean meteorological data for learning, and ocean physics for learning for each of specific points of the ocean according to a specific latitude and a specific longitude. In a state in which training data including data is used to generate a machine learning-based marine resource data prediction model that predicts the relationship between the learning marine meteorological data and the learning marine physical data and the learning marine resource data, marine data A step of obtaining, by a generating device, ocean meteorological data for testing and ocean physical data for testing for each of the grid points partitioning the predetermined area of the ocean by latitude and longitude; (b) The oceanic data generating apparatus may transmit the test ocean meteorological data and the test ocean physical data for each of the grid points in which the predetermined region of the ocean is partitioned by the latitude and the longitude. By inputting into a data prediction model, the marine resource data prediction model causes the test ocean meteorological data for each of the grid points and the test ocean at each of the grid points corresponding to the test ocean physical data. Predicting and outputting resource data; And (c) generating, by the ocean data generating apparatus, ocean data for the predetermined region of the ocean by referring to the ocean resource data for testing at the respective grid points. .

As an example, before step (a), when (a0) the learning data including the learning marine resource data, the learning ocean meteorological data, and the learning ocean physical data for each of the specific points of the ocean are obtained, And the ocean data generating apparatus inputs the learning ocean meteorological data and the training ocean physical data into the ocean resource data prediction model, causing the ocean resource data prediction model to cause the training ocean meteorological data at each of the specific points And generating output marine resource data for learning corresponding to the marine physical data for learning, and calculating a loss by comparing the output marine resource data for learning and the marine resource data for learning at each of the specific points, and the Generating the marine resource data prediction model for predicting a relationship between the learning ocean meteorological data, the learning ocean physical data, and the learning ocean resource data by performing back propagation using a loss A method further comprising; is disclosed.

As an example, in the step (a), when some ocean meteorological variable measured data generated by being measured at at least some of the grid points is acquired, the ocean data generating device may be configured to measure the partial ocean meteorological variable. Ocean meteorological data prediction model-The ocean meteorological data prediction model is a specific ocean meteorological variable generated by measuring at a specific grid point among a plurality of grid points partitioned by the latitude and the longitude of the arbitrary region of the ocean It is generated to predict arbitrary ocean meteorological data for each of the plurality of grid points using data-by inputting as and allowing the ocean meteorological data prediction model to refer to the actual measured data of some ocean meteorological variables for the test. A method is disclosed, characterized by generating ocean meteorological data.

As an example, in the step (a), when the partial ocean meteorological variable actual measurement data and specific resolution information are obtained, the ocean data generating apparatus predicts the ocean meteorological data based on the partial ocean meteorological variable actual measurement data and the specific resolution information By inputting as a model and causing the marine meteorological data prediction model to refer to the actual measurement data of some marine meteorological variables and the specific resolution information, generating and transmitting the test marine meteorological data matching the specific resolution information, Obtaining the test ocean meteorological data and the test ocean physical data corresponding to the resolution information, in the step (b), the ocean data generating device, the predetermined region of the ocean in the latitude and the longitude By inputting the test ocean meteorological data and the test ocean physical data for each of the grid points partitioned to correspond to a specific resolution into the ocean resource data prediction model, the ocean resource data prediction model causes the respective Predicting and outputting the test marine meteorological data for the grid points and the test marine resource data at each of the grid points corresponding to the test marine physical data, and in step (c), the The marine data generating apparatus is disclosed, wherein the marine data corresponding to the specific resolution for the predetermined region of the sea is generated by referring to the test marine resource data at each of the grid points. do.

As an example, in the step (b), the marine data generating device may perform a predetermined pre-processing process for the test marine meteorological data and the test marine physical data (i) a predetermined distance to each of the grid points. After determining the sampling grid points with reference to the standard, (ii) generating ocean meteorological sampling data and ocean physical sampling data corresponding to each of the sampling grid points among the test ocean meteorological data and the test ocean physical data , By inputting the marine meteorological sampling data and the marine physical sampling data into the marine resource data prediction model, causing the marine resource data prediction model to output marine resource sampling data at each of the sampling grid points, and the In step (c), the ocean data generating apparatus generates ocean sampling data for the predetermined region of the ocean by referring to the ocean resource sampling data for each of the sampling grid points. Is initiated.

As an example, the test ocean meteorological data and the test ocean physical data refer to ocean meteorological observation data and ocean physical observation data separately obtained for specific grid points that are at least some of the respective grid points, A method is disclosed, wherein each of the ocean weather forecast data and ocean physical prediction data obtained for each of the grid points is corrected and generated.

As an example, the learning ocean meteorological data and the test ocean meteorological data include information on at least some of air pressure, wind direction, wind speed, solar radiation, temperature, humidity, general condition, moon phase, wave height, and rainfall, and the learning ocean A method is disclosed, wherein the physical data and the ocean physical data for testing include information on at least some of the even, wave direction, wave speed, sound speed, density and surface sediment.

As an example, the learning marine resource data and the test marine resource data include information on at least some of water temperature, salinity, dissolved oxygen, nutrient salt information, zooplankton information, and fish stock information. do.

As an example, (d) when a GIS (geographic information system) marine data inquiry request is obtained from a user terminal, the marine data generating device transmits the marine data generated for the predetermined area of the ocean to the user terminal The method further comprising: causing the user terminal to display GIS marine data corresponding to the marine data.

As an example, in step (d), when a specific GIS marine data inquiry request including information on a specific location among the predetermined area of the ocean is obtained from the user terminal, the ocean data generating device Disclosed is a method comprising transmitting specific ocean data generated for a specific location to the user terminal, and causing the user terminal to display specific GIS ocean data corresponding to the specific ocean data.

According to another aspect of the present invention, there is provided an apparatus for generating marine data for generating high-resolution marine data, comprising: at least one memory storing instructions; And at least one processor configured to execute the instructions, wherein the processor comprises: (1) ocean resource data for learning, ocean weather data for learning, and ocean for learning for each of specific points of the ocean according to a specific latitude and a specific longitude In a state in which training data including physical data is used to generate a machine learning-based marine resource data prediction model that predicts the relationship between the learning marine meteorological data and the learning marine physical data and the learning marine resource data, the A process of acquiring test ocean meteorological data and test ocean physical data for each of the grid points divided by latitude and longitude in a predetermined area of the ocean; (2) inputting the test marine meteorological data and the test marine physical data for each of the grid points divided by the latitude and the longitude into the marine resource data prediction model, To cause the marine resource data prediction model to predict and output the test marine meteorological data for the respective grid points and the test marine resource data at the respective grid points corresponding to the test marine physical data. The process of doing; And (3) an apparatus for generating marine data that executes the instructions for performing a process of generating marine data for the predetermined region of the sea with reference to the test marine resource data at each of the grid points. .

As an example, before the (1) process, the processor acquires the learning data including the learning marine resource data, the learning ocean meteorological data, and the learning ocean physical data for each of the specific points of the ocean Then, by inputting the learning ocean meteorological data and the learning ocean physical data into the ocean resource data prediction model, the ocean resource data prediction model causes the training ocean meteorological data and the training ocean physical data at each of the specific points. Generate output marine resource data for learning corresponding to, and calculate a loss by comparing the output marine resource data for learning and the marine resource data for learning at each of the specific points, and backpropagation using the loss By performing (back propagation), the process of generating the marine resource data prediction model for predicting a relationship between the marine meteorological data for training, the marine physical data for training, and the marine resource data for training; further comprising: Disclosed is an apparatus for generating ocean data, characterized by.

As an example, in the process (1), when some ocean meteorological variable measured data generated by measuring at least some of the grid points is acquired, the partial ocean meteorological variable measured data Meteorological data prediction model-The ocean meteorological data prediction model uses specific ocean meteorological variable actual measurement data generated by actual measurement at a specific grid point among a plurality of grid points partitioned by the latitude and the longitude of the arbitrary region of the ocean. It is generated to predict arbitrary marine meteorological data for each of the plurality of grid points-by inputting as and causing the marine meteorological data prediction model to refer to the actual measured data of some marine meteorological variables, and the marine meteorological data for the test Disclosed is an apparatus for generating ocean data, characterized in that to generate a.

As an example, in the process (1), when the actual measurement data of some marine meteorological variables and specific resolution information are obtained, the processor inputs the actual measurement data of some marine meteorological variables and the specific resolution information into the marine meteorological data prediction model. By allowing the marine meteorological data prediction model to refer to the actual measurement data of some marine meteorological variables and the specific resolution information to generate and transmit the test marine meteorological data matching the specific resolution information, the specific resolution information The corresponding ocean meteorological data for testing and the ocean physical data for testing are obtained, and in the (2) process, each of the predetermined regions of the ocean is partitioned to correspond to the specific resolution by the latitude and the longitude. By inputting the test marine meteorological data for the grid points and the test marine physical data into the marine resource data prediction model, the marine resource data prediction model causes the test marine meteorological data for each of the grid points. Predict and output the test marine resource data at each of the grid points corresponding to the data and the test marine physical data, and in the (3) process, the test at each of the grid points Disclosed is an apparatus for generating ocean data, characterized in that generating the ocean data corresponding to the specific resolution for the predetermined region of the ocean with reference to ocean resource data.

As an example, in the (2) process, the processor refers to a predetermined distance reference for each of the grid points as a predetermined preprocessing process for the test ocean meteorological data and the test ocean physical data. To determine the sampling grid points, (ii) generating ocean meteorological sampling data and ocean physical sampling data corresponding to each of the sampling grid points among the test ocean meteorological data and the test ocean physical data, By inputting meteorological sampling data and the marine physical sampling data into the marine resource data prediction model, causing the marine resource data prediction model to output marine resource sampling data at each of the sampling grid points, and (3) In the process, an apparatus for generating marine data is disclosed, wherein marine sampling data for the predetermined area of the sea is generated by referring to the marine resource sampling data for each of the sampling grid points.

As an example, the test ocean meteorological data and the test ocean physical data refer to ocean meteorological observation data and ocean physical observation data separately obtained for specific grid points that are at least some of the respective grid points, A marine data generation apparatus is disclosed, wherein the marine weather forecast data and the marine physical prediction data obtained for each of the grid points are corrected and generated.

As an example, the learning ocean meteorological data and the test ocean meteorological data include information on at least some of air pressure, wind direction, wind speed, solar radiation, temperature, humidity, general condition, moon phase, wave height, and rainfall, and the learning ocean The physical data and the marine physical data for testing include information on at least a portion of an even (潮差), wave direction, wave speed, sound speed, density, and surface sediment.

As an example, the marine resource data for learning and the test marine resource data include information on at least some of water temperature, salinity, dissolved oxygen, nutrient salt information, zooplankton information, and fish stock information. The device is started.

As an example, (4) when the processor obtains a geographic information system (GIS) ocean data inquiry request from a user terminal, the processor transmits the ocean data generated for the predetermined area of the ocean to the user terminal, and the user A process for causing a terminal to display GIS ocean data corresponding to the ocean data is disclosed.

As an example, the processor, in the process (4), when a specific GIS ocean data inquiry request including information on a specific location of the predetermined area of the ocean is obtained from the user terminal, the specific location of the ocean A marine data generation apparatus is disclosed, characterized in that by transmitting the specific ocean data generated for the user terminal to the user terminal, the user terminal displays specific GIS ocean data corresponding to the specific ocean data.

In addition to this, a computer-readable recording medium for recording a computer program for executing the method of the present invention is further provided.

The present invention grasps the relationship between marine meteorological data and marine physical data based on information actually observed at a wide range and marine resource data observed at a limited point, and provides marine resource data for an empty area where observation is restricted. Has the effect of producing.

In addition, the present invention has the effect of providing integrated ocean data for the entire ocean space by integrating ocean resource data for an actually observed point and ocean resource data for a blank area where observation is restricted with ocean meteorological data and ocean physical data. There is.

The accompanying drawings, which are attached to be used in the description of the embodiments of the present invention, are only some of the embodiments of the present invention, and those of ordinary skill in the technical field to which the present invention pertains (hereinafter referred to as "common technician") Other drawings can be obtained based on these drawings without any further work being done.

1 schematically illustrates a problem of a conventional ocean data generation method that relies on fixed line and peak observation,

2 schematically illustrates an apparatus for generating marine data for generating high-resolution marine data according to an embodiment of the present invention,

3A and 3B schematically illustrate a process of generating a marine resource data prediction model according to an embodiment of the present invention,

4A and 4B schematically illustrate a process of generating high-resolution ocean data according to an embodiment of the present invention,

5 schematically shows ocean data generated for a grid corresponding to an unobserved point according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION The detailed description of the present invention to be described below refers to the accompanying drawings, which illustrate specific embodiments in which the present invention may be practiced in order to clarify the objects, technical solutions and advantages of the present invention. These embodiments are described in detail sufficient to enable a person skilled in the art to practice the present invention.

Further, throughout the detailed description and claims of the present invention, the word "comprises" and variations thereof are not intended to exclude other technical features, additions, components or steps. Other objects, advantages, and features of the present invention to those skilled in the art will appear, partly from the present disclosure, and partly from the practice of the present invention. The examples and drawings below are provided by way of example and are not intended to limit the invention.

Moreover, the present invention covers all possible combinations of the embodiments indicated herein. It is to be understood that the various embodiments of the present invention are different from each other, but need not be mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the present invention in relation to one embodiment. In addition, it is to be understood that the location or arrangement of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the present invention. Accordingly, the detailed description to be described below is not intended to be taken in a limiting sense, and the scope of the present invention, if appropriately described, is limited only by the appended claims, along with all scopes equivalent to those claimed by the claims. Like reference numerals in the drawings refer to the same or similar functions over several aspects.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings in order to enable those of ordinary skill in the art to easily implement the present invention.

FIG. 2 schematically illustrates an ocean data generation apparatus 200 for generating high-resolution ocean data according to an embodiment of the present invention. Referring to FIG. 2, the ocean data generation apparatus 200 generates high-resolution ocean data. It may include a memory 210 for storing instructions to be generated and a processor 220 for generating high-resolution ocean data in response to instructions stored in the memory 210.

Specifically, the marine data generating device 200 is typically a computing device (e.g., a device that may include components of a computer processor, memory, storage, input device and output device, and other existing computing devices; routers, switches, etc.). Using a combination of the same electronic communications device; electronic information storage systems such as network attached storage (NAS) and storage area networks (SAN)) and computer software (i.e., instructions that cause the computing device to function in a specific way). It may be to achieve the desired system performance.

In addition, the processor of the computing device may include a hardware configuration such as a micro processing unit (MPU) or a central processing unit (CPU), a cache memory, and a data bus. In addition, the computing device may further include an operating system and a software configuration of an application that performs a specific purpose.

However, it is not excluded that the computing device includes an integrated processor in which a medium, a processor, and a memory are integrated for implementing the present invention.

A method of generating high-resolution ocean data by the ocean data generating apparatus 200 configured as described above will be described below.

First, an ocean meteorological data prediction model that generates ocean meteorological data will be described, and a process of the ocean meteorological data prediction model generating ocean meteorological data for training or testing ocean meteorological data will be described.

The ocean meteorological data prediction model uses specific ocean meteorological variable measured data generated by being measured at a specific grid point among a plurality of grid points that divide an arbitrary region of the ocean by latitude and longitude, for each of the plurality of grid points. It can be created to predict any oceanic weather data.

In other words, the ocean meteorological data prediction model is actually measured at a specific grid point that is at least a part of the total grid points, even if the ocean meteorological variable is not measured for all of the plurality of grid points divided by latitude and longitude in an arbitrary region of the ocean. Arbitrary ocean meteorological data for all of the plurality of grid points may be predicted using the generated specific ocean meteorological variable actual measurement data.

For reference, the actual measured data of marine meteorological variables may include information on at least some of air pressure, wind direction, wind speed, solar radiation, temperature, humidity, general condition, moon phase, wave height, and rainfall, but is not limited thereto. Various variables can be included to predict the data.

In this way, when a marine meteorological data prediction model for predicting marine meteorological data is generated, some marine meteorological data generated by actually measuring at least some of the grid points divided by latitude and longitude of a certain area of the ocean When the measured variable data is acquired, the marine data generating device 200 inputs the measured data of some marine meteorological variables into the marine meteorological data prediction model, and makes the marine meteorological data prediction model refer to the actual measured data of some marine meteorological variables for learning. It can be used to generate marine meteorological data or marine weather data for testing. That is, the ocean meteorological data prediction model may generate data from remaining grid points including adjacent grid points by referring to data actually measured at some grid points.

Alternatively, for the test ocean meteorological data and the test ocean physical data, each of the ocean meteorological observation data and ocean physical observation data obtained separately for specific grid points that are at least some of the respective grid points are referenced, and each grid point Each of the ocean weather forecast data and ocean physical prediction data obtained for the fields may be corrected and generated. For example, the ocean data generating apparatus 200 may generate ocean meteorological data for testing by correcting ocean meteorological data predicted for all grid points with ocean meteorological data observed for at least some of the grid points. have.

Next, a process of generating a marine resource data prediction model that predicts the relationship between the marine meteorological data generated through the process described above and the marine physical data and the marine resource data will be described.

For reference, the marine meteorological data for learning and the marine meteorological data for testing may include information on at least some of air pressure, wind direction, wind speed, solar radiation, temperature, humidity, general condition, moon phase, wave height, and rainfall.

In addition, the ocean physical data for learning and the ocean physical data for testing may include information on at least a portion of a ridge, a wave direction, a wave velocity, a sound velocity, a density, and a surface sediment. However, it is not limited thereto, and information on flow rate, water depth, ocean currents and currents, sedimentary heavy metals, water depth flow data, suspended matter, branch, gravel, sand, coastal properties, rock properties, sedimentation environment, electrical conductivity, transparency, hydrogen ion concentration And the like.

In addition, the marine resource data for learning and the marine resource data for testing may include information on at least some of water temperature, salinity, dissolved oxygen, nutrient salt information, zooplankton information, and fish stock information.

Here, the fish stock information may include at least some of information on the fish species in the grid area, the amount of catch, the size of the caught individual, the type of aquatic product, the amount of aquatic product, and the size of the harvested aquatic product, but are limited thereto. no.

In addition, the marine resource data for learning and the marine resource data for testing are data including information on at least some of water temperature, salinity, dissolved oxygen, nutrient salt information, zooplankton information, and fish stock information at the sea level, which is a two-dimensional plane. However, it is not limited thereto.

That is, in a three-dimensional grid space including the sea level and the space below it, water temperature, salinity, dissolved oxygen, nutrient salt information, zooplankton information, and fish stock information values change according to changes in latitude, longitude and sea depth. I can.

On the other hand, since the machine learning method applied to the marine resource data prediction model is not limited to a specific method, linear regression, decision tree, multilayer perceptron, support vector machine, and deep neural network ( A deep neural network), a convolution neural network, a recurrent neural network (RNN), and a generative adversarial network (GAN) can be applied.

As an example, referring to the flowchart of FIG. 3A, when learning data including marine resource data for learning, marine meteorological data for learning, and marine physics data for learning are acquired (S301) for each of specific points in the ocean, ocean data is generated. The apparatus 200 inputs the learning ocean meteorological data and the learning ocean physical data into the ocean resource data prediction model (S302), and causes the ocean resource data prediction model to learn ocean meteorological data and training ocean physical data at each specific point. The output marine resource data for learning corresponding to is generated (S303), and the loss is calculated by comparing the learning output marine resource data and the learning marine resource data at each of specific points (S304), and the loss is used. By performing back propagation (S305), it is possible to generate a marine resource data prediction model that predicts a relationship between ocean meteorological data for training, ocean physical data for training, and ocean resource data for training.

As another example, according to the boosting regression tree method, the gradient of each individual tree can be learned and residuals can be reduced.

For example, referring to FIG. 3B, when the result of the point at which the water temperature is to be predicted is a hatched pattern box, tree 1 and tree 2 are generated as weak classifiers, and a large tree combined with each tree is selected as a strong classifier. , 2.3, which is the sum of the scores of each individual tree, can be the water temperature score at the point to be predicted.

As described above, in a state in which a marine resource data prediction model for predicting the relationship between marine meteorological data and marine physical data and marine resource data is generated, the marine data generating apparatus shows the flowchart of FIG. 4A for a process of generating marine data. It will be described in detail below with reference.

That is, the ocean data generating apparatus 200 may acquire test ocean meteorological data and test ocean physical data for each of the grid points partitioned by latitude and longitude in a predetermined area of the ocean (S401).

In addition, the marine data generating apparatus 200 inputs the test marine meteorological data and the test marine physical data for each of the grid points divided by latitude and longitude into a marine resource data prediction model ( S402), the marine resource data prediction model can predict and output the test marine meteorological data for each grid point and the test marine resource data at each grid point corresponding to the test marine physical data. Yes (S403).

In addition, the ocean data generating apparatus 200 may generate ocean data for a certain area of the ocean by referring to ocean resource data for testing at respective grid points (S404).

Above, the overall process was briefly described with reference to the flowchart of FIG. 3A and the flowchart of FIG. 4A, and a more detailed process will be described below with reference to FIG. 4B.

As an example, when learning data at limited points in which fixed line observation data and fisheries statistical data exist are acquired, the ocean data generating apparatus 200 performs a data standardization process for the corresponding learning data in order to construct a data set for machine learning. It can be rough. Such learning data includes, for example, (i) water temperature, salinity, dissolved oxygen content, nutrient salts, zooplankton population, aquatic product types, and marine resource data for learning, including at least some of the catch information, (ii) barometric pressure, wind direction, wind speed, solar radiation, Learning marine meteorological data including at least some of the temperature, humidity, moon phase, wave height, and rainfall information, and (iii) learning marine physics including at least some of the information on wave direction, wave velocity, sound velocity, density and surface sediment. Data may be included.

And, the ocean data generating apparatus 200, by inputting the learning data into the ocean resource data prediction model, the ocean resource data prediction model to generate the learning output ocean resource data corresponding to the ocean meteorological data for training and the ocean physical data for training You can do it.

In addition, the marine data generation apparatus 200 allows the marine resource data prediction model to compare the training output marine resource data as a predicted value with the learning marine resource data as a correct answer to calculate a loss, and perform backpropagation using the loss. By doing so, it is possible to learn to predict the relationship between the variables of the learning ocean meteorological data, the learning ocean physical data, and the learning ocean resource data.

As described above, in order to generate actual marine resource data for each of the grid points demarcating an arbitrary region of the ocean while the marine resource data prediction model is generated, the test marine meteorological data for each of the grid points and the test When the marine physical data is obtained, the marine data generating apparatus 200 may cause the marine resource data prediction model to predict and output the test marine resource data at respective grid points.

In other words, unlike the previous reliance on fixed-line observation data and fisheries statistical data at a limited point to generate marine resource data, the marine data generating apparatus 200 is based on high-resolution marine meteorological data and marine physical data. It will be possible to generate high-resolution marine resource data.

In addition, the marine data generating apparatus 200 may provide only the high-resolution marine resource data generated in this way to service users, but complex marine data in which marine meteorological data and marine physical data are referenced through GIS WEB development/expression service. You can also provide

As an embodiment of the present invention as described above, the ocean data generating apparatus 200 may generate ocean data to correspond to specific resolution information. That is, the ocean data generating apparatus 200 includes (i) corresponding to all of the grid points so as to correspond to ocean meteorological data and ocean physical data at all of the grid points divided by latitude and longitude in a certain area of the ocean. Ocean data may be generated, but (ii) ocean data at corresponding grid points may be generated by dividing a certain area of the ocean to match specific resolution information.

Specifically, when some ocean meteorological variable actual measurement data and specific resolution information are acquired, the ocean data generation apparatus 200 inputs the ocean meteorological data prediction model by inputting the ocean meteorological variable actual measurement data and specific resolution information. By allowing the prediction model to refer to the actual measurement data of some marine meteorological variables and specific resolution information to generate and transmit the test marine meteorological data matching the specific resolution information, the marine meteorological data for testing and the test corresponding to the specific resolution information Ocean physical data can be acquired.

In addition, the ocean data generating device 200 predicts ocean resource data for testing ocean meteorological data and testing ocean physical data for each grid point partitioned to correspond to a specific resolution in a certain area of the ocean with latitude and longitude. By inputting into the model, the marine resource data prediction model predicts and outputs the test marine meteorological data for each grid point and the test marine resource data at each grid point corresponding to the test marine physical data. can do.

In addition, the ocean data generating apparatus 200 may generate ocean data corresponding to a specific resolution for a predetermined region of the ocean by referring to ocean resource data for testing at respective grid points.

In another embodiment, in order to reduce the computational amount of the marine resource data prediction model and increase the computational speed, a predetermined pre-processing process may be performed on the oceanic meteorological data for testing and the oceanic physical data for testing.

That is, the marine data generating apparatus 200 (i) determines sampling grid points with reference to a predetermined distance reference for each of the grid points, and (ii) sampling among the test ocean meteorological data and the test ocean physical data After generating ocean meteorological sampling data and ocean physical sampling data corresponding to each of the grid points, the ocean meteorological sampling data and ocean physical sampling data are input into the ocean resource data prediction model, and the ocean resource data prediction model makes each sampling. The marine resource sampling data at the grid points may be output, and marine sampling data for a certain area of the ocean may be generated by referring to the marine resource sampling data for each of the sampling grid points.

On the other hand, when a GIS (geographic information system) marine data inquiry request is obtained from the user terminal, the marine data generation apparatus 200 transmits the marine data generated for a certain area of the ocean to the user terminal, thereby causing the user terminal to GIS ocean data corresponding to the data can be displayed.

As an example, when the marine data generation device 200 generates marine data over the entire sea area near the Korean peninsula and stores it for each grid in the GIS engine, when a GIS marine data inquiry request is obtained from the user terminal, the user can Ocean data can be retrieved.

At this time, when a specific GIS ocean data inquiry request including information on a specific location in a certain area of the ocean is obtained from the user terminal, the ocean data generating device 200 may use the specific ocean data generated for a specific location on the ocean. By transmitting to the terminal, the user terminal displays specific GIS ocean data corresponding to the specific ocean data at a specific location. You can also check GIS ocean data for your location.

Through this, as shown in FIG. 5, it is not only possible to check the discontinuous ocean observation information that existed only at the existing observation point (black box), but to check the high-resolution ocean data for each grid produced in response to a certain area. do. In addition, it not only generates ocean data for adjacent grids of the observed grid, but also generates ocean data for grids located at a distance from the observed grid, for example, grids in the ocean boundary region of neighboring countries, where observation was previously difficult. You can also create it.

Further, the embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded in the computer-readable recording medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in the computer software field. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magnetic-optical media such as floptical disks. media), and a hardware device specially configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of the program instructions include not only machine language codes such as those produced by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform processing according to the present invention, and vice versa.

In the above, the present invention has been described by specific matters such as specific elements and limited embodiments and drawings, but this is provided only to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments. , Anyone with ordinary knowledge in the technical field to which the present invention pertains can make various modifications and variations from these descriptions.

Therefore, the spirit of the present invention is limited to the above-described embodiments and should not be defined, and all modifications that are equally or equivalent to the claims as well as the claims to be described later fall within the scope of the spirit of the present invention. I would say.

Claims

In a method for generating high-resolution ocean data,

(a) Learning data including marine resource data for learning, marine meteorological data for learning, and marine physics data for learning are used for each of specific points in the ocean according to a specific latitude and a specific longitude, and the marine meteorological data for learning and the marine physics for learning are used. In a state in which a machine learning-based marine resource data prediction model that predicts the relationship between the data and the learning marine resource data is generated, the marine data generating device divides a certain area of the ocean into latitude and longitude. Acquiring test ocean meteorological data and test ocean physical data for each;

(b) The oceanic data generating apparatus may transmit the test ocean meteorological data and the test ocean physical data for each of the grid points in which the predetermined region of the ocean is partitioned by the latitude and the longitude. By inputting into a data prediction model, the marine resource data prediction model causes the test ocean meteorological data for each of the grid points and the test ocean at each of the grid points corresponding to the test ocean physical data. Predicting and outputting resource data; And

(c) generating, by the ocean data generating apparatus, ocean data for the predetermined region of the ocean by referring to the ocean resource data for testing at the respective grid points;

How to include.
The method of claim 1,

Before step (a),

(a0) When the learning data including the learning marine resource data, the learning ocean meteorological data, and the learning ocean physical data for each of the specific points of the ocean are obtained, the ocean data generating device By inputting meteorological data and the learning marine physical data into the marine resource data prediction model, the marine resource data prediction model causes the learning marine meteorological data at each of the specific points and the learning output corresponding to the learning marine physical data To generate marine resource data, calculate a loss by comparing the output marine resource data for learning and the marine resource data for learning at each of the specific points, and perform back propagation using the loss. Generating the marine resource data prediction model for predicting a relationship between the learning marine meteorological data, the learning marine physical data, and the learning marine resource data;

The method further comprising a.
The method of claim 1,

In step (a),

When some ocean meteorological variable actual measurement data generated by actual measurement at at least some of the respective grid points is obtained, the ocean data generating device may use the ocean meteorological data prediction model-the ocean The meteorological data prediction model includes the plurality of grid points using specific ocean meteorological variable measured data generated by being measured at a specific grid point among a plurality of grid points divided by the latitude and the longitude of the arbitrary region of the ocean. It is generated to predict arbitrary marine meteorological data for each-by inputting as-and allowing the marine meteorological data prediction model to refer to the actual measured data of some marine meteorological variables to generate the marine meteorological data for the test. How to.
The method of claim 3,

In step (a),

When the partial ocean meteorological variable actual measurement data and specific resolution information are obtained, the ocean data generating device inputs the partial ocean meteorological variable actual measurement data and the specific resolution information as the ocean meteorological data prediction model to predict the ocean weather data The test marine meteorological data corresponding to the specific resolution information is generated and transmitted by allowing the user to refer to the actual measurement data of some marine meteorological variables and the specific resolution information to generate and transmit the test marine meteorological data matching the specific resolution information. Acquire data and marine physical data for the test,

In step (b),

The oceanic data generating apparatus may generate the test ocean meteorological data and the test ocean physical data for each of the grid points divided into the predetermined region of the ocean to correspond to the specific resolution by the latitude and the longitude. By inputting into the marine resource data prediction model, the marine resource data prediction model causes the test marine meteorological data for each of the grid points and the respective grid points corresponding to the test marine physical data. Predict and output the test marine resource data,

In step (c),

The marine data generating apparatus, by referring to the test marine resource data at each of the grid points, and generating the marine data corresponding to the specific resolution for the predetermined region of the ocean.
The method of claim 1,

In step (b),

The marine data generating apparatus determines sampling grid points with reference to a predetermined distance reference for each of the grid points as a predetermined pre-processing process for the test marine meteorological data and the test marine physical data. , (ii) After generating ocean meteorological sampling data and ocean physical sampling data corresponding to each of the sampling grid points among the ocean meteorological data for testing and ocean physical data for testing, the ocean meteorological sampling data and the ocean physics Inputting sampling data into the marine resource data prediction model, causing the marine resource data prediction model to output marine resource sampling data at each of the sampling grid points,

In step (c),

Wherein the ocean data generating apparatus generates ocean sampling data for the predetermined region of the ocean by referring to the ocean resource sampling data for each of the sampling grid points.
The method of claim 1,

For the test ocean meteorological data and the test ocean physical data, each of the ocean meteorological observation data and ocean physical observation data separately obtained for specific grid points that are at least some of the respective grid points are referenced, and the respective A method, characterized in that each of the ocean weather forecast data and ocean physical prediction data obtained for the grid points are corrected and generated.
The method of claim 1,

The learning marine meteorological data and the test marine meteorological data include information on at least some of air pressure, wind direction, wind speed, solar radiation, temperature, humidity, general condition, moon phase, wave height, and rainfall,

The method of claim 1, wherein the learning ocean physical data and the test ocean physical data include information on at least a portion of an even (潮差), wave direction, wave velocity, sound velocity, density, and surface sediment.
The method of claim 1,

The learning marine resource data and the test marine resource data include information on at least some of water temperature, salinity, dissolved oxygen, nutrient salt information, zooplankton information, and fish stock information.
The method of claim 1,

(d) When a geographic information system (GIS) marine data inquiry request is obtained from a user terminal, the marine data generating device transmits the marine data generated for the predetermined area of the ocean to the user terminal, and the user Causing the terminal to display GIS marine data corresponding to the marine data;

The method further comprising a.
The method of claim 9,

In step (d),

When a specific GIS marine data inquiry request including information on a specific location in the predetermined area of the ocean is obtained from the user terminal, the ocean data generating device may generate the specific ocean data generated for the specific location on the ocean. By transmitting to the user terminal, the method characterized in that the user terminal to display specific GIS ocean data corresponding to the specific ocean data.
In the ocean data generation apparatus for generating high-resolution ocean data,

At least one memory for storing instructions; And

Comprising at least one processor configured to execute the instructions,

The processor includes (1) learning marine resource data for each of specific points of the ocean according to a specific latitude and a specific longitude, marine meteorological data for learning, and learning data including marine physical data for learning, and In a state in which a machine learning-based marine resource data prediction model that predicts the relationship between the learning marine physical data and the learning marine resource data is generated, each of the grid points divided by latitude and longitude of the ocean A process of acquiring test marine meteorological data and test marine physical data; (2) inputting the test marine meteorological data and the test marine physical data for each of the grid points divided by the latitude and the longitude into the marine resource data prediction model, To cause the marine resource data prediction model to predict and output the test marine meteorological data for the respective grid points and the test marine resource data at the respective grid points corresponding to the test marine physical data. The process of doing; And (3) an apparatus for generating marine data that executes the instructions for performing a process of generating marine data for the predetermined area of the sea by referring to the test marine resource data at each of the grid points.
The method of claim 11,

Before the above (1) process,

When the processor acquires the learning data including the learning marine resource data, the learning ocean meteorological data, and the learning ocean physical data for each of the specific points of the ocean, the learning ocean meteorological data and the learning ocean By inputting physical data into the marine resource data prediction model, causing the marine resource data prediction model to generate training output marine resource data corresponding to the learning marine meteorological data and the learning marine physical data at each of the specific points. And, by comparing the learning output marine resource data and the learning marine resource data at each of the specific points to calculate a loss, and performing a back propagation using the loss, A process of generating the marine resource data prediction model for predicting a relationship between the marine meteorological data and the learning marine physical data and the learning marine resource data;

Ocean data generation apparatus further comprising a.
The method of claim 11,

The processor,

In the above (1) process,

When some marine meteorological variable actual measurement data generated by actual measurement at at least some of the respective grid points is obtained, the marine meteorological data prediction model is used as the marine meteorological data prediction model. Arbitrary ocean meteorology for each of the plurality of grid points using specific ocean meteorological variable measured data generated by being measured at a specific grid point among a plurality of grid points partitioned by the latitude and the longitude of an arbitrary region of the ocean The marine data generating apparatus, characterized in that by inputting as-to generate the marine meteorological data for the test by making the marine meteorological data prediction model refer to the actual measurement data of some marine meteorological variables.
The method of claim 13,

The processor,

In the above (1) process,

When the partial ocean meteorological variable actual measurement data and specific resolution information are acquired, the partial ocean meteorological variable actual measurement data and the specific resolution information are input as the ocean meteorological data prediction model to cause the ocean meteorological data prediction model By referring to the actual measurement data and the specific resolution information to generate and transmit the test ocean meteorological data matching the specific resolution information, the test ocean meteorological data and the test ocean physics corresponding to the specific resolution information Acquire data,

In the above (2) process,

The test ocean meteorological data and the test ocean physical data for each of the grid points divided by the latitude and the longitude to correspond to the specific resolution in the ocean resource data prediction model By inputting, the marine resource data prediction model causes the test marine meteorological data for each of the grid points and the test marine resource data at each of the grid points corresponding to the test marine physical data. To predict and output,

In the above (3) process,

And generating the ocean data corresponding to the specific resolution for the predetermined region of the ocean by referring to the test ocean resource data at each of the grid points.
The method of claim 11,

The processor,

In the above (2) process,

As a predetermined preprocessing process for the test ocean meteorological data and the test ocean physical data, (i) determining sampling grid points with reference to a predetermined distance criterion for each of the grid points, and (ii) for the test After generating ocean meteorological sampling data and ocean physical sampling data corresponding to each of the sampling grid points among ocean meteorological data and ocean physical data for testing, the ocean meteorological sampling data and the ocean physical sampling data are converted to the ocean resource data Input to the prediction model, causing the marine resource data prediction model to output marine resource sampling data at each of the sampling grid points,

In the above (3) process,

And generating ocean sampling data for the predetermined area of the ocean by referring to the ocean resource sampling data for each of the sampling grid points.
The method of claim 11,

For the test ocean meteorological data and the test ocean physical data, each of the ocean meteorological observation data and ocean physical observation data separately obtained for specific grid points that are at least some of the respective grid points are referenced, and the respective Ocean data generation apparatus, characterized in that the ocean weather forecast data and ocean physical prediction data obtained with respect to the grid points are corrected and generated.
The method of claim 11,

The learning marine meteorological data and the test marine meteorological data include information on at least some of air pressure, wind direction, wind speed, solar radiation, temperature, humidity, general condition, moon phase, wave height, and rainfall,

The marine physical data for learning and the marine physical data for testing include information on at least a portion of an even, wave direction, wave speed, sound speed, density, and surface sediment.
The method of claim 11,

The marine resource data for learning and the test marine resource data include information on at least some of water temperature, salinity, dissolved oxygen, nutrient salt information, zooplankton information, and fish stock information.
The method of claim 11,

(4) When the processor obtains a geographic information system (GIS) ocean data inquiry request from the user terminal, the processor transmits the ocean data generated for the predetermined area of the ocean to the user terminal, causing the user terminal to A process for displaying GIS ocean data corresponding to the ocean data;

Ocean data generation apparatus further comprising a.
The method of claim 19,

The processor,

In the above (4) process,

When a specific GIS ocean data inquiry request including information on a specific location among the predetermined area of the ocean is obtained from the user terminal, specific ocean data generated for the specific location of the ocean is transmitted to the user terminal, Ocean data generating apparatus, characterized in that the user terminal to display specific GIS ocean data corresponding to the specific ocean data.