CN110942111B - Method and device for identifying strong convection cloud cluster - Google Patents
Method and device for identifying strong convection cloud cluster Download PDFInfo
- Publication number
- CN110942111B CN110942111B CN201911423710.4A CN201911423710A CN110942111B CN 110942111 B CN110942111 B CN 110942111B CN 201911423710 A CN201911423710 A CN 201911423710A CN 110942111 B CN110942111 B CN 110942111B
- Authority
- CN
- China
- Prior art keywords
- cloud
- identifying
- cluster
- strong convection
- cloud cluster
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/411—Identification of targets based on measurements of radar reflectivity
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
- G06F18/256—Fusion techniques of classification results, e.g. of results related to same input data of results relating to different input data, e.g. multimodal recognition
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Radar Systems Or Details Thereof (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method and a device for identifying a strong convection cloud cluster, wherein the method for identifying the strong convection cloud cluster comprises the following steps: based on three-dimensional radar echo data, identifying anvil cloud clusters, transition anvil cloud clusters, layer cloud clusters and convection cloud clusters according to the difference of development heights and strengths of different types of cloud clusters to obtain radar cloud classification results; acquiring brightness temperature data of a satellite cloud picture, extracting a characteristic region taking a grid point as a center aiming at each grid point, performing convolution and pooling operation on the characteristic region, and identifying 9 types of cloud clusters based on the result of the convolution and pooling operation to obtain a satellite cloud classification result; and performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics to obtain the strong convection cloud cluster. The identification tracking precision of the strong convection cloud cluster is effectively improved.
Description
Technical Field
The invention relates to the technical field of weather forecasting, in particular to a method and a device for identifying strong convection clouds.
Background
The cloud with strong Convection is generally a cloud including hail, strong wind, tornado, strong precipitation and the like generated by Deep wet Convection (DMC), and has the characteristics of burst property, short life history (short duration, generally ranging from several minutes to tens of hours), strong locality (small spatial scale, generally ranging from tens of kilometers to two hundred kilometers), easy disaster and the like. Due to its strong destructive nature, strong convection weather from strong convection clouds is generally listed as fourth-place disastrous weather second only after tropical cyclone, earthquake, flood disasters. Therefore, the strong convection cloud cluster is accurately identified and tracked, prediction and early warning of strong convection weather are facilitated, and serious casualties and economic losses caused by the strong convection weather are reduced.
The method has the problems that the threshold value is single and definition is undefined, the radar observation has limited effect before precipitation in convection cloud is formed, an extrapolation technology based on radar observation alone cannot effectively forecast the onset, development and death of a storm, and structures (a deep convection core, a stratospheric region and an anvil cloud region) in a deep convection system cannot be effectively distinguished. The new generation of static meteorological satellite observation has incomparable advantages in time, horizontal and vertical spatial resolution and spatial coverage degree, and plays an important role in short-time nowcasting of strong convection weather.
At present, a threshold method is mainly adopted for satellite identification of convection cloud. The threshold method is most commonly used, is a simple image segmentation method, and is particularly suitable for feature identification with obvious difference between a target and a background. For a long-wave infrared channel, an infrared bright temperature threshold method is generally adopted, and comprises a temperature threshold, a split window difference threshold, an area threshold and a temperature threshold. Among these, the temperature threshold is a simple segmentation that mainly detects convective cloud boundaries. Due to factors such as application purposes, difference of seasons and regions and the like, a unified standard method is difficult to determine the satellite threshold, and therefore the problem that the convection monomer is identified by the satellite threshold method exists.
Disclosure of Invention
In view of the above, the present invention provides a method and an apparatus for identifying a strong convection cloud cluster, so as to improve the accuracy of identifying and tracking the strong convection cloud cluster.
In a first aspect, an example of the present invention provides a method for identifying a strong convective cloud, including:
based on three-dimensional radar echo data, identifying anvil cloud clusters, transition anvil cloud clusters, layer cloud clusters and convection cloud clusters according to the difference of development heights and strengths of different types of cloud clusters to obtain radar cloud classification results;
acquiring brightness temperature data of a satellite cloud picture, extracting a characteristic region taking a grid point as a center aiming at each grid point, performing convolution and pooling operation on the characteristic region, and identifying 9 types of cloud clusters based on the result of the convolution and pooling operation to obtain a satellite cloud classification result;
and performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics to obtain the strong convection cloud cluster.
In combination with the first aspect, the present invention provides a first possible implementation manner of the first aspect, wherein,
based on three-dimensional radar echo data, according to the difference of different types of cloud cluster development height and intensity, the anvil cloud cluster is identified, including:
selecting three-dimensional radar echo data of a first height layer from the three-dimensional radar echo data;
and determining lattice points with radar echo intensity greater than zero according to the three-dimensional radar echo data of the first height layer, and identifying the anvil cloud cluster according to the determined lattice points.
With reference to the first aspect, an embodiment of the present invention provides a second possible implementation manner of the first aspect, where identifying the cloud cluster of transition anvils includes:
selecting three-dimensional radar echo data of a second height layer from the anvil cloud cluster;
and determining lattice points with radar echo intensity greater than zero according to the three-dimensional radar echo data of the second height layer, and obtaining a transition anvil cloud cluster according to the determined lattice points.
With reference to the second possible implementation manner of the first aspect, an embodiment of the present invention provides a third possible implementation manner of the first aspect, where the identifying the layer cloud includes:
and acquiring lattice points corresponding to the three-dimensional radar echo data with the combined reflectivity larger than a preset first threshold value of the combined reflectivity from the transition anvil cloud cluster to obtain a layer cloud cluster, wherein the combined reflectivity corresponding to each lattice point in the layer cloud cluster is larger than the first threshold value of the combined reflectivity.
With reference to the third possible implementation manner of the first aspect, an embodiment of the present invention provides a fourth possible implementation manner of the first aspect, where the identifying the convective cloud includes:
and identifying mature convective clouds from the transitional anvil cloud cluster and the layer cloud cluster according to a preset first identification algorithm, and identifying new or extinct convective clouds according to a preset second identification algorithm.
With reference to the first aspect and any one of the first to fourth possible implementation manners of the first aspect, an embodiment of the present invention provides a fifth possible implementation manner of the first aspect, where the method further includes:
acquiring adjacent time strong convection cloud clusters, and calculating the characteristic quantity of the adjacent time strong convection cloud clusters, wherein the characteristic quantity comprises: the center of gravity position, the area, the eccentricity, the lowest brightness temperature and the average brightness temperature;
and obtaining a value function value according to the characteristic value of each characteristic quantity and the weight of the characteristic quantity, and if the value function value is larger than a preset value function threshold, confirming that the strong convection cloud cluster at the next time is the tracking result of the strong convection cloud cluster at the previous time.
With reference to the first aspect and any one of the first to fourth possible implementation manners of the first aspect, an embodiment of the present invention provides a sixth possible implementation manner of the first aspect, where the method further includes:
and generating the motion trail of the strong convection cloud cluster according to the same strong convection cloud cluster confirmed in sequence, and predicting the future motion trail of the strong convection cloud cluster based on the motion trail and a preset prediction algorithm.
In a second aspect, an embodiment of the present invention further provides an apparatus for identifying a strong convection cloud, including:
the radar cloud picture processing module is used for identifying anvil cloud clusters, transition anvil cloud clusters, layer cloud clusters and convection cloud clusters according to the difference of development heights and strengths of different types of cloud clusters on the basis of three-dimensional radar echo data to obtain radar cloud classification results;
the satellite cloud picture processing module is used for acquiring brightness temperature data of a satellite cloud picture, extracting a characteristic region taking the grid point as a center aiming at each grid point, performing convolution and pooling operation on the characteristic region, and identifying 9 types of clouds based on the results of the convolution and pooling operation to obtain satellite cloud classification results;
and the strong convection cloud cluster acquisition module is used for performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics to acquire the strong convection cloud cluster.
In a third aspect, an embodiment of the present application provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the steps of the above method when executing the computer program.
In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, performs the steps of the method described above.
According to the method and the device for identifying the strong convection cloud cluster, provided by the embodiment of the invention, the anvil cloud cluster, the transition anvil cloud cluster, the layer cloud cluster and the convection cloud cluster are identified according to the difference of development heights and strengths of different types of cloud clusters on the basis of three-dimensional radar echo data, so that a radar cloud classification result is obtained; acquiring brightness temperature data of a satellite cloud picture, extracting a characteristic region taking a grid point as a center aiming at each grid point, performing convolution and pooling operation on the characteristic region, and identifying 9 types of cloud clusters based on the result of the convolution and pooling operation to obtain a satellite cloud classification result; and performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics to obtain the strong convection cloud cluster. Therefore, data of the satellite and the radar are fully utilized, the radar and the satellite fully consider vertical development characteristics and horizontal development characteristics of the convection cloud cluster, development characteristics of a strong convection cloud top are grasped, accurate cloud cluster classification results can be obtained through a cloud cluster classification technology of a convolutional neural network based on deep learning, the cloud cluster classification results and the radar are fused, and identification accuracy of the strong convection cloud cluster is effectively improved.
In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
Fig. 1 is a schematic flow chart of a method for identifying a strong convection cloud according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of an apparatus for identifying strong convective clouds according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a computer device 300 according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all the embodiments. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a method and a device for identifying a strong convection cloud cluster, which are described by embodiments below.
At present, the method for identifying the strong convection single body according to the three-dimensional radar echo data and the satellite data is mainly a threshold value method, the threshold value for identifying the strong convection cloud cluster is single, and the method is suitable for objects with obvious target and background differences.
According to the embodiment of the invention, by means of the detection advantages of radar and satellite, a threshold value method and a deep learning method are fused, the horizontal and vertical development conditions of the convection cloud cluster are comprehensively considered, the detailed cloud cluster classification result based on two kinds of data is obtained, the two classification results are fused, the detection advantages from top to bottom and from bottom to top are integrated, a more accurate strong convection region is identified, a corresponding tracking strategy is adopted for tracking the size of the cloud cluster in the strong convection region, and finally the strong convection cloud cluster identification tracking result based on the satellite radar fusion is obtained.
Fig. 1 shows a flowchart of a method for identifying a strong convection cloud provided by an embodiment of the present invention.
As shown in fig. 1, the method includes:
in the embodiment of the invention, the anvil cloud cluster, the transition anvil cloud cluster, the layer cloud cluster and the convection cloud cluster are identified according to the vertical development height and the development strength of different cloud cluster types according to three-dimensional radar data. Wherein, the coverage areas of the anvil cloud cluster, the transition anvil cloud cluster, the layer cloud cluster and the convection cloud cluster are reduced in sequence.
In the embodiment of the present invention, as an optional embodiment, the identifying the anvil cloud cluster based on three-dimensional radar echo data according to the difference between the development heights and strengths of different types of cloud clusters includes:
a11, selecting three-dimensional radar echo data of a first height layer from the three-dimensional radar echo data;
and A12, determining lattice points with radar echo intensity larger than zero according to the three-dimensional radar echo data of the first height layer, and identifying the anvil cloud cluster according to the determined lattice points.
In the embodiment of the present invention, as an optional embodiment, the height of the first height layer is 6km to 7.5km. Where the grid points are the horizontal resolution of the radar detection. And the radar echo intensity corresponding to each lattice point in the anvil cloud cluster is greater than zero.
In an embodiment of the present invention, as an optional embodiment, the identifying the cloud cluster of the transition anvil includes:
a21, selecting three-dimensional radar echo data of a second height layer from the anvil cloud cluster;
and A22, determining lattice points with the radar echo intensity larger than zero according to the three-dimensional radar echo data of the second height layer, and obtaining a transition anvil cloud cluster according to the determined lattice points.
In the embodiment of the invention, as an optional embodiment, the height of the second height layer is 2-6km, and the radar echo intensity corresponding to each lattice point in the transition anvil cloud cluster is greater than zero in the height range of 2-6 km.
In the embodiment of the invention, because the anvil cloud cluster and the transition anvil cloud cluster are positioned at the top of the convection cloud and the echo tops of the anvil cloud cluster and the transition anvil cloud cluster are higher, the anvil cloud cluster and the transition anvil cloud cluster are determined according to the height information reflected by the radar echo intensity corresponding to the three-dimensional radar echo data.
In embodiments of the present invention, the anvil cloud comprises a transition anvil cloud, and the two types of clouds comprise a laminar cloud and a convective cloud.
In an embodiment of the present invention, as an optional embodiment, the identifying the layer cloud cluster includes:
and acquiring lattice points corresponding to the three-dimensional radar echo data with the combined reflectivity larger than a preset combined reflectivity first threshold value from the transition anvil cloud cluster to obtain a layer cloud cluster, wherein the combined reflectivity corresponding to each lattice point in the layer cloud cluster is larger than the combined reflectivity first threshold value.
In the embodiment of the invention, the classical research shows that the maximum radar echo intensity of the cloud cluster is less than 40dBz, the combined reflectivity is generally greater than 10dBz, and the echo top height can be developed to be more than 6 km. Thus, as an alternative embodiment, the first threshold of the combined reflectivity may be set to 10dBZ, and in the transition anvil cloud cluster, the region composed of the grid points with the combined reflectivity greater than the first threshold of the combined reflectivity is selected as the layer cloud cluster. Within the layer cloud, the combined reflectivity of each grid point is greater than a combined reflectivity first threshold.
In an embodiment of the present invention, as an optional embodiment, the identifying, in the transition anvil cloud and the layer cloud, a convection cloud includes:
and identifying mature convective clouds from the transitional anvil cloud clouds and the laminar cloud clouds according to a preset first identification algorithm, and identifying new or dead convective clouds according to a preset second identification algorithm.
In an embodiment of the present invention, the transition anvil cloud cluster and the layer cloud cluster include a convection cloud cluster, and the convection cloud cluster includes: mature convective clouds and new or depleted convective clouds develop.
In the embodiment of the present invention, as an optional embodiment, a developed convection cloud is identified according to a preset first identification algorithm, where the first identification algorithm includes two criteria, where the criterion is:
and A31, obtaining lattice points corresponding to the three-dimensional radar echo data with the combined reflectivity larger than a preset combined reflectivity second threshold value to obtain a developed convection cloud cluster, wherein the combined reflectivity corresponding to each lattice point in the developed convection cloud cluster is larger than the combined reflectivity second threshold value, and the combined reflectivity second threshold value is larger than the combined reflectivity first threshold value.
In this embodiment, as an optional embodiment, the first threshold of the combined reflectivity is 10dbz, and the second threshold of the combined reflectivity is 35dbz.
And (5) judging the criterion two:
a321, acquiring a first convection cloud lattice point set with a combined reflectivity larger than a third threshold value of the combined reflectivity and smaller than a fourth threshold value of the combined reflectivity from a transition anvil cloud cluster and a layer cloud cluster;
in this embodiment of the present invention, as an optional embodiment, the third threshold value of the combined reflectance is 15dbz, and the fourth threshold value of the combined reflectance is 35dbz.
A322, in the first convection cloud lattice point set, for each lattice point, sequentially calculating a ratio of a square of a combined reflectivity of the lattice point to a preset first empirical threshold, a first difference between a preset second empirical threshold and the ratio, and a second difference between the combined reflectivity of the lattice point and the first difference;
in this embodiment of the present invention, as an optional embodiment, the first empirical threshold is set to 120, and the second empirical threshold is set to 15.
And A323, acquiring candidate grid points with a second difference value larger than zero, and constructing and developing mature convection clouds according to the candidate grid points.
In the embodiment of the invention, in the developed convection cloud cluster, the second difference value corresponding to each grid point is greater than zero.
In the embodiment of the present invention, the developed convective cloud can be determined by using one of the following two formulas:
CRdbz [ i, j ] > 35, or
In the formula (I), the compound is shown in the specification,
CRdbz is the combined reflectivity, and i and j are the grid point locations.
In an embodiment of the present invention, as an optional embodiment, identifying a new or disappeared convective cloud cluster according to a preset second identification algorithm includes:
a41, acquiring a second convection cloud lattice point set of which the combined reflectivity is greater than a fifth threshold value of the combined reflectivity;
in this embodiment of the present invention, as an optional embodiment, the fifth threshold of the combined reflectance is 40dbz.
A42, aiming at each lattice point in the second convection cloud lattice point set, acquiring a preset threshold number of adjacent lattice points around the lattice point as a center, and if all the adjacent threshold number of lattice points belong to a transition anvil cloud cluster or a layer cloud cluster, placing the lattice point in the lattice point set;
in this embodiment of the present invention, as an optional embodiment, the preset adjacent threshold is 8.
And A43, constructing a new born or lost convection cloud based on the grid point set.
In the embodiment of the invention, the new or dead convection cloud cluster is determined by the following formula:
wherein the content of the first and second substances,
bb [ ii, jj ] is an adjacent grid point with the grid point as the center;
bb [ ii, jj ] =2, or bb [ ii, jj ] =3 are 8 adjacent grid points, each belonging to a transition anvil cloud or a layer cloud.
102, acquiring brightness temperature data of a satellite cloud picture, extracting a feature area taking a grid point as a center aiming at each grid point, performing convolution and pooling operation on the feature area, and identifying 9 types of cloud clusters based on the result of the convolution and pooling operation to obtain a satellite cloud classification result;
in the embodiment of the invention, as an optional embodiment, the satellite data is sunflower 8 satellite channel brightness temperature data. The method comprises the steps of identifying convection clouds in a sunflower 8 Satellite Cloud picture by using an International Satellite weather program (ISCP) classification standard, wherein the ISCP classification standard divides the clouds into rolling clouds, rolling layer clouds, convection clouds, high-volume clouds, high-layer clouds, rain layer clouds, cloud clouds, laminated clouds and layer clouds, and divides Cloud phases into ice clouds, water clouds and ice-water mixed clouds.
In the embodiment of the invention, the characteristics of the satellite cloud pictures are mined by using a convolutional neural network containing convolution and pooling operations and combining with ISCP classification standards, and the sunflower 8 satellite cloud pictures are classified and predicted, so that various classified clouds of the sunflower 8 satellite cloud pictures are obtained.
In the embodiment of the invention, the brightness temperature data is used for feature extraction, and as an optional embodiment, the brightness temperature data of 10 channels of a sunflower 8 satellite are obtained, wherein the channels comprise: 3.9 μm channel, 6.2 μm channel, 7.0 μm channel, 7.3 μm channel, 8.6 μm channel, 9.6 μm channel, 10.4 μm channel, 11.2 μm channel, 12.3 μm channel, and 13.3 μm channel, each corresponding to the same scan range.
In the embodiment of the present invention, as an optional embodiment, the size of the feature area is: 9 × 9km.
In the embodiment of the invention, the two-dimensional channel images of the 9 × 9km characteristic region are multiplied by different two-dimensional matrixes, so that characteristic maps reflecting different characteristics of the images can be obtained. The two-dimensional matrix is called a convolution kernel, and different forms of convolution kernels can pertinently extract different information of the image to obtain different characteristic maps.
In the embodiment of the present invention, for each lattice point included in the feature region, the convolution value of the lattice point is derived from the product of the lattice point and 8 surrounding lattice points and a3 × 3 average convolution kernel matrix.
In the embodiment of the invention, the Convolutional Neural Networks (CNN) can be used for continuously training and adjusting the weight of the Convolutional kernel, thereby achieving the purpose of extracting different characteristics by the CNN.
In the embodiment of the invention, after convolution processing, the scale of the original satellite channel image can be reduced by using pooling operation, the number of pixels of the 9 x 9km sub-matrixes in the original satellite channel image is reduced in sequence by taking the maximum value or the average value and the like, and then the pixels are reduced to one dimension, so that low-order information can be integrated into high-order information which is input into a full connection layer.
In the embodiment of the invention, the convolutional neural network continuously extracts the characteristics of the original matrix from low order to high order by alternately using convolution and pooling operations, and fully explores the internal information of the original matrix. Finally, the information is related to the quantity to be inverted through a simple Back Propagation (BP) neural network. As an optional embodiment, the satellite cloud picture includes bright temperature data of a predetermined number of channels, and for each grid point, extracting a feature region centered on the grid point, and performing convolution and pooling operations on the feature region includes:
b11, aiming at each channel, extracting a characteristic region taking the lattice point as the center according to each lattice point in the channel to obtain a plurality of characteristic regions;
in the embodiment of the present invention, if the channel includes k × k grid points, each grid point corresponds to a feature region, and there are k × k feature regions in total. For each lattice point in the feature region, the convolution value for the lattice point is derived from the lattice point and the product of the 8 surrounding lattice points and a3 x 3 average convolution kernel matrix.
And B12, inputting a plurality of characteristic areas corresponding to the channel into a convolutional layer and a pooling layer in the convolutional neural network for processing to obtain the characteristics of the channel.
In the embodiment of the invention, k × k feature areas are input into convolution layers in a convolution neural network for convolution operation, the result of the convolution operation is output to a pooling layer for pooling operation and the like, and the features of the channel are obtained after multiple times of convolution, pooling and other operations.
In the embodiment of the invention, the convolution and pooling operations are carried out by taking the channel as a unit, and the convolution and pooling operations among the channels are independent. As an alternative embodiment, the embodiment of the present invention has 10 channels, and correspondingly 10 features are obtained. Wherein the features of each channel are comprised of features of the respective lattice points.
In this embodiment of the present invention, as an optional embodiment, identifying a convective cloud based on a result of the convolution and pooling operation includes:
b21, inputting the characteristics corresponding to the preset channel number into a full connection layer and a softmax function layer in the convolutional neural network for processing to obtain a classified cloud cluster to which each grid point in the satellite cloud picture belongs;
in embodiments of the present invention, the fully-connected layer is comprised of multiple layers, for example, comprising: the first layer full connection layer, the second layer full connection layer, \ 8230and the mth layer full connection layer.
In an embodiment of the present invention, classifying the cloud includes: rolling cloud, convection cloud, high-accretion cloud, high-rise cloud, raincloud, accretion cloud, laminated cloud and laminar cloud.
And B22, obtaining lattice points of the convection cloud which is the classified cloud cluster, obtaining the convection cloud according to the lattice points of the convection cloud which is the classified cloud cluster, and obtaining the convection cloud in which each lattice point belongs to the convection cloud.
In the embodiment of the invention, in the training process of the convolutional neural network, by continuously comparing the error between the value obtained by current calculation and the true value result, the derivative of each convolutional kernel and offset can be obtained, and each parameter is adjusted along the negative direction of the derivative, so that the overall accuracy of the network can be improved, the value of each convolutional kernel can be adjusted by means of an error feedback mechanism, the convolutional neural network is automatically improved, and the desired fitting effect is finally obtained.
In the embodiment of the present invention, an error feedback mechanism of the convolutional neural network is as follows:
for convolutional layer
In the formula (I), the compound is shown in the specification,
l is the current neural network layer, i and j represent different channels, M j Shows the filtering of the results of the previous layer (the first layer shows the combination of different satellite channels and then the selection of different characteristic maps), b shows the bias parameters, x shows the input of the current layer, k is the convolution kernel parameters,. Indicates the convolution operation, and f is the activation function. At this time, the next layer is the pooling layer, and the derivative of the input variable of the current layer
Comprises the following steps:
in the formula, β represents a coefficient of the pooling layer, u represents an output of the current layer, o represents a point-by-point operation, and up represents changing each element of the matrix into a sub-matrix (each element value of the sub-matrix is the same as the original element).
The derivatives of the net total error E with respect to the bias and convolution kernels are:
wherein u, v represent the row and column numbers of the elements in the matrix,
is a convolution kernel in (l-1) layer
Act to obtain
The element set of the element of the u-th row and the v-th column.
For a pooling layer:
wherein down represents pooling operations,
is a parameter of the coefficient(s),
is a bias parameter. When the next layer is a convolutional layer, the derivative of the input variable of the current layer
Comprises the following steps:
where conv denotes the convolution operation and rot180 denotes the 180 degree rotation of the matrix, therefore, the derivatives of the convolutional neural network total error E with respect to the bias and coefficient are:
and 103, performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics to obtain the strong convection cloud cluster.
In the embodiment of the invention, interpolation processing needs to be carried out on the sunflower 8 satellite cloud picture with lower resolution ratio when space-time matching is carried out, so that the resolution ratio of the sunflower 8 satellite cloud picture after interpolation processing is the same as that of the radar cloud picture.
And marking the cloud cluster type corresponding to the strong convection weather in the satellite cloud classification result, and merging the cloud cluster type marked out from the radar cloud picture and the cloud cluster type marked out from the satellite cloud picture to obtain the strong convection cloud cluster.
The method for identifying the strong convection cloud cluster in the embodiment of the invention is based on three-dimensional radar echo data, and identifies the anvil cloud cluster, the transition anvil cloud cluster, the layer cloud cluster and the convection cloud cluster according to the difference of the development heights and the strengths of different types of cloud clusters to obtain a radar cloud classification result; acquiring brightness temperature data of a satellite cloud picture, extracting a characteristic region taking a grid point as a center aiming at each grid point, performing convolution and pooling operation on the characteristic region, and identifying 9 types of cloud clusters based on the results of the convolution and pooling operation to obtain a satellite cloud classification result; and performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained by statistics to obtain the strong convection cloud cluster. Therefore, the advantages of satellite data and radar data are fully utilized, the vertical development characteristic and the horizontal development characteristic of the convection cloud cluster are fully considered, the accurate cloud cluster classification result can be obtained through the cloud cluster classification technology based on the deep learning convolutional neural network, the cloud cluster classification result and the deep learning convolutional neural network are integrated, the advantages of detection modes of the cloud cluster classification technology and the deep learning convolutional neural network can be fully utilized, the strong convection cloud cluster is more accurately identified, the identification result is more reliable, and the identification precision is higher.
In the embodiment of the present invention, after the strong convection cloud is determined, the strong convection cloud may be tracked and forecasted, so as an optional embodiment, the method further includes:
c11, acquiring adjacent time strong convection clouds, and extrapolating the strong convection clouds identified in the previous time;
in the embodiment of the invention, the strong convection cloud cluster of the previous time and the strong convection cloud cluster of the next time are respectively identified by obtaining the strong convection identification results of the previous time and the next time.
In the embodiment of the present invention, extrapolation is performed on the cloud cluster, which can be specifically referred to in the related art documents, and detailed description is omitted here.
And C12, calculating the overlapping degree of the extrapolated strong convection cloud cluster and the strong convection cloud cluster of the next time in the geographical position, and if the overlapping degree is greater than a preset overlapping degree threshold value, determining that the strong convection cloud cluster of the next time is the tracking result of the strong convection cloud cluster of the previous time.
In the embodiment of the present invention, as an optional embodiment, before performing overlap matching, a distance threshold may be preset according to a time difference between adjacent times and an estimated moving speed of the strong convection cloud, and only the strong convection cloud within the range of the set distance threshold is subjected to overlap matching, so as to avoid a misjudgment.
In the embodiment of the invention, if the overlapping degree is greater than the preset overlapping degree threshold value, the tracking is successful. In the embodiment of the invention, the first strong convection cloud at the time t is defined as C, A (C) is the area of the first strong convection cloud C, and C' is the second strong convection cloud at the time t-delta tA (C ') is the area of the second convection monomer C', V (C) is the moving speed of the first strong convection cloud C, and the third strong convection cloud
And obtaining the first strong convection cloud C after V (C) delta t, and calculating the overlapping degree of the first strong convection cloud C and the second strong convection cloud C' by using the following formula:
in the formula (I), the compound is shown in the specification,
o (C, C') is the degree of overlap;
In the embodiment of the present invention, if O (C, C') > is not less than Tov1, it indicates that the overlap matching is successful. Wherein, tov1 is an overlap threshold, i.e. matching of strong convection clouds can be successful only when the overlap area between two strong convection clouds is large enough. Thus, as an alternative embodiment, after obtaining the strong convective cloud of the adjacent time, before extrapolating the strong convective cloud of the previous time, the method further includes:
and acquiring the minimum value in the areas of the strong convection clouds of the adjacent times, and if the minimum value is larger than a preset area threshold value, executing the step of extrapolating the strong convection cloud of the previous time.
In the embodiment of the invention, the area threshold value can be set according to actual requirements.
In the embodiment of the invention, the overlapping matching based on the area overlapping method needs a larger overlapping area, and is not suitable for the overlapping matching with a smaller area of the strong convection cloud cluster. Thus, as another alternative embodiment, the method further comprises:
c21, acquiring adjacent time-series strong convection clouds, and calculating the characteristic quantity of the adjacent time-series strong convection clouds, wherein the characteristic quantity comprises: the center of gravity position, the area, the eccentricity, the lowest brightness temperature and the average brightness temperature;
and C22, obtaining a value function value according to the characteristic value of each characteristic quantity and the weight of the characteristic quantity, and if the value function value is greater than a preset value function threshold, determining that the strong convection cloud of the next time is the tracking result of the strong convection cloud of the previous time.
In the embodiment of the invention, for the small strong convection cloud cluster, the feature values corresponding to the feature quantities of the previous and subsequent times are extracted, for example, the gravity center position, the area and the eccentricity of the small strong convection cloud cluster of the previous and subsequent times, the minimum brightness temperature in the convection cloud cluster and the average brightness temperature in the convection cloud cluster respectively correspond to the small strong convection cloud cluster of the previous and subsequent times, corresponding weights are given to the 5 feature quantities, the value function value is obtained by calculation, and the overlap degree is represented by the value function value.
According to the embodiment of the invention, according to the size of the strong convection cloud cluster, the area overlapping method can be adopted for overlapping matching aiming at the strong convection cloud cluster with large area so as to track, and the characteristic quantity weighting mode is adopted for matching tracking aiming at the strong convection cloud cluster with small area, so that the tracking of the strong convection cloud cluster is more accurate, the tracking precision is improved, and the method has important guiding significance for early warning and forecasting of the strong convection cloud cluster. Certainly, in practical applications, for a strong convection cloud with a large area, a feature quantity weighting mode may also be adopted for matching and tracking.
In the embodiment of the present invention, as an optional embodiment, the value of the cost function is calculated by using the following formula:
in the formula (I), the compound is shown in the specification,
I j is a value of a cost function;
C i is the characteristic value of the ith characteristic quantity;
w i is the weight of the ith feature quantity;
F i,j is a membership function of the ith characteristic quantity;
m is the total number of the feature quantities.
In the embodiment of the present invention, the eigenvalue corresponding to each eigenvalue includes an eigenvalue of a previous strong convection cloud and an eigenvalue of a next strong convection cloud.
In the embodiment of the invention, the larger the value of the cost function is, the larger the similarity of the strong convection clouds in the previous and subsequent times is.
In the embodiment of the present invention, as an optional embodiment, the threshold of the cost function is set to be 0.6, and when the calculated value of the cost function is greater than 0.6, the pairing is considered to be successful, and the strong convection cloud of the next time is a result of the motion of the strong convection cloud of the previous time.
In this embodiment of the present invention, as an optional embodiment, after obtaining the strong convection cloud of the neighboring time, before calculating the feature quantity of the strong convection cloud of the neighboring time, the method further includes:
and acquiring the minimum value in the areas of the adjacent time strong convection clouds, and if the minimum value is not larger than a preset area threshold, executing the step of calculating the characteristic quantity of the adjacent time strong convection clouds.
Fig. 2 shows a schematic structural diagram of an apparatus for identifying a strong convection cloud provided by an embodiment of the present invention.
As shown in fig. 2, the apparatus includes:
the radar cloud picture processing module 201 is used for identifying anvil cloud clusters, transition anvil cloud clusters, layer cloud clusters and convection cloud clusters according to the difference of development heights and intensities of different types of cloud clusters based on three-dimensional radar echo data to obtain radar cloud classification results;
in this embodiment of the present invention, as an optional embodiment, the radar cloud processing module 201 includes:
a anvils cloud cluster identification unit (not shown in the figure) for selecting three-dimensional radar echo data of a first height layer from the three-dimensional radar echo data;
and determining lattice points with radar echo intensity greater than zero from the three-dimensional radar echo data of the first height layer, and identifying the anvil cloud cluster according to the determined lattice points.
In the embodiment of the invention, the height of the first height layer is 6km-7.5km.
In this embodiment of the present invention, as another optional embodiment, the radar cloud processing module 201 further includes:
the transition anvil cloud cluster identification unit selects three-dimensional radar echo data of a second height layer from the anvil cloud clusters;
and determining lattice points with radar echo intensity greater than zero according to the three-dimensional radar echo data of the second height layer, and obtaining a transition anvil cloud cluster according to the determined lattice points.
In this embodiment of the present invention, as a further optional embodiment, the radar cloud processing module 201 further includes:
and the layer cloud cluster identification unit is used for acquiring lattice points corresponding to the three-dimensional radar echo data with the combined reflectivity larger than a preset first threshold value of the combined reflectivity from the transition anvil cloud cluster to obtain a layer cloud cluster, wherein the combined reflectivity corresponding to each lattice point in the layer cloud cluster is larger than the first threshold value of the combined reflectivity.
In an embodiment of the present invention, the first threshold of the combined reflectivity is 10dbz.
In this embodiment of the present invention, as a further optional embodiment, the radar cloud processing module 201 further includes:
and the convection cloud cluster identification unit is used for identifying mature convection cloud clusters from the transition anvil cloud clusters and the layer cloud clusters according to a preset first identification algorithm and identifying new or dead convection cloud clusters according to a preset second identification algorithm.
In the embodiment of the present invention, as an optional embodiment, a developed convection cloud is identified according to a preset first identification algorithm, where the first identification algorithm includes two criteria, that is:
according to the first judgment criterion, grid points corresponding to three-dimensional radar echo data with combined reflectivity larger than a preset combined reflectivity second threshold are obtained, and a developed convection cloud cluster is obtained, wherein the combined reflectivity corresponding to each grid point in the developed convection cloud cluster is larger than the combined reflectivity second threshold, and the combined reflectivity second threshold is larger than the combined reflectivity first threshold.
In this embodiment, as an optional embodiment, the second threshold of the combined reflectivity is 35dbz, and the first threshold of the combined reflectivity is 10dbz.
According to a second judgment criterion, acquiring a first convection cloud lattice point set of which the combined reflectivity is greater than a third threshold value of the combined reflectivity and less than a fourth threshold value of the combined reflectivity from the transition anvil cloud cluster and the layer cloud cluster;
in the first convection cloud lattice point set, for each lattice point, sequentially calculating a ratio of a square of a combined reflectivity of the lattice point to a preset first experience threshold, a first difference of a preset second experience threshold to the ratio, and a second difference of the combined reflectivity of the lattice point to the first difference;
and acquiring candidate grid points with a second difference value larger than zero, and constructing and developing mature convection clouds according to the candidate grid points.
In this embodiment, as an optional embodiment, the third threshold of the combined reflectivity is 15dbz, the fourth threshold of the combined reflectivity is 35dbz, the first empirical threshold is set to 120, and the second empirical threshold is set to 15.
In an embodiment of the present invention, as an optional embodiment, identifying a new or disappeared convective cloud cluster according to a preset second identification algorithm includes:
acquiring a second convection cloud lattice point set of which the combined reflectivity is not less than a fifth threshold value of the combined reflectivity;
for each grid point in the second convection cloud grid point set, acquiring preset adjacent threshold grid points with the grid point as a center, and if the adjacent threshold grid points all belong to a transition anvil cloud cluster or a layer cloud cluster, placing the grid point in the grid point set;
and constructing a new born or lost and killed convective cloud cluster based on the lattice point set.
In this embodiment of the present invention, as an optional embodiment, the fifth threshold of the combined reflectivity is 40dbz, and the preset adjacent threshold is 8.
The satellite cloud picture processing module 202 is used for acquiring brightness temperature data of a satellite cloud picture, extracting a feature region taking the grid point as a center for each grid point, performing convolution and pooling operation on the feature region, and identifying 9 types of clouds based on the results of the convolution and pooling operation to obtain satellite cloud classification results;
as an optional embodiment, the satellite cloud picture includes bright temperature data of a predetermined number of channels, and for each grid point, extracting a feature region centered on the grid point, and performing convolution and pooling operations on the feature region includes:
aiming at each channel, extracting a characteristic region taking the lattice point as the center according to each lattice point in the channel to obtain a plurality of characteristic regions;
and inputting a plurality of characteristic areas corresponding to the channel into a convolutional layer and a pooling layer in the convolutional neural network for processing to obtain the characteristics of the channel.
In this embodiment of the present invention, as an optional embodiment, identifying a convective cloud based on the result of the convolution and pooling operations includes:
inputting the characteristics corresponding to the preset channel number into a full connection layer and a softmax function layer in a convolutional neural network for processing to obtain a classified cloud cluster to which each grid point in the satellite cloud picture belongs;
and acquiring grid points of the convection cloud according to the classification cloud, wherein each grid point belongs to the convection cloud in the obtained convection cloud.
And the strong convection cloud cluster acquisition module 203 is configured to perform space-time matching on the obtained radar cloud classification result and satellite cloud classification result, and perform coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics, so as to acquire a strong convection cloud cluster.
In the embodiment of the invention, interpolation processing needs to be carried out on the sunflower 8 satellite cloud picture with lower resolution ratio when space-time matching is carried out, so that the resolution ratio of the sunflower 8 satellite cloud picture after interpolation processing is the same as that of the radar cloud picture.
And marking the cloud cluster type corresponding to the strong convection weather in the satellite cloud classification result, and merging the cloud cluster type marked out from the radar cloud picture and the cloud cluster type marked out from the satellite cloud picture to obtain the strong convection cloud cluster.
In this embodiment of the present invention, as an optional embodiment, the apparatus further includes:
a tracking module (not shown in the figure) configured to obtain neighboring temporal strong convection clouds and calculate feature quantities of the neighboring temporal strong convection clouds, where the feature quantities include: the center of gravity position, area, eccentricity, minimum brightness temperature and average brightness temperature;
and obtaining a value function value according to the characteristic value of each characteristic quantity and the weight of the characteristic quantity, and if the value function value is larger than a preset value function threshold, confirming that the strong convection cloud cluster at the next time is the tracking result of the strong convection cloud cluster at the previous time.
In this embodiment, as another optional embodiment, the apparatus further includes:
the extrapolation matching module is used for acquiring the strong convection clouds of adjacent times and extrapolating the strong convection clouds of the previous time; and calculating the overlapping degree of the extrapolated strong convection cloud cluster and the strong convection cloud cluster at the next time on the geographical position, and if the overlapping degree is greater than a preset overlapping degree threshold value, confirming that the strong convection cloud cluster at the next time is the tracking result of the strong convection cloud cluster at the previous time.
In this embodiment, as a further optional embodiment, the apparatus further includes:
and the prediction module is used for generating the motion trail of the strong convection cloud cluster according to the same strong convection cloud cluster confirmed in sequence and predicting the future motion trail of the strong convection cloud cluster based on the motion trail and a preset prediction algorithm.
As shown in fig. 3, an embodiment of the present application provides a computer device 300 for executing the method for identifying a strong convection cloud in fig. 1, the device includes a memory 301, a processor 302, and a computer program stored in the memory 301 and executable on the processor 302, wherein the processor 302 implements the steps of the method for identifying a strong convection cloud when executing the computer program.
Specifically, the memory 301 and the processor 302 can be general-purpose memory and processor, and are not limited in particular, and when the processor 302 runs the computer program stored in the memory 301, the method for identifying the strong convection cloud can be performed.
Corresponding to the method for identifying strong convection clouds in fig. 1, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the steps of the above method for identifying strong convection clouds.
In particular, the storage medium can be a general-purpose storage medium, such as a removable disk, a hard disk, or the like, and when executed, the computer program on the storage medium can perform the above method for identifying strong convective clouds.
In the embodiments provided in the present application, it should be understood that the disclosed system and method may be implemented in other ways. The above-described system embodiments are merely illustrative, and for example, the division of the units is only one logical functional division, and there may be other divisions in actual implementation, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of systems or units through some communication interfaces, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one position, or may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus once an item is defined in one figure, it need not be further defined and explained in subsequent figures, and moreover, the terms "first", "second", "third", etc. are used merely to distinguish one description from another and are not to be construed as indicating or implying relative importance.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used for illustrating the technical solutions of the present application, but not limiting the same, and the scope of the present application is not limited thereto, and although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present application. Are intended to be covered by the scope of this application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method of identifying a strong convective cloud, comprising:
based on three-dimensional radar echo data, identifying anvil cloud clusters, transition anvil cloud clusters, layer cloud clusters and convection cloud clusters according to the difference of development heights and intensities of different types of cloud clusters to obtain radar cloud classification results;
acquiring brightness temperature data of a satellite cloud picture, extracting a characteristic region taking a grid point as a center aiming at each grid point, performing convolution and pooling operation on the characteristic region, and identifying 9 types of cloud clusters based on the result of the convolution and pooling operation to obtain a satellite cloud classification result;
and performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics to obtain the strong convection cloud cluster.
2. The method of claim 1, wherein identifying the anvil cloud based on the three-dimensional radar echo data from differences in the heights and intensities of different types of clouds comprises:
selecting three-dimensional radar echo data of a first height layer from the three-dimensional radar echo data;
and determining lattice points with radar echo intensity greater than zero according to the three-dimensional radar echo data of the first height layer, and identifying the anvil cloud cluster according to the determined lattice points.
3. The method of claim 1, wherein identifying the cloud of transition anvils comprises:
selecting three-dimensional radar echo data of a second height layer from the anvil cloud cluster;
and determining lattice points with radar echo intensity larger than zero aiming at the three-dimensional radar echo data of the second height layer, and obtaining a transition anvil cloud cluster according to the determined lattice points.
4. The method of claim 3, wherein identifying the layer cloud comprises:
and acquiring lattice points corresponding to the three-dimensional radar echo data with the combined reflectivity larger than a preset combined reflectivity first threshold value from the transition anvil cloud cluster to obtain a layer cloud cluster, wherein the combined reflectivity corresponding to each lattice point in the layer cloud cluster is larger than the combined reflectivity first threshold value.
5. The method of claim 4, wherein identifying the convective cloud comprises:
and identifying mature convective clouds from the transitional anvil cloud clouds and the laminar cloud clouds according to a preset first identification algorithm, and identifying new or dead convective clouds according to a preset second identification algorithm.
6. The method of any of claims 1 to 5, further comprising:
acquiring adjacent time strong convection cloud clusters, and calculating the characteristic quantity of the adjacent time strong convection cloud clusters, wherein the characteristic quantity comprises: the center of gravity position, area, eccentricity, minimum brightness temperature and average brightness temperature;
and obtaining a value function value according to the characteristic value of each characteristic quantity and the weight of the characteristic quantity, and if the value function value is larger than a preset value function threshold, confirming that the strong convection cloud cluster at the next time is the tracking result of the strong convection cloud cluster at the previous time.
7. The method according to any one of claims 1 to 5, further comprising:
and generating the motion trail of the strong convection cloud cluster according to the same strong convection cloud cluster confirmed in sequence, and predicting the future motion trail of the strong convection cloud cluster based on the motion trail and a preset prediction algorithm.
8. An apparatus for identifying strong convective clouds, comprising:
the radar cloud picture processing module is used for identifying anvil cloud clusters, transition anvil cloud clusters, layer cloud clusters and convection cloud clusters according to the difference of development heights and strengths of different types of cloud clusters on the basis of three-dimensional radar echo data to obtain radar cloud classification results;
the satellite cloud picture processing module is used for acquiring brightness temperature data of the satellite cloud picture, extracting a characteristic region with the grid point as the center aiming at each grid point, performing convolution and pooling operation on the characteristic region, and identifying 9 types of cloud clusters based on the results of the convolution and pooling operation to obtain a satellite cloud classification result;
and the strong convection cloud cluster acquisition module is used for performing space-time matching on the obtained radar cloud classification result and the satellite cloud classification result, and performing coincidence processing on the space-time matching result based on the cloud cluster type corresponding to the ground strong convection weather obtained through statistics to acquire the strong convection cloud cluster.
9. An electronic device, comprising: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory communicating over the bus when the electronic device is operating, the machine-readable instructions when executed by the processor performing the steps of the method of identifying strong convective clouds in any of claims 1 to 7.
10. A computer-readable storage medium, having stored thereon a computer program which, when being executed by a processor, is adapted to carry out the steps of the method of identifying strong convective clouds of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911423710.4A CN110942111B (en) | 2019-12-31 | 2019-12-31 | Method and device for identifying strong convection cloud cluster |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911423710.4A CN110942111B (en) | 2019-12-31 | 2019-12-31 | Method and device for identifying strong convection cloud cluster |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110942111A CN110942111A (en) | 2020-03-31 |
| CN110942111B true CN110942111B (en) | 2022-11-29 |
Family
ID=69912964
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911423710.4A Active CN110942111B (en) | 2019-12-31 | 2019-12-31 | Method and device for identifying strong convection cloud cluster |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110942111B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111474529B (en) * | 2020-06-10 | 2023-07-07 | 浙江省气象台 | Satellite inversion radar echo method, inversion radar echo system and navigation radar |
| CN112924974B (en) * | 2021-01-21 | 2022-11-25 | 中国气象科学研究院 | Method and device for identifying cloud cluster by using DBSCAN clustering algorithm and electronic equipment |
| CN113109816B (en) * | 2021-03-29 | 2023-10-03 | 广东工业大学 | Echo block tracking method, device and storage medium of radar echo image |
| CN115326207B (en) * | 2022-10-14 | 2023-02-10 | 四创科技有限公司 | Convection cloud identification method and terminal based on sliding window |
| CN116778354B (en) * | 2023-08-08 | 2023-11-21 | 南京信息工程大学 | Deep learning-based visible light synthetic cloud image marine strong convection cloud cluster identification method |
| CN118053077A (en) * | 2023-12-12 | 2024-05-17 | 兰州大学 | Method for detecting cloud-aerosol by using FY-4A full-disc data at all days |
| CN117518299B (en) * | 2024-01-05 | 2024-03-22 | 南京大学 | Classified strong convection proximity probability forecasting method, system, equipment and terminal |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5839089A (en) * | 1994-10-20 | 1998-11-17 | Kabushiki Kaisha Toshiba | Thunder cloud observation system |
| CN105022101A (en) * | 2015-06-23 | 2015-11-04 | 国家卫星气象中心 | Strong convective cloud tracking method |
| CN107993215A (en) * | 2017-11-27 | 2018-05-04 | 象辑知源（武汉）科技有限公司 | A kind of weather radar image processing method and system |
| CN108445464A (en) * | 2018-03-12 | 2018-08-24 | 南京恩瑞特实业有限公司 | Satellite radar inverting fusion methods of the NRIET based on machine learning |
| CN110414420A (en) * | 2019-07-25 | 2019-11-05 | 中国人民解放军国防科技大学 | Mesoscale convection system identification and tracking method based on infrared cloud picture of stationary satellite |
-
2019
- 2019-12-31 CN CN201911423710.4A patent/CN110942111B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5839089A (en) * | 1994-10-20 | 1998-11-17 | Kabushiki Kaisha Toshiba | Thunder cloud observation system |
| CN105022101A (en) * | 2015-06-23 | 2015-11-04 | 国家卫星气象中心 | Strong convective cloud tracking method |
| CN107993215A (en) * | 2017-11-27 | 2018-05-04 | 象辑知源（武汉）科技有限公司 | A kind of weather radar image processing method and system |
| CN108445464A (en) * | 2018-03-12 | 2018-08-24 | 南京恩瑞特实业有限公司 | Satellite radar inverting fusion methods of the NRIET based on machine learning |
| CN110414420A (en) * | 2019-07-25 | 2019-11-05 | 中国人民解放军国防科技大学 | Mesoscale convection system identification and tracking method based on infrared cloud picture of stationary satellite |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110942111A (en) | 2020-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110942111B (en) | Method and device for identifying strong convection cloud cluster | |
| JP6331114B2 (en) | POSITION ESTIMATION DEVICE, POSITION ESTIMATION METHOD, AND CONTROL PROGRAM | |
| McRoberts et al. | Using remotely sensed data to construct and assess forest attribute maps and related spatial products | |
| Zhang et al. | Detecting horizontal and vertical urban growth from medium resolution imagery and its relationships with major socioeconomic factors | |
| CN112001339A (en) | Pedestrian social distance real-time monitoring method based on YOLO v4 | |
| CN108038846A (en) | Transmission line equipment image defect detection method and system based on multilayer convolutional neural networks | |
| CN107463901A (en) | Multiple dimensioned Regional Flooding disasters danger remote sensing evaluation method and system | |
| CN109917394B (en) | Short-term intelligent extrapolation method based on weather radar | |
| CN107479065B (en) | Forest gap three-dimensional structure measuring method based on laser radar | |
| CN113469278B (en) | Strong weather target identification method based on deep convolutional neural network | |
| CN104237890B (en) | The heavy rain identification that one is caused by " train effect " and forecasting procedure | |
| CN104463164A (en) | Tree canopy structure information extraction method based on rib method and crown height ratio | |
| CN108254750B (en) | Down-blast intelligent identification early warning method based on radar data | |
| JP2019211342A (en) | Weather analyzer, weather analysis method, and program | |
| CN113553764A (en) | Mountain fire prediction method based on deep learning network | |
| CN110717496B (en) | Complex scene tree detection method based on neural network | |
| CN111323788B (en) | Building change monitoring method and device and computer equipment | |
| CN115691049A (en) | Convection birth early warning method based on deep learning | |
| Yang et al. | Flood detection based on unmanned aerial vehicle system and deep learning | |
| CN111259736A (en) | Real-time pedestrian detection method based on deep learning in complex environment | |
| Sardar et al. | Ensemble learning with CNN and BMO for drought prediction | |
| CN114386654A (en) | Multi-scale numerical weather forecasting mode fusion weather forecasting method and device | |
| CN111624606B (en) | Radar image rainfall identification method | |
| Pillai et al. | Fine-Tuned EfficientNetB4 Transfer Learning Model for Weather Classification | |
| CN116012618A (en) | Weather identification method, system, equipment and medium based on radar echo diagram |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |