Mountain area surface atmospheric visibility inversion method and system and computer equipment

Abstract

The invention relates to the field of meteorological inversion, and provides a mountainous area surface atmospheric visibility inversion method and system and a computer device, through designing a fog zone visibility model, fog zone visibility is calculated based on an Attention attention mechanism and a BP neural network, identification accuracy of a fog zone is improved, interpolation calculation is carried out through an inverse distance weighted interpolation algorithm, and the accuracy of the fog zone visibility is improved. The method obtains the visibility of the target area, accurately reflects the distribution trend of the atmospheric visibility, further improves the recognition accuracy, carries out the weight calculation according to the meteorological elements and elevation data of the target area, considers the generation and maintenance conditions of the fog area to enhance the prediction of the change of the fog area, also considers the actual landform of the mountainous area, and improves the recognition accuracy. The influence of the complex terrain of the mountainous area is avoided, the weight fusion inversion is finally carried out, the final visibility inversion result is obtained, the inversion result is closer to the actual state of the mountainous area surface, and the accuracy and applicability of the mountainous area surface atmospheric visibility inversion method are improved.

Description

Technical Field

[0001] The present invention relates to the field of meteorological inversion, and particularly to a method, a system and a computer device for inverting the atmospheric visibility on the surface of mountainous areas. Background Art

[0002] Meteorological visibility is a physical quantity representing the visual distance of the human eye, including daytime visibility and nighttime visibility. In fact, it is the meteorological optical range. The main objective factors affecting visibility include: atmospheric transparency, and the brightness contrast between the target and the background (referring to the horizon sky background). Among them, atmospheric transparency is the direct factor affecting visibility. The air molecules and aerosol particles contained in the atmosphere can weaken the energy of the light passing through the atmosphere. The more water vapor and impurities in the air, the more turbid the air, the worse the atmospheric transparency, and the worse the meteorological visibility. With the rapid development of meteorological observations, the requirements for the accuracy of meteorological observations are gradually increasing, especially for meteorological visibility.

[0003] In the prior art, visibility observations are mainly realized by establishing artificial meteorological stations. However, the current observation method has the disadvantages of insufficient number of observation stations and low spatial resolution, and it is difficult to meet the high spatial resolution requirements for visibility in meteorological work. Moreover, the existing enhancement schemes are difficult to represent the correct visibility distribution, and can only vaguely reflect the visibility distribution trend, but are not accurate enough to meet the refined monitoring requirements of meteorology. Especially for the surface visibility in mountainous areas, it is seriously affected by haze weather. The boundary of the fog is restricted by the terrain, and the terrain has a great influence on the formation and maintenance of the fog. Most of the existing inversion methods are designed for the sea surface and land plains, and most of them do not take into account this important influencing factor of the terrain. Therefore, there is a lack of targeted optimization for complex mountainous areas, and the visibility inversion effect of haze formation and dissipation in mountainous areas is poor.

[0004] Therefore, how to design a method for inverting atmospheric visibility to avoid the influence of complex terrain in mountainous areas and improve the accuracy and applicability of inversion. Summary of the Invention

[0005] Based on this, a method, system and computer device for inverting the atmospheric visibility of mountainous surface proposed by the present invention design a fog area visibility model, calculate the fog area visibility based on the Attention mechanism and BP neural network, improve the recognition accuracy of the fog area, and then perform interpolation calculation through the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, which accurately reflects the distribution trend of the atmospheric visibility, further improves the recognition accuracy, and calculates the weights according to the meteorological elements and elevation data of the target area. It not only considers the generation and maintenance conditions of the fog area to enhance the prediction of the fog area changes, but also considers the actual topography and landforms of the mountainous area to avoid the influence of the complex terrain in the mountainous area. Finally, weight fusion inversion is performed to obtain the final visibility inversion result, making the inversion result closer to the actual state of the mountainous surface. The present invention improves the accuracy and applicability of the method for inverting the atmospheric visibility of mountainous surface.

[0006] A method for inverting the atmospheric visibility of mountainous surface proposed by the present invention includes: Obtain multi-source meteorological data of the target area, calculate the multi-source meteorological data of the target area according to the fog area visibility model to obtain the fog area visibility. The multi-source meteorological data includes meteorological station data and satellite remote sensing data. The fog area visibility model includes an Attention part and a BP neural network part. The Attention part performs attention enhancement processing based on the attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on the BP neural network; Then perform interpolation calculation processing on the multi-source meteorological data to obtain the visibility of the target area. The interpolation calculation processing is based on the inverse distance weighted interpolation algorithm; Calculate the meteorological element weight and elevation data weight of the target area. The meteorological element weight is based on the relative humidity weight and wind speed weight of the target area, and the elevation data weight is based on the elevation characteristics and slope characteristics of the fog area; Perform weight fusion visibility inversion processing according to the fog area visibility, target area visibility, meteorological element weight and elevation data weight to obtain the final visibility inversion result.

[0007] In summary, according to the above method for retrieving the atmospheric visibility of mountainous areas, by designing a visibility model for foggy areas, calculating the visibility of foggy areas based on the Attention mechanism and BP neural network, the recognition accuracy of foggy areas is improved. Then, interpolation calculation is performed through the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, which accurately reflects the distribution trend of atmospheric visibility and further improves the recognition accuracy. Additionally, weight calculation is carried out based on the meteorological elements and elevation data of the target area. This not only considers the generation and maintenance conditions of foggy areas to enhance the prediction of fog area changes but also takes into account the actual terrain and landforms in mountainous areas, avoiding the influence of complex mountain terrains. Finally, weight fusion inversion is performed to obtain the final visibility inversion result, making the inversion result closer to the actual state of the mountainous surface. The present invention improves the accuracy and applicability of the method for retrieving the atmospheric visibility of mountainous areas. Specifically, multi-source meteorological data of the target area is obtained, and the multi-source meteorological data of the target area is calculated according to the fog area visibility model to obtain the visibility of the fog area. The multi-source meteorological data includes meteorological station data and satellite remote sensing data. The fog area visibility model includes an Attention part and a BP neural network part. The Attention part performs attention enhancement processing based on the attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on the BP neural network, improving the recognition accuracy of foggy areas. Then, interpolation calculation processing is performed on the multi-source meteorological data to obtain the visibility of the target area. The interpolation calculation processing is based on the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, which accurately reflects the distribution trend of atmospheric visibility and further improves the recognition accuracy. The meteorological element weight and elevation data weight of the target area are calculated. The meteorological element weight is based on the relative humidity weight and wind speed weight of the target area, and the elevation data weight is based on the elevation characteristics and slope characteristics of the fog area. This not only considers the generation and maintenance conditions of foggy areas to enhance the prediction of fog area changes but also takes into account the actual terrain and landforms in mountainous areas, avoiding the influence of complex mountain terrains. According to the fog area visibility, target area visibility, meteorological element weight, and elevation data weight, weight fusion visibility inversion processing is performed to obtain the final visibility inversion result, making the inversion result closer to the actual state of the mountainous surface. The present invention improves the accuracy and applicability of the method for retrieving the atmospheric visibility of mountainous areas.

[0008] Further, the step of calculating the visibility of the fog area by calculating the multi-source meteorological data of the target area according to the fog area visibility model specifically includes: After obtaining the multi-source meteorological data of the target area, calculate the multi-source meteorological data of the target area according to the fog area visibility model to obtain the fog area visibility. The fog area visibility model includes an Attention part and a BP neural network part. The multi-source meteorological data includes meteorological station data and satellite remote sensing data. The meteorological station data includes temperature, dew point temperature, and wind speed. The satellite remote sensing data includes visible light data, mid-infrared data, and long-wave infrared data; The Attention part divides the multi-source meteorological data into multiple single features, and each single feature corresponds uniquely to any meteorological data in the multi-source meteorological data; Obtain the query vector, key vector, and value vector of the single feature, and perform dot product operation and weighted summation processing on the query vector, key vector, and value vector according to the attention mechanism to obtain the associated weighted output of all single features. The associated weighted output represents the weighted output after any single feature is associated with all single features. The specific formula of the attention mechanism is as follows: , , where, represents the attention degree of the th single feature pair to the th single feature, , and c respectively represent the ordinal numbers of different single features, q , k , v respectively represent the query vector, key vector, and value vector, Tr represents transpose, represents the associated weighted output of the single feature; The BP neural network part performs error calculation and iterative optimization processing according to the associated weighted output of all single features.

[0009] Further, the steps of the BP neural network part performing error calculation and iterative optimization processing according to the associated weighted output of all single features specifically include: The BP neural network part performs layer-by-layer weighted summation processing on the associated weighted output of all single features according to the network layer order to obtain an error value. The error value is the arithmetic deviation between the visibility estimation value and the observed true value; Judge whether the error value meets the acceptable accuracy. If the error value meets the acceptable accuracy, output the neuron output value of the current network layer. If the error value does not meet the acceptable accuracy, perform iterative optimization of the neuron weights and biases; The specific formula of the BP neural network part is as follows: , , , wherein, a represents the neuron output value, l represents the number of network layers, I and J respectively represent the ordinal numbers of different neurons, N represents the total number of neurons, w represents the neuron weight, b represents the neuron bias, represents the learning rate, L represents the loss function value of the output network layer.

[0010] Further, the step of further performing interpolation calculation processing on the multi-source meteorological data to obtain the visibility of the target area specifically includes: Obtain the hourly minimum visibility data of the meteorological stations in the target area from the multi-source meteorological data, and perform interpolation calculation on the minimum visibility data according to the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, and the visibility of the target area is grid data; The specific formula of the inverse distance weighted interpolation algorithm is as follows: , , wherein, represents the visibility interpolation prediction result of the inverse distance weight, D represents the number of reference observation points, represents the hourly minimum visibility observation value of the target area, represents the weight coefficient of the observation point of the hourly minimum visibility data to the estimated interpolation point, represents the distance attenuation coefficient, represents the distance between the point to be interpolated and the reference observation point.

[0011] Further, the step of calculating the meteorological element weight and the elevation data weight of the target area specifically includes: Obtain the relative humidity according to the temperature and dew point temperature of the target area, and the relative humidity is grid data. The formula for obtaining the relative humidity is as follows: , wherein, RH represents the relative humidity, Td represents the dew point temperature, T represents the temperature; Calculate the relative humidity weight according to the relative humidity, and the specific formula for the relative humidity weight is as follows: , wherein, represents the relative humidity weight; Calculate the wind speed weight according to the wind speed in the target area, and the specific formula for the wind speed weight is as follows: , wherein, represents the wind speed weight, represents the wind speed in the target area; Calculate the meteorological element weight of the target area according to the relative humidity weight and the wind speed weight, and the specific formula for the meteorological element weight is as follows: , wherein, represents the meteorological element weight; Calculate the elevation data weight based on the elevation characteristics and slope characteristics of the fog area.

[0012] Further, the steps of calculating the elevation data weight based on the elevation characteristics and slope characteristics of the fog area specifically include: Obtain the original elevation data in the target area, and then calculate the slope data according to the digital elevation model. The digital elevation model is based on the third-order inverse distance squared weight difference algorithm of the geographic information model, and the geographic information model is constructed based on the terrain information and positioning information of the target area; Calculate the elevation data weight according to the original elevation data and the slope data, and the specific formula for the elevation data weight is as follows: , wherein, represents the elevation data weight, represents the original elevation data, represents the slope data.

[0013] Further, the steps of performing weighted fusion visibility inversion processing according to the fog area visibility, the target area visibility, the meteorological element weight, and the elevation data weight to obtain the final visibility inversion result specifically include: Perform weighted fusion visibility inversion processing according to the fog area visibility, the target area visibility, the meteorological element weight, and the elevation data weight to obtain the final visibility inversion result; The specific formula for obtaining the final visibility inversion result is as follows: , wherein, represents the final visibility inversion result, Indicates the inversion prediction result of the visibility in the fog area, Indicates the interpolation prediction result of the visibility with inverse distance weighting, Indicates the elevation data weight, Indicates the meteorological element weight.

[0014] A mountain surface atmospheric visibility inversion system proposed by the present invention includes: A fog area visibility calculation module, which is used to obtain multi-source meteorological data of the target area, calculate the multi-source meteorological data of the target area according to the fog area visibility model to obtain the fog area visibility. The multi-source meteorological data includes meteorological station data and satellite remote sensing data. The fog area visibility model includes an Attention part and a BP neural network part. The Attention part performs attention enhancement processing based on the attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on the BP neural network; A target area visibility calculation module, which is used to perform interpolation calculation processing on the multi-source meteorological data again to obtain the target area visibility. The interpolation calculation processing is based on the inverse distance weighting interpolation algorithm; A weight calculation module, which is used to calculate the meteorological element weight and elevation data weight of the target area. The meteorological element weight is based on the relative humidity weight and wind speed weight of the target area, and the elevation data weight is based on the elevation characteristics and slope characteristics of the fog area; An inversion result module, which is used to perform weighted fusion visibility inversion processing according to the fog area visibility, target area visibility, meteorological element weight and elevation data weight to obtain the final visibility inversion result.

[0015] The present invention also provides a storage medium, which stores one or more programs. When the programs are executed by a processor, the mountain surface atmospheric visibility inversion method as described above is implemented.

[0016] The present invention also provides a computer device, which includes a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer programs stored in the memory, the mountain surface atmospheric visibility inversion method as described above is implemented. Description of the Drawings

[0017] Figure 1 Is the flowchart of the mountain surface atmospheric visibility inversion method proposed in the first embodiment of the present invention; Figure 2 Is the structural schematic diagram of the mountain surface atmospheric visibility inversion system proposed in the second embodiment of the present invention; Figure 3Schematic diagram of the fog visibility model of the mountain surface atmospheric visibility inversion method proposed in the first embodiment of the present invention.

[0018] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. Specific Embodiments

[0019] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0020] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly on the other element or there can also be a middle element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be a middle element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for the purpose of illustration.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0022] Please refer to Figure 1 , which shows the flowchart of the mountain surface atmospheric visibility inversion method proposed in the first embodiment of the present invention. This mountain surface atmospheric visibility inversion method includes steps S01 to S04, where: Step S01: Obtain multi-source meteorological data of the target area, and calculate the multi-source meteorological data of the target area according to the fog visibility model to obtain the fog visibility; It should be noted that in this embodiment, the multi-source meteorological data includes meteorological station data and satellite remote sensing data. The fog visibility model includes an Attention part and a BP neural network part. The Attention part performs attention enhancement processing based on the attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on the BP neural network. For the specific structure setting of the fog visibility model, please refer to Figure 3, in the figure, A1, A2, A3, and A4 represent the input data of the Attention part, B1, B2, B3, and B4 represent the output data after the attention enhancement process based on the attention mechanism, and C1, C2, C3, and C4 represent the output data of the Attention part, that is, the input data of the BP neural network part; After obtaining the multi-source meteorological data of the target area, calculate the multi-source meteorological data of the target area according to the fog visibility model to obtain the fog visibility. The fog visibility model includes an Attention part and a BP neural network part. The multi-source meteorological data includes meteorological station data and satellite remote sensing data. The meteorological station data includes temperature, dew point temperature, and wind speed. The satellite remote sensing data includes visible light data, mid-infrared data, and long-wave infrared data; The multi-source meteorological data in this embodiment includes artificial meteorological station data: temperature, dew point temperature, wind speed, and remote sensing data of the geostationary meteorological satellite FY4B: visible light with a wavelength of 0.65 microns in channel 2, mid-infrared with a wavelength of 3.75 microns in channel 7, and long-wave infrared with a wavelength of 10.8 microns in channel 13; The Attention part divides the multi-source meteorological data into multiple single features, and each single feature uniquely corresponds to any meteorological data in the multi-source meteorological data; Obtain the query vector, key vector, and value vector of the single feature, and perform dot product operation and weighted summation processing on the query vector, key vector, and value vector according to the attention mechanism to obtain the associated weighted output of all single features. The associated weighted output represents the weighted output after any single feature is associated with all single features. The specific formula of the attention mechanism is as follows: , , Among them, represents the th single feature pair's attention to the th single feature, , and c respectively represent the ordinals of different single features, q , k , v respectively represent the query vector, key vector, and value vector, Tr represents transpose, represents the associated weighted output of the single feature; The BP neural network part performs error calculation and iterative optimization processing according to the associated weighted output of all single features; The BP neural network part performs layer-by-layer weighted summation processing on the associated weighted outputs of all single features according to the network layer order to obtain an error value, where the error value is the arithmetic deviation between the visibility estimation value and the observed true value; In this embodiment, the acceptable accuracy is defined as less than or equal to 100m; Judge whether the error value meets the acceptable accuracy. If the error value meets the acceptable accuracy, output the neuron output value of the current network layer. If the error value does not meet the acceptable accuracy, perform iterative optimization of the neuron weights and biases; The specific formula of the BP neural network part is as follows: , , , Among them, a represents the neuron output value, l represents the number of network layers, I and J respectively represent the ordinal numbers of different neurons, N represents the total number of neurons, w represents the neuron weight, b represents the neuron bias, represents the learning rate, L represents the loss function value of the output network layer.

[0023] Step S02: Then perform interpolation calculation processing on the multi-source meteorological data to obtain the visibility of the target area; It should be noted that in this embodiment, the interpolation calculation processing is based on the inverse distance weighted interpolation algorithm; Obtain the minimum visibility data within one hour of the meteorological stations in the target area from the multi-source meteorological data, and perform interpolation calculation on the minimum visibility data according to the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, where the visibility of the target area is grid data; The specific formula of the inverse distance weighted interpolation algorithm is as follows: , , Among them, represents the visibility interpolation prediction result of the inverse distance weight, D represents the number of reference observation points, represents the observed value of the minimum visibility within one hour of the target area, represents the weight coefficient of the observation point of the minimum visibility data to the estimated interpolation point, represents the distance attenuation coefficient, Represents the distance between the interpolation point to be interpolated and the reference observation point. In this embodiment, the inverse distance weight is 5 points.

[0024] Step S03: Calculate the meteorological element weight and elevation data weight of the target area; It should be noted that in this embodiment, the meteorological element weight is based on the relative humidity weight and wind speed weight of the target area, and the elevation data weight is based on the elevation characteristics and slope characteristics of the fog area; In this embodiment, based on the favorable and unfavorable meteorological element conditions for the occurrence and maintenance of the cooling fog with the highest frequency, a weight judgment is made. Among them, the favorable meteorological conditions are that the relative humidity is greater than 90% and the wind speed is less than 8 m / s. Generally speaking, reaching the favorable meteorological conditions does not necessarily mean that there will be fog, but if these meteorological conditions are not met, it is difficult for fog to appear. Compared with visibility observations, the number of observation stations for humidity and wind speed is much larger. Taking Jiangxi Province as an example, the number of visibility observation stations is 92, while the number of observation stations with both humidity and wind speed is 2538. Therefore, using humidity and wind to set weights can effectively improve the spatial resolution and the adaptability of the method of the present invention; Obtain the relative humidity according to the temperature and dew point temperature of the target area. The relative humidity is grid data. The formula for obtaining the relative humidity is as follows: , Among them, RH Represents the relative humidity, Td Represents the dew point temperature, T Represents the temperature; In this embodiment, when the relative humidity threshold for fog distribution is set such that the relative humidity is above 98%, the condition is extremely good, and the weight can be set to 1. When the relative humidity is between 90% and 98%, the condition is better, and the weight decreases as the humidity decreases. When the relative humidity is below 90%, it is considered that the condition is poor and there will be no fog; Calculate the relative humidity weight according to the relative humidity. The specific formula for the relative humidity weight is as follows: , Among them, Represents the relative humidity weight; In this embodiment, when the wind speed threshold for heavy fog distribution is considered, when the wind speed is below 4 m / s, the condition is extremely good, and the weight can be set to 1. When the relative humidity is between 4 m / s and 8 m / s, the condition is better, and the weight decreases as the wind speed increases. When the wind speed is above 8 m / s, it is considered that the condition is poor and there will be no fog; Calculate the wind speed weight according to the wind speed of the target area. The specific formula for the wind speed weight is as follows: , Among them, Represents the wind speed weight, Represents the wind speed in the target area; Calculate the meteorological element weight of the target area according to the relative humidity weight and the wind speed weight. The specific formula for the meteorological element weight is as follows: , where, Represents the meteorological element weight; Calculate the elevation data weight based on the elevation characteristics and slope characteristics of the fog area; Obtain the original elevation data in the target area, and then calculate the slope data according to the digital elevation model. The digital elevation model is based on the third-order inverse distance squared weight difference algorithm of the geographic information model, and the geographic information model is constructed based on the terrain information and positioning information of the target area; In this embodiment, based on the fog spatial distribution characteristics being close to the ground and having a low vertical extension height, usually from dozens of meters to three hundred meters, and due to its stable atmospheric stratification conditions, it is difficult to climb to higher terrains in areas with large slopes. Therefore, the distribution of fog has characteristics closely related to elevation and terrain slope. When the elevation is below 300 m, it is considered conducive to heavy fog. When the elevation is between 300 m and 1000 m, it is considered that both slope and elevation have an impact on the distribution. The logical relationship is that the higher the elevation, the smaller the weight, and it is also related to the slope. In places with large slopes, it is difficult for fog to climb. Therefore, when the slope is too large, set to 25 degrees, the weight needs to be appropriately reduced. When the elevation is greater than 1000 m, the fog reaches a height where it is difficult to develop, and the wind speed above 1000 m is relatively large, making it difficult to form large-scale fog, and only small-scale patchy fog appears. Set the weight to 0; Calculate the elevation data weight according to the original elevation data and the slope data. The specific formula for the elevation data weight is as follows: , where, Represents the elevation data weight, Represents the original elevation data, Represents the slope data.

[0025] Step S04: Perform weighted fusion visibility inversion processing according to the fog area visibility, the target area visibility, the meteorological element weight, and the elevation data weight to obtain the final visibility inversion result; It should be noted that in this embodiment, weighted fusion visibility inversion processing is performed according to the fog area visibility, the target area visibility, the meteorological element weight, and the elevation data weight to obtain the final visibility inversion result; In this embodiment, based on the characteristics of historical heavy fog and low visibility weather processes, there are often sudden changes in visibility values at the fog boundary. For each grid point, a logical judgment is made. When the product of the meteorological element weight and the elevation data weight is greater than 0.8, the result of the fog area visibility inversion model is adopted. When the product of the meteorological element weight and the elevation data weight is between 0.5 and 0.8, the visibility result of the target area calculated by interpolation is adopted. When the product of the meteorological element weight and the elevation data weight is less than 0.5, it is considered that fog is unlikely to occur and the probability of low atmospheric visibility is low, so it is set as high visibility. In this embodiment, only below 1000 m is considered, so it is set as 1000 m; The specific formula for obtaining the final visibility inversion result is as follows: , where, represents the final visibility inversion result, represents the fog area visibility inversion prediction result, represents the visibility interpolation prediction result of the inverse distance weight, represents the elevation data weight, represents the meteorological element weight.

[0026] In summary, according to the above-mentioned method for inverting the atmospheric visibility of mountainous areas, by designing a visibility model for foggy areas, calculating the visibility of foggy areas based on the Attention mechanism and BP neural network, the recognition accuracy of foggy areas is improved. Then, interpolation calculation is performed through the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, which accurately reflects the distribution trend of atmospheric visibility and further improves the recognition accuracy. Moreover, weight calculation is carried out according to the meteorological elements and elevation data of the target area. Not only the generation and maintenance conditions of foggy areas are considered to enhance the prediction of foggy area changes, but also the actual topography and landforms of mountainous areas are considered to avoid the influence of complex terrains in mountainous areas. Finally, weight fusion inversion is performed to obtain the final visibility inversion result, making the inversion result closer to the actual state of the mountainous surface. The present invention improves the accuracy and applicability of the method for inverting the atmospheric visibility of mountainous areas. Specifically, multi-source meteorological data of the target area are obtained, and the multi-source meteorological data of the target area are calculated according to the visibility model of foggy areas to obtain the visibility of foggy areas. The multi-source meteorological data include meteorological station data and satellite remote sensing data. The visibility model of foggy areas includes an Attention part and a BP neural network part. The Attention part performs attention enhancement processing based on the attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on the BP neural network, improving the recognition accuracy of foggy areas. Then, interpolation calculation processing is performed on the multi-source meteorological data to obtain the visibility of the target area. The interpolation calculation processing is based on the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, which accurately reflects the distribution trend of atmospheric visibility and further improves the recognition accuracy. The meteorological element weight and elevation data weight of the target area are calculated. The meteorological element weight is based on the relative humidity weight and wind speed weight of the target area, and the elevation data weight is based on the elevation characteristics and slope characteristics of foggy areas. Not only the generation and maintenance conditions of foggy areas are considered to enhance the prediction of foggy area changes, but also the actual topography and landforms of mountainous areas are considered to avoid the influence of complex terrains in mountainous areas. According to the visibility of foggy areas, the visibility of the target area, the meteorological element weight, and the elevation data weight, weight fusion visibility inversion processing is performed to obtain the final visibility inversion result, making the inversion result closer to the actual state of the mountainous surface. The present invention improves the accuracy and applicability of the method for inverting the atmospheric visibility of mountainous areas.

[0027] Please refer to Figure 2 , which shows the structural schematic diagram of the system for inverting the atmospheric visibility of mountainous areas proposed in the third embodiment of the present invention. The system includes: The fog area visibility calculation module 10 is used to obtain multi-source meteorological data of the target area, calculate the multi-source meteorological data of the target area according to the fog area visibility model to obtain the fog area visibility. The multi-source meteorological data includes meteorological station data and satellite remote sensing data. The fog area visibility model includes an Attention part and a BP neural network part. The Attention part performs attention enhancement processing based on the attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on the BP neural network; The target area visibility calculation module 20 is used to perform interpolation calculation processing on the multi-source meteorological data to obtain the target area visibility. The interpolation calculation processing is based on the inverse distance weighted interpolation algorithm; The weight calculation module 30 is used to calculate the meteorological element weight and elevation data weight of the target area. The meteorological element weight is based on the relative humidity weight and wind speed weight of the target area, and the elevation data weight is based on the elevation characteristics and slope characteristics of the fog area; The inversion result module 40 is used to perform weighted fusion visibility inversion processing according to the fog area visibility, target area visibility, meteorological element weight and elevation data weight to obtain the final visibility inversion result.

[0028] The present invention also proposes a computer storage medium, on which one or more programs are stored. When the program is executed by a processor, the above-mentioned mountain surface atmospheric visibility inversion method is implemented.

[0029] The present invention also proposes a computer device, including a memory and a processor. The memory is used to store a computer program, and the processor is used to execute the computer program stored on the memory to implement the above-mentioned mountain surface atmospheric visibility inversion method.

[0030] Those skilled in the art can understand that the logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus or device), or used in combination with these instruction execution systems, apparatus or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus or device.

[0031] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0032] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques known in the art can be used: a discrete logic circuit having logic gate circuits for implementing logical functions on data signals, an application-specific integrated circuit having appropriate combinational logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0033] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0034] The above-described embodiments merely represent several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims

1. A method for inverting surface atmospheric visibility in mountainous areas, characterized in that: include: Acquire multi-source meteorological data of a target area, and calculate the multi-source meteorological data of the target area according to a fog area visibility model to obtain visibility in the fog area, wherein the multi-source meteorological data includes meteorological station data and satellite remote sensing data, and the fog area visibility model includes an Attention part and a BP neural network part, wherein the Attention part performs attention enhancement processing based on an attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on a BP neural network; Then, interpolation calculation processing is performed on the multi-source meteorological data to obtain visibility of the target area, and the interpolation calculation processing is based on an inverse distance weighted interpolation algorithm; Calculating meteorological element weights and elevation data weights of the target area, wherein the meteorological element weights are based on relative humidity weights and wind speed weights of the target area, and the elevation data weights are based on elevation features and slope features of the fog area; A weighted fusion visibility inversion process is performed according to the visibility in the fog area, the visibility in the target area, the weight of meteorological elements and the weight of elevation data to obtain a final visibility inversion result.

2. The method for inverting the surface atmospheric visibility in mountainous areas according to claim 1, characterized in that: The step of calculating the multi-source meteorological data of the target area according to the fog area visibility model to obtain the visibility of the fog area specifically includes: After obtaining multi-source meteorological data of the target area, the multi-source meteorological data of the target area are calculated according to a fog area visibility model to obtain fog area visibility, wherein the fog area visibility model includes an attention part and a BP neural network part, the multi-source meteorological data includes meteorological station data and satellite remote sensing data, the meteorological station data includes temperature, dew point temperature and wind speed, and the satellite remote sensing data includes visible light data, mid-infrared data and long-wave infrared data; The Attention part divides the multi-source meteorological data into a plurality of single features, each of which uniquely corresponds to any meteorological data in the multi-source meteorological data; The query vector, key vector and value vector of the single feature are obtained, and the query vector, key vector and value vector are subjected to dot product operation and weighted summation according to the attention mechanism to obtain the associated weighted output of all single features, wherein the associated weighted output represents the weighted output after any single feature is associated with all single features. The specific formula of the attention mechanism is as follows: ， ， in, Indicates The first The attention paid to a single feature, , and c They represent the ordinal numbers of different single features, q , k , v Represent the query vector, key vector, and value vector respectively. Tr represents transpose, Represents the associated weighted output of a single feature; The BP neural network part performs error calculation and iterative optimization processing according to the associated weighted outputs of all the single features.

3. The method for inverting the surface atmospheric visibility in mountainous areas according to claim 2, characterized in that: The BP neural network part performs error calculation and iterative optimization processing according to the associated weighted outputs of all the single features, specifically including: The BP neural network part performs weighted summation of the associated weighted outputs of all single features layer by layer in the order of network layers to obtain an error value, which is the arithmetic deviation between the visibility estimation value and the observed true value; Determine whether the error value meets the acceptable accuracy, if the error value meets the acceptable accuracy, output the neuron output value of the current network layer, if the error value does not meet the acceptable accuracy, perform iterative optimization of neuron weights and biases; The specific formula of the BP neural network part is as follows: ， ， ， in, a represents the neuron output value, l Indicates the number of network layers, I and J Represent the ordinal numbers of different neurons, N represents the total number of neurons, w represents the neuron weight, b represents the neuron bias, represents the learning rate, L Represents the loss function value of the output network layer.

4. The method for inverting the surface atmospheric visibility in mountainous areas according to claim 1, characterized in that: The step of performing interpolation calculation processing on the multi-source meteorological data to obtain visibility of the target area specifically includes: Obtain the hourly minimum visibility data of the meteorological station in the target area from the multi-source meteorological data, and perform interpolation calculation on the minimum visibility data according to the inverse distance weighted interpolation algorithm to obtain the visibility of the target area, wherein the visibility of the target area is grid point data; The specific formula of the inverse distance weighted interpolation algorithm is as follows: ， ， in, Represents the inverse distance weighted visibility interpolation prediction result, D represents the number of reference observation points, It represents the minimum visibility observation value in the target area within the hour. The weight coefficient of the observation point representing the minimum visibility data within the hour to the estimated interpolation point, represents the distance attenuation coefficient, Indicates the distance between the point to be interpolated and the reference observation point.

5. The method for inverting the surface atmospheric visibility in mountainous areas according to claim 1, characterized in that: The step of calculating the meteorological element weights and elevation data weights of the target area specifically includes: The relative humidity is obtained according to the temperature and dew point temperature of the target area. The relative humidity is grid data. The formula for obtaining the relative humidity is as follows: ， in, RH Relative humidity, Td Indicates the dew point temperature, T Indicates temperature; The relative humidity weight is calculated according to the relative humidity, and the specific formula of the relative humidity weight is as follows: ， in, represents the relative humidity weight; The wind speed weight is calculated according to the wind speed of the target area. The specific formula of the wind speed weight is as follows: ， in, represents the wind speed weight, Indicates the wind speed in the target area; The meteorological element weight of the target area is calculated according to the relative humidity weight and the wind speed weight. The specific formula of the meteorological element weight is as follows: ， in, Indicates the weight of meteorological elements; The weight of elevation data is calculated based on the elevation and slope characteristics of the fog area.

6. The method for inverting the surface atmospheric visibility in mountainous areas according to claim 5, characterized in that: The step of calculating the elevation data weight based on the elevation characteristics and slope characteristics of the fog area specifically includes: Obtaining original elevation data in the target area, and then calculating slope data according to an elevation digital model, wherein the elevation digital model is based on a third-order inverse distance square weighted difference algorithm of a geographic information model, and the geographic information model is constructed based on terrain information and positioning information of the target area; The elevation data weight is calculated based on the original elevation data and slope data. The specific formula of the elevation data weight is as follows: ， in, represents the weight of elevation data, represents the original elevation data, Represents slope data.

7. The method for inverting the surface atmospheric visibility in mountainous areas according to claim 1, characterized in that: The step of performing weighted fusion visibility inversion processing according to the visibility in the fog area, the visibility in the target area, the weight of meteorological elements and the weight of elevation data to obtain the final visibility inversion result specifically includes: According to the visibility in fog area, the visibility in target area, the weight of meteorological elements and the weight of elevation data, the weighted fusion visibility inversion processing is performed to obtain the final visibility inversion result; The specific formula for obtaining the final visibility inversion result is as follows: ， in, represents the final visibility inversion result, represents the inversion prediction result of visibility in fog area, Represents the inverse distance weighted visibility interpolation prediction result, represents the weight of elevation data, Represents the weight of meteorological elements.

8. A mountainous surface atmospheric visibility inversion system, characterized in that: include: A fog area visibility calculation module is used to obtain multi-source meteorological data of a target area, and calculate the multi-source meteorological data of the target area according to a fog area visibility model to obtain fog area visibility, wherein the multi-source meteorological data includes meteorological station data and satellite remote sensing data, and the fog area visibility model includes an Attention part and a BP neural network part, wherein the Attention part performs attention enhancement processing based on an attention mechanism, and the BP neural network part performs error calculation and iterative optimization processing based on a BP neural network; A target area visibility calculation module, used for further performing interpolation calculation processing on the multi-source meteorological data to obtain the visibility of the target area, wherein the interpolation calculation processing is based on an inverse distance weighted interpolation algorithm; A weight calculation module, used for calculating the meteorological element weight and the elevation data weight of the target area, wherein the meteorological element weight is based on the relative humidity weight and the wind speed weight of the target area, and the elevation data weight is based on the elevation characteristics and slope characteristics of the fog area; The inversion result module is used to perform weighted fusion visibility inversion processing according to the visibility in the fog area, the visibility in the target area, the weight of the meteorological elements and the weight of the elevation data to obtain the final visibility inversion result.

9. A storage medium, characterized in that: The storage medium stores one or more programs, which, when executed by a processor, implement the method for inverting the surface atmospheric visibility in a mountainous area as described in any one of claims 1 to 7.

10. A computer device, characterized in that: The computer device comprises a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer program stored in the memory, it implements the method for inverting the surface atmospheric visibility in mountainous areas as described in any one of claims 1-7.

Images