Atmospheric correction method and device for radiation signal in coastal waters under solar zenith angle

Abstract

The present application provides an atmospheric correction method and device for radiation signals in coastal waters under the solar zenith angle. The method provided by the present application includes: using an ultraviolet-near infrared atmospheric correction combined model to generate noon remote sensing reflectance satellite data based on the observation data of coastal waters; using the maximum cross-correlation method to obtain adjusted noon remote sensing reflectance satellite data based on the spatial resolution and temporal resolution of the data; calculating the Rayleigh correction radiance based on the Rayleigh scattering look-up table corresponding to the observation data, and matching the Rayleigh correction radiance and the adjusted noon remote sensing reflectance satellite data based on the time information and location information to obtain a training set; inputting the radiation signal of the coastal waters to be corrected into the model trained based on the training set to determine the water-leaving radiance of the radiation signal. The method and device provided by the present application achieve atmospheric correction of radiation signals in coastal waters under a high solar zenith angle for turbid coastal waters.

Description

Technical Field

[0001] This application relates to the technical field of remote sensing and atmospheric correction, and particularly to an atmospheric correction method and device for radiation signals in coastal waters under the solar zenith angle. Background Art

[0002] Since gases and aerosols in the atmosphere can cause scattering and absorption effects on the radiation signals received by sensors, the information of target objects in the image is interfered with and distorted. Through atmospheric correction, these atmospheric effects can be removed or reduced, and the radiation values observed by the sensors can be converted into the true reflectance of ground objects. This can ensure that remote sensing data more accurately reflects surface features, and thus improve the reliability and accuracy of applications such as land cover, water quality monitoring, and ecosystem analysis. Especially in water body remote sensing, accurate atmospheric correction is of great significance for distinguishing water bodies and suspended substances and monitoring water quality changes.

[0003] Currently, the geostationary ocean color sensor is usually used to observe radiation signals, which observes once per hour from dawn to dusk. Inevitably, it will encounter challenges at extremely high solar zenith angles in the morning and evening. Especially in winter, more than 50% of the data at the central observation position of the Global Ocean Observing Index-II is collected at high solar zenith angles (greater than 70 degrees), and there is even more high-latitude observation data at extreme solar zenith angles. Such high solar zenith angle observations by geostationary satellite ocean color sensors have greatly affected the reflectance at the top of the atmosphere. It should be noted that as the solar zenith angle increases, due to the contributions of Rayleigh scattering and aerosol scattering, the atmospheric path radiance will increase, and when the amount of light penetrating the water body rapidly decreases, the specular reflection of the sea surface will also increase. Therefore, the backscattering signal carrying water body information will decrease. Due to the weak illumination at high solar zenith angles, the optical path of Rayleigh scattering and aerosol scattering is extended, and the bidirectional reflection effect is enhanced. In addition, due to the increased influence of the earth's curvature, traditional atmospheric correction models are not very suitable for high solar zenith angle observations.

[0004] The atmospheric correction model adopted by traditional near-infrared iteration schemes estimates the aerosol radiance inaccurately under high solar zenith angles (SZA), resulting in a large loss of effective products when the geostationary satellite ocean color sensor observes at dawn or dusk. To solve this problem, an atmospheric correction model has been developed, which is applicable to the open sea observed by the first geostationary satellite ocean color imager (GOCI) at high solar zenith angles. This model is constructed based on a stable open sea dataset. Due to the large dynamic changes in water body properties and the lack of reliable measurements during the noon period, this model is not applicable to turbid coastal areas and is not very suitable for coastal waters. The atmospheric correction model for coastal waters has become a research gap. Therefore, there is an urgent need for a method to achieve atmospheric correction of radiation signals in coastal waters at high solar zenith angles for turbid coastal waters. Summary of the Invention

[0005] In view of this, the present application provides an atmospheric correction method and device for the radiation signal of coastal waters under the solar zenith angle, which is used to achieve the atmospheric correction of the radiation signal of coastal waters under a high solar zenith angle for turbid coastal waters.

[0006] Specifically, the present application is implemented through the following technical solutions:

[0007] The first aspect of the present application provides an atmospheric correction method for the radiation signal of coastal waters under the solar zenith angle, and the method includes:

[0008] Using an ultraviolet-near infrared atmospheric correction combined model, generating noon remote sensing reflectance satellite data based on the observation data of coastal waters; the noon remote sensing reflectance satellite data represents the water-leaving radiance of the radiation signal in the coastal waters;

[0009] Using the maximum cross-correlation method, deleting the pixels with a flow velocity exceeding the flow velocity threshold in the noon remote sensing reflectance satellite data based on the spatial resolution and temporal resolution of the noon remote sensing reflectance satellite data, and obtaining the adjusted noon remote sensing reflectance satellite data;

[0010] Determining the Rayleigh scattering radiance based on the Rayleigh scattering lookup table corresponding to the observation data, calculating the Rayleigh correction radiance based on the Rayleigh scattering radiance, and matching the Rayleigh correction radiance and the adjusted noon remote sensing reflectance satellite data based on the time information and position information to obtain a training set;

[0011] Training the ultraviolet-near infrared atmospheric correction combined model based on the training set, inputting the Rayleigh correction radiance corresponding to the radiation signal of the coastal waters to be corrected into the trained ultraviolet-near infrared atmospheric correction combined model, and determining the water-leaving radiance of the radiation signal based on the output noon remote sensing reflectance satellite data.

[0012] The second aspect of the present application provides an atmospheric correction device for the radiation signal of coastal waters under the solar zenith angle, and the device includes a generation module, an adjustment module, a matching module, and a determination module; wherein,

[0013] The generation module is used to generate noon remote sensing reflectance satellite data based on the observation data of coastal waters by using an ultraviolet-near infrared atmospheric correction combined model; the noon remote sensing reflectance satellite data represents the water-leaving radiance of the radiation signal in the coastal waters;

[0014] The adjustment module is configured to use the maximum cross-correlation method to delete the pixels with flow velocity exceeding the flow velocity threshold in the noon remote sensing reflectance satellite data based on the spatial resolution and temporal resolution of the noon remote sensing reflectance satellite data, so as to obtain the adjusted noon remote sensing reflectance satellite data;

[0015] The matching module is configured to determine the Rayleigh scattering radiance based on the Rayleigh scattering lookup table corresponding to the observation data, calculate the Rayleigh-corrected radiance based on the Rayleigh scattering radiance, and match the Rayleigh-corrected radiance and the adjusted noon remote sensing reflectance satellite data based on the time information and position information to obtain a training set;

[0016] The determination module is configured to train the ultraviolet-near infrared atmospheric correction combined model based on the training set, input the Rayleigh-corrected radiance corresponding to the coastal water radiation signal to be corrected into the trained ultraviolet-near infrared atmospheric correction combined model, and determine the water-leaving radiance of the radiation signal based on the output noon remote sensing reflectance satellite data.

[0017] The atmospheric correction method and device for the radiation signal of coastal waters under the solar zenith angle provided by this application. In the first aspect, by combining the near-infrared atmospheric correction model and the ultraviolet atmospheric correction model, the observation data of coastal waters are processed based on the ultraviolet-near-infrared atmospheric correction combined model to generate the satellite data of the remote sensing reflectance at noon. Since coastal waters include both clear waters and turbid waters, doing so not only gives play to the advantage of the near-infrared atmospheric correction model in better considering the influence of suspended solids and other particulate matters in the water body on the radiation signal when dealing with turbid waters, improving the accuracy of correction; but also gives play to the advantage that the ultraviolet atmospheric correction model is suitable for the correction of clear waters, can effectively process the optical properties of the water body, and ensure accurate reflectance under the condition of small light absorption. By combining the two models, various water body conditions can be processed, the accuracy of the satellite data of the remote sensing reflectance at noon can be improved, and it has strong adaptability. The most suitable model can be selected for processing according to the characteristics of the actual observation data, so as to obtain more reliable satellite data of the remote sensing reflectance at noon. Moreover, this ultraviolet-near-infrared atmospheric correction combined model can be applied to different types of waters, not limited to specific clear or turbid waters, making it have a wider applicability. In the second aspect, the flow velocity of each pixel is calculated according to the spatial resolution and temporal resolution of the satellite data of the remote sensing reflectance at noon, and the pixels with a flow velocity exceeding the flow velocity threshold are removed. In this way, large-flow pixels can be effectively filtered out, ensuring that within the temporal resolution of the satellite data of the remote sensing reflectance at noon, the pixels with slower flow velocities remain in the original pixel block, thus avoiding the rapid changes in the water body caused by strong water currents. By removing the pixels with faster flow velocities, the influence of instantaneous changes caused by water flow on the analysis of observation data is reduced, thereby improving the stability of the data. The pixels with slower flow velocities usually represent a more stable state of the water body, which helps to more accurately reflect the optical properties and other relevant information of the water body. Moreover, the flow velocity threshold can be flexibly adjusted according to different water body characteristics to adapt to various environmental conditions and application requirements. Description of the Drawings

[0018] Figure 1 It is a flowchart of the first embodiment of the atmospheric correction method for the radiation signal of coastal waters under the solar zenith angle provided by this application;

[0019] Figure 2 It is a schematic structural diagram of the first embodiment of the atmospheric correction device for the radiation signal of coastal waters under the solar zenith angle provided by this application. Detailed Embodiments

[0020] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0021] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms "a", "the", and "said" used in this application are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0022] It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0023] Specific embodiments are given below to introduce the technical solutions of this application in detail.

[0024] Figure 1 It is a flowchart of the first embodiment of the atmospheric correction method for the radiation signal in coastal waters under the solar zenith angle provided for this application. Please refer to Figure 1 , the method provided in this embodiment may include:

[0025] S101. Using an ultraviolet-near infrared atmospheric correction combined model, generate satellite data of noon remote sensing reflectance based on the observation data of coastal waters; the satellite data of noon remote sensing reflectance characterizes the water-leaving radiance of the radiation signal in the coastal waters.

[0026] Specifically, the ultraviolet-near infrared atmospheric correction combined model is used to perform atmospheric correction on the radiation signal of coastal waters, and determines the water-leaving radiance of the radiation signal in coastal waters based on the observation data of coastal waters. Among them, the water-leaving radiance removes the water body reflection characteristics affected by the atmosphere and can more accurately reflect the spectral characteristics of the water body. The ultraviolet-near infrared atmospheric correction combined model includes an input layer, a first hidden layer, a second hidden layer, and an output layer connected in sequence. Among them, the activation functions of the first hidden layer and the second hidden layer are hyperbolic tangent activation functions.

[0027] Furthermore, coastal waters are usually turbid waters, and the observation data of coastal waters include water body reflectance, pigment concentration, suspended sediment concentration, water temperature data, meteorological related data, etc. The noon remote sensing reflectance satellite data refers to the remote sensing reflectance satellite data at noon. At noon, the sun is near the zenith, and the solar zenith angle is usually small (close to 0°). The incident angle of sunlight is close to vertical, which can minimize the influence of shadow effect and illumination change, provide relatively stable observation conditions, and the solar radiation intensity at noon is usually high, which can improve the signal intensity of reflectance and reduce the influence of atmospheric scattering on reflectance. Therefore, the noon remote sensing reflectance satellite data is closer to the true radiation signal of the water body itself and can better represent the water-leaving radiance of the radiation signal in coastal waters.

[0028] When specifically implemented, generating the noon remote sensing reflectance satellite data based on the observation data of coastal waters by using the ultraviolet-near infrared atmospheric correction combined model includes:

[0029] (1) The input layer receives the observation data, preprocesses the observation data, and transfers the preprocessed observation data to the first hidden layer; the preprocessed observation data includes observation angle information and Rayleigh-corrected radiance of multiple bands.

[0030] Specifically, the preprocessed observation data includes observation angle information (such as solar zenith angle, observation zenith angle, etc.) and Rayleigh-corrected radiance of multiple bands.

[0031] When specifically implemented, the input layer receives the observation data of coastal waters detected by remote sensing instruments (usually including the radiation signal of the water body and related observation information), extracts the observation angle information in the observation data, and uses the Rayleigh scattering correction algorithm to correct the original radiation signal, removes the influence caused by atmospheric scattering, obtains the Rayleigh-corrected radiance of multiple bands, and finally standardizes or normalizes the extracted observation angle information and Rayleigh-corrected radiance, and transfers the preprocessed observation data to the first hidden layer.

[0032] (2) Each neuron in the first hidden layer performs weighted summation on the preprocessed observation data, extracts the features of the preprocessed observation data, and each neuron outputs first features of different scales and transfers them to the second hidden layer. The first features characterize the correlation between the observation angle information and the Rayleigh-corrected radiance.

[0033] Specifically, the first hidden layer contains multiple neurons. Each neuron is used to extract the first feature of the observation data at one scale, and the first features extracted by each neuron are different. The first features characterize the correlation between the observation angle information and the Rayleigh-corrected radiance in the observation data and reflect the data characteristics under different observation conditions.

[0034] In specific implementation, each neuron in the first hidden layer performs a weighted sum of the preprocessed observation data based on the neuron weights (the neuron weights of each neuron are obtained through training and learning, and the neuron weights of different neurons are different), applies the hyperbolic tangent activation function to perform a non-linear transformation on the result of the weighted sum, introduces non-linear features, and obtains the first features of different scales.

[0035] (3) Each neuron in the second hidden layer performs a weighted sum and transformation on the first features of different scales output by the first hidden layer. Each neuron in the second hidden layer outputs a second feature, and the scales of the second features output by each neuron are different. The second features are transmitted to the output layer; the second features represent the correlation between the observation angle information and the Rayleigh-corrected radiance, and the data accuracy of the second features is higher than that of the first features.

[0036] Specifically, the second hidden layer is similar to the first hidden layer. The second hidden layer also contains multiple neurons. Each neuron is used to extract the second features of the first features at a deeper level, and the sizes of the second features extracted by each neuron are different. The second features also represent the correlation between the observation angle information and the Rayleigh-corrected radiance in the observation data. Compared with the first features, the second features have higher data accuracy and more deeply reflect the data features under different observation conditions.

[0037] In specific implementation, each neuron in the second hidden layer performs a weighted sum of the first features of different scales output by the first hidden layer based on the neuron weights, applies the hyperbolic tangent activation function to perform a non-linear transformation on the result of the weighted sum, introduces non-linear features, and obtains the second features of different scales.

[0038] (4) The output layer performs a weighted sum on each second feature and obtains the satellite data of the noon remote sensing reflectance output by each neuron through non-linear transformation; the number of neurons is the same as the number of bands.

[0039] Specifically, the number of neurons in the output layer is the same as the number of bands included in the input observation data, that is, each neuron corresponds to the reflectance output of one band, and the output layer can process the information of multiple bands simultaneously.

[0040] In specific implementation, multiple neurons in the output layer respectively perform a weighted sum on each scale of the second features, and apply a non-linear transformation (such as softmax transformation or sigmoid transformation) to convert the sum result into the satellite data of the noon remote sensing reflectance, ensure that the satellite data of the noon remote sensing reflectance output is within a reasonable range, and obtain the satellite data of the noon remote sensing reflectance output by multiple neurons.

[0041] Optionally, the ultraviolet-near-infrared atmospheric correction combined model includes a near-infrared atmospheric correction model and an ultraviolet atmospheric correction model. The near-infrared atmospheric correction model adopts the dark pixel assumption, that is, the radiance leaving the water in the near-infrared band is zero, resulting in low accuracy in highly turbid waters. The ultraviolet atmospheric correction model adopts the dark pixel assumption in the ultraviolet band, that is, the radiance leaving the water in the near-ultraviolet band is zero, which is applicable to turbid waters and not applicable to clear ocean waters. Since in this application, coastal waters include both clear waters and turbid waters, therefore, the near-infrared atmospheric correction model and the ultraviolet atmospheric correction model are combined to generate the satellite data of the noon remote sensing reflectance.

[0042] The method for generating the satellite data of the noon remote sensing reflectance based on the observation data of the coastal waters by using the ultraviolet-near-infrared atmospheric correction combined model includes: using the near-infrared atmospheric correction model to calculate the satellite data of the noon remote sensing reflectance of the first band based on the observation data of the first band, and obtaining the satellite data of the noon remote sensing reflectance of the second band based on the satellite data of the noon remote sensing reflectance of the first band; the wavelength of the second band is greater than that of the first band; the proportion of clear waters in the satellite data of the noon remote sensing reflectance of the second band is greater than that of turbid waters; using the ultraviolet atmospheric correction model to calculate the satellite data of the noon remote sensing reflectance of the third band; the wavelength of the third band is less than that of the first band; the proportion of turbid waters in the satellite data of the noon remote sensing reflectance of the third band is greater than that of clear waters; using the logistic regression method to fit the first band, the second band, and the third band to determine the weight of the first band, the weight of the second band, and the weight of the third band, and performing weighted fusion on the satellite data of the noon remote sensing reflectance based on the weights to obtain the satellite data of the noon remote sensing reflectance.

[0043] Specifically, the first band, the second band, and the third band are set according to actual needs, and in this embodiment, no limitation is imposed on this. It should be noted that the wavelength of the third band < the wavelength of the first band < the wavelength of the second band. The first band and the second band are usually near-infrared bands, and the third band is usually an ultraviolet band.

[0044] In specific implementation, the observed data of the first band is used, and the near-infrared atmospheric correction model is applied to calculate the satellite data of the noon remote sensing reflectance of the first band. Since the near-infrared band is sensitive to the optical properties of water bodies, it can effectively remove the atmospheric influence. Based on the satellite data of the noon remote sensing reflectance of the first band, the satellite data of the noon remote sensing reflectance of the second band is derived by using an appropriate algorithm (for example, the color ratio method or model fitting) (realized through the correlation of spectral characteristics). Further, the ultraviolet atmospheric correction model is used to process the observed data and calculate the satellite data of the noon remote sensing reflectance of the third band. Since the third band is the ultraviolet band, the relationship between its spectral characteristics and the components of water bodies and the atmosphere may be different, and the data of the ultraviolet band can provide different spectral information. The logistic regression method is used to fit the satellite data of the noon remote sensing reflectance of the first band, the second band, and the third band to determine the weights of each band. Based on the determined weights, the satellite data of the noon remote sensing reflectance of the first band, the second band, and the third band are weighted and fused to obtain the final satellite data of the noon remote sensing reflectance. This weighted fusion better integrates the information from different bands and improves the overall quality and reliability of the data.

[0045] S102. Using the maximum cross-correlation method, based on the spatial resolution and temporal resolution of the satellite data of the noon remote sensing reflectance, delete the pixels in the satellite data of the noon remote sensing reflectance whose flow velocity exceeds the flow velocity threshold, and obtain the adjusted satellite data of the noon remote sensing reflectance.

[0046] Specifically, the flow velocity threshold is set according to actual needs, and in this embodiment, it is not limited thereto. For example, in one embodiment, the flow velocity threshold is . The flow velocity threshold is determined according to the spatial resolution of the satellite data of the noon remote sensing reflectance, and the spatial resolution ranges from 250 meters at the sub-satellite point to 500 meters for the entire scene. This flow velocity threshold effectively filters out the high-flow pixels, ensuring that within the 1-hour time interval of the satellite data of the noon remote sensing reflectance, the pixels with slower flow velocities remain in the original pixel blocks, thus avoiding the rapid changes in water bodies caused by strong water currents.

[0047] In specific implementation, the step of using the maximum cross-correlation method to delete the pixels in the satellite data of the noon remote sensing reflectance whose flow velocity exceeds the flow velocity threshold based on the spatial resolution and temporal resolution of the satellite data of the noon remote sensing reflectance includes: determining the time interval based on the temporal resolution of the satellite data of the noon remote sensing reflectance; performing spatial alignment on the satellite data of the noon remote sensing reflectance of adjacent time intervals; calculating the cross-correlation corresponding to the pixels at the same position in the satellite data of the noon remote sensing reflectance of the adjacent time intervals, and determining the actual distance based on the pixel displacement corresponding to the maximum value and the spatial resolution of the satellite data of the noon remote sensing reflectance; and determining the flow velocity of each pixel based on the actual distance and the time interval.

[0048] Specifically, an adjacent time interval (e.g., 1 hour) is determined according to the temporal resolution (e.g., data acquisition frequency) of the midday remote sensing reflectance satellite data. For all the midday remote sensing reflectance satellite data, the midday remote sensing reflectance satellite data of adjacent time intervals are spatially aligned (using image registration technology to correct the midday remote sensing reflectance satellite data at different times so that the images of the same water body area at different times can be accurately aligned), ensuring that the midday remote sensing reflectance satellite data acquired at adjacent time intervals correspond to the same position in space. Further, for the pixels at the same position in the midday remote sensing reflectance satellite data of adjacent time intervals, their cross-correlation is calculated (cross-correlation can quantify the similarity between two images), and the maximum cross-correlation value is determined. The pixel displacement corresponding to the maximum cross-correlation value is determined as the actual pixel displacement. The product of the actual pixel displacement and the spatial resolution of the midday remote sensing reflectance satellite data is determined as the actual distance, and based on the quotient of the actual distance and the determined time interval, the flow velocity of each pixel is calculated.

[0049] It should be noted that the spatial resolution of the midday remote sensing reflectance satellite data is not an exact value and varies dynamically within a variable range.

[0050] Optionally, the method further includes: determining the sensor type based on the characteristics of the coastal waters, the working bands, and the characteristics of the sensor; determining the range of the spatial resolution based on the spatial resolution of each working band of the sensor and the external observation conditions; verifying the spatial resolution based on the distinguishability of different features in the observation data, and adjusting the range of the spatial resolution according to the verification result.

[0051] In specific implementation, appropriate sensors are selected considering the characteristics of coastal waters (such as optical characteristics, turbidity, phytoplankton distribution, etc.). For example, for water body monitoring, sensors with visible light, near-infrared, and ultraviolet bands may be selected to comprehensively analyze water quality and biological characteristics. Based on parameters such as the sensitivity, dynamic range, and spectral resolution of the selected sensors, suitable sensors are further screened. Based on the spatial resolution of each working band of the sensor, its applicability is analyzed, that is, the spatial resolution affects the ability to clearly distinguish the water body from the surrounding environment and monitor small water body characteristics (such as floating objects and algae distribution), and external factors such as atmospheric conditions, light intensity, and marine environment are considered, which will all affect the observation effect of the sensor. For example, in turbid waters, a higher spatial resolution may be required to improve the target detection ability. Further, discriminative analysis is carried out according to the spectral characteristics and reflection characteristics of different features (such as water bodies, floating objects, bottom topography, etc.). The detection effects of different spatial resolutions on different water body characteristics are tested. According to the detection results, if it is found that a specific feature cannot be effectively identified at the current spatial resolution, the resolution range needs to be adjusted. For example, if a higher resolution is required to detect specific water quality changes, the sensor parameters need to be adjusted or other sensors need to be selected.

[0052] S103. Determine the Rayleigh scattering radiance based on the Rayleigh scattering lookup table corresponding to the observation data, calculate the Rayleigh correction radiance based on the Rayleigh scattering radiance, and match the Rayleigh correction radiance and the adjusted satellite data of the noon remote sensing reflectance based on the time information and position information to obtain a training set.

[0053] Specifically, the Rayleigh scattering lookup table records the Rayleigh scattering radiance under different observation data, and different observation data correspond to different time information and position information. The Rayleigh correction radiance is calculated based on the Rayleigh scattering radiance, and the Rayleigh correction radiance is equal to the total radiance minus the Rayleigh scattering radiance, the white cap radiance, and the glint radiance. In this application, the ultraviolet-near-infrared atmospheric correction combined model performs atmospheric correction on the radiation signal of coastal waters based on the Rayleigh correction radiance. Therefore, first, the Rayleigh correction radiance needs to be calculated based on the Rayleigh scattering radiance (the white cap radiance and the glint radiance can be calculated according to the standard procedure).

[0054] In specific implementation, the determination process of the Rayleigh scattering lookup table includes:

[0055] (1) Calculate the Rayleigh optical depth of each band in the observation data based on the observation data.

[0056] In specific implementation, the Rayleigh optical depth of each band can be calculated according to the following formula:

[0057] ;

[0058] Among them, the is the Rayleigh optical thickness of the band; the is the spectral response function; the is the average extraterrestrial solar irradiance; the is the Rayleigh optical thickness of the band.

[0059] (2) Divide the Rayleigh atmosphere based on the physical properties of the Rayleigh atmosphere to obtain different altitude layers, and determine the relevant parameters of each altitude layer based on the optical characteristics, radiation transfer principle, and climate environment factors of the altitude layer.

[0060] Specifically, the physical properties of the Rayleigh atmosphere are different at different altitudes. In this step, the Rayleigh atmosphere is divided based on the physical properties of the Rayleigh atmosphere at different altitudes to obtain different altitude layers. For each altitude layer, determine the relevant parameters (including the observation zenith angle, solar zenith angle, wind speed, etc.) according to the chemical characteristics, radiation transfer principle, and climate environment factors.

[0061] As an optional embodiment, the dividing the Rayleigh atmosphere based on the physical properties of the Rayleigh atmosphere to obtain different altitude layers specifically includes: determining a plurality of altitude thresholds according to the atmosphere mechanism, performing a first division on the Rayleigh atmosphere according to the plurality of altitude thresholds to obtain a first altitude division result; determining the boundary positions of each altitude layer in the first altitude division result as the positions to be optimized; obtaining the climate environment factors of each position point of the positions to be optimized, and establishing a distribution map of the environmental factors in the boundary region; determining the radiation transfer influence margin of each point at the boundary position based on the distribution map of the environmental factors, and adjusting the altitude of the corresponding point based on the influence margin to complete the optimization of the first altitude division result and obtain different altitude layers.

[0062] Specifically, the influence margin refers to the degree of influence of environmental factors (such as temperature, humidity, etc.) on radiation propagation.

[0063] In specific implementation, first, based on the physical characteristics of the Rayleigh atmosphere (such as the variation laws of atmospheric pressure, temperature, gas density, etc. with height), an appropriate height threshold is determined through known atmospheric data (such as the International Standard Atmosphere Model) or through experimental data. Each height threshold represents an important physical change point, such as the temperature inversion layer or the key position of air pressure change. Further, according to the determined multiple height thresholds, the first division is carried out to divide the Rayleigh atmosphere into multiple different height layers. The range of each height layer is defined by the upper and lower height thresholds, and the environmental conditions such as climate, air pressure, and temperature in each layer can be considered relatively uniform within the same layer. In the result of the first division, each height layer has clear upper and lower boundaries. Then, these upper and lower boundaries are determined as the positions to be optimized. For each position to be optimized, the climate environmental factors near this position are obtained, including temperature, air pressure, humidity, wind speed, cloud density, etc. The obtained climate environmental factor data is visualized and plotted into an environmental factor distribution map. The distribution map shows how the climate characteristics change at different positions (i.e., the boundary positions to be optimized) and indicates the potential impact of radiation transmission at each position. The environmental factor distribution map may include a temperature distribution map, an air pressure distribution map, a humidity distribution map, etc. Based on the environmental factor distribution map, radiation transmission simulation is carried out to calculate the radiation transmission impact margin at each position to be optimized. Through a radiation transmission model (such as the Radiation Transfer Equation RTT), the radiation transmission intensity and impact at each position to be optimized are evaluated. A position with a larger impact margin means that the environmental factors at this position have a greater impact on radiation transmission, and it may be necessary to adjust the height layer boundary at this position. According to the calculated radiation transmission impact margin, the height of the position to be optimized is adjusted. If the environmental factors at a certain position have a greater impact on radiation transmission, the height of this position can be considered to be adjusted upward or downward to optimize the radiation transmission effect. For example, if at a certain boundary position, factors such as temperature and humidity have a greater impact on radiation transmission, the height of this position can be adjusted to reduce the interference of radiation transmission. By evaluating the impact margin, the height of each position to be optimized is gradually adjusted.

[0064] The method provided in this embodiment optimizes the division of altitude layers based on the physical properties of the Rayleigh atmosphere. By introducing multiple altitude thresholds and optimizing on this basis, it can significantly improve the understanding and modeling of the Rayleigh atmosphere structure. The initial division was carried out based on the mechanism of the atmosphere. By determining multiple altitude thresholds, the Rayleigh atmosphere was divided for the first time to form a preliminary altitude layer structure. However, the physical properties of the Rayleigh atmosphere are very complex, and a simple preliminary division often cannot fully reflect the subtle differences in radiation transfer, climate change, etc. among different altitude layers. Therefore, this preliminary division may not provide sufficient reference for accurate calculations. By optimizing the boundary positions of each altitude layer, it is possible to better combine the actual climate conditions with the physical properties of the atmosphere, avoiding the possible neglect of environmental factors or overly simple assumptions in traditional methods. Specifically, by analyzing the climate environmental factors at the boundary positions and establishing a distribution map of environmental factors, it is possible to accurately reveal the influence margin of radiation transfer between different altitude layers. The concept of this margin means that in some regions, due to the influence of factors such as climate, humidity, and air pressure, the behavior of radiation transfer may change significantly. The optimized altitude layer division will adjust the altitude of each position according to the distribution of these environmental factors, so as to achieve an accurate description of the radiation transfer characteristics of different altitude layers. First of all, it improves the accuracy of the atmosphere division, making the modeling of the atmosphere more in line with the actual situation, and thus improving the accuracy of the simulation results. For example, in applications such as satellite communication and remote sensing detection, the accuracy of the radiation transfer model directly affects the signal transmission effect and data quality. The refined altitude layer division can effectively reduce the errors caused by inaccurate division and improve the reliability of transmission prediction. Secondly, this optimization method can flexibly cope with different climate conditions in practical applications, enhancing the adaptability of the model. Especially in complex climate environments, it can adapt to sudden meteorological changes by adjusting the boundary positions of each altitude layer in real time, thereby improving the prediction ability of the atmosphere characteristics under extreme weather conditions.

[0065] For example, the Rayleigh atmosphere is divided into 32 altitude layers up to a height of 100 km, increasing by 1 km per layer below 25 km, increasing by 5 km per layer from 25 to 50 km, and two upper layers (50 to 70 km and 70 to 100 km). The optical thickness of each altitude layer is proportional to the pressure curve of the US standard atmosphere model. For the chemical characteristics, radiation transfer principle, and climate environmental factors of each altitude layer, the observed zenith angle (OZA) is set from 0 to 84.2 degrees, with an increment of approximately 0.5 degrees. The solar zenith angle (SZA) is set from 0 degrees to 88 degrees, with an increment of 2 degrees. The wind speed is set from 0 millisecond -1 to 30 millisecond -1 s, with an increment of 5 millisecond -1 .

[0066] (3) Input the Rayleigh optical thickness and related parameters into the ocean-atmosphere coupled radiative transfer model to construct a Rayleigh scattering lookup table.

[0067] Specifically, when implemented, the ocean-atmosphere coupled radiative transfer model expands the vector radiative transfer equation by Fourier series to obtain an azimuth-independent equation; discretizes the solar zenith angle in the azimuth-independent equation, calculates the radiative transfer situation at each solar zenith angle, and obtains a vector radiative transfer matrix equation; uses the doubling method to solve the vector radiative transfer matrix equation to obtain the distribution of vector radiative signals;

[0068] Based on the reflection, refraction characteristics of light at the ocean-atmosphere interface and interference factors, adjust the distribution of the vector radiative signals, and construct a Rayleigh scattering lookup table based on the distribution.

[0069] Optionally, after inputting the Rayleigh optical thickness and related parameters into the ocean-atmosphere coupled radiative transfer model to construct a Rayleigh scattering lookup table, the method further includes: inputting the Rayleigh optical thickness and related parameters into a radiative transfer model based on the Monte Carlo method to construct a target Rayleigh scattering lookup table; comparing the differences between the Rayleigh scattering radiative rates and the actual Rayleigh scattering radiative rates of the Rayleigh scattering lookup table and the target Rayleigh scattering lookup table under different observation data, and adjusting the Rayleigh scattering lookup table based on the differences.

[0070] Specifically, the radiative transfer model based on the Monte Carlo method determines the scattering characteristics and reflection characteristics of photons based on the input Rayleigh optical thickness and related parameters; emits the photons from a detector, and simulates the scattering of the photons in the atmosphere and the reflection on the sea surface based on the scattering characteristics and reflection characteristics of the photons; calculates the related parameters of the scattering of the photons in the atmosphere based on the scattering phase matrix, and when the photons reach the sea surface, calculates the related parameters of the reflection of the photons in the target direction based on the Fresnel formula; the target direction is the same as the solar direction; constructs a target Rayleigh scattering lookup table based on the related parameters of the scattering of the photons in the atmosphere and the related parameters of the reflection of the photons in the target direction.

[0071] Further, calculate the differences between the Rayleigh scattering radiative rates and the actual Rayleigh scattering radiative rates of the Rayleigh scattering lookup table constructed based on the ocean-atmosphere coupled radiative transfer model under different observation data and the differences between the Rayleigh scattering radiative rates and the actual Rayleigh scattering radiative rates of the Rayleigh scattering lookup table constructed based on the Monte Carlo method under different observation data, compare the two differences, select the one with the smaller difference as the Rayleigh scattering lookup table, and adjust the Rayleigh scattering lookup table based on the comparison result (the difference of the Rayleigh scattering lookup table constructed based on the ocean-atmosphere coupled radiative transfer model is smaller).

[0072] Further, the Rayleigh scattering radiance corresponding to the observed data is found by looking up the constructed Rayleigh scattering lookup table, and the Rayleigh-corrected radiance is calculated based on the Rayleigh scattering radiance (subtracting the Rayleigh scattering radiance, the white cap radiance, and the flash radiance from the total radiance). After obtaining the Rayleigh-corrected radiance and the adjusted satellite data of the midday remote sensing reflectance, before matching the two based on the time information and location information, it is necessary to ensure that the change in the optical properties of the water body is minimized. In this case, the satellite data of the midday remote sensing reflectance must be spatially uniform to eliminate the frontal region and stray clouds, and temporally uniform to eliminate pixels with significant inherent changes in optical properties due to algal blooms or strong currents. The spatial uniformity is evaluated by using the coefficient of variation within the pixel block. The coefficient of variation is calculated based on the value of each pixel, the total number of pixels, and the average value, and pixels with a coefficient of variation exceeding 0.15 are excluded. Temporally, the coefficient of variation is calculated for a specified number of observed data, and pixels with a coefficient of variation exceeding 0.15 are excluded. After adjusting the spatial and temporal consistency of the satellite data of the midday remote sensing reflectance, the adjusted satellite data of the midday remote sensing reflectance is matched with the Rayleigh-corrected radiance according to the time information (the time information of the two differs by a preset time), and the coefficient of variation is calculated for the matched data, and data with a coefficient of variation exceeding 0.15 are excluded to obtain a training set (the input of the training set is the Rayleigh-corrected radiance, and the output of the training set is the satellite data of the midday remote sensing reflectance).

[0073] For example, if the preset time is 3 hours, in this step, the Rayleigh-corrected radiance at 07:15 is corresponding to the satellite data of the midday remote sensing reflectance at 10:15, the Rayleigh-corrected radiance at 08:15 is corresponding to the satellite data of the midday remote sensing reflectance at 11:55, and the Rayleigh-corrected radiance at 16:15 is corresponding to the satellite data of the midday remote sensing reflectance at 13:15.

[0074] S104. Train the ultraviolet-near infrared atmospheric correction combined model based on the training set, input the Rayleigh-corrected radiance corresponding to the radiation signal of the coastal water area to be corrected into the trained ultraviolet-near infrared atmospheric correction combined model, and determine the water-leaving radiance of the radiation signal based on the output satellite data of the midday remote sensing reflectance.

[0075] In specific implementation, the training set is divided into a training data set and a test data set according to a preset ratio. For example, 70% of the training set is used as the training data set to train the ultraviolet-near-infrared atmospheric correction combined model, and 30% of the training set is used as the test data set to verify the performance of the trained ultraviolet-near-infrared atmospheric correction combined model. In addition, during the training process, k-fold cross-validation is used to prevent the trained model from overfitting. After obtaining the trained ultraviolet-near-infrared atmospheric correction combined model, based on the radiance signal of the coastal water area to be corrected, its corresponding Rayleigh scattering radiance and total radiance are determined, and then the corresponding Rayleigh correction radiance is determined. The Rayleigh correction radiance corresponding to the radiance signal of the coastal water area to be corrected is input into the trained ultraviolet-near-infrared atmospheric correction combined model, and the model outputs the corresponding satellite data of the noon remote sensing reflectance, that is, the water-leaving radiance of the radiance signal of the coastal water area to be corrected.

[0076] In the method provided by this embodiment, on the first hand, by combining the near-infrared atmospheric correction model and the ultraviolet atmospheric correction model, the observation data of coastal waters are processed based on the ultraviolet-near-infrared atmospheric correction combined model to generate noon remote sensing reflectance satellite data. Since coastal waters include both clear waters and turbid waters, this approach not only takes advantage of the fact that the near-infrared atmospheric correction model better considers the influence of suspended solids and other particulate matters in the water body on the radiation signal when dealing with turbid waters, improving the accuracy of correction, but also takes advantage of the fact that the ultraviolet atmospheric correction model is suitable for the correction of clear waters and can effectively process the optical properties of the water body to ensure accurate reflectance under relatively low light absorption conditions. By combining the two models, various water body conditions can be handled, the accuracy of noon remote sensing reflectance satellite data can be improved, and the adaptability is strong. The most suitable model can be selected for processing according to the characteristics of the actual observation data, thereby obtaining more reliable noon remote sensing reflectance satellite data. Moreover, this ultraviolet-near-infrared atmospheric correction combined model can be applied to different types of waters, not limited to specific clear or turbid waters, making it more widely applicable. On the second hand, according to the spatial resolution and temporal resolution of the noon remote sensing reflectance satellite data, the flow velocity of each pixel is calculated, and the pixels with flow velocity exceeding the flow velocity threshold are removed. This can effectively filter out high-flow pixels, ensuring that within the temporal resolution of the noon remote sensing reflectance satellite data, the pixels with slower flow velocities remain in the original pixel blocks, thus avoiding the rapid changes in the water body caused by strong water currents. By removing the pixels with faster flow velocities, the influence of instantaneous changes caused by water flow on the analysis of observation data is reduced, thereby improving the stability of the data. Pixels with slower flow velocities usually represent a more stable state of the water body, which helps to more accurately reflect the optical properties and other relevant information of the water body. Moreover, the flow velocity threshold can be flexibly adjusted according to different water body characteristics to adapt to various environmental conditions and application requirements. On the third hand, when constructing the Rayleigh scattering look-up table, both the ocean-atmosphere coupled radiation transfer model and the radiation transfer model based on the Monte Carlo method are used. First, an initial Rayleigh scattering look-up table is constructed based on the ocean-atmosphere coupled radiation transfer model, and then a target Rayleigh scattering look-up table for cross-validation is constructed based on the radiation transfer model of the Monte Carlo method. The initial Rayleigh scattering look-up table is continuously optimized through the target Rayleigh scattering look-up table, so that the finally obtained Rayleigh scattering look-up table can more clearly reflect the relationship between the Rayleigh scattering radiance and the observation data. When determining the Rayleigh correction radiance based on the Rayleigh scattering radiance and constructing a training set to train the ultraviolet-near-infrared atmospheric correction combined model subsequently, the performance of the ultraviolet-near-infrared atmospheric correction combined model is improved, and the accuracy of the atmospheric correction of the radiation signal in coastal waters is enhanced.

[0077] Corresponding to the embodiment of the atmospheric correction method for the radiation signal in the coastal waters at a previous solar zenith angle, the present application also provides an embodiment of an atmospheric correction device for the radiation signal in the coastal waters at a solar zenith angle.

[0078] Figure 2 It is a schematic structural diagram of the first embodiment of the atmospheric correction device for the radiation signal in the coastal waters at a solar zenith angle provided by the present application. Please refer to Figure 2 The device provided in this embodiment includes a generation module 210, an adjustment module 220, a matching module 230, and a determination module 240; among them,

[0079] The generation module 210 is used to generate noon remote sensing reflectance satellite data based on the observation data of the coastal waters by using the ultraviolet-near infrared atmospheric correction combined model; the noon remote sensing reflectance satellite data represents the water-leaving radiance of the radiation signal in the coastal waters;

[0080] The adjustment module 220 is used to delete the pixels with a flow velocity exceeding the flow velocity threshold in the noon remote sensing reflectance satellite data based on the spatial resolution and time resolution of the noon remote sensing reflectance satellite data by using the maximum cross-correlation method, and obtain the adjusted noon remote sensing reflectance satellite data;

[0081] The matching module 230 is used to determine the Rayleigh scattering radiance based on the Rayleigh scattering look-up table corresponding to the observation data, calculate the Rayleigh correction radiance based on the Rayleigh scattering radiance, and match the Rayleigh correction radiance and the adjusted noon remote sensing reflectance satellite data based on the time information and position information to obtain a training set;

[0082] The determination module 240 is used to train the ultraviolet-near infrared atmospheric correction combined model based on the training set, input the Rayleigh correction radiance corresponding to the radiation signal of the coastal waters to be corrected into the trained ultraviolet-near infrared atmospheric correction combined model, and determine the water-leaving radiance of the radiation signal based on the output noon remote sensing reflectance satellite data.

[0083] The device of this embodiment can be used to execute Figure 1 the steps of the method embodiment shown, and the specific implementation principle and process are similar, which will not be elaborated here.

[0084] For the specific implementation process of the functions and roles of each unit in the above device, please refer to the implementation process of the corresponding steps in the above method, which will not be elaborated here.

[0085] For the apparatus embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the descriptions of the method embodiments. The apparatus embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0086] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A method for atmospheric correction of coastal waters radiation signals under solar zenith angle, characterized in that: The method comprises: Using an ultraviolet-near infrared atmospheric correction combined model, noon remote sensing reflectance satellite data are generated based on the observation data of coastal waters; the noon remote sensing reflectance satellite data represent the water-leaving radiance of the radiation signal in the coastal waters; Using a maximum cross-correlation method, based on the spatial resolution and temporal resolution of the noon remote sensing reflectivity satellite data, pixels in the noon remote sensing reflectivity satellite data with a flow velocity exceeding a flow velocity threshold are deleted to obtain adjusted noon remote sensing reflectivity satellite data; Determine Rayleigh scattering radiance based on the Rayleigh scattering lookup table corresponding to the observation data, calculate Rayleigh corrected radiance based on the Rayleigh scattering radiance, and match the Rayleigh corrected radiance with the adjusted noon remote sensing reflectance satellite data based on time information and location information to obtain a training set; The ultraviolet-near infrared atmospheric correction combined model is trained based on the training set, the Rayleigh-corrected radiance corresponding to the coastal waters radiation signal to be corrected is input into the trained ultraviolet-near infrared atmospheric correction combined model, and the water-leaving radiance of the radiation signal is determined based on the output noon remote sensing reflectance satellite data; The ultraviolet-near infrared atmospheric correction combined model includes a near infrared atmospheric correction model and an ultraviolet atmospheric correction model; the method of using the ultraviolet-near infrared atmospheric correction combined model to generate noon remote sensing reflectivity satellite data based on observation data of coastal waters includes: The near-infrared atmospheric correction model is used to calculate the noon remote sensing reflectance satellite data of the first band based on the observation data of the first band, and the noon remote sensing reflectance satellite data of the second band is obtained based on the noon remote sensing reflectance satellite data of the first band; the wavelength of the second band is greater than that of the first band; and the proportion of clear water in the noon remote sensing reflectance satellite data of the second band is greater than that of turbid water; The ultraviolet atmosphere correction model is used to calculate the noon remote sensing reflectance satellite data of the third band; the wavelength of the third band is shorter than that of the first band; the proportion of turbid water areas in the noon remote sensing reflectance satellite data of the third band is greater than that of clear water areas; The first band, the second band, and the third band are fitted by a logistic regression method to determine the weight of the first band, the weight of the second band, and the weight of the third band. The noon remote sensing reflectivity satellite data are weighted fused based on the weights to obtain the noon remote sensing reflectivity satellite data.

2. The method according to claim 1, characterized in that The ultraviolet-near infrared atmospheric correction combined model comprises an input layer, a first hidden layer, a second hidden layer and an output layer connected in sequence; the method of using the ultraviolet-near infrared atmospheric correction combined model to generate noon remote sensing reflectivity satellite data based on the observation data of coastal waters comprises: The input layer receives the observation data, preprocesses the observation data, and passes the preprocessed observation data to the first hidden layer; the preprocessed observation data includes observation angle information and Rayleigh-corrected radiances of multiple bands; Each neuron in the first hidden layer performs weighted summation on the preprocessed observation data to extract features of the preprocessed observation data, and each neuron outputs first features of different scales, which are transmitted to the second hidden layer, wherein the first features represent the correlation between the observation angle information and the Rayleigh-corrected radiance; Each neuron in the second hidden layer performs weighted summation and transformation on the first features of different scales output by the first hidden layer, and each neuron in the second hidden layer outputs a second feature, and the scales of the second features output by each neuron are different, and the second feature is transmitted to the output layer; the second feature represents the correlation between the observation angle information and the Rayleigh-corrected radiance, and the data accuracy of the second feature is higher than that of the first feature; The output layer performs weighted summation on each second feature, and obtains the noon remote sensing reflectance satellite data output by each neuron through nonlinear transformation; the number of the neurons is the same as the number of the bands.

3. The method according to claim 1, characterized in that The method of using the maximum cross correlation method to delete pixels in the noon remote sensing reflectivity satellite data whose flow velocity exceeds a flow velocity threshold based on the spatial resolution and temporal resolution of the noon remote sensing reflectivity satellite data includes: Determining a time interval based on the temporal resolution of the noon remote sensing reflectivity satellite data; Spatial alignment of noon remote sensing reflectance satellite data at adjacent time intervals; Calculate the cross-correlation corresponding to the pixels at the same position in the noon remote sensing reflectivity satellite data of the adjacent time intervals, and determine the actual distance based on the pixel displacement corresponding to the maximum value and the spatial resolution of the noon remote sensing reflectivity satellite data; The flow velocity of each pixel is determined based on the actual distance and the time interval.

4. The method according to claim 1, characterized in that: The method further comprises: Determining the sensor type based on the characteristics of the coastal waters and the operating band and characteristics of the sensor; Determining the range of spatial resolution based on the spatial resolution of each working band of the sensor and external observation conditions; The spatial resolution is verified based on the resolution of different features in the observed data, and the range of the spatial resolution is adjusted according to the verification result.

5. The method according to claim 1, characterized in that The process of determining the Rayleigh scattering lookup table includes: Based on the observation data, calculating the Rayleigh optical thickness of each band in the observation data; The Rayleigh atmosphere is divided based on its physical properties to obtain different altitude layers, and the relevant parameters of each altitude layer are determined based on the optical properties, radiation transmission principles and climate and environmental factors of the altitude layer; the relevant parameters include the observation zenith angle, the solar zenith angle and the wind speed; The Rayleigh optical thickness and related parameters are input into an ocean-atmosphere coupled radiation transfer model to construct a Rayleigh scattering lookup table.

6. The method according to claim 5, characterized in that After inputting the Rayleigh optical thickness and related parameters into the ocean-atmosphere coupled radiation transfer model to construct a Rayleigh scattering lookup table, the method further includes: Inputting the Rayleigh optical thickness and related parameters into a radiation transfer model based on the Monte Carlo method to construct a target Rayleigh scattering lookup table; The Rayleigh scattering lookup table and the target Rayleigh scattering lookup table are compared to determine the difference between the Rayleigh scattering radiance under different observation data and the actual Rayleigh scattering radiance, and the Rayleigh scattering lookup table is adjusted based on the difference.

7. The method according to claim 5, characterized in that The step of inputting the Rayleigh optical thickness and related parameters into an ocean-atmosphere coupled radiation transfer model to construct a Rayleigh scattering lookup table comprises: The ocean-atmosphere coupled radiative transfer model obtains the azimuth-independent equation by expanding the vector radiative transfer equation through Fourier series; Discretizing the solar zenith angle in the azimuth-independent equation, calculating the radiation transfer at each solar zenith angle, and obtaining a vector radiation transfer matrix equation; Solving the vector radiation transfer matrix equation using the doubling method to obtain the distribution of the vector radiation signal; Based on the reflection and refraction characteristics of light on the ocean-atmosphere interface and interference factors, the distribution of the vector radiation signal is adjusted, and a Rayleigh scattering lookup table is constructed based on the distribution.

8. The method according to claim 6, characterized in that The step of inputting the Rayleigh optical thickness and related parameters into a radiation transfer model based on a Monte Carlo method to construct a target Rayleigh scattering lookup table includes: The radiative transfer model based on the Monte Carlo method determines the scattering and reflection characteristics of photons based on the input Rayleigh optical thickness and related parameters; The photons are emitted from the detector, and the scattering of the photons in the atmosphere and the reflection of the photons on the sea surface are simulated based on the scattering characteristics and reflection characteristics of the photons; Calculating relevant parameters of the photon scattering in the atmosphere based on the scattering phase matrix, and calculating relevant parameters of the photon reflection in the target direction based on the Fresnel formula when the photon reaches the sea surface; the target direction is consistent with the direction of the sun; A target Rayleigh scattering lookup table is constructed based on the relevant parameters of the photon scattering in the atmosphere and the relevant parameters of the photon reflection in the target direction.

9. An atmospheric correction device for the radiation signal of coastal waters under the solar zenith angle, characterized in that: The device includes a generating module, an adjusting module, a matching module and a determining module; wherein, The generating module is used to generate noon remote sensing reflectance satellite data based on the observation data of the coastal waters by using the ultraviolet-near infrared atmospheric correction combined model; the noon remote sensing reflectance satellite data represents the water-leaving radiance of the radiation signal in the coastal waters; The adjustment module is used to delete pixels in the noon remote sensing reflectivity satellite data whose flow velocity exceeds a flow velocity threshold value based on the spatial resolution and temporal resolution of the noon remote sensing reflectivity satellite data by using a maximum cross correlation method to obtain adjusted noon remote sensing reflectivity satellite data; The matching module is used to determine the Rayleigh scattering radiance based on the Rayleigh scattering lookup table corresponding to the observation data, calculate the Rayleigh corrected radiance based on the Rayleigh scattering radiance, and match the Rayleigh corrected radiance with the adjusted noon remote sensing reflectance satellite data based on time information and location information to obtain a training set; The determination module is used to train the ultraviolet-near infrared atmospheric correction combined model based on the training set, input the Rayleigh-corrected radiance corresponding to the coastal waters radiation signal to be corrected into the trained ultraviolet-near infrared atmospheric correction combined model, and determine the water-leaving radiance of the radiation signal based on the output noon remote sensing reflectance satellite data; The ultraviolet-near infrared atmospheric correction combined model includes a near infrared atmospheric correction model and an ultraviolet atmospheric correction model; the method of using the ultraviolet-near infrared atmospheric correction combined model to generate noon remote sensing reflectivity satellite data based on observation data of coastal waters includes: The near-infrared atmospheric correction model is used to calculate the noon remote sensing reflectance satellite data of the first band based on the observation data of the first band, and the noon remote sensing reflectance satellite data of the second band is obtained based on the noon remote sensing reflectance satellite data of the first band; the wavelength of the second band is greater than that of the first band; and the proportion of clear water in the noon remote sensing reflectance satellite data of the second band is greater than that of turbid water; The ultraviolet atmosphere correction model is used to calculate the noon remote sensing reflectance satellite data of the third band; the wavelength of the third band is shorter than that of the first band; the proportion of turbid water areas in the noon remote sensing reflectance satellite data of the third band is greater than that of clear water areas; The first band, the second band, and the third band are fitted by a logistic regression method to determine the weight of the first band, the weight of the second band, and the weight of the third band. The noon remote sensing reflectivity satellite data are weighted fused based on the weights to obtain the noon remote sensing reflectivity satellite data.

Images