A method for coupled radiation simulation of atmosphere and three-dimensional surface based on photon tracing

Abstract

The present invention discloses a method for simulating the coupled radiation of the atmosphere and three-dimensional surface based on photon tracing, comprising the following steps: (1) reading in the input parameters of the light source, atmospheric scene, and surface scene, and performing calculation and processing on the atmospheric input parameters; (2) emitting a photon from the light source into the scene, simulating the traveling distance of the photon according to the calculation result of step (1), and determining the position of the collision point; (3) updating the position, direction, and energy information of the photon at the collision point, and collecting the energy contribution; (4) calculating the next direction of the photon according to the photon position, and continuing to trace the photon until the termination condition is met; (5) emitting a new photon, repeating the tracing process, and after reaching the set number of photons, obtaining the converged radiance result at the top of the atmosphere. This method simulates the coupling of surface and atmospheric radiation transmission by tracing the direction and position information of photons, which is closer to the actual situation, and thus more accurately simulates the influence of the atmosphere on the surface radiation transmission signal.

Description

Technical Field

[0001] The present invention relates to the field of quantitative remote sensing simulation, and particularly to a method for simulating the coupled radiation of the atmosphere and three-dimensional surface based on photon tracing. It is of great significance in aspects such as quantitatively analyzing the influence of atmospheric coupling on remote sensing signals and researching surface vegetation ecology, etc. Background Art

[0002] Remote sensing measurement is an important information acquisition tool for exploring the state of ground objects by remotely receiving electromagnetic radiation information from ground objects through sensors. Obtaining characteristic information such as the directional reflectivity of surface objects can obtain rich information on object structures and categories. Establishing a three-dimensional surface scene model can directly simulate and calculate the reflection spectrum of the scene by modifying the input scene state parameters, which is of great significance for quantitatively studying the influence of the three-dimensional structure of the surface scene on remote sensing signals.

[0003] For remote sensing observations, the influence of the Earth's atmosphere on signals is a key issue that cannot be ignored. The reflectivity characteristics of the surface scene at a specific observation angle depend to a certain extent on the current atmospheric conditions. The current main method is to use the remote sensing signal at the top of the atmosphere after atmospheric correction to obtain the spectral signal at the height of the top of the surface scene. However, this requires making various approximate assumptions about the atmospheric conditions, and these assumptions about the atmospheric conditions will bring large errors in non-typical situations; on the other hand, existing atmospheric coupling models mostly accumulate the contributions of the surface scene and the atmosphere at the height of the bottom of the atmosphere, and then simulate the coupling of the surface scene and the atmosphere by calculating a limited number of interactions, which cannot restore the real physical process, and in most cases, the atmospheric proximity effect is ignored, resulting in limited accuracy of the obtained results.

[0004] Therefore, the most convenient and most promising method for quantitatively studying the relationship between remote sensing signals and atmospheric properties and three-dimensional surface scenes is to establish a model that can truly restore the coupled radiation transmission of the three-dimensional surface scene and the atmosphere, which is of great significance and application value for better understanding and analyzing the radiation transmission process of the three-dimensional surface scene and the design and development of actual sensor indicators, etc.

[0005] It should be noted that this part aims to provide background or context for the present invention stated in the claims. The description herein is not admitted to be prior art merely because it is included in this part. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for simulating the coupled radiation of the atmosphere and three-dimensional surface based on photon tracing for the problem of many approximations and inaccurate simulation results in the current simulation process of the radiation transmission of a three-dimensional surface scene considering the atmosphere.

[0007] The technical solution of the present invention is as follows: Through the Monte Carlo idea, the atmospheric part and the surface part are parametrically described. During the simulation process, the direction, position, and energy information of each photon are tracked, and its position information is used to determine whether the photon is currently in the atmosphere or in the three-dimensional ground scene. Correspondingly, the optical parameters of the atmosphere or the optical and structural parameters of the surface scene are used to calculate the travel distance and direction of the photon. At the position where each photon collides with the elements in the scene, the energy contribution in the observation direction of the sensor is collected and accumulated. Through the simulation process of a certain number of photons, the convergent radiance at the top of the atmosphere is obtained. The specific steps are as follows:

[0008] (1) Read in the input parameters of the light source, atmospheric scene, and surface scene, and calculate and process the atmospheric input parameters;

[0009] (2) Emit a photon from the light source into the scene, simulate the travel distance of the photon according to the calculation results of step (1), and determine the collision point position;

[0010] (3) Update the photon position, direction, and energy information at the collision point, and collect the energy contribution of the photon to the sensor;

[0011] (4) According to the photon position, calculate the next direction of the photon, and continue to track the photon until the termination condition is reached;

[0012] (5) Emit a new photon from the light source, repeat the tracking process, and after reaching the set number of photons, obtain the convergent radiance result at the top of the atmosphere.

[0013] Among them, in step (1), the three-dimensional structure parameters, spectral parameters of the light source and surface scene, and spectral parameters of the atmosphere are read in, and the atmospheric input parameters are calculated and processed:

[0014] The three-dimensional surface scene is described by the method of combining geometric patches. The size, position, and orientation information of each triangle or circle are described using three-dimensional coordinates. At the same time, the hyperspectral reflectivity and transmittance information of each patch are input. The splicing of geometric patches can describe three-dimensional surface scenes of any complexity.

[0015] Due to the vertical heterogeneity of the Earth's atmosphere, the atmosphere above the three-dimensional surface scene is modeled as 12 height layers with unequal intervals. The atmosphere within each layer is horizontally uniform and has corresponding optical parameters. For the atmosphere at each height, it is classified into atmospheric molecules, aerosols, and clouds according to type. The extinction coefficients and single-scattering albedos of atmospheric molecules, aerosols, and clouds at each layer in each simulation band are input respectively. The double Henyey-Greenstein asymmetry factor parameters of aerosols and clouds are correspondingly input to describe their scattering phase functions.

[0016] The sun is the incident light source for the coupled scenario of the atmosphere and the three-dimensional surface. The input geometric parameters include the solar zenith angle and azimuth angle. The light source is set at the top of the atmosphere, and its size is the same as the boundary size of the underlying three-dimensional surface scene. The input spectral parameter is the spectral energy distribution of the incident sunlight.

[0017] Preprocess the input atmospheric optical parameters. Generate a lookup table for the scattering phase function of atmospheric molecules using Rayleigh scattering; calculate the lookup tables for the scattering phase functions of clouds and aerosols using double Henyey-Greenstein parameters for different bands and different altitude layers; adjust the actual height of each atmospheric layer according to the atmospheric elevation; synthesize the absorption coefficients, single-scattering albedos, and scattering phase function lookup tables of different atmospheric components at each height position for each band, and store the results separately as calculation parameters directly used in the subsequent simulation process.

[0018] Among them, in step (2), a photon is emitted from the light source into the scene. According to the calculation results of step (1), simulate the distance traveled by the photon and determine the collision point position: According to the set light source parameters, determine the starting position, direction, and energy of the photon entering the scene from the top of the atmosphere.

[0019] The light source surface is set at the top of the atmosphere. Use uniformly distributed random numbers to randomly determine the starting position of each photon on the light source surface. The direction of the photon is determined by the input solar zenith angle and azimuth angle, and the initial energy of each photon is determined by the input spectral parameters:

[0020]

[0021] In the formula, N is the number of photons to be traced, S represents the bottom area of the simulation scene, and λ is the wavelength of the current simulation; after the photon enters the scene, it will first enter the atmosphere module and propagate along a straight line in the set initial direction. The calculation method of the distance traveled by the light in the scene will be determined according to the position of the photon.

[0022] According to the position of the photon, determine whether it is inside the atmosphere or inside the three-dimensional surface scene. If the photon is in the atmosphere, find the equivalent extinction coefficient of the atmosphere at the current position by altitude layer and calculate the free path. The free path S traveled by the photon in the atmosphere is jointly determined by a random number and the equivalent extinction coefficient of the layer it is in:

[0023]

[0024] rand is a uniformly distributed random number between 0 and 1, β atm is the equivalent extinction coefficient of the atmosphere at the current position calculated in step (1), iz represents the atmospheric layer where the photon is currently located, λ is the wavelength of the current simulation; after traveling the free path S, the photon collides with the atmospheric components.

[0025] If the photon is in the surface scene, simulate the propagation of the photon in the current direction until it collides with a geometric patch. Perform geometric intersection based on the three-dimensional coordinates of the triangle or disc, and calculate the triangle or disc number where the light ray collides and the position of the collision point.

[0026] Among them, in step (3), update the photon position, direction, and energy information at the collision point, and collect the energy contribution of the photon to the sensor: at the collision point, the energy carried by the photon will be attenuated correspondingly at the collision point:

[0027] Q out (λ) = Q in (λ)·ω(λ)

[0028] Among them, Q out and Q in represent the photon energy after and before the collision respectively. ω is the single-scattering albedo, which is the equivalent atmospheric albedo at the current position in the atmosphere, and in the surface scene, it is the sum of the reflectivity and transmittance of the current collision triangle or disc.

[0029] The new scattering direction distribution of the photon after collision in the atmosphere satisfies the distribution of the scattering phase function. Use the scattering phase function lookup table calculated in step (1) at the current position to extract and calculate the new propagation direction; the scattering of the photon on the surface of the surface scene patch is assumed to satisfy the Lambert's law, and calculate the new scattering direction according to the patch orientation information.

[0030] At the position where each collision occurs, use the photon diffusion method to calculate the energy contribution of the cumulative collision event to the set observation sensor; if there is no occlusion on the path from the collision point position to the sensor, calculate the probability of the light ray scattering in the direction of the sensor according to the photon incident direction and the position information of the collision point, and collect the energy proportionally:

[0031] Q collect (Ω,λ) = Q out (λ)·P(Ω,λ)·exp(-L a ·β atm )

[0032] P is the probability of the light ray scattering in the direction of the sensor, and L a represents the length of the atmospheric path from the collision point to the sensor position. The last exponential term in the formula describes the atmospheric extinction over this distance.

[0033] Among them, in step (4), according to the photon position, calculate the next direction of the photon and continue to track the photon until the termination condition is reached: the photon will leave the collision point in a new direction, continue to propagate along a straight line and repeat the collision process. During the entire tracking process, the photon carries position, direction, and energy information. Use the position information of the photon to determine whether it is in the atmosphere or the surface scene at present. When the photon is in the atmosphere, in the subsequent propagation process, use the atmospheric phase function lookup table to randomly extract the new direction of the photon, use the path length calculation method and collision processing method in the atmosphere to continue tracking the photon, update the information carried by the photon, and collect the sensor contribution; when the photon is in the surface scene, randomly extract the new direction of the photon according to the normal direction of the patch where the current collision occurs and the Lambert's law, use the path length calculation method and collision processing method in the surface scene to continue tracking the photon, update the information carried by the photon, and collect the sensor contribution.

[0034] After a certain number of collisions, the energy carried by the photon will become relatively small. Once the photon energy is less than the threshold, use "Russian roulette" to terminate the tracking process of the photon in an energy-conserving and unbiased manner.

[0035] When the photon enters the surface scene from the bottom of the atmosphere, it has specific direction and position information, which can well simulate the anisotropy of the sky diffuse light after atmospheric scattering. Once the photon scatters and leaves the upper boundary of the surface scene through collisions in the surface scene, using the height information and specific propagation direction of the photon, the photon can be made to return to the atmosphere to continue tracking, well reproducing the cyclic interaction of the photon between the two parts at the junction of the atmosphere and the surface scene, achieving a complete coupling of the atmosphere and the surface scene below the atmosphere.

[0036] Among them, in step (5), emit new photons and repeat the tracking process. After reaching the set number of photons, obtain the converged radiance result at the top of the atmosphere: Using the Monte Carlo idea, in the simulation process of a single photon, describe each uncertain physical process as a probability process according to its physical distribution, and use the method of random number sampling to determine the specific events that occur in this simulation. Each position described by a random method strictly follows its physical distribution law.

[0037] Through a large number of repeated simulations (i.e., using a certain number of simulated photon numbers), each position described by a random event is fully sampled according to its distribution. Finally, the radiance signal received in the observation direction of the sensor at the top of the atmosphere can be obtained by accumulating the energy contributions to the observation sensor each time:

[0038]

[0039] The meanings of the symbols in the above formula are the same as those in the previous text.

[0040] The advantages of the present invention compared with the prior art are as follows: Based on physical principles, the transmission process of photons in a scene containing the atmosphere is restored, and the cyclic interaction process of photons at the junction of the atmosphere and the surface scene is truly reproduced, achieving the coupling of the radiation processes of the atmosphere and the surface scene, reducing the assumptions and approximations in describing this process, and thus reducing the simulation error of the top-of-atmosphere radiance.

[0041] It has the following advantages: (1) Through computer simulation, within a limited calculation time, remote sensing quantities of interest such as the directional reflectance factor at different heights, arbitrary viewing angles, selectable wavelength ranges, and different atmospheric and surface parameter settings can be calculated. It has diverse functions, and the modifiable parameters are rich and representative. (2) The three-dimensional ground object scene with any degree of complexity can be described by using the method of three-dimensional coordinate geometric patches, and the construction of the three-dimensional scene and the radiation simulation process are independent of each other. The former can introduce the latest development results of computer graphics and has a broad application scenario. (3) The model established using this method is a computer simulation model, which has higher accuracy than the modeling methods of other common remote sensing analytical models and geometric optical models. At the same time, a variety of acceleration algorithms are introduced in the simulation process, and the calculation time is effectively controlled. With the improvement of computer performance, the calculation efficiency will be further improved. Brief Description of the Drawings

[0042] Figure 1 It is a flowchart of a method for coupled radiation simulation of the atmosphere and three-dimensional surface based on photon tracing according to the present invention. Detailed Embodiments

[0043] To better illustrate the method for coupled radiation simulation of three-dimensional surface and atmosphere based on photon tracing involved in the present invention, radiation transfer simulation is carried out using hyperspectral three-dimensional vegetation canopy data and atmospheric data in the visible to short-wave infrared wavelength range, and the top-of-atmosphere reflectance in the vertical viewing direction is calculated. The specific implementation steps are as follows:

[0044] (1) Read in the input parameters of the light source, atmospheric scene, and surface scene, and perform calculation and processing on the atmospheric input parameters: The input parameters of the three-dimensional surface scene include its structural parameters and spectral parameters. Read in the structural parameter file and the reflectance and transmittance spectral files of the corn canopy scene at the jointing stage (described by splicing triangular patches of different sizes). Read in the soil structure parameter file and the reflectance spectrum file describing the soil.

[0045] Read in the atmospheric spectral parameters in the corresponding wavelength range, use the standard mid-latitude summer atmospheric type as the atmospheric input parameter, divide it into 12 stratified heights with a denser lower end and a sparser upper end according to height intervals, and the wavelength range can be 400 - 2500 nm, but not limited to this;

[0046] The sun is the incident light source for the coupled scenario of the atmosphere and the three-dimensional surface. The incident solar zenith angle is set to 30 degrees and the azimuth angle is set to 0 degrees. The light source is set at the top of the atmosphere, and its size is the same as the boundary size of the three-dimensional surface scene below. The input spectral parameter is the spectral energy distribution of incident sunlight.

[0047] Preprocess the input atmospheric optical parameters, generate the scattering phase function of atmospheric molecules using Rayleigh scattering; calculate the scattering phase function look-up tables of clouds and aerosols using double Henyey-Greenstein parameters for different bands and different altitude layers; adjust the actual height of each atmospheric layer according to the atmospheric elevation; perform equivalent synthesis on the absorption coefficients, single-scattering albedos, and scattering phase function look-up tables of different atmospheric components at each height position for each band; and store them separately as calculation parameters for direct use in subsequent simulation processes.

[0048] (2) Emit a photon into the scene from the light source, simulate the distance the photon travels according to the calculation results in step (1), and determine the collision point position: The light source surface is set at the top of the atmosphere, and a random position is randomly determined on the light source surface using a uniform random number as the starting point of the photon, and it enters the scene in the direction of 30-degree zenith angle and 0-degree azimuth angle. The initial energy of each photon is determined by the input spectral parameters:

[0049]

[0050] In the formula, N is the number of photons to be traced, S represents the bottom area of the simulation scene, and λ is the wavelength of the current simulation; after the photon enters the scene, it will first enter the atmosphere module and propagate along a straight line in the set initial direction. The calculation method of the distance the light travels in the scene will be determined according to the position of the photon.

[0051] According to the position of the photon, determine whether it is inside the atmosphere or inside the corn canopy. If the photon is in the atmosphere, find the equivalent extinction coefficient of the atmosphere at the current position by altitude layer and calculate the free path.

[0052] The free path S that the photon travels in the atmosphere is jointly determined by a random number and the extinction coefficient of the layer it is in:

[0053]

[0054] rand is a uniform random number between 0 and 1, β atm is the equivalent extinction coefficient of the atmosphere at the current position calculated in step (1), iz represents the atmospheric layer where the photon is currently located, λ is the wavelength of the current simulation; after traveling the free path S, the photon collides with the atmospheric components.

[0055] If the photon is in the surface scene, simulate the propagation of the photon in the current direction until it collides with a geometric patch. Perform geometric intersection based on the three-dimensional coordinates of the triangle or disc to calculate the number of the triangle or disc where the light ray collides and the position of the collision point.

[0056] (3) Update the photon position, direction, and energy information at the collision point, and collect the energy contribution of the photon to the sensor: At the collision point, the energy carried by the photon will be attenuated correspondingly at the collision point:

[0057] Q out (λ) = Q in (λ)·ω(λ)

[0058] where Q out and Q in represent the photon energy after and before the collision respectively. ω is the single-scattering albedo, which is the equivalent atmospheric albedo at the current position in the atmosphere and the sum of the reflectivity and transmissivity of the current collision triangle or disc in the surface scene.

[0059] The new scattering direction distribution of the photon after collision in the atmosphere satisfies the distribution of the scattering phase function. Use the scattering phase function lookup table calculated in step (1) at the current position to extract and calculate the new propagation direction; the scattering of the photon on the surface of the surface scene patch is assumed to satisfy the Lambert's law, and the new scattering direction is calculated according to the patch orientation information.

[0060] At the position where each collision occurs, use the photon diffusion method to calculate the energy contribution of the cumulative collision event to the set observation sensor; if there is no occlusion on the path from the collision point position to the sensor, calculate the probability of the light ray scattering in the direction of the sensor according to the photon incident direction and the position information of the collision point, and collect the energy proportionally:

[0061] Q collect (Ω,λ) = Q out (λ)·P(Ω,λ)·exp(-L a ·β atm )

[0062] P is the probability of the light ray scattering in the direction of the sensor, and L a represents the length of the atmospheric path from the collision point to the sensor position. The last exponential term in the formula describes the atmospheric extinction over this distance.

[0063] (4) Calculate the next direction of the photon based on its position, and continue to track the photon until the termination condition is reached: the photon will leave the collision point in the new direction, continue to propagate along a straight line, and repeat the collision process; during the entire tracking process, the photon carries position, direction, and energy information. Use the position information of the photon to determine whether it is currently in the atmosphere or the surface scene. When the photon is in the atmosphere, randomly extract the new direction of the photon using the atmospheric phase function lookup table during the subsequent propagation process, and use the path length calculation method and collision handling method in the atmosphere to continue tracking the photon, update the information carried by the photon, and collect the sensor contributions; when the photon is in the surface scene, randomly extract the new direction of the photon according to the normal direction of the patch where the current collision occurs and Lambert's law, and use the path length calculation method and collision handling method in the surface scene to continue tracking the photon, update the information carried by the photon, and collect the sensor contributions.

[0064] After a certain number of collisions, the energy carried by the photon will become relatively small. Once the photon energy is less than the threshold, use "Russian roulette" to terminate the tracking process of the photon in an energy-conserving and unbiased manner.

[0065] When the photon enters the surface scene from the bottom of the atmosphere, it has specific direction and position information, which can well simulate the anisotropy of the sky diffuse light after atmospheric scattering. Once the photon scatters and leaves the upper boundary of the surface scene through collisions in the surface scene, using the height information and specific propagation direction of the photon, the photon can be made to return to the atmosphere for continued tracking, well reproducing the cyclic interaction of the photon between the two parts at the junction of the atmosphere and the surface scene, achieving a complete coupling of the atmosphere and the surface scene below the atmosphere.

[0066] (5) Emit new photons and repeat the tracking process. After reaching the set number of photons, obtain the converged radiance result at the top of the atmosphere: Using the Monte Carlo idea, in the simulation process of a single photon, describe each uncertain physical process (the distance traveled by the photon, the direction of scattering after collision) as a probability process according to its physical distribution, and use the method of random number sampling to determine the specific events that occur in this simulation. Each position described using the random method strictly follows its physical distribution law.

[0067] Through the simulation with N = 100,000 photons, each position described using random events is fully sampled according to its distribution. Finally, the final overall three-dimensional coupling result of the surface scene and the atmosphere can be obtained by accumulating the energy contributions to the observation sensor each time, that is, the radiance signal received in the observation direction of the sensor at the top of the atmosphere:

[0068]

[0069] The meanings of the symbols in the above formula are the same as those in the previous text. Without multi-threaded optimization, the computing time on a personal laptop (CPU Intel Core i7-7500U@2.70GHz) is approximately 5.2 minutes.

Claims

1. A method for simulating the coupled radiation of the atmosphere and three-dimensional surface based on photon tracing, characterized in that, It includes the following steps: (1) Read in the input parameters of the light source, atmospheric scene, and surface scene, and perform computational processing on the atmospheric input parameters, including: The three-dimensional surface scene is described using a combination of geometric patches. The size, position, and orientation information of each triangle or circle are described using three-dimensional coordinates. At the same time, the hyperspectral reflectance and transmittance information of each patch are input. The splicing of geometric patches can describe three-dimensional surface scenes of any complexity; Perform computational processing on the atmospheric input parameters. Due to the vertical heterogeneity of the Earth's atmosphere, the atmosphere above the three-dimensional surface scene is modeled as 12 altitude layers with unequal intervals. The atmosphere within each layer is horizontally uniform and has corresponding optical parameters. For the atmosphere at each altitude, it is classified into atmospheric molecules, aerosols, and clouds according to type. The extinction coefficients and single-scattering albedos of atmospheric molecules, aerosols, and clouds at each layer for each simulated band are input respectively. The double Henyey-Greenstein asymmetry parameters of aerosols and clouds are correspondingly input to describe their scattering phase functions; The sun is the incident light source for the coupled atmosphere and three-dimensional surface scene. The input geometric parameters include the solar zenith angle and azimuth angle. The light source is set at the top of the atmosphere, and its size is the same as the boundary size of the underlying three-dimensional surface scene. The input spectral parameter is the spectral energy distribution of the incident sunlight; Preprocess the input atmospheric optical parameters; Generate a lookup table for the scattering phase function of atmospheric molecules using Rayleigh scattering; Calculate the lookup tables for the scattering phase functions of clouds and aerosols using the double Henyey-Greenstein parameters for different bands and different altitude layers; Adjust the actual layer height of each atmospheric layer according to the atmospheric elevation; Perform equivalent synthesis on the absorption coefficients, single-scattering albedos, and scattering phase function lookup tables of different atmospheric components at each height position for each band; and store them separately as calculation parameters directly used in the subsequent simulation process; (2) Emit a photon from the light source into the scene, simulate the distance traveled by the photon according to the calculation results of step (1), and determine the collision point position; (3) Update the photon position, direction, and energy information at the collision point, and collect the energy contribution of the photon to the sensor; (4) According to the photon position, use the corresponding method to calculate the next direction of the photon, and continue to track the photon until the termination condition is reached; (5) Emit new photons from the light source, repeat the tracking process, and after reaching the required number of photons, obtain the converged radiance result at the top of the atmosphere.

2. The method for simulating the coupled radiation of the atmosphere and the three-dimensional surface based on photon tracing according to claim 1, wherein Among them, in step (2), a photon is emitted from the light source into the scene, and the distance traveled by the photon is simulated according to the calculation results of step (1), and the collision point position is determined: The light source surface is set at the top of the atmosphere. Each photon's starting position is randomly determined on the light source surface using uniformly distributed random numbers. The direction of the photon is determined by the input solar zenith angle and azimuth angle, and the initial energy of each photon is determined by the input spectral parameter: In the formula, N is the number of photons to be traced, S represents the bottom area of the simulation scene, and λ is the wavelength of the current simulation; After photons enter the scene, they will first enter the atmosphere module and propagate in a straight line along the set initial direction. The calculation method for the distance that the light travels in the scene will be determined according to the position of the photons. During the subsequent propagation process, if the photons are in the atmosphere, the extinction coefficient at the current position will be found by stratifying according to height, and the free path will be calculated. The free path d that the photons travel in the atmosphere is jointly determined by a random number and the extinction coefficient of the layer where they are located: rand is a uniform random number between 0 and 1, and β atm is the atmospheric equivalent extinction coefficient at the current position calculated in step (1), iz represents the atmospheric layer where the photon is currently located, and λ is the wavelength of the current simulation; after traveling a free path d, the photon collides with the atmospheric components; If the photons are in the surface scene, it is simulated that the photons propagate in the current direction until they collide with a geometric patch. Geometric intersection is performed based on the three-dimensional coordinates of the triangle or disc, and the number of the triangle or disc where the light collides and the position of the collision point are calculated.

3. A method for simulating the coupled radiation of the atmosphere and three-dimensional surface based on photon tracing according to claim 2, characterized in that In step (3), the position, direction, and energy information of the photons are updated at the collision point, and the energy contribution of the photons to the sensor is collected. The energy carried by the photons will decay correspondingly at the collision point: Q out ψ(λ) = Q in ψ(λ)·ω(λ) where Q out and Q in represent the photon energy after and before the collision respectively, ω is the single-scattering albedo, which is the equivalent atmospheric albedo at the current position in the atmosphere and the sum of the reflectance and transmittance of the current collision geometry patch in the surface scene; The distribution of the new scattering direction after the photons collide in the atmosphere satisfies the distribution of the scattering phase function. The lookup table of the scattering phase function at the current position calculated in step (1) is used to extract and calculate the new propagation direction. It is assumed that the scattering that occurs on the surface of the surface scene patch of the photons satisfies Lambert's law, and the new scattering direction is calculated according to the patch orientation information. At the position where each collision occurs, the photon diffusion method is used to calculate the energy contribution of the cumulative collision events to the set observation sensor. If there is no occlusion on the path from the collision point position to the sensor, then according to the incident direction of the photons and the position information of the collision point, the probability of the light scattering in the direction of the sensor is calculated, and the energy is collected proportionally. Q collect (Ω,λ) = Q out (λ)·P(Ω,λ)·exp(-L a ·β atm ) P is the probability of light scattering towards the sensor, and L a represents the atmospheric path length from the collision point to the sensor position. The last exponential term in the formula describes the atmospheric extinction over this distance.

4. A method for simulating the coupled radiation of the atmosphere and three-dimensional surface based on photon tracing according to claim 1, wherein In step (4), according to the position of the photons, the corresponding method is used to calculate the next destination of the photons, and the photons are continuously tracked until the termination condition is reached: The photons leave the collision point in the new direction, continue to propagate in a straight line and repeat the collision process. During the entire tracking process, the photons carry position, direction, and energy information. The position information of the photons is used to determine whether they are currently in the atmosphere or the surface scene, and the corresponding method is used to update the information carried by the photons and collect the sensor contribution. After a certain number of collisions, the energy carried by the photons will become relatively small. Once the photon energy is less than the threshold, the Russian roulette is used to terminate the tracking process of the photons conserving energy and unbiased. When the photons enter the surface scene from the bottom of the atmosphere, they have specific direction and position information, and can well simulate the anisotropy of the sky diffuse light after atmospheric scattering. Once the photons scatter and leave the upper boundary of the surface scene through collisions in the surface scene, using the height information and the specific propagation direction of the photons, the photons can be made to return to the atmosphere again to continue tracking, and can well reproduce the cyclic interaction of the photons between the two parts at the junction of the atmosphere and the surface scene, achieving complete coupling of the atmosphere and the surface scene under the atmosphere.

5. A method for simulating the coupled radiation of the atmosphere and three-dimensional surface based on photon tracing according to claim 3, characterized in that, In step (5), new photons are emitted and the tracking process is repeated. After reaching the required number of photons, a convergent simulation result is obtained: Using the Monte Carlo idea, during the simulation of a single photon, each uncertain physical process is described as a probability process according to its physical distribution, and the specific events occurring in this simulation are determined using the method of random number sampling; every position described using the random method strictly follows its physical distribution law; Through a large number of repeated simulation processes, each position described using random events is fully sampled according to its distribution. Finally, the final overall three-dimensional surface scene and atmospheric coupling result can be obtained by accumulating the energy contributions to the observation sensor each time, that is, the radiance signal received in the observation direction of the sensor at the top of the atmosphere:

Images