Method, device and electronic device for constructing surface reflectance spectrum model

Abstract

The present invention relates to the field of remote sensing measurement technology, and in particular to a method, device and electronic device for constructing a surface reflectance spectrum model. The method for constructing a surface reflectance spectrum model provided by the present invention matches the simulated reflectance of the top of the atmosphere calculated by the surface reflection model and the radiation transmission model with the actual reflectance of the top of the atmosphere, and continuously changes the initial value of the surface reflectance model by the gradient descent method until the deviation between the simulated reflectance and the actual reflectance meets the preset deviation condition, and then the surface reflection model finally obtained is used as the surface reflectance spectrum model of the current environment. Based on the obtained surface reflectance spectrum model, the surface reflectance corresponding to any wavelength can be calculated. Compared with the existing site measurement method, the method of the present application has low cost and short cycle, and can greatly improve the frequency of site calibration.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing measurement technology, and in particular to a method, device and electronic equipment for constructing a surface reflectance spectral model. Background Art

[0002] Remote sensing satellites use the ground radiation reference field spectrum model to calibrate remote sensors, which is a common method for on-orbit calibration of satellite-borne remote sensors. Before the satellite is launched, the remote sensor is calibrated with ground equipment. However, due to the influence of the launch process and the space environment, the calibration coefficients before launch are no longer applicable, and the satellite needs to be calibrated on-orbit. Field calibration is a commonly used satellite on-orbit alternative calibration method. At the moment of satellite transit, uniform and stable targets on the ground are synchronously observed, and surface reflectance spectrum data and atmospheric parameters are measured. As input parameters of the atmospheric radiation transmission model, the radiation spectrum information at the top of the atmosphere is calculated and compared with the data observed by the on-board remote sensor to obtain the calibration coefficient. This calibration method has the advantages of full aperture, full field of view, and full dynamic range. At the same time, it can realize the calibration of satellite remote sensors throughout their life cycle and analyze the changes of their remote sensor calibration coefficients over time. Accurate ground radiation reference field spectral modeling plays a key role in improving the accuracy of on-orbit alternative calibration of remote sensors, especially for satellite remote sensors that do not have on-board calibration equipment or whose on-board calibration equipment cannot operate normally due to various reasons.

[0003] Traditional field calibration is to measure the reflectivity information of the ground synchronously by artificial running when the satellite passes by, transfer the radiation benchmark to the top of the atmosphere by atmospheric correction method, and then use the reflectivity basis method to obtain the calibration coefficient of the satellite optical sensor. Most of the sites that meet the requirements of spatial uniformity and temporal stability are located in remote areas such as Gobi and deserts. It takes a lot of manpower, material and financial resources to conduct a complete test, and it poses a great threat to the life safety of observers. Some sites are deep in uninhabited areas and the observation team cannot even reach them. Most of the measured data at the sites are greatly affected by weather conditions, which leads to data delay and insufficient data samples to judge the decay of the payload. At the same time, with the increase in the number and types of space remote sensors and the improvement of calibration accuracy requirements, traditional field calibration is difficult to meet the current requirements. At present, due to the long calibration cycle and low frequency of field radiation calibration, foreign countries usually only conduct field calibration on satellites twice a year, and only once in China.

[0004] At present, the surface reflectance spectrum required for site calibration is mainly obtained through field test measurements. The instruments used for field measurements are mainly ASD field spectrometers and SVC spectroradiometers. At present, the surface reflectance is mainly obtained through field tests, which is costly and time-consuming to conduct a test. Summary of the invention

[0005] The present invention provides a method, device and electronic device for constructing a surface reflectance spectral model, so as to solve the technical problems of high cost and long period in obtaining the surface reflectance through actual field measurement in the prior art.

[0006] In one aspect, the present invention provides a method for constructing a surface reflectance spectral model, comprising:

[0007] The simulated reflectivity corresponding to the surface reflectivity at the top of the atmosphere after the surface reflectivity is transmitted through the atmosphere is obtained through a preset surface reflection model and a radiation transmission model; the surface reflection model is used to calculate the surface reflectivity of any wavelength;

[0008] Obtaining the actual reflectivity of the surface reflectivity at the top of the atmosphere;

[0009] Determine whether the deviation between the simulated reflectivity and the actual reflectivity meets a preset deviation condition; if so, use the surface reflectivity model as the surface reflectivity spectral model of the current environment;

[0010] If not, the surface reflection model is corrected, and the simulated reflectivity corresponding to the surface reflectivity at the top of the atmosphere is obtained again based on the corrected surface reflection model and the radiation transfer model, and it is re-determined whether the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, and the corrected surface reflection model is used as the surface reflectivity spectral model of the current environment.

[0011] According to a method for constructing a surface reflectance spectral model provided by the present invention, the method of obtaining the simulated reflectance corresponding to the top of the atmosphere after the surface reflectance is transmitted through the atmosphere by using a preset surface reflection model and a radiation transmission model comprises:

[0012] Determining the surface reflectivity corresponding to the wavelength according to the surface reflection model;

[0013] The radiation transfer model is used to determine the simulated reflectivity corresponding to the top of the atmosphere after the reflectivity is transmitted through the atmosphere based on the surface reflectivity.

[0014] According to a method for constructing a surface reflectance spectral model provided by the present invention, correcting the surface reflectance model until the deviation between the simulated reflectance and the actual reflectance meets a preset deviation condition includes:

[0015] The model parameters of the surface reflection model are corrected by the gradient descent method so that the deviation between the simulated reflectivity obtained after the reflectivity is processed by the corrected surface reflection model and the radiation transmission model and the actual reflectivity meets the preset deviation condition.

[0016] According to a method for constructing a surface reflectance spectral model provided by the present invention, the surface reflectance model includes the following formula (1):

[0017]

[0018] In the above formula (1), represents the wavelength of the reflected light, Represents the surface reflectivity of reflected light; A , B , , These are all model parameters of the surface reflection model;

[0019] in, A Represents the difference between the maximum and minimum values ​​of the surface reflectance spectrum; B Indicates the magnitude of the surface reflectance spectrum at the turning point; Indicates how fast the surface reflectance spectrum changes with wavelength; Indicates the wavelength position where the rate of change of the surface reflectance spectrum with wavelength is the largest.

[0020] According to a method for constructing a surface reflectance spectral model provided by the present invention, the method of correcting the model parameters of the surface reflectance model using the gradient descent method includes:

[0021] Get the simulated reflectivity change corresponding to the reflectivity ( ) A , B , , The corresponding change values ​​are ;

[0022] The are respectively set to each time the surface reflection model is corrected, the model parameters A , B , , The corresponding step length.

[0023] According to a method for constructing a surface reflectance spectral model provided by the present invention, judging whether the deviation between the simulated reflectance and the actual reflectance meets a preset deviation condition includes:

[0024] Using a preset evaluation function to determine whether the deviation between the simulated reflectivity and the actual reflectivity meets a preset deviation condition;

[0025] The evaluation function is:

[0026]

[0027] In the above formula (2), represents the evaluation function; The wavelength is The reflectivity of corresponds to the actual reflectivity of the top of the atmosphere; The wavelength is The reflectivity of corresponds to the simulated reflectivity at the top of the atmosphere; N Indicates the number of spectral channels;

[0028] The larger the value of is, the greater the deviation between the simulated reflectivity and the actual reflectivity is. The smaller the value of is, the smaller the deviation between the simulated reflectivity and the actual reflectivity is. If the value of satisfies the preset deviation condition, it is determined that the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition.

[0029] According to a method for constructing a surface reflectance spectral model provided by the present invention, when initializing the model parameters of a preset surface reflectance model, the model parameters A , B , , The initial values ​​are set to: A =0.100, B =0.2200, =0.0100, =520.0.

[0030] According to a method for constructing a surface reflectance spectral model provided by the present invention, the step sizes are respectively set as: , , , .

[0031] According to a method for constructing a surface reflectance spectral model provided by the present invention, the method of correcting the model parameters of the surface reflectance model using the gradient descent method further includes:

[0032] Get the rate of change respectively ;in, ;

[0033] Get the evaluation function Relative to The vector of partial derivatives of ;

[0034] For the vector Normalize and get the gradient value ;

[0035] According to the gradient value The input parameters of the evaluation function are modified.

[0036] On the other hand, the present invention also provides a device for constructing a surface reflectance spectral model, comprising:

[0037] A processing module, used to obtain the simulated reflectivity corresponding to the top of the atmosphere after the surface reflectivity is transmitted through the atmosphere through a preset surface reflection model and a radiation transmission model; the surface reflection model is used to calculate the surface reflectivity of any wavelength;

[0038] An acquisition module, used for acquiring the actual reflectivity of the surface reflectivity at the top of the atmosphere;

[0039] A judgment module, used to judge whether the deviation between the simulated reflectivity and the actual reflectivity meets a preset deviation condition; if so, the surface reflectivity model is used as the surface reflectivity spectral model of the current environment;

[0040] A correction module, configured to correct the surface reflection model when the deviation between the simulated reflectivity and the actual reflectivity does not satisfy a preset deviation condition, and send the corrected surface reflection model to the processing module;

[0041] The processing module is also used to re-obtain the simulated reflectivity corresponding to the surface reflectivity at the top of the atmosphere according to the revised surface reflection model and radiation transfer model;

[0042] The judgment module is also used to re-judge whether the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, and then the corrected surface reflectance model is used as the surface reflectance spectral model of the current environment.

[0043] On the other hand, the present invention also provides a method for obtaining surface reflectivity, comprising:

[0044] Obtain the wavelength to be measured;

[0045] Inputting the wavelength into a pre-constructed surface reflectance spectrum model to obtain the surface reflectance;

[0046] The surface reflectance spectral model is obtained by any of the above-mentioned methods for constructing the surface reflectance spectral model.

[0047] On the other hand, the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a method for constructing a surface reflectance spectral model as described in any one of the above-mentioned methods is implemented.

[0048] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for constructing a surface reflectance spectral model as described in any one of the above-mentioned methods is implemented.

[0049] On the other hand, the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the method for constructing a surface reflectance spectral model as described in any one of the above.

[0050] The method for constructing a surface reflectivity spectral model provided by the present invention matches the simulated reflectivity obtained through the surface reflection model and the radiation transmission model with the actual reflectivity of the top of the atmosphere, so as to continuously change the parameters of the surface reflection model until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, and then the surface reflection model finally obtained is used as the surface reflectivity spectral model of the current environment. By obtaining the surface reflectivity spectral model, the corresponding surface reflectivity can be obtained according to any wavelength. Compared with the existing site measurement method, the method of the present application has low cost and short cycle, and can greatly improve the frequency of site calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 It is a schematic flow chart of a method for constructing a surface reflectance spectral model provided by the present invention;

[0053] Figure 2 It is a reflectivity curve diagram obtained by fitting the XCDH_W surface reflectivity spectrum using the surface reflectivity spectrum model provided by the present invention;

[0054] Figure 3 It is a schematic diagram of the gradient descent method provided by the present invention;

[0055] Figure 4 It is a schematic diagram of the reflectivity of typical deserts in the northwest study area of ​​the present invention;

[0056] Figure 5 It is a schematic diagram of the change of the evaluation function of the present invention with the number of iterations;

[0057] Figure 6 This is one of the comparison graphs of the simulated reflectivity reconstructed by the present invention and the actual reflectivity results observed by Tropomi (i.e., Sentinel-5P);

[0058] Figure 7 This is the second comparison diagram of the simulated reflectivity reconstructed by the present invention and the actual reflectivity observed by Tropomi;

[0059] Figure 8 3 is a comparison diagram of the simulated reflectivity reconstructed by the present invention and the actual reflectivity observed by Tropomi;

[0060] Fig. 9 4 is a comparison diagram of the simulated reflectivity reconstructed by the present invention and the actual reflectivity observed by Tropomi;

[0061] Fig.10 is a schematic diagram comparing the surface reflectivity reconstructed by the present invention and the actual surface emissivity measured by MODIS products;

[0062] Fig.11 It is a schematic diagram of the structure of a device for constructing a surface reflectance spectrum model of the present invention;

[0063] Fig.12 is a flow chart of a method for obtaining the surface reflectivity of reflected light of the present invention;

[0064] Fig.13 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] In order to overcome the technical problems of high cost and long period of calibration through field measurement in the prior art, this embodiment uses a method of constructing a spectral model of surface reflectivity to match the simulated reflectivity obtained through the surface reflection model and the radiation transmission model with the actual reflectivity of the top of the atmosphere, and then uses the gradient descent method to continuously correct the surface reflection model until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, indicating that the simulated reflectivity obtained by the current surface reflection model is close to the actual reflectivity, and the surface reflection model finally obtained is used as the surface reflectivity spectral model of the current environment. In this way, by obtaining the surface reflectivity spectral model, the corresponding surface reflectivity can be obtained according to any wavelength, so there is no need to spend huge costs and cycles to go to the site for calibration measurement, saving a lot of manpower costs and time costs.

[0067] Embodiment 1:

[0068] like Figure 1 This embodiment provides a method for constructing a surface reflectance spectral model, which specifically includes:

[0069] Step 101: Obtain the simulated reflectivity corresponding to the top of the atmosphere after the surface reflectivity is transmitted through the atmosphere through a preset surface reflection model and radiation transmission model. The surface reflection model is used to calculate the surface reflectivity of any wavelength.

[0070] In step 101, the surface reflection model includes the following formula (1):

[0071] In the above formula (1), represents the wavelength, Represents the surface reflectivity; A , B , , are all model parameters of the surface reflection model. Among them, A Represents the difference between the maximum and minimum values ​​of the surface reflectance spectrum; B Indicates the magnitude of the surface reflectance spectrum at the turning point; Indicates how fast the surface reflectance spectrum changes with wavelength; Indicates the wavelength position where the rate of change of the surface reflectance spectrum with wavelength is the largest.

[0072] Deserts are often selected as reference radiation fields for satellite alternative calibration due to their good uniformity, stability, and low water vapor and ozone content. This application is based on a large number of studies on the spectral characteristics of desert surface reflectance. Within the spectral range, the reflectance spectrum values ​​of the same desert surface at different solar zenith angles, observation zenith angles and relative azimuth angles are different, but the spectral shapes and trends are the same; the reflectance spectra of different desert surfaces have the same shape, but the reflectance magnitude and the position of the spectral turning point and the rate of change with wavelength are different. Therefore, through multiple experiments and summaries, the surface reflectance spectrum model represented by the above formula (1) is obtained.

[0073] In this embodiment, when the model parameters of the preset surface reflection model are initialized, the model parameters A , B , , The initial values ​​are set to: A =0.100, B =0.2200, =0.0100, =520.0.

[0074] Among them, for any wavelength, the corresponding surface reflectivity can be determined through the initialized surface reflection model.

[0075] In this embodiment, a simulated surface reflectivity can be obtained through the initialized surface reflectivity model. However, since the current surface reflectivity model is not necessarily optimal, the currently obtained surface reflectivity is not necessarily close to the actually obtained surface reflectivity.

[0076] This embodiment proposes a relative error measurement method for measuring the error between the surface reflectivity obtained through the surface reflection model and the actually measured surface reflectivity.

[0077] Specifically, the relative root mean square RMSRE is calculated by the following formula (3) to measure the error between the surface reflectivity obtained by the surface reflection model and the actual measured surface reflectivity. The smaller the RMSRE value, the closer the surface reflectivity obtained by the surface reflection model and the actual measured surface reflectivity are, and vice versa.

[0078] (3)

[0079] In the above formula (3), The wavelength is The measured reflectivity of The wavelength is The corresponding simulated reflectance; N represents the number of spectral channels.

[0080] Figure 2 The results of fitting the surface reflectance spectrum observed in the west of Xiaochaidamu Lake in the 400nm-1000nm band using the model of the present application are shown. Please refer to the following Table 1, which lists the four model parameters of the fitting and the fitting degree R value and RMSE value. The four-parameter model can accurately describe the characteristics of the desert surface reflectance spectrum. By empirically summarizing the characteristics of the desert surface reflectance, the actual reflectance of the surface is represented by a mathematical expression (1) that conforms to the characteristics of the surface reflectance spectrum, and then the parameters in the expression are determined by the top of the atmosphere reflectance obtained by the satellite and the atmospheric transmission model. It is considered that the reconstruction of the surface reflectance spectrum has been achieved. The desert surface spectral reflectance is represented by the above four model parameters. Constructing a four-parameter model can greatly reduce the amount of calculations in the reconstruction process.

[0081] Table 1 Fitting of field test data using the model

[0082]

[0083] The preset radiation transfer model is used to determine the simulated reflectivity corresponding to the top of the atmosphere after the reflectivity is transmitted through the atmosphere based on the surface reflectivity.

[0084] In this embodiment, according to the transmission characteristics of reflectivity in the atmosphere, MODTRAN (radiative transfer model) can be used to simulate and calculate the top of atmosphere reflectivity for known surface reflectivity, and the inverse problem of solving the corresponding surface reflectivity spectrum for known top of atmosphere reflectivity is usually solved by an iterative method. MODTRAN in this embodiment can also be replaced by other radiative transfer models (such as 6S radiative transfer model), and the full name of 6S model is second simulation of the satellite signal in the solar spectrum.

[0085] Step 102: Obtain the actual reflectivity of the top of the atmosphere corresponding to the surface reflectivity.

[0086] In this embodiment, the actual reflectivity of the top of the atmosphere obtained by Tropomi (i.e., Sentinel-5P).

[0087] Step 103: determine whether the deviation between the simulated reflectivity and the actual reflectivity meets a preset deviation condition; if so, use the current surface reflectivity model as the surface reflectivity spectral model of the current environment.

[0088] After obtaining the surface reflectance spectrum model, the surface reflectance of any wavelength can be obtained. Compared with the existing field measurement method, the method of the present application has low cost and short cycle, and can greatly increase the frequency of field calibration.

[0089] Specifically, in step 103, when determining whether the deviation between the simulated reflectivity and the actual reflectivity satisfies a preset deviation condition, a preset evaluation function is used to determine whether the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition.

[0090] The evaluation function of this embodiment is:

[0091]

[0092] In the above formula (2), represents the evaluation function; The wavelength is The reflectivity of corresponds to the reflectivity observed by satellite at the top of the atmosphere; The wavelength is The simulated reflectivity of the reflected light at the top of the atmosphere; N Indicates the number of spectral channels.

[0093] The larger the value of is, the greater the deviation between the simulated reflectivity and the actual reflectivity is. The smaller the value of is, the smaller the deviation between the simulated reflectivity and the actual reflectivity is. If the value of satisfies the preset deviation condition, it means that the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition. If the value of is less than 5%, it means that the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition.

[0094] Step 104: If not, correct the surface reflection model, re-obtain the simulated reflectivity corresponding to the surface reflected light at the top of the atmosphere based on the corrected surface reflection model and the radiation transmission model, and re-judge whether the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition; until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, the corrected surface reflection model is used as the surface reflectivity spectral model of the current environment.

[0095] In step 104, when correcting the surface reflection model in this embodiment, the gradient descent method is specifically used to correct the model parameters of the surface reflection model so that after the reflected light is processed by the corrected surface reflection model and the radiation transmission model, the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition.

[0096] When different surface reflection model parameters are selected, the calculated evaluation function value will also be different. Error evaluation function value is a four-parameter spectral model with desert surface reflectance When the initial parameters are given, the gradient descent iteration method can quickly find the local optimal solution of the parameters.

[0097] For example, for a multivariate function , solving its gradient vector, we can obtain the direction of change of the independent variable corresponding to the fastest increase in the function value at each input point. On the contrary, the reverse direction of the gradient is the direction of change of the independent variable corresponding to the fastest decrease in the function value. If the initial input parameters are known, the reverse direction of the gradient can be calculated to know the direction in which the function becomes smaller, thereby finding the local optimal solution.

[0098] Figure 3 This shows how to use gradient descent to find the local optimal solution for a binary function. Given an initial input vector , calculate the evaluation function and the gradient at that location , we can determine the direction of parameter change and change the input vector to , when choosing the appropriate ratio After the value is obtained, the local optimal solution can be found through multiple iterations location.

[0099] After determining the input parameter vector After analyzing the change direction of the surface reflectance spectrum, the influence of the four parameters on the evaluation function is analyzed, and then the The value is If the value is too small, the number of iterations required to change the initial parameters to the local optimal solution will be large, which will greatly increase the amount of calculation. When the value is too large, the local minimum may be skipped and other local solutions may be entered.

[0100] In this embodiment, appropriate observation geometry, atmospheric parameters and reflectance spectrum model four parameters are selected ( ) is used as a benchmark. By changing the four parameters of the surface reflectance model, the sensitivity of the MODTRAN forward simulation results to the changes of the four parameters of the surface reflectance spectrum is analyzed according to the size of the simulation results before and after the parameter changes.

[0101] The changes in simulation results are measured by the average value of relative deviation:

[0102]

[0103] is the surface reflectance spectrum parameter vector The wavelength in the MODTRAN simulation results is The reflectivity of the top of the atmosphere at . is the surface reflectance spectral parameter vector ( ) in the MODTRAN simulation results. The reflectivity of the top of the atmosphere at Indicates the deviation twice.

[0104] When the parameters of the surface reflection model change, the reflectivity of the top of the atmosphere simulated by MODTRAN will also change accordingly. The greater the parameter change, the greater the change in the simulated reflectivity of the top of the atmosphere. The parameter change size when the simulated reflectivity of the top of the atmosphere changes by about 1% is selected as the step size reference value in the iteration process ( In other words, the change in simulated reflectivity corresponding to the reflected light is ( ) A , B , , The corresponding change values ​​are ;Will Set to each time the surface reflection model is corrected, the model parameters A , B , , The corresponding step length.

[0105] Since the transformation rate of the evaluation function with each parameter is not at the same order of magnitude, the evaluation function needs to be further optimized during the iteration using the gradient descent method. ;in, ; Get the evaluation function Relative to The vector of partial derivatives of ; for vector Normalize and get the gradient value ; According to the gradient value Correct the input parameters of the evaluation function.

[0106] Specifically, ,Will As the input of the evaluation function, the evaluation function is re-determined as follows:

[0107]

[0108] in is the top of atmosphere reflectivity simulated by MODTRAN, is the actual top of atmosphere reflectivity observed by the satellite. Using the evaluation function of formula (5) above, the parameters of the surface reflectivity spectrum in the MODTRAN simulation process are changed, and the approximate values ​​of the partial derivatives of the four parameters of the evaluation function are calculated by using the change of the small step size evaluation function. , normalize the vector to get , and then use the gradient value to change the input parameters for , recalculate the evaluation function and gradient to achieve a fast iteration process of the gradient descent method.

[0109] Embodiment 2:

[0110] In this embodiment, several calibration sites with clear weather conditions on September 27, 2019 in Northwest my country are selected. Table 2 shows the observation geometry of the calibration sites using Tropomi, and Table 3 shows the corresponding atmospheric parameters.

[0111]

[0112]

[0113] like Figure 4 For this example, the MODIS (Moderate Resolution Imaging Spectroradiometer) surface bidirectional reflectance function product is used to summarize the parameter range of typical desert reflectance in the northwest study area of ​​my country (Libya 4 does not belong to the surface of the northwest study area). Figure 4 The approximate ranges of the four model parameters for determining the desert surface reflectance spectrum during the reconstruction process are shown in Table 4 below.

[0114]

[0115] This embodiment selects four parameters: appropriate observation geometry, atmospheric parameters, and reflectance spectrum model As a benchmark, by changing the four parameters of the surface reflectance spectrum model, the sensitivity of the MODTRAN forward simulation results to the changes of the four parameters of the surface reflectance spectrum is analyzed according to the size of the simulation results before and after the parameter changes.

[0116] The initial values ​​of the model parameters of the surface reflection model obtained through analysis in this embodiment are shown in Table 5.

[0117]

[0118] When the parameters of the surface reflection model change, the reflectivity of the top of the atmosphere simulated by MODTRAN will also change accordingly. The greater the parameter change, the greater the change in the simulated reflectivity of the top of the atmosphere. The parameter change size when the simulated reflectivity of the top of the atmosphere changes by about 1% is selected as the step size reference value in the iterative process. The step sizes of the four parameters in this embodiment are selected as follows:

[0119]

[0120]

[0121]

[0122] Since the transformation rate of the evaluation function with each parameter is not at the same order of magnitude, the evaluation function needs to be modified in the process of iterating using the gradient descent method.

[0123] set up ,Will As an input to the evaluation function, the evaluation function is re-determined.

[0124] The surface reflectance model of this embodiment selects the initial surface reflectance parameters of Table 5 above, calculates the surface reflectance, combines the corresponding observation geometry (Table 2) and atmospheric parameters (Table 3), uses MODTRAN to simulate the corresponding top of atmosphere reflectance (i.e., simulated reflectance), and uses the evaluation function to compare it with the actual top of atmosphere reflectance observed by Tropomi. If the deviation between the two does not meet the preset deviation condition, then change the four related parameters of the surface reflectance model in the MODTRAN simulation process, and use the change of the small step size evaluation function to calculate the approximate values ​​of the partial derivatives of the four parameters of the evaluation function. , normalize the vector to get , and then use the gradient value to change the input parameters for , recalculate the evaluation function and gradient, and implement the iterative process of gradient descent method accordingly, thus avoiding the problem of evaluation function changing with time. The change rates of the four parameters are not at the same level, which leads to the problem that the iteration process is too affected by a single parameter and the iteration process is slow. The iteration step size can be changed continuously during the entire iteration process. At the beginning, a relatively large The value makes the parameter approach the local optimal solution as quickly as possible, and can be gradually reduced as the number of iterations increases. To avoid the influence of local optimal solutions, the initial values ​​can be randomly selected multiple times, and the set of results with the lowest evaluation function is selected as the final result.

[0125] Figure 5 The change of the evaluation function with the increase of the number of iterations when reconstructing TKLM1 is given in FIG. The evaluation function generally decreases with the number of iterations. At the beginning, the evaluation function changes very quickly and converges quickly to the local optimal solution. Then the evaluation function decreases slowly. The initial value is selected within the selected four-parameter range. Generally, it can be stabilized after 20 to 50 iterations. Table 6 shows the four-parameter results of the surface reflectance spectral model and the corresponding evaluation function results finally obtained by the method of this embodiment.

[0126]

[0127] Figure 6 , Figure 7 , Figure 8 and Fig. 9 The comparison between the simulated reflectivity of the top of the atmosphere obtained by simulation and the actual reflectivity of the top of the atmosphere observed by Tropomi is shown respectively. Figure 6 , Figure 7 , Figure 8 and Fig. 9 S2 represents the simulated reflectivity of the top of the atmosphere, S1 represents the actual reflectivity of the top of the atmosphere observed by Tropomi, and S3 represents the error between the simulated reflectivity and the actual reflectivity. Among them, the simulated reflectivity of the top of the atmosphere of each surface obtained by reconstruction gradually decreases between 400nm-600nm and gradually increases between 600nm-800nm, which is in line with the actual situation. The simulated reflectivity of the top of the atmosphere of each surface obtained by reconstruction is very close to the actual reflectivity of the top of the atmosphere observed by satellite. The evaluation function of the simulated reflectivity of the top of the atmosphere of each surface obtained by reconstruction is within 2%. Excluding the influence of the spectral absorption peak, the relative error in the entire band is around 5%.

[0128] The bidirectional scattering distribution function model coefficients of MODIS (Moderate Resolution Imaging Spectroradiometer) on each surface were extracted, and the actual measured surface reflectance under the same observation geometry was calculated and compared with the surface reflectance obtained by reconstructing the surface reflectance spectral model (see Table 8 and Figure 10). Fig.10The surface reflectance obtained by the reconstructed surface reflectance spectral model at the four calibration fields is compared with the MODIS surface reflectance product. It can be seen that the deviation between the surface reflectance obtained by the reconstructed surface reflectance spectral model and the MODIS surface reflectance product is within 5%. It can be seen that the surface reflectance spectral model constructed by the method of the present application is relatively accurate.

[0129] Fig.10 The sites are distributed in different locations in the northwest research area of ​​my country and are far away from each other. Field measurement of the optical characteristics of these surfaces will consume a lot of manpower and material resources. The construction method of this embodiment is used to reconstruct the typical surface reflectance spectral model in the study area, which not only plays an important role in site radiation calibration, but also can compare and analyze the optical characteristics of various surfaces in the study area.

[0130] Embodiment three:

[0131] This embodiment provides a device for constructing a surface reflectance spectrum model. Please refer to Fig.11 The construction device includes: a processing module 201, an acquisition module 202, a judgment module 203 and a correction module 204.

[0132] Among them, the processing module 201 is used to obtain the simulated reflectivity corresponding to the top of the atmosphere after the surface reflected light is transmitted through the atmosphere through the preset surface reflection model and radiation transmission model; the surface reflection model is used to calculate the surface reflectivity of reflected light of any wavelength. The acquisition module 202 is used to obtain the actual reflectivity of the reflected light at the top of the atmosphere. The judgment module 203 is used to judge whether the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition; if so, the initialized surface reflection model is used as the surface reflectivity spectral model of the current environment. The correction module 204 is used to correct the surface reflection model when the deviation between the simulated reflectivity and the actual reflectivity does not meet the preset deviation condition, and send the corrected surface reflection model to the processing module, specifically, send the model parameters of the corrected surface reflection model to the processing module 201, and the processing module 201 sets the corrected model parameters in the parameters of the surface reflection model.

[0133] The processing module 201 is also used to re-obtain the simulated reflectivity corresponding to the surface reflected light at the top of the atmosphere based on the revised surface reflection model and radiation transmission model; the judgment module 203 is also used to re-judge whether the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, the revised surface reflection model is used as the surface reflectivity spectral model of the current environment.

[0134] Among them, the implementation method of the above-mentioned modules in this embodiment is the same as that in the above-mentioned embodiment 1, and will not be repeated here.

[0135] Embodiment 4:

[0136] This embodiment provides a method for obtaining surface reflectivity. Fig.12 , the surface reflectivity acquisition method includes:

[0137] Step 301: Obtain the wavelength to be measured.

[0138] Step 302: Input the wavelength into a pre-constructed surface reflectance spectrum model to obtain the surface reflectance of the reflected light.

[0139] The surface reflectance spectral model is obtained by the method for constructing the surface reflectance spectral model provided in the first embodiment.

[0140] Embodiment five:

[0141] Fig.13 An example of a physical structure diagram of an electronic device is shown in FIG. Fig.13 As shown, the electronic device may include: a processor 401, a communication interface 402, a memory 403 and a communication bus 404, wherein the processor 401, the communication interface 402 and the memory 403 communicate with each other via the communication bus 404. The processor 401 may call the logic instructions in the memory 403 to execute the method for constructing the surface reflectance spectral model provided in the first embodiment.

[0142] In addition, the logic instructions in the above-mentioned memory 403 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0143] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the method for constructing a surface reflectance spectral model provided by the above-mentioned methods.

[0144] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the method for constructing a surface reflectance spectral model provided by the above-mentioned methods.

[0145] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0146] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a surface reflectance spectral model, characterized in that: include: The simulated reflectivity corresponding to the surface reflectivity at the top of the atmosphere after the surface reflectivity is transmitted through the atmosphere is obtained through a preset surface reflection model and a radiation transmission model; the surface reflection model is used to calculate the surface reflectivity of any wavelength; Obtaining the actual reflectivity of the surface reflectivity at the top of the atmosphere; Determine whether the deviation between the simulated reflectivity and the actual reflectivity meets a preset deviation condition; if so, use the surface reflectivity model as the surface reflectivity spectral model of the current environment; If not, the surface reflection model is corrected, and the simulated reflectivity corresponding to the surface reflectivity at the top of the atmosphere is obtained again according to the corrected surface reflection model and the radiation transfer model, and it is re-determined whether the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition; until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, the corrected surface reflection model is used as the surface reflectivity spectral model of the current environment; wherein, correcting the surface reflection model includes: using the gradient descent method to correct the model parameters of the surface reflection model, so that after the surface reflectivity is processed by the corrected surface reflection model and the radiation transfer model, the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition; the surface reflection model includes the following formula (1): , in the above formula (1), represents the wavelength, Represents the surface reflectivity; A , B , , are all model parameters of the surface reflection model; among them, A Represents the difference between the maximum and minimum values ​​of the surface reflectance spectrum; B Indicates the magnitude of the surface reflectance spectrum at the turning point; Indicates how fast the surface reflectance spectrum changes with wavelength; Indicates the wavelength position where the surface reflectivity spectrum changes with wavelength at the maximum rate; the model parameters of the surface reflection model corrected by the gradient descent method include: obtaining the simulated atmospheric top reflectivity change amount corresponding to the reflectivity is ( ), the A , B , , The corresponding change values ​​are ; are respectively set to each time the surface reflection model is corrected, the model parameters A , B , , Corresponding step length; judging whether the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition includes: using a preset evaluation function to judge whether the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition; when initializing the model parameters of the preset surface reflection model, the model parameters A , B , , The initial values ​​are set to: A =0.100, B =0.2200, =0.0100, =520.0; the step sizes are respectively set to: , , , The method of using the gradient descent method to correct the model parameters of the surface reflection model also includes: obtaining the change rate ;in, ; Get the evaluation function Relative to The vector of partial derivatives of ; for the vector Normalize and get the gradient value According to the gradient value The input parameters of the evaluation function are modified.

2. The method for constructing a surface reflectance spectral model according to claim 1, characterized in that: The method of obtaining the simulated reflectivity corresponding to the top of the atmosphere after the surface reflectivity is transmitted through the atmosphere by using the preset surface reflection model and radiation transmission model includes: Determine the corresponding surface reflectivity according to the surface reflection model; The radiation transfer model is used to determine the simulated reflectivity corresponding to the top of the atmosphere after the reflectivity is transmitted through the atmosphere based on the surface reflectivity.

3. The method for constructing a surface reflectance spectral model according to claim 1, characterized in that: The evaluation function is: ； In the above formula (2), represents the evaluation function; The wavelength is The reflectivity of corresponds to the actual reflectivity of the top of the atmosphere; The wavelength is The reflectivity of corresponds to the simulated reflectivity at the top of the atmosphere; N Indicates the number of spectral channels; The larger the value of is, the greater the deviation between the simulated reflectivity and the actual reflectivity is. The smaller the value of is, the smaller the deviation between the simulated reflectivity and the actual reflectivity is. If the value of satisfies the preset deviation condition, it is determined that the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition.

4. A device for constructing a surface reflectance spectral model, characterized in that: A method for constructing a surface reflectance spectral model applicable to any one of claims 1 to 3, comprising: A processing module, used to obtain the simulated reflectivity corresponding to the top of the atmosphere after the surface reflectivity is transmitted through the atmosphere through a preset surface reflection model and a radiation transmission model; the surface reflection model is used to calculate the surface reflectivity of any wavelength; An acquisition module, used for acquiring the actual reflectivity of the surface reflectivity at the top of the atmosphere; A judgment module, used to judge whether the deviation between the simulated reflectivity and the actual reflectivity meets a preset deviation condition; if so, the surface reflectivity model is used as the surface reflectivity spectral model of the current environment; A correction module, configured to correct the surface reflection model when the deviation between the simulated reflectivity and the actual reflectivity does not satisfy a preset deviation condition, and send the corrected surface reflection model to the processing module; The processing module is also used to re-obtain the simulated reflectivity corresponding to the surface reflectivity at the top of the atmosphere according to the revised surface reflection model and radiation transfer model; The judgment module is also used to re-judge whether the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, until the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition, and then the corrected surface reflection model is used as the surface reflection spectral model of the current environment; wherein, correcting the surface reflection model includes: using the gradient descent method to correct the model parameters of the surface reflection model, so that after the surface reflectivity is processed by the corrected surface reflection model and the radiation transmission model, the deviation between the simulated reflectivity and the actual reflectivity meets the preset deviation condition; the surface reflection model includes the following formula (1): , in the above formula (1), represents the wavelength, Represents the surface reflectivity; A , B , , are all model parameters of the surface reflection model; among them, A Represents the difference between the maximum and minimum values ​​of the surface reflectance spectrum; B Indicates the magnitude of the surface reflectance spectrum at the turning point; Indicates how fast the surface reflectance spectrum changes with wavelength; Indicates the wavelength position where the surface reflectivity spectrum changes with wavelength at the maximum rate; the model parameters of the surface reflection model corrected by the gradient descent method include: obtaining the simulated atmospheric top reflectivity change amount corresponding to the reflectivity is ( ), the A , B , , The corresponding change values ​​are ; are respectively set to each time the surface reflection model is corrected, the model parameters A , B , , Corresponding step length; judging whether the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition includes: using a preset evaluation function to judge whether the deviation between the simulated reflectivity and the actual reflectivity satisfies the preset deviation condition; when initializing the model parameters of the preset surface reflection model, the model parameters A , B , , The initial values ​​are set to: A =0.100, B =0.2200, =0.0100, =520.0; the step sizes are respectively set to: , , , The method of using the gradient descent method to correct the model parameters of the surface reflection model also includes: obtaining the change rate ;in, ; Get the evaluation function Relative to The vector of partial derivatives of ; for the vector Normalize and get the gradient value According to the gradient value The input parameters of the evaluation function are modified.

5. A method for obtaining surface reflectivity, characterized in that: include: Obtain the wavelength to be measured; Inputting the wavelength into a pre-constructed surface reflectance spectrum model to obtain the surface reflectance; The surface reflectance spectral model is obtained by the method for constructing a surface reflectance spectral model according to any one of claims 1 to 3.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for constructing the surface reflectance spectral model as described in any one of claims 1 to 3 is implemented.

Images