Atmospheric correction parameter estimation system, atmospheric correction parameter estimation method, atmospheric correction parameter estimation program, aerosol optical thickness estimation system, aerosol optical thickness estimation method, and aerosol optical thickness estimation program

Abstract

By correcting for the effects of atmospheric aerosols and thin clouds in the observed spectra obtained from photographs of the Earth's surface taken from outside the atmosphere, the Earth's surface reflectance is estimated with high accuracy. [Solution] An atmospheric correction parameter estimation system 1 comprises: an observation spectrum acquisition unit 11 that acquires an observation spectrum by photographing the ground surface from outside the atmospheric layer; an atmospheric layer parameter candidate generation unit 14 that generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer; a BOA reflectance calculation unit 13 that calculates the BOA reflectance from the candidate values ​​of the ground surface parameters; a TOA reflectance calculation unit 15 that calculates the TOA reflectance from the BOA reflectance and the candidate values ​​of the atmospheric layer parameters; and an atmospheric correction parameter estimation unit 16 that estimates the atmospheric layer parameter with the greatest degree of agreement as an atmospheric correction parameter based on the degree of agreement between the TOA reflectance calculated from multiple candidate values ​​of the ground surface parameters and atmospheric layer parameters and the observation spectrum.

Description

Technical Field

[0001] The present invention relates to an atmospheric correction parameter estimation system, an atmospheric correction parameter estimation method, an atmospheric correction parameter estimation program, an aerosol optical depth estimation system, an aerosol optical depth estimation method, and an aerosol optical depth estimation program.

Background Art

[0002] In order to stably supply food to people around the world, technologies for accurately obtaining the cultivated area and yield of crops worldwide are required.

[0003] In that regard, Patent Document 1 discloses a technique for predicting the yield of paddy rice cultivated in a field using satellite observation data.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Techniques for predicting crop yields using observation spectra obtained by photographing the earth's surface from outside the atmosphere by an earth observation optical satellite or the like are known. However, since such observation spectra are strongly affected by atmospheric aerosols and thin clouds, the prediction accuracy of crop yields has not reached a practical level. There is a need for a technique for accurately estimating the influence of aerosols and thin clouds in the atmosphere over agricultural land and correcting this atmospheric influence with high accuracy.

[0006] One of the objectives of this invention is to estimate the surface reflectance with high accuracy by correcting for the effects of aerosols and thin clouds in the atmospheric layer from observational spectra obtained by photographing the Earth's surface from outside the atmospheric layer, and to improve the quality of satellite image data by generating corrected image data (or surface reflectance data) from observational spectral data with atmospheric effects (noise) removed. [Means for solving the problem]

[0007] To achieve the above objective, an atmospheric correction parameter estimation system according to one aspect of the present invention includes: an observation spectrum acquisition unit that acquires an observation spectrum obtained by photographing the ground surface from outside the atmospheric layer; a ground surface parameter candidate generation unit that generates a plurality of candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface; an atmospheric layer parameter candidate generation unit that generates a plurality of candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer; a BOA reflectance calculation unit that calculates the BOA reflectance from the candidate values ​​of the ground surface parameters; a TOA reflectance calculation unit that calculates the TOA reflectance from the BOA reflectance and the candidate values ​​of the atmospheric layer parameters; and an atmospheric correction parameter estimation unit that estimates the atmospheric layer parameter that maximizes the degree of agreement between the TOA reflectance calculated from the plurality of candidate values ​​of the ground surface parameters and the atmospheric layer parameters and the observation spectrum, as an atmospheric correction parameter.

[0008] The ground surface parameter candidate generation unit may be characterized by generating candidate values ​​such that the values ​​of the ground surface parameters are continuous over time.

[0009] The atmospheric layer parameter candidate generation unit may be characterized by generating candidate values ​​such that the values ​​of the atmospheric layer parameters are spatially continuous.

[0010] The atmospheric correction parameter estimation unit may estimate the atmospheric correction parameters by calculating the degree of agreement between the TOA reflectance calculated from a plurality of candidate values ​​of the ground surface parameter and the atmospheric layer parameter and the observed spectrum for each location, and by searching for the combination that maximizes the sum of the degrees of agreement in a predetermined area. Here, the predetermined area may be defined, for example, by (a) a predetermined mesh (e.g., 100m square or 1km square), (b) a management area such as a field polygon, or (c) a moving window (e.g., k×k pixels). Furthermore, the search for the combination is not limited to searching for all combinations, but may be performed as a stepwise search, a nearest neighbor search, or a constrained optimization.

[0011] The ground surface parameter candidate generation unit may repeatedly generate multiple new ground surface parameters based on the multiple ground surface parameters and the degree of agreement calculated by the atmospheric correction parameter estimation unit, and the differences between the multiple ground surface parameters generated may decrease as the generation process is repeated.

[0012] The atmospheric layer parameter candidate generation unit may repeatedly generate multiple new atmospheric layer parameters based on the multiple atmospheric layer parameters and the degree of agreement calculated by the atmospheric correction parameter estimation unit, and the differences between the multiple atmospheric layer parameters generated may decrease as the generation process is repeated.

[0013] The BOA reflectance calculation unit may estimate the BOA reflectance by synthesizing elemental spectra at a ratio specified by the ground surface parameters.

[0014] The TOA reflectance calculation unit may calculate the TOA reflectance by calculating the radiative transfer function using the atmospheric layer parameters.

[0015] The aforementioned ground surface parameters may include a vegetation cover rate, which is the percentage of the ground surface covered by vegetation.

[0016] The surface parameters may include the chlorophyll concentration, which is a photosynthetic pigment of the plants covering the surface.

[0017] The surface parameters may include the soil surface moisture content, which is the moisture content of the surface layer of the surface.

[0018] The surface parameters may include the soil saturation amount, which is the saturation amount of the surface.

[0019] The surface parameters may include the crop heading rate, which is the ratio of the heading of the crops covering the surface.

[0020] The surface parameters may include the crop ripening rate, which is the degree of ripening of the crops covering the surface.

[0021] The atmospheric parameters may include the aerosol optical depth, which indicates the transmittance of incident light by the suspended particles in the atmosphere.

[0022] The atmospheric parameters may include the aerosol Ångström exponent, which indicates the wavelength characteristics of the transmittance of incident light by the suspended particles in the atmosphere.

[0023] The atmospheric parameters may include the aerosol backscattering ratio, which indicates the backscattering ratio of the incident light by the suspended particles in the atmosphere without reaching the surface.

[0024] The atmospheric parameters may include the optical depth of the cloud between the sun and the surface, which indicates the transmittance of incident light by the cloud in the atmosphere.

[0025] The atmospheric parameters may include the optical depth of the cloud between the surface and the satellite, which indicates the transmittance of incident light by the cloud in the atmosphere.

[0026] The aforementioned atmospheric parameters may include the backscattering rate by clouds, which indicates the ratio of sunlight incident by the sun that is reflected by clouds in the atmosphere without reaching the Earth's surface.

[0027] The aforementioned atmospheric layer parameters may include the absorptivity due to water vapor, which indicates the absorptivity of incident light by water vapor in the atmosphere.

[0028] To achieve the above objective, an atmospheric correction parameter estimation method according to another aspect of the present invention is performed by a computer, comprising: an observation spectrum acquisition step of acquiring an observation spectrum obtained by photographing the ground surface from outside the atmospheric layer; a ground surface parameter candidate generation step of generating a plurality of candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface; an atmospheric layer parameter candidate generation step of generating a plurality of candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer; a BOA reflectance calculation step of calculating the BOA reflectance from the candidate values ​​of the ground surface parameters; a TOA reflectance calculation step of calculating the TOA reflectance from the BOA reflectance and the candidate values ​​of the atmospheric layer parameters; and an atmospheric correction parameter estimation step of estimating the atmospheric layer parameter that maximizes the degree of agreement between the TOA reflectance calculated from the plurality of candidate values ​​of the ground surface parameters and the atmospheric layer parameters and the observation spectrum.

[0029] To achieve the above objective, an atmospheric correction parameter estimation program according to another aspect of the present invention causes a computer to execute the following commands: an observation spectrum acquisition command to acquire an observation spectrum obtained by photographing the ground surface from outside the atmospheric layer; a ground surface parameter candidate generation command to generate multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface; an atmospheric layer parameter candidate generation command to generate multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer; a BOA reflectance calculation command to calculate the BOA reflectance from the candidate values ​​of the ground surface parameters; a TOA reflectance calculation command to calculate the TOA reflectance from the BOA reflectance and the candidate values ​​of the atmospheric layer parameters; and an atmospheric correction parameter estimation command to estimate the atmospheric layer parameter that maximizes the degree of agreement between the TOA reflectance calculated from the multiple candidate values ​​of the ground surface parameters and the atmospheric layer parameters and the observation spectrum.

[0030] To achieve the above objective, an aerosol optical thickness estimation system according to another aspect of the present invention comprises: an observation spectrum acquisition unit that acquires an observation spectrum obtained by photographing the ground surface from outside the atmospheric layer; a ground surface parameter candidate generation unit that generates a plurality of candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface; an atmospheric layer parameter candidate generation unit that generates a plurality of candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer and include at least the optical thickness of the aerosol; a BOA reflectance calculation unit that calculates the BOA reflectance from the candidate values ​​of the ground surface parameters; a TOA reflectance calculation unit that calculates the TOA reflectance from the BOA reflectance and the candidate values ​​of the atmospheric layer parameters; an atmospheric correction parameter estimation unit that estimates the atmospheric layer parameter that maximizes the degree of agreement between the TOA reflectance calculated from the plurality of candidate values ​​of the ground surface parameters and the atmospheric layer parameters and the observation spectrum, and an output unit that outputs at least the optical thickness of the aerosol from among the estimated atmospheric correction parameters.

[0031] To achieve the above objective, a method for estimating the optical thickness of an aerosol according to another aspect of the present invention is performed by a computer and includes: an observation spectrum acquisition step of acquiring an observation spectrum obtained by photographing the ground surface from outside the atmospheric layer; a ground surface parameter candidate generation step of generating a plurality of candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface; an atmospheric layer parameter candidate generation step of generating a plurality of candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer and include at least the optical thickness of the aerosol; a BOA reflectance calculation step of calculating the BOA reflectance from the candidate values ​​of the ground surface parameters; a TOA reflectance calculation step of calculating the TOA reflectance from the BOA reflectance and the candidate values ​​of the atmospheric layer parameters; an atmospheric correction parameter estimation step of estimating the atmospheric layer parameter that has the greatest degree of agreement with the TOA reflectance calculated from the plurality of candidate values ​​of the ground surface parameters and the atmospheric layer parameters as an atmospheric correction parameter; and an output step of outputting at least the optical thickness of the aerosol from among the estimated atmospheric correction parameters.

[0032] To achieve the above objective, an aerosol optical thickness estimation program according to another aspect of the present invention causes a computer to execute the following commands: an observation spectrum acquisition command to acquire an observation spectrum obtained by photographing the ground surface from outside the atmospheric layer; a ground surface parameter candidate generation command to generate a plurality of candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface; an atmospheric layer parameter candidate generation command to generate a plurality of candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer and include at least the optical thickness of the aerosol; a BOA reflectance calculation command to calculate the BOA reflectance from the candidate values ​​of the ground surface parameters; a TOA reflectance calculation command to calculate the TOA reflectance from the BOA reflectance and the candidate values ​​of the atmospheric layer parameters; an atmospheric correction parameter estimation command to estimate the atmospheric layer parameter that maximizes the degree of agreement between the TOA reflectance calculated from the plurality of candidate values ​​of the ground surface parameters and the atmospheric layer parameters and the observation spectrum; and an output command to output at least the optical thickness of the aerosol from among the estimated atmospheric correction parameters. Computer programs can be provided via download over networks such as the Internet, or by being recorded on various non-transitory, computer-readable recording media such as CD-ROMs, flash memory, and hard disk drives. [Effects of the Invention]

[0033] By obtaining observational spectra from images of the Earth's surface taken from outside the atmosphere, it is possible to estimate the Earth's surface reflectance with high accuracy by correcting for atmospheric effects. [Brief explanation of the drawing]

[0034] [Figure 1] This is an example of a diagram showing the configuration of the atmospheric correction parameter estimation system according to the present invention. [Figure 2] This is an example of a functional block diagram of the atmospheric correction parameter estimation device in the above-mentioned atmospheric correction parameter estimation system. [Figure 3] This is an example of a flowchart executed by the atmospheric correction parameter estimation system described above. [Modes for carrying out the invention]

[0035] The embodiments for carrying out the present invention will be described below with reference to the figures. All figures are illustrative. In the following detailed description, certain details are mentioned for illustrative purposes and to facilitate a full understanding of the disclosed embodiments. However, the embodiments are not limited to these specific details. Also, for the sake of simplification of the drawings, well-known structures and apparatus are shown schematically.

[0036] Figure 1 shows an example of the configuration diagram of the atmospheric correction parameter estimation system 1. The atmospheric correction parameter estimation system 1 is a system that estimates parameters that correct the influence of the atmosphere on the observed spectrum, i.e., atmospheric correction parameters, based on the observed spectrum obtained from outside the atmospheric layer and multiple candidate values ​​of reflectance at the atmospheric layer surface (hereinafter also referred to as "TOA (Top Of Atmosphere) reflectance") for the area R to be evaluated. It can also be configured as a ground surface reflectance estimation system that estimates the reflectance of the ground surface (hereinafter also referred to as "BOA (Bottom Of Atmosphere) reflectance") by correcting the observed spectrum using parameters obtained by leveling the atmospheric correction parameters estimated for each location over a continuous region. The estimated ground surface reflectance can be used to estimate the growth state of plants on the ground surface and the yield of crops. When configured as a ground surface reflectance estimation system, the estimated ground surface reflectance (BOA reflectance) may be generated as a set of reflectance values ​​for each wavelength band, or as image data corresponding to the pixel arrangement. The generated ground surface reflectance (BOA reflectance) may be stored in a memory unit (e.g., memory unit 20), output to and displayed on the user interface device 3300, or transmitted to an external device via a network NW. The atmospheric correction parameter estimation system 1 is, for example, an atmospheric correction parameter estimation device 100 and an artificial satellite 500 that are connected to each other via a network NW.

[0037] The observation spectrum is so-called satellite data, captured from outside the atmosphere by the Satellite 500. Satellite 500 is an Earth observation optical satellite, such as Sentinel-2 Landsat 8 / 9, but the specific model is not limited to this. Furthermore, the observation spectrum used to estimate atmospheric correction parameters is not limited to that obtained from Satellite 500, but may be any appropriate observation spectrum obtained from outside the atmospheric layer, or even an observation spectrum obtained by a space elevator, etc. The resolution of the observation spectrum is, for example, about 30 m, and preferably 10 m or less.

[0038] The observational spectrum acquired by the atmospheric correction parameter estimation system 1 includes wavelength bands in which the reflectance differs significantly from other wavelength bands, i.e., red-edge wavelengths, i.e., near-infrared (NIR) wavelengths, such as 700 nm, 740 nm, and 865 nm.

[0039] ● Overview of the control system Figure 2 shows the functional blocks of the atmospheric correction parameter estimation device 100 included in the atmospheric correction parameter estimation system 1. The atmospheric correction parameter estimation device 100 may be a hardware configuration or it may be configured on the cloud. Furthermore, the configuration shown in Figure 2 is illustrative, and one component may encompass another component, and the functional parts of each component may be possessed by other components.

[0040] The user interface device 300 only needs to have an input unit and a display unit for the operator, and may be implemented by the functions of the control unit 401. Alternatively, the user interface device 300 may be a personal computer, and information may be input and displayed on a web UI via a web browser installed on the personal computer.

[0041] ● Functional section of the atmospheric correction parameter estimation system As shown in Figure 2, the atmospheric correction parameter estimation device 100 is equipped with storage devices such as RAM (Random Access Memory) and ROM (Read Only Memory) for performing information processing, and a computing device such as a CPU (Central Processing Unit). As a result, it comprises at least the following functional blocks as software resources: a spectrum acquisition unit 11, a ground surface parameter candidate generation unit 12, a BOA reflectance calculation unit 13, an atmospheric layer parameter candidate generation unit 14, a TOA reflectance calculation unit 15, and an atmospheric correction parameter estimation unit 16.

[0042] The spectrum acquisition unit 11 is a functional unit that acquires time-series multispectral data in a predetermined area R as an observed spectrum. In this description, the observed spectrum is calculated as reflectance by dividing the measured value at a specific location by the amount of solar radiation. The spectrum acquisition unit 11 acquires observed spectra at multiple points in time, obtained from satellite 500 or the like, and links them with time information. As mentioned above, the spectrum acquisition unit 11 acquires observed spectra that include data at wavelengths such as the red edge, i.e., NIR (near-infrared band), such as 700 nm, 740 nm, and 865 nm. The timing of acquiring multiple observed spectra may be appropriate, and the accumulated time-series observed spectra may be acquired all at once, or they may be acquired at different points in time. The spectrum acquisition unit 11 may acquire, for example, time-series observed spectra spanning multiple years. The observed spectrum may be a value obtained by performing radiative calibration on radiance data acquired by sensors such as satellite 500 and converting it to TOA reflectance considering solar incidence conditions (e.g., solar zenith angle). Furthermore, depending on the band response characteristics of the sensor, the reflectance may be treated as a value obtained by integrating or averaging over a predetermined wavelength band. In addition, the TOA reflectance may be calculated or corrected by considering the observed geometry (e.g., observed zenith angle, azimuth angle difference).

[0043] The ground surface parameter candidate generation unit 12 is a functional unit that generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface. Ground surface parameters are parameters necessary for calculating the BOA reflectance, and once the ground surface parameters are determined, the BOA reflectance can be calculated.

[0044] Here, the BOA reflectance is determined by multiple elements that reflect the state quantities of the ground surface in the field area. More specifically, the BOA reflectance can be said to be calculated by combining the elemental spectra of each element at a synthesis ratio appropriate to the state. In the case of rice crop, the elemental spectra include, for example, the soil spectrum, the stem and leaf spectrum reflecting absorption and reflection in the stems and leaves, the panicle spectrum reflecting absorption and reflection in the panicles, the yellowed stem and leaf spectrum reflecting absorption and reflection in yellowed stems and leaves, or the yellowed panicle spectrum reflecting absorption and reflection in yellowed stems and leaves. Therefore, the ground surface parameter candidate generation unit 12 generates candidate values ​​for ground surface parameters defined in accordance with the state quantities of the ground surface that affect the synthesis ratio of the elemental spectra. The state quantities of the ground surface in a field change significantly over time, but the variations in these changes have properties unique to the field. In this way, by selecting parameters and estimating the BOA reflectance based on the premise that the ground surface is a field, the BOA reflectance in the field can be determined with greater accuracy.

[0045] In other words, surface parameters include, for example, vegetation cover rate, leaf area index (LAI), chlorophyll concentration, carotenoid concentration, soil surface moisture content, soil water content, crop heading rate, and crop grain filling rate.

[0046] Vegetation cover is the percentage of the ground surface covered by vegetation. Vegetation patterns include, for example, at least vegetated and unvegetated states. Vegetation patterns may also be expressed as vegetation quantity, which is a numerical representation of the degree of vegetation. Vegetation quantity is an index that shows the amount of plants in each pixel of the ground surface. A state in which the vegetation quantity is above a predetermined threshold may be defined as a vegetated state, and a state in which the vegetation quantity is below that threshold may be defined as an unvegetated state. The threshold may be, for example, 0. Vegetation cover has the following relationship with the reflectance of near-infrared light, for example, as shown in equation (1). R NIR =R NIR,L × Vegetation cover rate + R NIR,S × (1 - vegetation cover rate) …(1) R NIRR is the reflectance of the observed near-infrared light. NIR,L R is the reflectance of near-infrared light in the stem and leaf spectrum. NIR,S This is the reflectance of near-infrared light in the soil spectrum.

[0047] Vegetation cover can be described as the combined ratio of the soil spectrum and the plant-specific spectrum. Therefore, the atmospheric correction parameter estimation device 100 acquires the spectrum of the ground, i.e., the soil surface, at the site in question over time during periods when there is no vegetation, and estimates the wavelength characteristics of the soil surface reflectance (hereinafter also referred to as the "soil spectrum"). Furthermore, plants absorb and reflect the spectrum of light at specific wavelengths. That is, for example, plants absorb blue and red light, reflect green light, and strongly reflect near-infrared light. In other words, the higher the vegetation cover, the lower the reflectance of red light and the higher the reflectance of green light and near-infrared light compared to the soil spectrum.

[0048] Chlorophyll concentration is the concentration of chlorophyll, a photosynthetic pigment in terrestrial plants, within the plant body. The spectrum of plants covering the ground surface changes depending on the chlorophyll concentration. As the chlorophyll concentration increases, the red and blue reflectance decreases. Therefore, by defining the chlorophyll concentration of plants covering the ground surface as one of the ground surface parameters and considering that the spectrum can change based on chlorophyll concentration, the BOA reflectance can be estimated more accurately.

[0049] Carotenoid concentration refers to the concentration of carotenoids, which are photosynthetic pigments in land plants, within the plant body. The spectrum of plants covering the ground surface changes depending on the carotenoid concentration. As the carotenoid concentration increases, the blue reflectance decreases and the red reflectance increases. Therefore, by defining the carotenoid concentration of plants covering the ground surface as one of the ground surface parameters and considering that the spectrum can change based on the carotenoid concentration, the BOA reflectance can be estimated more accurately.

[0050] Soil surface moisture content is a parameter that indicates the amount of moisture in the top layer of the soil surface. The higher the surface moisture content of the soil surface, the more uniformly the reflectance decreases, regardless of wavelength. Therefore, by defining soil surface moisture content as one of the ground surface parameters and considering that the BOA reflectance can change based on the soil surface moisture content, the BOA reflectance can be estimated more accurately.

[0051] Soil water content is the amount of water present on the ground surface and is a parameter that indicates the depth of the water content or the presence or absence of water content. The water content includes at least both watered and unwatered states. A watered state is a state in which the ground surface is covered with liquid water. An unwatered state refers to a state in which the ground surface is not covered with liquid water. Liquid water has the characteristic of absorbing light with a wavelength of 740 nm and not absorbing light with a wavelength of 700 nm. Therefore, depending on whether or not there is water content, the reflectance of light with a wavelength of 740 nm decreases compared to the reflectance of light with a wavelength of 700 nm in the area in question.

[0052] Furthermore, in flooded conditions, the reflectivity decreases at all wavelengths as a result of the microscopic voids in the soil being filled with water. Therefore, the soil water content may be a parameter that takes into account the decrease in reflectivity at all wavelengths from blue light to near-infrared light.

[0053] Crop heading rate and crop ripening rate are parameters used to consider the crop's growth stage when estimating BOA reflectance. Here, the BOA reflectance varies depending on the growth stage of the crop covering the ground surface. A growth stage is a classification of the crop's growth process according to its characteristics. For example, in the case of rice, the growth stages are classified into vegetative growth, panicle formation, heading, and ripening. The heading and ripening stages are examples of the reproductive growth stage.

[0054] During the reproductive growth phase, a characteristic spectrum different from that of the vegetative growth phase is generated. The spectrum characteristically present during the vegetative growth phase varies depending on the type of crop. For example, in rice and wheat, during the vegetative growth phase, the BOA reflectance is a composite spectrum of the soil spectrum and the stem and leaf spectrum from the plant's stems and leaves. However, during the heading and ripening phases, yellowed stem and leaf spectra and yellowed panicle spectra are further synthesized as panicles and yellowing occur. Therefore, the crop heading rate, which indicates the ratio of the spectrum characteristically present during the heading phase, is used as one of the soil surface parameters. Similarly, the crop ripening rate, which indicates the ratio of the spectrum characteristically present during the ripening phase, is also used as one of the soil surface parameters. For crops other than rice and wheat, similar processing is possible by including one or more parameters that indicate the ratio of the spectrum characteristically present during the reproductive growth phase as soil surface parameters.

[0055] The crop heading rate is the ratio of crops that have headed to cover the soil surface, and is a parameter that indicates the ratio of the area of ​​the crop's panicle to the area of ​​its leaves. The heading time is represented by a composite spectrum of the leaf spectrum and the panicle spectrum, but the value obtained by subtracting the red reflectance from the green reflectance of the panicle spectrum is greater than the value obtained by subtracting the red reflectance from the green reflectance of the leaf spectrum. Therefore, as the crop heading rate parameter increases, the value obtained by subtracting the red reflectance from the green reflectance in the calculated BOA reflectance becomes larger.

[0056] Crop filling rate is the degree of filling of crops covering the soil surface, and in rice and wheat fields, it is a parameter that indicates the proportion of yellowing of the panicles. The filling period is the period from when photosynthetic products are translocated to the grains of the panicle until they ripen. As filling progresses and the filling period nears its end, the proportion of the yellowed stem / leaf spectrum and yellowed panicle spectrum increases. Nitrogen is translocated to the grains along with the products of photosynthesis. In addition, nitrogen is produced from the decomposition of chlorophyll in the stems, leaves, and panicles, so the chlorophyll concentration decreases as filling progresses. As a result, the stems, leaves, and panicles become predominantly yellow due to carotenoids and turn yellow. Furthermore, the ratio of the yellowing panicle spectrum during the ripening stage may be set for each rice variety, and the candidate ground surface parameter generation unit 12 may be referenced as a constraint condition in generating candidate values ​​for crop ripening rate.

[0057] The ground surface parameter candidate generation unit 12 generates candidate values ​​such that the values ​​of the ground surface parameters are continuous over time. Because crops grow on the ground surface, it is highly likely that the values ​​change continuously over time as growth progresses gradually. Also, changes in soil moisture content change continuously over time. Therefore, by taking this into consideration in the estimation of ground surface parameters and referencing the ground surface parameters at the immediately preceding point in time to create candidate values ​​that are continuous over time, unnecessary calculations of the search space can be omitted, and the computational load (CPU load) and memory usage of the computer can be reduced while efficiently searching for values ​​of ground surface parameters. Furthermore, the configuration for generating candidate values ​​for ground surface parameters based on temporal continuity may be used in combination with the configuration for generating candidate values ​​for atmospheric layer parameters based on spatial continuity, as described later.

[0058] The ground surface parameter candidate generation unit 12 may generate values ​​in appropriate increments. For example, when the ground surface parameter candidate generation unit 12 considers a value between 0.40 and 0.50 as a candidate value, it may first generate candidate values ​​in increments of 0.05, such as {0.40 0.45 0.50}. Furthermore, if the atmospheric correction parameter estimation unit 16, described later, estimates that 0.45 is optimal, the next calculation will generate candidate values ​​in finer increments of 0.01 around 0.45, such as {0.44 0.45 0.46}. The increment size for generating candidate values ​​may decrease as the number of calculations increases. In other words, the ground surface parameter candidate generation unit 12 repeatedly generates multiple new ground surface parameters based on the agreement between multiple ground surface parameters and the degree of agreement calculated by the atmospheric correction parameter estimation unit 16, and the differences between the multiple ground surface parameters generated decrease as the generation is repeated. In this way, by gradually narrowing the difference between candidate values, we can efficiently narrow down the ground surface parameters. In this way, by gradually narrowing the step size (increment) or search range of the generated candidate values, the convergence speed to the optimal solution can be improved and computational resources can be saved. This search process may be performed in combination with known optimization methods such as gradient descent, Bayesian optimization, genetic algorithms, or particle swarm optimization.

[0059] Furthermore, the ground surface parameter candidate generation unit 12 may generate candidate values ​​for ground surface parameters while considering crop growth. Since crops grow over time, the candidate values ​​for ground surface parameters are those that can be taken when the crop has grown more than the ground surface parameters at the previous point in time. With this configuration, by selecting candidate values ​​with a certain degree of accuracy in advance, the values ​​of ground surface parameters can be searched efficiently.

[0060] Furthermore, the ground surface parameter candidate generation unit 12 may generate candidate values ​​by considering the results estimated by a growth simulator of crops growing on the ground surface. The growth simulator, for example, has a growth model for each crop and estimates the state of the crop by applying the amount of solar radiation irradiated to the crop to the growth model. For example, if the amount of solar radiation is relatively high, values ​​that can be taken when the degree of growth is greater will be selected as candidate values. With such a configuration, the values ​​of the ground surface parameters can be searched more efficiently by considering the degree of crop growth.

[0061] Furthermore, the memory unit 20 stores in advance a correspondence between a growth stage, one or more elements that occur in the field area during that growth stage, and the element spectrum of those elements. The ground surface parameter candidate generation unit 12 may then determine the element spectrum to be synthesized based on the growth stage estimated by the growth simulator.

[0062] The BOA reflectance calculation unit 13 is a functional unit that calculates the BOA reflectance from candidate values ​​of ground surface parameters. The BOA reflectance calculation unit 13 estimates the BOA reflectance by synthesizing elemental spectra at a ratio specified by the ground surface parameters.

[0063] Elemental spectra include, for example, crop spectra, stem and leaf spectra, panicle spectra, yellowed panicle spectra, dark leaf spectra, light leaf spectra, yellowed stem and leaf spectra, soil spectra, dry soil spectra, wet soil spectra, and flooded soil spectra.

[0064] A crop spectrum is a spectrum obtained when the observation location is completely covered by the crop. Crop spectra can be synthesized from elemental spectra specific to each crop variety and type (e.g., thin foliage spectrum, dense foliage spectrum, panicle spectrum, yellowed panicle spectrum, and yellowed foliage spectrum).

[0065] A stem and leaf spectrum is the spectrum of a crop in a state consisting only of stems and leaves. Stem and leaf spectra can be synthesized from element spectra specific to each crop variety and type (for example, dark leaf spectrum, light leaf spectrum, and yellowed leaf spectrum).

[0066] A dark leaf colored stem and leaf spectrum represents a state where the crop consists only of stems and leaves, and where the chlorophyll concentration in the stems and leaves (for example, a value normalized to 0-1) is at its highest. The dark leaf spectrum of the stem and leaf is specific to a particular crop variety or species. This spectrum can be obtained in advance from past observations of that crop variety or species.

[0067] A light-colored stem and leaf spectrum represents a state where the crop consists only of stems and leaves, and where the chlorophyll concentration in the stems and leaves (for example, a value normalized to 0-1) is at its lowest. Light-colored stem and leaf spectra are specific to a particular crop variety or species. These spectra can be obtained in advance from past observations of that crop variety or species.

[0068] The yellowed stem and leaf spectrum represents the state in which the carotenoid concentration (for example, a value normalized to 0-1) of the stem and leaves is highest. Yellowing foliage spectra are specific to a particular crop variety or species. They can be obtained in advance from past observations of that crop variety or species.

[0069] The ear spectrum is the spectrum of a crop when the ears have fully emerged. The panicle spectrum is a spectrum specific to a particular crop variety or species. It can be obtained in advance from past observations of that crop variety or species.

[0070] The yellowing panicle spectrum represents the state where yellowing is most advanced after the crop's panicles have fully emerged (for example, when the grains have turned yellow). The yellowing panicle spectrum is a spectrum specific to a particular crop variety or species. It can be obtained in advance from past observations of that crop variety or species.

[0071] Soil spectra are spectra obtained when there is no vegetation at the observation site, and can be synthesized from elemental spectra specific to the observation site (e.g., dry soil spectra, wet soil spectra, flooded soil spectra).

[0072] The dry soil spectrum represents the state in which the surface moisture content of the soil is lowest. Dry soil spectra are unique to the observation location and can be obtained by extracting the spectrum showing the highest reflectance among the past ground surface reflectances at that location.

[0073] The wet soil spectrum represents the state where the surface moisture content of the soil is highest. The wet soil spectrum is a spectrum unique to the observation location and can be obtained by extracting the spectrum showing the lowest reflectance among the past ground surface reflectances at the observation location.

[0074] A flooded soil spectrum is the spectrum of a soil in which there is no vegetation and liquid water is present at the deepest depth on the soil surface. The flooded soil spectrum is a spectrum unique to the observation location and can be obtained by extracting the spectrum with the lowest reflectance at 740 nm compared to the reflectance at 700 nm among the past ground surface reflectances at the observation location.

[0075] The ground surface reflectance (BOA reflectance) can be calculated using the following equation (2), which is the ratio indicated by the soil spectrum and crop spectrum, and the ground surface parameter, vegetation cover. Ground surface reflectance (BOA reflectance) = (1 - vegetation cover) × soil spectrum + vegetation cover × crop spectrum ... (2)

[0076] Crop spectra can be calculated from stem and leaf spectra, panicle spectra, and yellowed panicle spectra, using the ratios indicated by soil surface parameters such as crop heading rate and crop ripening rate.

[0077] If the crop heading rate is 0, the crop spectrum will be the same as the stem and leaf spectrum.

[0078] The stem and leaf spectrum can be calculated using the following equation (3) from the elemental spectra, which are the stem and leaf spectra for dark-colored and light-colored leaves, and the value of the chlorophyll concentration, which is a ground surface parameter. Stem and leaf spectrum = Chlorophyll concentration × concentrated stem and leaf spectrum + (1 - chlorophyll concentration) × dilute stem and leaf spectrum ... (3) Here, to simplify the calculation formula during spectral synthesis, the chlorophyll concentration is normalized so that the variety-specific maximum is 1 and the minimum is 0.

[0079] The crop spectrum can be calculated from the stem and leaf spectrum and the panicle spectrum using the following formula (4) when the crop heading rate is a positive value and the crop ripening rate is 0. Crop spectrum = (1 - crop heading rate) × stem and leaf spectrum + crop heading rate × panicle spectrum ... (4)

[0080] If the crop ripening rate is positive, the crop spectrum can be calculated using the panicle spectrum and the yellowed panicle spectrum by the following formula (5). Crop spectrum = (1 - crop ripening rate) × panicle spectrum + crop ripening rate × yellowed panicle spectrum ... (5)

[0081] When the soil water content is 0, the soil spectrum can be calculated using the following equation (6) from the dry soil spectrum and the wet soil spectrum, using the normalized value of the surface soil moisture content, which is a ground surface parameter. Soil spectrum = (1 - surface soil moisture content) × dry soil spectrum + surface soil moisture content × wet soil spectrum ... (6) Here, to simplify the calculation formula during spectrum synthesis, the soil surface moisture content is normalized so that the maximum moisture content of the soil at the observation site is 1 and the minimum moisture content is 0.

[0082] If the soil water content is a positive value, the soil spectrum can be calculated from the wet soil spectrum and the flooded soil spectrum using the normalized value of the soil water content, which is a surface parameter, by the following equation (7). Soil spectrum = (1 - soil water content) × wet soil spectrum + soil water content × flooded soil spectrum ... (7) Here, to simplify the calculation formula during spectral synthesis, the soil water content is normalized so that the maximum water depth at the observation site is 1 and the minimum water depth is 0.

[0083] In this way, the ground surface reflectance (BOA reflectance) can be calculated by combining elemental spectra (e.g., dark leaf spectrum, light leaf spectrum, panicle spectrum, yellowed panicle spectrum, yellowed leaf spectrum, dry soil spectrum, wet soil spectrum, and flooded soil spectrum, etc.) according to the ratios indicated by ground surface parameters (e.g., vegetation cover rate, crop heading rate, crop ripening rate, chlorophyll concentration, carotenoid concentration, soil surface moisture content, soil flooding content, etc.).

[0084] In other words, the ground surface reflectance (BOA reflectance) can be calculated by hierarchically combining elemental spectra. Specifically, the soil spectrum can be synthesized from dry soil spectra, moist soil spectra, and flooded soil spectra; the foliage spectrum can be synthesized from dark-leaved foliage spectra and light-leaved foliage spectra; and the crop spectrum can be synthesized from foliage spectra, panicle spectra, and yellowed panicle spectra. As a result, the ground surface reflectance (BOA reflectance) can be synthesized from crop spectra and soil spectra.

[0085] The atmospheric layer parameter candidate generation unit 14 is a functional unit that generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer. Atmospheric layer parameters are parameters necessary for calculating the TOA reflectance.

[0086] Light emitted from the sun passes through the atmosphere and enters the Earth's surface. Therefore, by determining atmospheric parameters, the spectrum of sunlight reaching the Earth's surface can be calculated. The spectrum that reaches the surface is reflected by the Earth's surface, passes through the atmosphere again, and is observed by satellite 500. In addition, some of the light emitted from the sun does not reach the Earth's surface but is reflected in the atmosphere and reaches satellite 500. Therefore, by using atmospheric parameters and BOA reflectance to calculate the reflectance of the light reflected back from the Earth's surface, and by using atmospheric parameters to calculate the reflectance of the light reflected in the atmosphere and returning to satellite 500, the TOA reflectance can be calculated by adding these together.

[0087] Atmospheric layer parameters are defined in relation to elements of the atmospheric layer that affect the observed spectrum. Specifically, atmospheric layer parameters include, for example, parameters relating to changes in the backscattering rate of the atmospheric layer, changes in the transmittance of the atmospheric layer, and changes in the wavelength characteristics of the atmospheric layer. Parameters relating to changes in the backscattering rate of the atmospheric layer include, for example, the backscattering rate of aerosols and clouds. Parameters relating to changes in the transmittance of the atmospheric layer include, for example, the optical thickness of aerosols, the optical thickness of clouds, the absorptive rate by water vapor, and the absorptive rate by ozone molecules. Parameters relating to changes in the wavelength characteristics of atmospheric layer transmittance include, for example, the angstrom index of aerosols. Furthermore, the optical thickness of clouds may be defined as different parameters for the optical thickness between the sun and the Earth's surface and the optical thickness between the Earth's surface and the satellite.

[0088] The aerosol backscattering rate is an index that indicates the ratio of light that does not reach the Earth's surface but travels in the opposite direction to the incident light as a result of multiple scattering of incident solar light by aerosols, which are suspended particles in the atmosphere. It can also be called the backscattering coefficient of aerosol scattering. The aerosol backscattering rate is expressed by the following equation (8). Aerosol backscattering rate = Amount of light traveling in the opposite direction to the incident light as a result of scattering by aerosols / Amount of incident light ... (8)

[0089] The angstrom index of an aerosol is an index that shows the wavelength characteristics of the transmittance of suspended particles in the atmosphere. Generally, the amount of scattering by suspended particles is greater at shorter wavelengths and smaller at longer wavelengths. Furthermore, the amount of scattering differs depending on the particle size; for example, for atmospheric molecules that make up the atmosphere, such as nitrogen and oxygen molecules, it is inversely proportional to the fourth power of the wavelength. The angstrom index A has the following relationship with wavelength λ as shown in equation (9). JPEG0007866812000002.jpg1145...(9) α ext is the extinction coefficient of the aerosol, and the amount obtained by integrating the extinction coefficient in the vertical direction of the atmosphere is the optical thickness of the aerosol.A It is a constant. The smaller the size of the suspended particles, the larger the angstrom index; the larger the size, the smaller the angstrom index. If the particles are molecular size, the index is 4; if the particles are the size of water or ice, the index is 0. Since the angstrom index of typical aerosols is known to be around 1.2, the initial value of the parameter can be set to around 1.2.

[0090] The backscattering rate by clouds is an indicator that shows the ratio of light that does not reach the Earth's surface but travels in the opposite direction to the incident light as a result of being scattered by the numerous particles that make up the cloud. The backscattering rate by clouds is expressed by the following equation (10). Backscattering rate by clouds = Amount of light traveling in the opposite direction to the incident light as a result of scattering by clouds / Amount of incident light ... (10)

[0091] The absorptivity due to water vapor is an indicator of the absorption rate of incident light by water vapor in the atmosphere. Although water molecules have their own inherent wavelength characteristics, the absorptivity due to water vapor varies depending on the concentration of water vapor in the atmosphere. Therefore, by considering the absorptivity due to water vapor as a parameter, the TOA reflectance can be estimated more precisely.

[0092] The absorptivity by ozone molecules is an indicator of the absorption rate of incident light by ozone molecules in the atmosphere. Although ozone molecules have their own unique wavelength characteristics, the absorptivity by ozone varies depending on the ozone concentration in the atmosphere. Therefore, by considering the absorptivity by ozone as a parameter, the TOA reflectance can be estimated more precisely.

[0093] The optical thickness of an aerosol is an indicator of the transmittance of incident light through suspended particles in the atmosphere. The optical thickness τ of an aerosol is expressed by the following formula (11) using the transmittance T according to a general calculation method. τ = - ln (T) ···(11)

[0094] The optical thickness of a cloud is an indicator of the transmittance of incident light through the cloud in the atmosphere. The method for calculating the optical thickness of a cloud is the same as for aerosols, and can be determined using equation (11). Note that the scattering amount of ice and water particles that make up the cloud does not depend on the wavelength. Thin clouds with relatively high transmittance, such as stratus clouds, can be accurately estimated to correct for cloud scattering, or in other words, eliminate the cloud's influence, and estimate the ground surface reflectance if their optical thickness can be accurately determined.

[0095] There are three methods for accurately separating aerosols from clouds: using temporal variation, using spatial distribution, and using wavelength characteristics. In the case of satellite data, it is often not possible to obtain time-delay data, so spatial distribution and wavelength characteristics are used. The method using spatial distribution separates aerosols from clouds by taking advantage of the fact that clouds with larger particle sizes have less spatial continuity with respect to aerosols with smaller particle sizes.

[0096] One method that utilizes wavelength characteristics involves identifying particles with a wavelength dependence of around 1.2 (angstrom exponent) as aerosols, and identifying particles with little wavelength dependence (exponent near 0) due to Mie scattering characteristics, where reflection and scattering occur uniformly across all wavelengths, as clouds. By utilizing this difference in physical characteristics, it is possible to accurately separate and estimate aerosols and thin clouds in a mixed state. Furthermore, the atmospheric layer parameter candidate generation unit 14 may change the generation range, step size, or smoothing strength of candidate values ​​for each component, such as imposing strong spatial continuity on candidate values ​​corresponding to aerosols and weak spatial continuity on candidate values ​​corresponding to thin clouds.

[0097] The atmospheric layer parameter candidate generation unit 14 generates candidate values ​​such that the values ​​of atmospheric layer parameters are spatially continuous. That is, for example, the atmospheric layer parameter candidate generation unit 14 may generate candidate values ​​by referring to candidate values ​​of atmospheric layer parameters at nearby locations. When obtaining candidate values ​​at nearby locations, the atmospheric layer parameter candidate generation unit 14 may define an area partitioned by a predetermined size and then refer to candidate values ​​in an area adjacent to the area to be estimated. This area may be, for example, 100 m square. Alternatively, the system may refer to candidate values ​​at a predetermined distance from the point to be estimated.

[0098] The atmospheric layer parameter candidate generation unit 14 may generate candidate values ​​in appropriate increments. For example, when generating candidate values ​​for the optical thickness of an aerosol between 0.11 and 0.13, candidate values ​​may first be generated in increments of 0.01, such as {0.11 0.12 0.13}. Furthermore, if the atmospheric correction parameter estimation unit 16, described later, estimates that 0.12 is optimal, in the next calculation, candidate values ​​in finer increments of 0.005 around 0.12, such as {0.115 0.120 0.125}, are generated as candidate values. The increment size for generating candidate values ​​may decrease as the number of calculations increases. That is, the atmospheric layer parameter candidate generation unit 14 repeatedly generates multiple new atmospheric layer parameters based on multiple atmospheric layer parameters and the degree of agreement calculated by the atmospheric correction parameter estimation unit 16, and the differences between the multiple atmospheric layer parameters generated decrease as the generation is repeated. In this way, by gradually narrowing the difference between candidate values, the number of iterative calculations required to find the optimal solution can be reduced, saving computer computational resources while efficiently narrowing down the atmospheric layer parameters.

[0099] The TOA reflectance calculation unit 15 is a functional unit that calculates the TOA reflectance from the BOA reflectance and candidate values ​​of atmospheric layer parameters. The TOA reflectance calculation unit 15 calculates the TOA reflectance by calculating the radiative transfer function using candidate values ​​of atmospheric layer parameters. In a simplified manner, by approximating the atmospheric layer as a single layer, the TOA reflectance can be calculated by the following equation (12). TOA reflectance = Backscattering rate of the atmosphere + BOA reflectance × (Transmittance of the atmosphere between the sun and the Earth's surface) × (Transmittance of the atmosphere between the Earth's surface and the satellite) ... (12) The radiative transfer function is not limited to the single-layer approximation; it may be calculated using multi-layer atmospheric models, models that consider multiple scattering and absorption, etc. The calculation may be performed by any of the following methods: (i) sequential calculation using a radiative transfer model, (ii) referencing lookup tables (LUTs) for each wavelength band, or (iii) using an approximation formula.

[0100] The atmospheric backscattering rate is the ratio of the amount of solar incident light that, as a result of scattering by the atmosphere, does not reach the Earth's surface but instead passes into space. In this case, the atmospheric backscattering rate is the sum of the backscattering rates by atmospheric molecules such as nitrogen molecules, oxygen molecules, and water vapor, the backscattering rate by aerosols, and the backscattering rate by clouds. The transmittance of the atmospheric layer is determined by e^(-optical thickness of the atmospheric layer) based on the definition of optical thickness. The backscattering rate by atmospheric molecules is uniquely calculated based on the temperature and pressure at the time the observation spectrum is acquired.

[0101] The optical thickness of the atmosphere is the sum of the optical thickness of atmospheric molecules, the optical thickness due to aerosols, and the optical thickness due to clouds. By calculating the transmittance of the atmosphere using the optical thickness between the sun and the Earth's surface, and the optical thickness of the atmosphere between the Earth's surface and the satellite, the transmittance of the atmosphere between the sun and the Earth's surface can be determined.

[0102] The atmospheric correction parameter estimation unit 16 estimates atmospheric correction parameters based on the degree of agreement between the TOA reflectance calculated from multiple candidate values ​​of the ground surface parameters and atmospheric layer parameters and the observed spectrum. That is, the atmospheric correction parameter estimation unit 16 repeatedly calculates the TOA reflectance by changing the ground surface parameters and atmospheric layer parameters in various ways, and calculates the degree of agreement between the calculated TOA reflectance and the observed spectrum. When a combination that spatially maximizes this degree of agreement is found, the ground surface parameters and atmospheric layer parameters in that combination are estimated as atmospheric correction parameters for the target location.

[0103] To calculate the degree of agreement, some or all of the near-infrared reflectance, red reflectance, green reflectance, and blue reflectance are used, depending on the characteristics of the sensor being observed, such as the Satellite-500.

[0104] The degree of agreement is an index that indicates the statistical similarity or distance between the observed spectrum and the TOA reflectance calculated from candidate values ​​of surface parameters and atmospheric layer parameters. Specifically, it is calculated using the root mean square error (RMSE), mean absolute error, or spectral angle mapper of the difference in reflectance in each wavelength band. The atmospheric correction parameter estimation unit 16 searches for the combination that minimizes these error indices or maximizes similarity indices such as the correlation coefficient, and identifies this combination as the one with the maximum degree of agreement.

[0105] Furthermore, when calculating the degree of agreement, the calculation accuracy may be improved based on calculated values ​​for a specific wavelength band or between wavelength bands. For example, the degree of agreement of indicators that characterize atmospheric conditions or vegetation conditions, such as the difference between blue reflectance and green reflectance, or the difference between near-infrared reflectance and red reflectance, may be evaluated. In addition, the degree of agreement may be calculated by multiplying the short-wavelength band (e.g., blue reflectance), which is strongly affected by atmospheric scattering, by a larger weighting coefficient than other wavelength bands. The frequency bands of each wavelength adopted may be any known bands as appropriate.

[0106] Alternatively, the atmospheric correction parameter estimation unit 16 may estimate atmospheric correction parameters by searching for the combination that spatially maximizes the degree of agreement between the TOA reflectance calculated from multiple candidate values ​​of the ground surface parameters and atmospheric layer parameters and the observed spectrum. That is, for example, the atmospheric correction parameter estimation unit 16 may calculate the degree of agreement for each location and then search for the combination that maximizes the sum of the degrees of agreement in a predetermined area.

[0107] By determining the atmospheric correction parameter that maximizes the degree of agreement in this way, estimated values ​​for both the ground surface parameter and the atmospheric layer parameter can be obtained simultaneously. Therefore, for example, the atmospheric correction parameter estimation device 100 can also estimate the optical thickness of aerosols, which is one of the atmospheric layer parameters. In other words, the atmospheric correction parameter estimation device 100 can also be called an aerosol optical thickness estimation device. In this case, the aerosol optical thickness estimation device may output and display the estimated aerosol optical thickness to a user interface device 300 or the like.

[0108] Furthermore, a ground surface reflectance estimation device 200 can also be configured using an atmospheric correction parameter estimation device 100, a leveling unit 101, and a ground surface reflectance estimation unit 102.

[0109] The leveling unit 101 is a functional unit that leveles the atmospheric correction parameters of neighboring regions. For example, atmospheric correction parameters of relatively close regions, such as a 1km square area, are highly likely to be approximately the same. Therefore, the leveling unit 101 changes each atmospheric correction parameter so that the difference between each atmospheric correction parameter of a region within a predetermined range falls within a predetermined range. This process allows for more accurate estimation of atmospheric correction parameters. For example, the leveling unit 101 may update the atmospheric correction parameters based on the mean or median of the estimated values ​​within the predetermined range. Alternatively, the atmospheric correction parameters may be leveled using a smoothing filter such as a Gaussian filter. Furthermore, the atmospheric correction parameters may be leveled by optimizing an objective function that includes a regularization term that reduces the difference between neighboring regions. Alternatively, the atmospheric correction parameters may be leveled by imposing spatial constraints using a Markov random field (MRF) and optimizing an objective function that includes a regularization term that reduces the difference between neighboring regions.

[0110] The ground surface reflectance estimation unit 102 estimates the ground surface reflectance using atmospheric correction parameters estimated by the leveling unit 101 and the observed spectrum. The estimated ground surface reflectance is obtained from the observed spectrum acquired by the spectrum acquisition unit 11, with the effects of atmospheric aerosols and thin clouds removed, thus more accurately reflecting the state of the ground surface. Specifically, the ground surface reflectance estimation unit 102 performs image processing to correct the pixel values ​​of the observed spectrum using the atmospheric correction parameters, generating and outputting high-precision ground surface reflectance image data from which atmospheric effects have been removed. The image data thus generated is stored as electronic information in a storage device or displayed on a display. Therefore, the state of crops growing on the ground surface can be accurately estimated using this ground surface reflectance. Consequently, the type and yield of the crops can be estimated more accurately.

[0111] ● Flowchart for estimating atmospheric correction parameters As shown in Figure 3, first, the spectrum acquisition unit 11 acquires the time-series observed spectrum (step S11). Next, the ground surface parameter candidate generation unit 12 generates candidate values ​​for the ground surface parameters (step S12). Next, the BOA reflectance calculation unit 13 calculates the BOA reflectance using the candidate values ​​for the ground surface parameters (step S13). Next, the atmospheric layer parameter candidate generation unit 14 generates candidate values ​​for the atmospheric layer parameters (step S14). Next, the TOA reflectance calculation unit 15 calculates the TOA reflectance using the BOA reflectance and the candidate values ​​for the atmospheric layer parameters (step S15). Next, the atmospheric correction parameter estimation unit 16 calculates the degree of agreement between the observed spectrum and the TOA reflectance (step S16). Next, the combination of ground surface parameters and atmospheric layer parameters that maximizes the degree of agreement is found, and the atmospheric layer parameters are stored as atmospheric correction parameters (step S17).

[0112] The atmospheric correction parameter estimation device 100 repeats steps S12 to S17 multiple times, repeatedly storing the parameter that yields the highest degree of agreement. Steps S12 to S17 may be repeated for a predetermined number of times, or the process may be stopped when a predetermined condition is met. For example, the process may be stopped when the degree of agreement or the value or change of the atmospheric correction parameter falls within a predetermined range. The obtained atmospheric correction parameters, namely the ground surface parameter and atmospheric layer parameter, are output to and displayed on the user interface device 300.

[0113] (Technically remarkable effects of the present invention) In the atmospheric correction parameter estimation system according to the present invention, the ground surface reflectance can be estimated with high accuracy by correcting for the effects of aerosols and thin clouds in the atmospheric layer from observational spectra obtained by photographing the ground surface from outside the atmospheric layer.

Claims

1. An observation spectrum acquisition unit that acquires observation spectra by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation unit generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, An atmospheric layer parameter candidate generation unit generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer, A BOA reflectance calculation unit calculates the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A TOA reflectance calculation unit calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation unit estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, as an atmospheric correction parameter. Equipped with, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. The ground surface parameter candidate generation unit generates candidate values ​​such that the vegetation cover rate is continuous over time. The aforementioned atmospheric layer parameters include the optical thickness of aerosols, which indicates the transmittance of incident light through suspended particles in the atmosphere. The atmospheric layer parameter candidate generation unit generates candidate values ​​such that the optical thickness of the aerosol is spatially continuous. Characterized by, Atmospheric correction parameter estimation system.

2. An observation spectrum acquisition unit that acquires an observation spectrum of the Earth's surface taken from outside the atmospheric layer, A ground surface parameter candidate generation unit generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, An atmospheric layer parameter candidate generation unit generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer, A BOA reflectance calculation unit calculates the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A TOA reflectance calculation unit calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation unit estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, as an atmospheric correction parameter. Equipped with, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. The ground surface parameter candidate generation unit generates candidate values ​​such that the vegetation cover rate is continuous over time. The aforementioned atmospheric parameters include the backscattering rate due to aerosols, which indicates the ratio of sunlight incident on the Earth's surface that is reflected by suspended particles in the atmosphere without reaching the Earth's surface. The atmospheric layer parameter candidate generation unit generates candidate values ​​such that the backscattering rate of the aerosol is spatially continuous. Characterized by, Atmospheric correction parameter estimation system.

3. The atmospheric correction parameter estimation unit, The degree of agreement between the TOA reflectance calculated from multiple candidate values ​​of the ground surface parameter and the atmospheric layer parameter and the observed spectrum is calculated for each location, and the atmospheric correction parameter is estimated by searching for the combination that maximizes the sum of the degree of agreement in a predetermined region. An atmospheric correction parameter estimation system according to claim 1 or 2.

4. The ground surface parameter candidate generation unit is, Based on the multiple ground surface parameters and the degree of agreement calculated by the atmospheric correction parameter estimation unit, multiple new ground surface parameters are repeatedly generated, and the differences between the multiple generated ground surface parameters decrease with each generation. An atmospheric correction parameter estimation system according to claim 1 or 2.

5. The atmospheric layer parameter candidate generation unit, Based on the multiple atmospheric layer parameters and the degree of agreement calculated by the atmospheric correction parameter estimation unit, a new set of atmospheric layer parameters is repeatedly generated, and the difference between the generated atmospheric layer parameters decreases with each generation. An atmospheric correction parameter estimation system according to claim 1 or 2.

6. The BOA reflectance calculation unit is The BOA reflectance is estimated by synthesizing elemental spectra at the ratio specified by the aforementioned ground surface parameters. An atmospheric correction parameter estimation system according to claim 1 or 2.

7. The TOA reflectance calculation unit is The TOA reflectance is calculated by calculating the radiative transfer function using the aforementioned atmospheric layer parameters. An atmospheric correction parameter estimation system according to claim 1 or 2.

8. The aforementioned ground surface parameters are, This includes chlorophyll concentration, which is the concentration of photosynthetic pigments in plants covering the soil surface. An atmospheric correction parameter estimation system according to claim 1 or 2.

9. The aforementioned ground surface parameters are, This includes the surface moisture content of the soil, which is the amount of moisture in the surface layer of the ground. An atmospheric correction parameter estimation system according to claim 1 or 2.

10. The aforementioned ground surface parameters are, This includes the amount of water inundated on the ground surface, including soil water content. An atmospheric correction parameter estimation system according to claim 1 or 2.

11. The aforementioned ground surface parameters are, This includes the crop heading rate, which is the percentage of crops covering the soil surface that have headed. An atmospheric correction parameter estimation system according to claim 1 or 2.

12. The aforementioned ground surface parameters are, This includes the crop ripening rate, which is the degree of ripening of crops covering the soil surface. An atmospheric correction parameter estimation system according to claim 1 or 2.

13. The aforementioned atmospheric layer parameters are, This includes the optical thickness of the aerosol, which indicates the transmittance of incident light through suspended particles in the atmosphere. An atmospheric correction parameter estimation system according to claim 2.

14. The aforementioned atmospheric layer parameters are, This includes the angstrom index of aerosols, which shows the wavelength characteristics of the transmittance of incident light through suspended particles in the atmosphere. An atmospheric correction parameter estimation system according to claim 1 or 2.

15. The aforementioned atmospheric layer parameters are, This includes the backscattering rate by aerosols, which indicates the ratio of incident solar light that is reflected by suspended particles in the atmosphere before reaching the Earth's surface. An atmospheric correction parameter estimation system according to claim 1.

16. The aforementioned atmospheric layer parameters are, This includes the optical thickness of the cloud between the sun and the Earth's surface, which indicates the transmittance of incident light through clouds in the atmosphere. An atmospheric correction parameter estimation system according to claim 1 or 2.

17. The aforementioned atmospheric layer parameters are, This includes the optical thickness of the cloud between the Earth's surface and the satellite, which indicates the transmittance of incident light through clouds in the atmosphere. An atmospheric correction parameter estimation system according to claim 1 or 2.

18. The aforementioned atmospheric layer parameters are, This includes the backscattering rate by clouds, which indicates the ratio of sunlight incident on the Earth's surface that is reflected by clouds in the atmosphere without reaching the Earth's surface. An atmospheric correction parameter estimation system according to claim 1 or 2.

19. The aforementioned atmospheric layer parameters are, This includes the absorptivity due to water vapor, which indicates the absorptivity of incident light by water vapor in the atmosphere. An atmospheric correction parameter estimation system according to claim 1 or 2.

20. Observation spectrum acquisition step, which involves obtaining an observation spectrum by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation step generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation step generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer. A BOA reflectance calculation step, which calculates the BOA reflectance from the candidate values ​​of the ground surface parameters, A TOA reflectance calculation step, which calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation step is performed, in which the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the above surface parameter and the above atmospheric layer parameter and the observed spectrum is estimated as the atmospheric correction parameter, based on the degree of agreement between the TOA reflectance and the observed spectrum, This is done by a computer, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation step, candidate values ​​are generated such that the vegetation cover rate is continuous over time. The aforementioned atmospheric layer parameters include the optical thickness of aerosols, which indicates the transmittance of incident light through suspended particles in the atmosphere. The atmospheric layer parameter candidate generation step generates candidate values ​​such that the optical thickness of the aerosol is spatially continuous. Method for estimating atmospheric correction parameters.

21. Observation spectrum acquisition command, which acquires observation spectra by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation command generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation command generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer. A command to calculate the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A command for calculating TOA reflectance, which calculates TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation command estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the aforementioned ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, and Have the computer run it, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation command, candidate values ​​are generated so that the vegetation cover rate is continuous over time. The aforementioned atmospheric layer parameters include the optical thickness of aerosols, which indicates the transmittance of incident light through suspended particles in the atmosphere. The atmospheric layer parameter candidate generation command is an atmospheric correction parameter estimation program that generates candidate values ​​such that the optical thickness of the aerosol is spatially continuous.

22. An observation spectrum acquisition unit that acquires observation spectra by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation unit generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, An atmospheric layer parameter candidate generation unit generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer and include at least the optical thickness of aerosols; A BOA reflectance calculation unit calculates the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A TOA reflectance calculation unit calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation unit estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, as an atmospheric correction parameter. An output unit that outputs at least the optical thickness of the aerosol from the estimated atmospheric correction parameters, Equipped with, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. The ground surface parameter candidate generation unit generates candidate values ​​such that the vegetation cover rate is continuous over time. The atmospheric layer parameter candidate generation unit generates candidate values ​​such that the optical thickness of the aerosol is spatially continuous. An optical thickness estimation system for aerosols.

23. Observation spectrum acquisition step, which involves obtaining an observation spectrum by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation step generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation step, which generates multiple candidate values ​​for atmospheric layer parameters that represent the state of the atmospheric layer and include at least the optical thickness of aerosols, A BOA reflectance calculation step, which calculates the BOA reflectance from the candidate values ​​of the ground surface parameters, A TOA reflectance calculation step, which calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation step is performed, in which the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the above surface parameter and the above atmospheric layer parameter and the observed spectrum is estimated as the atmospheric correction parameter, based on the degree of agreement between the TOA reflectance and the observed spectrum, An output step that outputs at least the optical thickness of the aerosol from among the estimated atmospheric correction parameters, This is done by a computer, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation step, candidate values ​​are generated such that the vegetation cover rate is continuous over time. The atmospheric layer parameter candidate generation step is an aerosol optical thickness estimation method that generates candidate values ​​such that the optical thickness of the aerosol is spatially continuous.

24. Observation spectrum acquisition command, which acquires observation spectra by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation command generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation command generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer and include at least the optical thickness of aerosols, A command to calculate the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A command for calculating TOA reflectance, which calculates TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation command estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the aforementioned ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, and Among the estimated atmospheric correction parameters, an output command that outputs at least the optical thickness of the aerosol, Have the computer run it, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation command, candidate values ​​are generated so that the vegetation cover rate is continuous over time. In the command to generate candidate atmospheric layer parameters, candidate values ​​are generated such that the optical thickness of the aerosol is spatially continuous. A program for estimating the optical thickness of aerosols.

25. An observation spectrum acquisition step, which involves acquiring an observation spectrum obtained by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation step generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation step generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer. A BOA reflectance calculation step, which calculates the BOA reflectance from the candidate values ​​of the ground surface parameters, A TOA reflectance calculation step, which calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation step is performed, in which the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the above surface parameter and the above atmospheric layer parameter and the observed spectrum is estimated as the atmospheric correction parameter, based on the degree of agreement between the TOA reflectance and the observed spectrum, This is done by a computer, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation step, candidate values ​​are generated such that the vegetation cover rate is continuous over time. The aforementioned atmospheric parameters include the backscattering rate due to aerosols, which indicates the ratio of sunlight incident on the Earth's surface that is reflected by suspended particles in the atmosphere without reaching the Earth's surface. In the atmospheric layer parameter candidate generation step, candidate values ​​are generated such that the backscattering rate of the aerosol is spatially continuous. Method for estimating atmospheric correction parameters.

26. An observation spectrum acquisition command for acquiring an observation spectrum of the Earth's surface photographed from outside the atmospheric layer, A ground surface parameter candidate generation command generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation command generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer. A command to calculate the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A command for calculating TOA reflectance, which calculates TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation command estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the aforementioned ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, and Have the computer run it, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation command, candidate values ​​are generated so that the vegetation cover rate is continuous over time. The aforementioned atmospheric parameters include the backscattering rate due to aerosols, which indicates the ratio of sunlight incident on the Earth's surface that is reflected by suspended particles in the atmosphere without reaching the Earth's surface. In the command to generate candidate atmospheric layer parameters, candidate values ​​are generated such that the backscattering rate of the aerosol is spatially continuous. Atmospheric correction parameter estimation program.

27. ​​An observation spectrum acquisition unit that acquires an observation spectrum of the Earth's surface taken from outside the atmospheric layer, A ground surface parameter candidate generation unit generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, An atmospheric layer parameter candidate generation unit generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer and include at least the optical thickness of aerosols; A BOA reflectance calculation unit calculates the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A TOA reflectance calculation unit calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation unit estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, as an atmospheric correction parameter. An output unit that outputs at least the optical thickness of the aerosol from the estimated atmospheric correction parameters, Equipped with, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. The ground surface parameter candidate generation unit generates candidate values ​​such that the vegetation cover rate is continuous over time. The aforementioned atmospheric parameters include the backscattering rate due to aerosols, which indicates the ratio of sunlight incident on the Earth's surface that is reflected by suspended particles in the atmosphere without reaching the Earth's surface. The atmospheric layer parameter candidate generation unit generates candidate values ​​such that the backscattering rate of the aerosol is spatially continuous. An optical thickness estimation system for aerosols.

28. An observation spectrum acquisition step, which involves acquiring an observation spectrum obtained by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation step generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation step, which generates multiple candidate values ​​for atmospheric layer parameters that represent the state of the atmospheric layer and include at least the optical thickness of aerosols, A BOA reflectance calculation step, which calculates the BOA reflectance from the candidate values ​​of the ground surface parameters, A TOA reflectance calculation step, which calculates the TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation step is performed, in which the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the above surface parameter and the above atmospheric layer parameter and the observed spectrum is estimated as the atmospheric correction parameter, based on the degree of agreement between the TOA reflectance and the observed spectrum, An output step that outputs at least the optical thickness of the aerosol from among the estimated atmospheric correction parameters, This is done by a computer, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation step, candidate values ​​are generated such that the vegetation cover rate is continuous over time. The aforementioned atmospheric parameters include the backscattering rate due to aerosols, which indicates the ratio of sunlight incident on the Earth's surface that is reflected by suspended particles in the atmosphere without reaching the Earth's surface. In the atmospheric layer parameter candidate generation step, candidate values ​​are generated such that the backscattering rate of the aerosol is spatially continuous. A method for estimating the optical thickness of an aerosol.

29. An observation spectrum acquisition command for acquiring an observation spectrum obtained by photographing the Earth's surface from outside the atmospheric layer, A ground surface parameter candidate generation command generates multiple candidate values ​​for ground surface parameters, which are variables representing the state of the ground surface, and An atmospheric layer parameter candidate generation command generates multiple candidate values ​​for atmospheric layer parameters, which are variables representing the state of the atmospheric layer and include at least the optical thickness of aerosols, A command to calculate the BOA reflectance from candidate values ​​of the aforementioned ground surface parameters, A command for calculating TOA reflectance, which calculates TOA reflectance from the BOA reflectance and candidate values ​​of the atmospheric layer parameters, An atmospheric correction parameter estimation command estimates the atmospheric layer parameter that maximizes the agreement between the TOA reflectance calculated from the aforementioned ground surface parameter and the atmospheric layer parameter and the observed spectrum, based on the degree of agreement between the TOA reflectance and the observed spectrum, and Among the estimated atmospheric correction parameters, an output command that outputs at least the optical thickness of the aerosol, Have the computer run it, The aforementioned ground surface parameters include the vegetation cover rate, which is the percentage of the ground surface covered by vegetation. In the ground surface parameter candidate generation command, candidate values ​​are generated so that the vegetation cover rate is continuous over time. The aforementioned atmospheric parameters include the backscattering rate due to aerosols, which indicates the ratio of sunlight incident on the Earth's surface that is reflected by suspended particles in the atmosphere without reaching the Earth's surface. In the command to generate candidate atmospheric layer parameters, candidate values ​​are generated such that the backscattering rate of the aerosol is spatially continuous. A program for estimating the optical thickness of aerosols.

Images