A remote sensing method for methane emission detection based on hyperspectral imaging
By combining hyperspectral imaging technology with the coordinated processing of terrain and wind field data, the accuracy problem of remote sensing detection of methane emissions under complex terrain and non-uniform wind fields was solved, and stable inversion of the location and rate of methane emissions was achieved.
Patent Information
- Application Number
- CN202611620336.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-20
- Publication Date
- 2026-11-17
AI Technical Summary
Existing remote sensing methods for methane emissions struggle to accurately retrieve the location and rate of methane emission sources under complex terrain and non-uniform wind field conditions, leading to unstable detection results. In particular, under conditions of terrain variation and non-uniform wind field, the direction of the plume slice is inconsistent with the transport direction, resulting in distorted concentration distribution and missed emission sources.
By collaboratively processing hyperspectral radiance cubes, solar observation geometric parameters, surface elevation data, and synchronous wind field data, inter-band spectral curvature correction, radiometric calibration, methane absorption feature extraction, atmospheric path length layering and stripping, plume direction slice reconstruction, and source intensity inversion are performed to generate methane emission locations and rates.
Under complex terrain and non-uniform wind field conditions, it improves the accuracy and stability of methane emission detection, and can effectively identify weak emission sources and deduce accurate emission locations and rates.
Smart Images

Figure CN123111995A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of remote sensing detection technology, specifically to a remote sensing detection method for methane emissions based on hyperspectral imaging. Background Technology
[0002] Existing hyperspectral remote sensing methods for methane emission detection generally utilize spaceborne, airborne, or ground-based hyperspectral imagers to acquire radiance data of the target area. After radiometric calibration and atmospheric correction, the absorption characteristics of methane in the shortwave infrared band are extracted. The methane column concentration or enhancement value distribution is obtained through methods such as absorption depth, equivalent width, spectral angle, or matched filtering. For the emission source location and rate, existing methods mostly use empirical formulas combining concentration distribution with wind speed, Gaussian diffusion models, or mass balance methods for estimation. Surface elevation data for complex terrain areas are sometimes incorporated into the atmospheric correction process, but usually a uniform correction is made using the scene-wide average elevation or standard atmospheric path; synchronous wind field data are also mostly used as input parameters for diffusion simulations or only for wind speed scaling of inversion results. Variations in surface elevation and solar observation geometry in complex terrain regions cause significant differences in the lengths of incident and observation light paths between pixels. The same methane concentration exhibits different absorption valley depths and half-widths (HWHM) under different elevation layers and geometric conditions. Even after unified scene correction, the absorption intensity still includes variations caused by terrain path length, leading to shifts in high-concentration methane column areas and missed detections of weak emission sources. Existing plume slicing methods generally divide the plume according to geographic north, fixed grids, or image row and column directions, without establishing downwind and crosswind directions based on the synchronous wind field. Under non-uniform wind field conditions, the principal axis of the plume concentration gradient does not coincide with the slice direction, distorting the concentration distribution within the slice. The centroid displacement and area changes of isoconcentration closed boundaries cannot accurately reflect the transport process, resulting in unstable source location and emission rate inversion results. The deviation in methane absorption intensity caused by differences in atmospheric path length under complex terrain has not been effectively suppressed, and the deviation between the plume slice direction and the transport direction under non-uniform wind fields has not been corrected, affecting the accuracy of methane emission remote sensing detection in complex scenes in terms of location and rate inversion. Summary of the Invention
[0003] This invention aims to provide a method for remote sensing detection of methane emissions based on hyperspectral imaging. By co-processing hyperspectral radiance cubes, solar observation geometric parameters, surface elevation data, and synchronous wind field data, it performs inter-band spectral curvature correction, radiometric calibration, methane absorption feature extraction, atmospheric path length layer stripping, plume direction slice reconstruction, and source intensity inversion to obtain the methane emission location and rate, thereby reducing the interference of complex terrain and non-uniform wind fields on methane remote sensing detection.
[0004] To achieve the above objectives, the present invention provides the following technical solution: a remote sensing method for methane emission detection based on hyperspectral imaging, comprising:
[0005] Collect hyperspectral radiance cubes, solar observation geometric parameters, surface elevation data, and synchronous wind field data for the target area;
[0006] Inter-band spectral curvature correction and radiometric calibration are performed on the hyperspectral radiance cube to generate an absolute radiance spectral sequence;
[0007] Extract the methane absorption characteristic band set from the absolute radiance spectral sequence and construct the absorption valley morphology parameters;
[0008] Based on the solar observation geometric parameters and the surface elevation data, atmospheric path length layering and stripping are performed on the absorption valley morphology parameters to generate path-corrected absorption intensity.
[0009] Based on the synchronous wind field data, the path-corrected absorption intensity is reconstructed by plume direction slices to generate methane plume concentration slices.
[0010] Source intensity inversion is performed on the methane plume concentration slice to output the methane emission location and methane emission rate.
[0011] Preferably, the hyperspectral radiance cube undergoes inter-band spectral curvature correction and radiometric calibration to generate an absolute radiance spectral sequence, including:
[0012] Read the dark current frame and gain parameters of the hyperspectral radiance cube, replace the bad pixels according to the spectral continuity of the neighborhood, and generate a repair radiance cube;
[0013] The spectral curvature between adjacent narrow bands is calculated for the repaired radiance cube band by band, and the inter-band linear stretching is performed according to the spectral curvature to generate a flat field radiance cube.
[0014] The irradiance of the flat-field radiance cube is normalized based on the solar zenith angle in the solar observation geometric parameters to generate an absolute radiance spectral sequence.
[0015] Preferably, a set of methane absorption characteristic bands is extracted from the absolute radiance spectral sequence to construct absorption valley morphology parameters, including:
[0016] Multiple narrow bands within the methane absorption band are selected from the absolute radiance spectral sequence to form a methane absorption characteristic band group;
[0017] A local continuum is fitted pixel by pixel to the methane absorption characteristic band group, and the methane absorption residual spectrum is obtained after subtracting the local continuum.
[0018] Valley depth, half-width at half-maximum (HWHM), valley bottom asymmetry, and peak position shift are extracted from the methane absorption residual spectrum to form absorption valley morphology parameters.
[0019] Preferably, the atmospheric path length layering and stripping of the absorption valley morphology parameters based on the solar observation geometric parameters and the surface elevation data are performed to generate path-corrected absorption intensity, including:
[0020] The target area is divided into multiple elevation layers based on the surface elevation data, and the optical path projection length of each elevation layer is calculated according to the solar observation geometric parameters.
[0021] The valley depth and half-width at half-maximum in the absorption valley morphology parameters are decomposed layer by layer according to the optical path projection length to generate layered absorption components.
[0022] The overlapping contribution of the water vapor absorption band and the carbon dioxide absorption band is removed from the layered absorption component to generate path-corrected absorption intensity.
[0023] Preferably, the path-corrected absorption intensity is reconstructed into plume direction slices based on the synchronous wind field data to generate methane plume concentration slices, including:
[0024] Based on the synchronous wind field data, establish the downwind and crosswind directions, and project the path-corrected absorption intensity onto the downwind and crosswind directions to generate the wind field coordinate system absorption intensity;
[0025] The absorption intensity of the wind field coordinate system is sliced at equal intervals along the crosswind direction to generate a crosswind absorption profile.
[0026] The crosswind absorption profiles are sequentially spliced along the downwind direction to generate methane plume concentration slices.
[0027] Preferably, source intensity inversion is performed on the methane plume concentration slice to output the methane emission location and methane emission rate, including:
[0028] Isoconcentration closed boundaries are extracted from the methane plume concentration slices, and the centroid displacement of adjacent isoconcentration closed boundaries is calculated;
[0029] The transport time of the isoconcentration closed boundary is determined based on the wind speed and centroid displacement in the synchronous wind field data.
[0030] The methane emission location and methane emission rate are output based on the transport time and the area change of the isoconcentration closed boundary.
[0031] Preferably, the method involves calculating the spectral curvature between adjacent narrow bands for each band of the repaired radiance cube, and performing inter-band linear stretching according to the spectral curvature to generate a flat-field radiance cube, including:
[0032] In the repair radiance cube, select adjacent narrow band pairs with no methane absorption and calculate the spectral ratio curve of the adjacent narrow band pairs.
[0033] The spectral ratio curve is segmented and median filtered to generate inter-band spectral curvature.
[0034] The inter-band spectral curvature is applied to each band of the repaired radiance cube to generate a flat-field radiance cube.
[0035] Preferably, the methane absorption characteristic band group is fitted with a local continuum pixel by pixel, and the methane absorption residual spectrum is obtained after subtracting the local continuum, including:
[0036] Multiple continuum anchor bands are selected on both sides of the methane absorption characteristic band group, and the multiple continuum anchor bands are segmented and linearly connected to generate a local continuum.
[0037] The relative absorption spectrum is generated by dividing each band value in the methane absorption characteristic band group by the continuum value of the local continuum at the corresponding band position.
[0038] The relative absorption spectrum is zero-mean-normalized to obtain the methane absorption residual spectrum.
[0039] Preferably, the target area is divided into multiple elevation layers based on the surface elevation data, and the optical path projection length of each elevation layer is calculated according to the solar observation geometric parameters, including:
[0040] The target area is divided into multiple elevation layers according to the elevation contour lines in the surface elevation data;
[0041] Obtain the solar zenith angle, observation zenith angle, and relative azimuth angle from the solar observation geometric parameters;
[0042] The incident light path direction and the reflected light path direction of each elevation layer are determined based on the solar zenith angle, the observed zenith angle, and the relative azimuth angle.
[0043] The path length of the incident light path and the path length of the reflected light path within the elevation layer are added together to obtain the projected light path length.
[0044] Preferably, the method involves establishing the downwind and crosswind directions based on the synchronized wind field data, and projecting the path-corrected absorption intensity onto the downwind and crosswind directions to generate the wind field coordinate system absorption intensity, including:
[0045] The wind vectors in the synchronous wind field data are averaged over time to generate an average wind vector;
[0046] The direction of the average wind vector is determined as the downwind direction, and the downwind direction is vertically rotated to obtain the crosswind direction.
[0047] A wind field coordinate system is established using the downwind direction and the crosswind direction as coordinate axes;
[0048] The position of each pixel in the path-corrected absorption intensity is transformed to the wind field coordinate system to generate the wind field coordinate system absorption intensity.
[0049] The technical effects and advantages provided by the present invention in the above technical solution are as follows:
[0050] A layered stripping method based on atmospheric path length from solar observation geometry and surface elevation data was used to correct the morphological parameters of absorption valleys. The target area was divided into multiple elevation layers according to the elevation contour lines in the surface elevation data. The incident and reflected light paths for each elevation layer were determined based on the solar zenith angle, observation zenith angle, and relative azimuth angle. The path lengths of the incident and reflected light paths within each elevation layer were summed to obtain the projected light path length. The valley depth and half-width at half-maximum (HWHM) of the absorption valley morphological parameters were decomposed layer by layer according to the projected light path length, generating layered absorption components. The overlapping contribution of the water vapor and carbon dioxide absorption bands was removed from these layered absorption components to obtain the path-corrected absorption intensity. After this layered stripping, the distortions in absorption depth and width caused by elevation and observation geometry variations under complex terrain conditions were reduced, and the correspondence between the path-corrected absorption intensity and methane column concentration became more stable. Weak emission plumes could still be effectively identified under undulating surface conditions such as mountainous and hilly areas. Synchronous wind field data was used to reconstruct the plume direction slices of the path-corrected absorption intensity. The wind vectors in the synchronous wind field data are time-averaged, and the direction of the average wind vector is taken as the downwind direction. The downwind direction is then vertically rotated to obtain the crosswind direction, establishing a wind field coordinate system. Each pixel position in the path-corrected absorption intensity is transformed to the wind field coordinate system, generating the wind field coordinate system absorption intensity. The wind field coordinate system absorption intensity is equidistantly sliced along the crosswind direction to generate a crosswind absorption profile. These crosswind absorption profiles are then sequentially stitched together along the downwind direction to generate a methane plume concentration slice. In the wind field coordinate system, the plume slice direction is consistent with the atmospheric transport direction. The crosswind profile retains the concentration gradient structure of the plume in the vertical transport direction, and the slices stitched along the downwind direction reflect the diffusion and dilution process of the plume along the transport path. Based on this slice, isoconcentration closed boundaries are extracted, and the displacement of adjacent centroids is calculated. The transport time is determined based on wind speed and centroid displacement. Finally, the methane emission location and emission rate are output by combining the change in the area of the isoconcentration closed boundaries, making the source location and rate inversion results under non-uniform wind field conditions more stable. Attached Figure Description
[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.
[0052] Figure 1 This is a flowchart of a remote sensing method for methane emission detection based on hyperspectral imaging;
[0053] Figure 2 This is a flowchart of the extraction process for methane absorption valley morphology parameters;
[0054] Figure 3 This is a flowchart of a method for generating methane plume concentration slices based on a wind field coordinate system. Detailed Implementation
[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0056] See Figure 1 This invention provides a remote sensing method for methane emission detection based on hyperspectral imaging, comprising: acquiring hyperspectral radiance cubes, solar observation geometric parameters, surface elevation data, and synchronous wind field data of a target area; performing inter-band spectral curvature correction and radiometric calibration on the hyperspectral radiance cubes to generate an absolute radiance spectral sequence; extracting methane absorption characteristic band groups from the absolute radiance spectral sequence to construct absorption valley morphology parameters; performing atmospheric path length layering and stripping on the absorption valley morphology parameters based on the solar observation geometric parameters and the surface elevation data to generate path-corrected absorption intensities; performing plume direction slicing and recombining on the path-corrected absorption intensities based on the synchronous wind field data to generate methane plume concentration slices; and performing source intensity inversion on the methane plume concentration slices to output the methane emission location and methane emission rate.
[0057] In practice, a hyperspectral radiance cube of the target area acquired by a hyperspectral imager is obtained. Simultaneously, the dark current frame and gain parameters corresponding to the hyperspectral radiance cube are read. The dark current frame is the dark signal image acquired by the hyperspectral imager under shading conditions, and the gain parameters are the gain coefficients of the detectors in each band of the hyperspectral imager. The radiance value of each pixel in the hyperspectral radiance cube is subtracted from the corresponding pixel value in the dark current frame. Then, the radiance value after subtracting the dark current frame is divided by the corresponding pixel value in the gain parameters, completing dark current subtraction and gain correction. Bad pixel identification is performed on the hyperspectral radiance cube after dark current subtraction and gain correction. The bad pixel identification method is as follows: the difference between the radiance values of adjacent narrow bands at the same spatial location in the hyperspectral radiance cube after dark current subtraction and gain correction is calculated along the spectral dimension. Pixels at spatial locations where the difference between adjacent narrow band radiance values is greater than a set threshold are marked as bad pixels. For spatially located pixels marked as bad pixels, two narrow-band pixels adjacent to the bad pixels and not marked as bad pixels are selected along the spectral dimension. The radiance values of the two narrow-band pixels are used to perform linear interpolation on the bad pixels. The linear interpolation result is used as the replacement radiance value of the bad pixels to generate a repair radiance cube.
[0058] In practice, adjacent narrow band pairs without methane absorption are selected within the restored radiance cube. Each adjacent narrow band pair consists of multiple adjacent narrow bands located outside the methane absorption band. The radiance values of the two narrow bands in each adjacent narrow band pair are divided spatially to obtain a spectral ratio curve. This spectral ratio curve is then divided into multiple continuous, non-overlapping segments along the band direction. The median of the spectral ratio within each segment is taken, and the medians of all segments are arranged in band order to generate the inter-band spectral curvature.
[0059] In practice, the inter-band spectral curvature is applied to each band of the restored radiance cube. The radiance value of each band in the restored radiance cube is then linearly stretched along the spectral dimension to correct the spectral curvature between adjacent narrow bands, generating a flat-field radiance cube. The linear stretching method involves multiplying the radiance value of each band in the restored radiance cube by the inter-band spectral curvature corresponding to the band to be stretched, resulting in the stretched flat-field radiance cube.
[0060] In practice, the solar zenith angle is extracted from the solar observation geometric parameters. The solar zenith angle is the angle between the solar incident direction and the zenith direction of the target area. The irradiance is normalized by multiplying the radiance value of each pixel in the flat-field radiance cube by the reciprocal of the cosine of the solar zenith angle, thus generating an absolute radiance spectral sequence.
[0061] In specific implementation, please refer to Figure 2Multiple narrow bands within the methane absorption band are selected from the absolute radiance spectral sequence to form a methane absorption characteristic band group. The methane absorption band is located in the short-wave infrared region, and the methane absorption characteristic band group contains continuous narrow bands covering the methane absorption valley. The spacing between these narrow bands is determined by the spectral sampling interval of the hyperspectral imager. Each narrow band in the methane absorption characteristic band group corresponds to a center wavelength position, and all narrow bands are arranged in wavelength order.
[0062] In practice, a local continuum is fitted pixel-by-pixel to the methane absorption characteristic band group. Multiple continuum anchor bands are selected on both sides of the methane absorption characteristic band group; these anchor bands are those where no methane absorption occurs outside the two ends of the methane absorption characteristic band group. Each pair of adjacent continuum anchor bands is linearly connected to form a piecewise linearly connected continuum curve, which constitutes a local continuum. Specifically, a first continuum anchor band group and a second continuum anchor band group are selected on both sides of the wavelength range where the methane absorption characteristic band group is located. The first continuum anchor band group is located on the short-wave side of the methane absorption characteristic band group, and the second continuum anchor band group is located on the long-wave side. Linear interpolation is performed on each anchor band in the first continuum anchor band group and each corresponding anchor band in the second continuum anchor band group to obtain the local continuum value corresponding to each band position in the methane absorption characteristic band group.
[0063] In practice, the relative absorption spectrum is generated by dividing each band value in the methane absorption characteristic band group by the continuum value of the local continuum at the corresponding band position. Each value in the relative absorption spectrum is the ratio of the band value in the methane absorption characteristic band group to the continuum value of the local continuum at the corresponding band position. Zero-mean normalization is then applied to the relative absorption spectrum to obtain the methane absorption residual spectrum. The zero-mean normalization method is as follows: calculate the average value of all values in the relative absorption spectrum, and subtract the average value from each value in the relative absorption spectrum to obtain the methane absorption residual spectrum.
[0064] In practice, valley depth, full width at half maximum (FWHM), valley asymmetry, and peak shift are extracted from the methane absorption residual spectrum to form absorption valley morphology parameters. Valley depth is extracted by searching for the lowest point of the absorption valley in the methane absorption residual spectrum and taking the absolute value of the corresponding value as the valley depth. FWHM is extracted by finding positions in the methane absorption residual spectrum where the value reaches half the valley depth in both the short-wavelength and long-wavelength directions from the lowest point of the absorption valley, and taking the wavelength interval between the short-wavelength and long-wavelength half-value positions as the FWHM. Valley asymmetry is extracted by the ratio of the absolute value of the maximum slope from the lowest point of the absorption valley in the short-wavelength direction to the absolute value of the maximum slope in the long-wavelength direction in the methane absorption residual spectrum as the valley asymmetry. Peak shift is extracted by determining the band position of the lowest point of the absorption valley in the methane absorption residual spectrum and taking the difference between the band position of the lowest point of the methane absorption residual spectrum and the position of the standard methane absorption peak as the peak shift.
[0065] In practice, valley depth, half-width at half-maximum, valley bottom asymmetry, and peak position shift are combined into absorption valley morphology parameters, which are used to describe the morphological characteristics of absorption valleys in the methane absorption residual spectrum pixel by pixel.
[0066] In practice, the target area is divided into multiple elevation layers based on surface elevation data. The surface elevation data consists of elevation raster data covering the target area, with each raster cell containing a surface elevation value. The target area is further divided into multiple elevation layers according to the elevation contour lines in the surface elevation data, which are contour lines with the same elevation value. When dividing the elevation layers, elevation intervals are set, and the elevation range from the minimum to the maximum surface elevation in the target area is divided according to these intervals. Each elevation interval forms an elevation layer, and surface elevation values within the same elevation layer fall within the corresponding elevation interval. Each elevation layer contains all raster cells in the target area whose surface elevation values fall within that elevation interval.
[0067] In practical implementation, the solar zenith angle, observation zenith angle, and relative azimuth angle are obtained from the solar observation geometric parameters. The solar zenith angle is the angle between the solar incident direction and the zenith direction of the target area; the observation zenith angle is the angle between the hyperspectral imager observation direction and the zenith direction of the target area; and the relative azimuth angle is the angle between the projections of the solar incident direction and the hyperspectral imager observation direction onto the horizontal plane of the target area. The incident and reflected light path directions for each elevation layer are determined based on the solar zenith angle, observation zenith angle, and relative azimuth angle. The incident light path direction is determined by the solar zenith angle and solar azimuth angle, while the reflected light path direction is determined by the observation zenith angle and observation azimuth angle. The solar azimuth angle and observation azimuth angle are obtained from the solar observation geometric parameters. Both the incident and reflected light path directions are direction vectors in three-dimensional space.
[0068] In practice, the path lengths of the incident and reflected light paths within each elevation layer are added together to obtain the projected light path length. The path length of the incident light path within an elevation layer is the spatial distance traveled by the incident solar rays as they pass through that elevation layer, and the path length of the reflected light path within an elevation layer is the spatial distance traveled by the reflected light received by the hyperspectral imager as it passes through that elevation layer. The path lengths are determined based on the elevation layer thickness, solar zenith angle, observation zenith angle, and relative azimuth angle.
[0069] In practice, the valley depth and half-width at half-maximum (HWHM) parameters of the absorption valley morphology are decomposed layer by layer according to the optical path projection length to generate layered absorption components. The layer-by-layer decomposition is performed using the following formula:
[0070]
[0071] in, Indicates the first Layered absorption components of each elevation layer Indicates the first Valley depth in the morphological parameters of absorption valleys at different elevation levels Indicates the first The half-width at half-maximum (WHM) of the absorption valley morphology parameters of each elevation layer Indicates the first The optical path projection length of each elevation layer Elevation level numbering, The range of values is arrive , This represents the total number of elevation levels. The method of obtaining it is from the absorption valley morphology parameters according to the first... Valley depth is extracted from the spatial location corresponding to each elevation level. The method of obtaining it is from the absorption valley morphology parameters according to the first... Extract the half-width at half-height (WHM) of the spatial location corresponding to each elevation level. The method of obtaining it is to calculate the first [value] based on solar observation geometric parameters and surface elevation data. The optical path projection length of each elevation layer. and The product of is the first The product of the absorption valley intensity and wavelength width at each elevation level For the first The stratified absorption components of each elevation layer.
[0072] In practice, the above-mentioned layer-by-layer decomposition formula is applied to all elevation layers one by one. The valley depth and half-width in the absorption valley morphology parameters are decomposed layer by layer according to the optical path projection length to generate the layered absorption components corresponding to each elevation layer. The layered absorption components corresponding to all elevation layers are combined to form the layered absorption components.
[0073] In practice, the overlap contribution of the water vapor absorption band and the carbon dioxide absorption band is removed from the stratified absorption component to generate the path-corrected absorption intensity. The water vapor absorption band corresponds to one set of known center wavelength ranges, and the carbon dioxide absorption band corresponds to another set of known center wavelength ranges. Values within the center wavelength range corresponding to the water vapor absorption band are extracted from the stratified absorption component, and the average value of these values is calculated. This average value is taken as the water vapor overlap contribution. Similarly, values within the center wavelength range corresponding to the carbon dioxide absorption band are extracted from the stratified absorption component, and the average value of these values is calculated. This average value is taken as the carbon dioxide overlap contribution. The overlap contribution of both the water vapor and carbon dioxide overlap contributions is subtracted from the values in the overlapping region of the stratified absorption component that lies within both the center wavelength ranges of the water vapor and carbon dioxide absorption bands to obtain the path-corrected absorption intensity. Finally, the water vapor overlap contribution is subtracted from the values in the stratified absorption component that lie within the center wavelength range of the water vapor absorption band but not within the center wavelength range of the carbon dioxide absorption band to obtain the path-corrected absorption intensity. The path-corrected absorption intensity is obtained by subtracting the carbon dioxide overlap contribution from the values of the layered absorption components that fall within the center wavelength range of the carbon dioxide absorption band but not within the center wavelength range of the water vapor absorption band. The removal of water vapor and carbon dioxide overlap contributions is performed pixel-by-pixel to generate the path-corrected absorption intensity of all pixels in the target area.
[0074] In specific implementation, please refer to Figure 3 To obtain all wind vectors from the synchronous wind field data, the synchronous wind field data contains wind vector observations at multiple times. Each wind vector observation consists of an eastward wind speed component and a northward wind speed component. The eastward wind speed components at all times in the synchronous wind field data are summed to obtain the total eastward wind speed component. This total eastward wind speed component is then divided by the total number of wind vector observation times in the synchronous wind field data to obtain the average eastward wind speed component. Similarly, the northward wind speed components at all times in the synchronous wind field data are summed to obtain the total northward wind speed component. This total northward wind speed component is then divided by the total number of wind vector observation times in the synchronous wind field data to obtain the average northward wind speed component. The average eastward wind speed component and the average northward wind speed component together form the average wind vector. The average wind vector is calculated using the following formula:
[0075]
[0076] in, Represents the average wind vector. It includes the average eastward wind speed component and the average northward wind speed component; This indicates the total number of wind vector observation times included in the synchronous wind field data. It is a positive integer; Number the wind vector observation times. Take from arrive All integers; Indicates the first Wind vector at any given moment, Including the The eastward and northward wind speed components at each moment.
[0077] In practice, the direction of the average wind vector is determined as the downwind direction. The direction of the average wind vector is determined by the signs of the average eastward and average northward wind speed components. The crosswind direction is obtained by rotating the downwind direction vertically by 90 degrees. The vertical rotation direction is specified as clockwise, and the direction obtained after rotating the downwind direction 90 degrees clockwise is the crosswind direction. A wind field coordinate system is established with the downwind and crosswind directions as coordinate axes. The origin of the wind field coordinate system is selected as the position of the center of the lower left pixel of the path correction absorption intensity coverage area in the target region in the original image coordinates. The downwind direction is the positive direction of the vertical axis of the wind field coordinate system, and the crosswind direction is the positive direction of the horizontal axis of the wind field coordinate system.
[0078] In practice, the position of each pixel in the path correction absorption intensity is transformed to the wind field coordinate system. Each pixel in the path correction absorption intensity has an original x-coordinate and an original y-coordinate in the original image coordinate system. The x-coordinate and y-coordinate of the origin of the wind field coordinate system in the original image coordinate system are obtained, and the angle between the downwind direction and the x-axis of the original image coordinate system is obtained. For the _th_ pixel in the path correction absorption intensity... The first pixel will be the first pixel. Subtracting the x-coordinate of the origin from the original x-coordinate of the nth pixel yields the x-coordinate of the nth pixel. The difference in the x-coordinate between the pixel and the origin of the wind field coordinate system; [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] Subtracting the origin's ordinate from the original ordinate of the nth pixel yields the nth pixel's ordinate. The difference between the ordinate of the first pixel and the origin of the wind field coordinate system. Multiplying the difference between the x-coordinate of the first pixel and the origin of the wind field coordinate system by the cosine of the angle between the downwind direction and the x-axis of the original image coordinate system, and adding the first pixel's x-coordinate... Multiplying the difference between the ordinate of the first pixel and the origin of the wind field coordinate system by the sine of the angle between the downwind direction and the horizontal axis of the original image coordinate system, we obtain the th pixel. The crosswind direction coordinates of the first pixel in the wind field coordinate system. Multiply the difference between the x-coordinate of the first pixel and the origin of the wind field coordinate system by the sine of the angle between the downwind direction and the x-axis of the original image coordinate system, take the opposite value, and add the first pixel's x-coordinate. Multiplying the difference between the ordinate of the pixel and the origin of the wind field coordinate system by the cosine of the angle between the downwind direction and the horizontal axis of the original image coordinate system, we obtain the th pixel. The coordinates of the i-th pixel in the downwind direction in the wind field coordinate system. The crosswind and downwind coordinates of the first pixel in the wind field coordinate system are related to the first... The path-corrected absorption intensity values corresponding to the nth pixel are combined to generate the nth... The absorption intensity in the wind field coordinate system of each pixel. A coordinate transformation is performed on all pixels in the path correction absorption intensity to generate the absorption intensity in the wind field coordinate system.
[0079] In practice, the absorption intensity of the wind field coordinate system is sliced at equal intervals along the crosswind direction. The crosswind direction is the horizontal axis of the wind field coordinate system. A fixed distance along the crosswind direction is set between the crosswind slice intervals. A crosswind slice interval is defined at every interval, from the minimum to the maximum crosswind coordinate value. Each crosswind slice interval corresponds to one crosswind slice. All pixels whose crosswind coordinates fall within the same crosswind slice interval are arranged in ascending order of their downwind coordinates to form a crosswind absorption profile. Each crosswind slice yields one crosswind absorption profile, and all crosswind slices yield the complete crosswind absorption profile.
[0080] In practice, crosswind absorption profiles are sequentially stitched together along the downwind direction to generate methane plume concentration slices. The downwind direction is the vertical axis of the wind field coordinate system. All crosswind absorption profiles are arranged in ascending order of their downwind coordinates. These arranged crosswind absorption profiles are then sequentially arranged to form methane plume concentration slices. The methane plume concentration slices are two-dimensional spatial distribution data. Each location in the methane plume concentration slice corresponds to a crosswind coordinate and a downwind coordinate in the wind field coordinate system, and each location in the methane plume concentration slice corresponds to a path-corrected absorption intensity value.
[0081] In practice, an isoconcentration sequence is established within the methane plume concentration slice. This sequence comprises multiple concentration values divided at fixed intervals from the minimum to the maximum concentration value within the methane plume concentration slice. These fixed intervals are determined based on the dynamic concentration range within the methane plume concentration slice. For each concentration value in the isoconcentration sequence, all spatial locations with a concentration equal to that value are extracted from the methane plume concentration slice. Adjacent spatial locations are connected to form isoconcentration lines. The isoconcentration lines forming closed loops are defined as isoconcentration closure boundaries, with each concentration value corresponding to one isoconcentration closure boundary. For each isoconcentration closure boundary, the number of spatial locations contained within it is counted. This number is then multiplied by the area of each individual spatial location in the wind field coordinate system to obtain the area of the isoconcentration closure boundary.
[0082] In practice, for each isoconcentration closed boundary, the crosswind and downwind coordinates of all spatial locations on the isoconcentration closed boundary are extracted. The crosswind coordinates of all spatial locations on the isoconcentration closed boundary are summed one by one to obtain the total crosswind coordinates. This total is then divided by the number of spatial locations on the isoconcentration closed boundary to obtain the centroid crosswind coordinates of the isoconcentration closed boundary. Similarly, the downwind coordinates of all spatial locations on the isoconcentration closed boundary are summed one by one to obtain the total downwind coordinates. This total is then divided by the number of spatial locations on the isoconcentration closed boundary to obtain the centroid downwind coordinates of the isoconcentration closed boundary. The centroid crosswind and downwind coordinates of the centroid of the isoconcentration closed boundary constitute the centroid coordinates of the isoconcentration closed boundary. The centroid coordinates are calculated for each concentration value corresponding to an isoconcentration closed boundary in the isoconcentration value sequence.
[0083] In practice, the isoconcentration closed boundaries are arranged in ascending order of isoconcentration values. For two adjacent isoconcentration closed boundaries, the crosswind coordinate of the centroid of the later closed boundary is subtracted from the crosswind coordinate of the previous closed boundary to obtain the difference in crosswind coordinates of the centroids; the downwind coordinate of the centroid of the later closed boundary is subtracted from the downwind coordinate of the centroid of the previous closed boundary to obtain the difference in downwind coordinates of the centroids. The two crosswind coordinate differences are multiplied together, and the two downwind coordinate differences are multiplied together. The two products are added together, and the square root of the sum is taken to obtain the centroid displacement of the two adjacent isoconcentration closed boundaries.
[0084] In practice, the transport time of the isoconcentration closed boundary is determined based on the wind speed and centroid displacement in the synchronous wind field data. The wind speed in the synchronous wind field data is the magnitude of the average wind vector, which is determined by the average eastward and average northward wind speed components. The average eastward wind speed components are multiplied together, and the average northward wind speed components are multiplied together. The two products are added together, and the square root of the sum is taken to obtain the wind speed in the synchronous wind field data. The centroid displacement of two adjacent isoconcentration closed boundaries is then subtracted from the wind speed in the synchronous wind field data to obtain the transport time of the two adjacent isoconcentration closed boundaries.
[0085] In practice, the methane emission location and rate are output based on the transport time and the area change of the isoconcentration closed boundary. The methane emission rate is calculated using the following formula:
[0086]
[0087] in, Indicates the methane emission rate, This represents the area of the closed boundary with the higher concentration among two adjacent closed boundaries of equal concentration. This represents the area of the closed boundary with the lower concentration among two adjacent closed boundaries of equal concentration. This indicates the transport time between two adjacent closed boundaries of equal concentration. The method for obtaining the concentration is as follows: extract the area of the closed boundary with higher concentration from the area of the closed boundary with equal concentration. The closed boundary with higher concentration is the closed boundary with the larger concentration value among two adjacent closed boundaries with equal concentration. The method for obtaining the concentration is as follows: extract the area of the lower concentration closed boundary from the area of the closed boundary with equal concentration. The lower concentration closed boundary is the closed boundary with the smaller concentration value among two adjacent closed boundaries with equal concentration. The method of obtaining it is: remove the centroid of two adjacent isoconcentration closed boundaries and obtain it by synchronizing the wind speed in the wind field data. This represents the change in area between two adjacent closed boundaries with equal concentrations. The methane emission rate is obtained by dividing the change in area by the transport time.
[0088] In practice, the location of methane emissions is determined by the centroid coordinates of the lower-concentration closed boundary among two adjacent equal-concentration closed boundaries. The methane emission rate and location are then combined and output to obtain the methane emission location and rate within the target region.
[0089] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.
Claims
1. A remote sensing method for methane emission detection based on hyperspectral imaging, characterized in that, include: Collect hyperspectral radiance cubes, solar observation geometric parameters, surface elevation data, and synchronous wind field data for the target area; The hyperspectral radiance cube is subjected to inter-band spectral curvature correction and radiometric calibration to generate an absolute radiance spectral sequence. Methane absorption characteristic band set was extracted from the absolute radiance spectral sequence to construct absorption valley morphology parameters; Based on the solar observation geometric parameters and the surface elevation data, atmospheric path length layering and stripping are performed on the absorption valley morphology parameters to generate path-corrected absorption intensity. Based on the synchronous wind field data, the path-corrected absorption intensity is reconstructed by plume direction slices to generate methane plume concentration slices. Source intensity inversion is performed on the methane plume concentration slice to output the methane emission location and methane emission rate.
2. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 1, characterized in that, The hyperspectral radiance cube is subjected to inter-band spectral curvature correction and radiometric calibration to generate an absolute radiance spectral sequence, including: Read the dark current frame and gain parameters of the hyperspectral radiance cube, replace the bad pixels according to the spectral continuity of the neighborhood, and generate a repair radiance cube; The spectral curvature between adjacent narrow bands is calculated for the repaired radiance cube band by band, and the inter-band linear stretching is performed according to the spectral curvature to generate a flat field radiance cube. The irradiance of the flat-field radiance cube is normalized based on the solar zenith angle in the solar observation geometric parameters to generate an absolute radiance spectral sequence.
3. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 1, characterized in that, Methane absorption characteristic band set was extracted from the absolute radiance spectral sequence, and absorption valley morphology parameters were constructed, including: Multiple narrow bands within the methane absorption band are selected from the absolute radiance spectral sequence to form a methane absorption characteristic band group; A local continuum is fitted pixel by pixel to the methane absorption characteristic band group, and the methane absorption residual spectrum is obtained after subtracting the local continuum. Valley depth, half-width at half-maximum (HWHM), valley bottom asymmetry, and peak position shift were extracted from the methane absorption residual spectrum to form absorption valley morphology parameters.
4. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 1, characterized in that, Based on the solar observation geometric parameters and the surface elevation data, atmospheric path length layering and stripping are performed on the absorption valley morphology parameters to generate path-corrected absorption intensity, including: The target area is divided into multiple elevation layers based on the surface elevation data, and the optical path projection length of each elevation layer is calculated according to the solar observation geometric parameters. The valley depth and half-width at half-maximum in the absorption valley morphology parameters are decomposed layer by layer according to the optical path projection length to generate layered absorption components. The overlapping contribution of the water vapor absorption band and the carbon dioxide absorption band is removed from the layered absorption component to generate path-corrected absorption intensity.
5. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 1, characterized in that, Based on the synchronous wind field data, the path-corrected absorption intensity is reconstructed into plume direction slices to generate methane plume concentration slices, including: Based on the synchronous wind field data, establish the downwind and crosswind directions, and project the path-corrected absorption intensity onto the downwind and crosswind directions to generate the wind field coordinate system absorption intensity; The absorption intensity of the wind field coordinate system is sliced at equal intervals along the crosswind direction to generate a crosswind absorption profile. The crosswind absorption profiles are sequentially spliced along the downwind direction to generate methane plume concentration slices.
6. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 1, characterized in that, Source intensity inversion is performed on the methane plume concentration slices to output the methane emission location and methane emission rate, including: Isoconcentration closed boundaries are extracted from the methane plume concentration slices, and the centroid displacement of adjacent isoconcentration closed boundaries is calculated; The transport time of the isoconcentration closed boundary is determined based on the wind speed and centroid displacement in the synchronous wind field data. The methane emission location and methane emission rate are output based on the transport time and the area change of the isoconcentration closed boundary.
7. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 2, characterized in that, The spectral curvature between adjacent narrow bands is calculated band by band for the repaired radiance cube, and inter-band linear stretching is performed according to the spectral curvature to generate a flat-field radiance cube, including: In the repair radiance cube, select adjacent narrow band pairs with no methane absorption and calculate the spectral ratio curve of the adjacent narrow band pairs. The spectral ratio curve is segmented and median filtered to generate inter-band spectral curvature. The inter-band spectral curvature is applied to each band of the repaired radiance cube to generate a flat-field radiance cube.
8. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 3, characterized in that, A local continuum is fitted pixel-by-pixel to the methane absorption characteristic band group, and the methane absorption residual spectrum is obtained after subtracting the local continuum, including: Multiple continuum anchor bands are selected on both sides of the methane absorption characteristic band group, and the multiple continuum anchor bands are segmented and linearly connected to generate a local continuum. The relative absorption spectrum is generated by dividing each band value in the methane absorption characteristic band group by the continuum value of the local continuum at the corresponding band position. The relative absorption spectrum is zero-mean-normalized to obtain the methane absorption residual spectrum.
9. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 4, characterized in that, The target area is divided into multiple elevation layers based on the surface elevation data, and the optical path projection length of each elevation layer is calculated according to the solar observation geometric parameters, including: The target area is divided into multiple elevation layers according to the elevation contour lines in the surface elevation data; Obtain the solar zenith angle, observation zenith angle, and relative azimuth angle from the solar observation geometric parameters; The incident light path direction and the reflected light path direction of each elevation layer are determined based on the solar zenith angle, the observed zenith angle, and the relative azimuth angle. The path length of the incident light path and the path length of the reflected light path within the elevation layer are added together to obtain the projected light path length.
10. The method for remote sensing detection of methane emissions based on hyperspectral imaging according to claim 5, characterized in that, Based on the synchronized wind field data, the downwind and crosswind directions are established, and the path-corrected absorption intensity is projected onto the downwind and crosswind directions to generate the wind field coordinate system absorption intensity, including: The wind vectors in the synchronous wind field data are averaged over time to generate an average wind vector; The direction of the average wind vector is determined as the downwind direction, and the downwind direction is vertically rotated to obtain the crosswind direction. A wind field coordinate system is established using the downwind direction and the crosswind direction as coordinate axes; The position of each pixel in the path-corrected absorption intensity is transformed to the wind field coordinate system to generate the wind field coordinate system absorption intensity.
