AUTOMATED DETECTION AND QUANTIFICATION OF GAS EMISSIONS

TECHNICAL FIELD

[0001] The present disclosure, in various embodiments, relates generally to a method for automatically detecting and/or quantifying gas emissions. More particularly, the method relates to automatically detecting and/or quantifying gas emissions from satellite short- wavelength infrared images.

BACKGROUND

[0002] Methane gas emissions are increasingly subject to regulations worldwide. For example, oil and gas operations are required to install controls that keep methane gas from leaking out of oil and gas equipment by the US Environmental Protection Agency. Therefore, the detection of methane gas emissions is important for environmental and climate considerations and for health and security purposes. Information relating to method gas emissions may be useful to producers of methane gas and to regulatory agencies and governments worldwide.

BRIEF SUMMARY

[0003] Aspects of the present disclosure include a method comprising: obtaining an infrared image of an area of interest from an overhead image acquisition device, the overhead image acquisition device comprising an infrared sensor, detecting a variation of infrared light intensity within the infrared image, and correlating an amount of the variation in infrared light intensity within the infrared image to a gas concentration of a gas emission located in the area of interest depicted in the infrared image.

[0004] Certain preferred but non-limiting features of the method described above are the following, taken individually or in combination:

[0005] the infrared sensor detects at least two infrared spectral bands, the at least two infrared spectral bands comprising an active band and a reference band, wherein the gas emission at least partially absorbs infrared light within the active band, and wherein the gas emission is absorbed to a lesser extent in the reference band than in the active band;

[0006] correlating an amount of the variation in infrared light intensity within the infrared image comprises calculating an amount of infrared light absorption in each pixel of the infrared image, wherein calculating the amount of infrared absorption comprises calculating a linear combination of logarithms of each infrared spectral band captured by the infrared sensor;

[0007] coefficients of the logarithms of the at least two infrared spectral bands more affected by the gas emission are positive and the coefficients of the logarithms of the spectral bands less affected by the gas emission are negative;

[0008] the coefficients of the logarithms of the SWIR bands are computed by an optimization algorithm based on the correlation of bands and on the gas absorption spectra so as to maximize a signal -to-noise ratio of a processed infrared image;

[0009] the method further comprises normalizing the infrared image using the at least two infrared spectral bands detected by the infrared sensor, wherein normalizing the processed infrared image comprises normalizing the amount of attenuation in the infrared image to a scale extending in a range from 0 to 1;

[00010] normalizing the processed infrared image comprises applying a non-local means denoising algorithm to simulate a background of the gas emission;

[00011] the method further comprises applying an automatic thresholding procedure to the normalized and processed infrared image to distinguish the amount of attenuation attributable to the gas emission relative to a background of the normalized infrared image;

[00012] the method further comprises calculating an amount of infrared light absorption in each pixel of the normalized and processed image from the normalized infrared image;

[00013] the method further comprises calculating the dimming function of the infrared sensor, the dimming function being a ratio of an amount of infrared light collected by the active band and/or reference band of the sensor in an environment containing the gas emission and an amount of infrared light collected by the active band and/or reference band of the sensor in an environment being substantially free of the gas emission;

[00014] the method further comprises calculating a concentration of the gas emission in each pixel from the determined optical path length such that the amount of attenuation in the infrared image is correlated to the quantity of the gas emission visible in the infrared image;

[00015] the infrared image is a shortwave infrared image; and/or

[00016] the gas emission is a methane gas emission. BRIEF DESCRIPTION OF THE DRAWINGS

[00017] FIGS. 1 A-1C are graphs plotting transmittance as a function of wavelength of a infrared light transmitted through methane gas at 1 atmosphere (atm) having an optical path length of 1 cm, 2 cm, and 3 cm, respectively.

[00018] FIG. 2 is a frequency response curve for an infrared sensor plotting sensitivity of the infrared sensor to eight short wave infrared bands and methane transmittance as a function of wavelength.

[00019] FIGS. 3 A-3C are graphs plotting a dimming function calculated according to a method of the present disclosure as a function of various short-wave infrared spectral bands.

[00020] FIGS. 4A-4C are graphs plotting a dimming function calculated according to a method of the present disclosure as a function of optical path length through methane gas for various short-wave infrared spectral bands.

[00021] FIG. 5 is a schematic infrared image of an area of interest obtained from an overhead image acquisition device.

[00022] FIG. 6 illustrates the area of interest of FIG. 5 for which an active band/reference band image has been obtained.

[00023] FIG. 7 illustrates the area of interest of FIG. 5 in which a background hidden by a gas emission is simulated in a denoising step.

[00024] FIG. 8 illustrates the area of interest of FIG. 5 after the denoising step.

[00025] FIG. 9 illustrates the area of interest of FIG. 5 after an automatic thresholding step.

[00026] FIG. 10 illustrates the area of interest of FIG. 5 for which a concentration of the emitted gas in each pixel of the image has been determined.

[00027] FIG. 11 illustrates confidence intervals for the determined gas emission concentration.

[00028] FIG. 12 illustrates an average error plot as a function of the determined gas emission concentration.

[00029] FIG. 13 is a schematic diagram of the general architecture of a system for performing the disclosed method.

[00030] The illustrations presented herein are merely idealized and/or schematic representations which are employed to describe embodiments of the present invention. DETAILED DESCRIPTION

[00031] The goal of the method of the present disclosure is to provide automated detection and/or quantification of emissions of a predetermined gas by short wavelength infrared (SWIR) imaging obtained from at least one overhead image acquisition device. The method is based on the gas absorption spectra and frequency response of the image acquisition device’s on board infrared sensor. Using this method, gas emissions (e.g., releases) may be detected and/or quantified globally and communicated to, by way of example and not limitation, gas emission producers and regulators.

[00032] The present disclosure provides an automated method for extracting gas emission concentration information from infrared images. In some embodiments, the method comprises extracting methane gas emission concentration information from at least one short wave infrared (SWIR) image. Gas emission concentration information may be extracted from infrared images by correlating a variation in infrared light intensity within the infrared image to a gas concentration of a gas emission located in an area of interest depicted in the infrared image.

[00033] The method may comprise calculating (e.g., determining) a gas emission concentration for at least one infrared spectroscopy band in the SWIR image. The gas emission concentration may be quantified given that some percentage of infrared light is absorbed by (i.e., not transmitted through) the emitted gas. The gas emission concentration that may be quantified from the SWIR image further depends upon the extent of attenuation in the SWIR image. The extent of attenuation in the image may be computed by using a non-local means denoising algorithm as described in further detail below.

[00034] While embodiments of the present disclosure may be described with reference to methane gas emissions, the method is not so limited. The method may be used to detect and/or quantify emissions of any gas emission that at least partially absorbs infrared light.

[00035] The method may comprise obtaining at least one SWIR image depicting an area of interest from at least one overhead image acquisition device. The overhead image acquisition device may be a satellite. In some embodiments, the overhead image acquisition device may be a DigitalGlobe WorldView-3 satellite (hereinafter “the WV3 satellite”), a Sentinel-5P satellite, a Landsat 8 satellite, and/or a Sentinel-2 satellite.

[00036] The area of interest may comprise at least one gas emission plume for which a gas concentration may be determined according to the method set forth herein. Optionally, the method may comprise an initial step of identifying a gas plume using a machine learning algorithm. A deep learning algorithm may be trained to detect a shape of a gas plume within an infrared image. For instance, the method may utilize a semantic segmentation task that identifies a gas plume shape The deep learning algorithm may be trained using a classical U-NET architecture.

[00037] The overhead image acquisition device comprises at least one infrared sensor. The method may employ any infrared sensor that detects at least two wavelengths of infrared light including an active wavelength for which the gas emission absorbs infrared light and a reference wavelength for which the gas emission either does not absorb infrared light or for which the gas emission absorbs infrared light to a lesser extent than the active wavelength. As used herein, the term “active band” refers to an infrared band having the active wavelength such that infrared light is absorbed by the gas emission in the active band. As used herein, the term “reference band” refers to another infrared band (e.g., an infrared band different from the active band) having the reference wavelength such that infrared light is either fully transmitted through the gas emission in the reference band or such that infrared light is absorbed to a lesser extent than in the active band.

[00038] The intensity of infrared light transmitted through a material is related to a concentration of the material through which the infrared light travels and may be calculated using the Beer-Lambert law. Beer-LambeiTs law is shown by the following formula, wherein A is absorbance, 8 is the molar attenuation coefficient or absorptivity of the material through which light is passed, and c is the concentration of the material through which light is passed:

A = elc

Absorbance (A) is inversely proportional to transmittance (T) according to the following formula:

A = —log ₁₀ T

[00039] Transmittance refers to the fraction of light at a given wavelength that is transmitted when passing through the material. Transmittance is a function of wavelength and the optical path length (/). As can be seen from FIGS. 1A-1C, transmittance is a strictly decreasing function of the optical path length.

[00040] In the present disclosure, methane gas is the material of interest through which the infrared light travels. Therefore, with respect to the Beer-Lambert law, the absorption coefficient of methane (T=296K, P=lbar) is used to calculate transmittance and/or absorbance from the obtained SWIR images in the examples herein. [00041] FIGS. 1 A-1C illustrate the transmittance of infrared light through methane gas as a function of wavelength for optical path lengths of 1, 2, and 3 cm, respectively. Transmittance is plotted in a range extending from 0 to 1, wherein a transmittance value of 1 indicates that light is fully transmitted through the methane gas (e.g., no light is absorbed by the methane gas) and a transmittance value of 0 indicates that no light is transmitted through the methane gas (e.g., light is fully absorbed by the methane gas). As can be seen from FIGS. 1 A-1C, transmittance decreases as the optical path length increases.

[00042] The method may further comprise obtaining at least one relative radiometric response curve for at least one infrared band captured by the infrared sensor of the overhead imaging device. In some embodiments, the method may comprise obtaining a plurality of relative radiometric response curves for at least one active band and for at least one reference band. For example, the method may comprise obtaining a first relative radiometric response curve for the active band and a second relative radiometric response curve for the reference band of the infrared sensor.

[00043] By way of example, the infrared sensor of the WV3 satellite captures eight SWIR bands (SWIR1-SWIR8) as outlined in the table below. As previously stated, while the method is described with reference to the SWIR sensor of the WV3, the method is not so limited. Rather, the method of the present disclosure may employ other infrared sensors that capture fewer than or more than eight SWIR bands. By way of further example, the Sentinel-5P satellite includes a sensor capturing two SWIR bands: SWIR-1 (1590-1675nm) and SWIR-3 (2305-2385nm), and the Sentinel-2 satellite includes a sensor capturing two SWIR bands having a central wavelength of about 1600 nm and about 2200 nm, respectively.

[00044] The relative radiometric response curve for each of the eight SWIR-bands (SWIR1-SWIR8) captured by the WV3 satellite sensor is illustrated in FIG. 2. FIG. 2 is a graph plotting sensitivity of the sensor together with transmittance of methane (when the optical path length is 1 cm) as a function of the wavelength of infrared light. Assuming limited interference from other atmospheric gases, the attenuation of infrared light detected in the sensor’s SWIR bands corresponding to methane absorption can be attributed to methane only.

[00045] The method may further comprise calculating (e.g., determining) an amount of light collected by at least one SWIR band of the sensor in a fully transparent environment. More particularly, the method may comprise calculating an amount of infrared light collected by the at least one active band in a fully transparent environment and an amount of light collected by the at least one reference band in the fully transparent environment.

[00046] For a given SWIR band (e.g., the active band or reference band), the calculated value for a given SWIR band is referred to herein as a “reference effective sensor frequency response”. For a given SWIR band of the sensor, the integral of the response curve is a measure of the amount of light that is collected by the sensor within the given SWIR band in a fully transparent environment (e.g., wherein no light is absorbed).

[00047] The method may further comprise calculating (e.g., determining) an effective sensor frequency response for at least one SWIR band captured by the sensor in an environment containing the emitted gas. More particularly, the method may comprise calculating the amount of infrared light collected by the at least one active band in the emitted gas and an amount of light collected by the at least one reference band in the emitted gas.

[00048] As used herein, the “effective sensor frequency response” refers to an amount of infrared light collected by the at least one SWIR band (e.g., the active band or reference band) of the sensor in an environment containing the emitted gas. The effective sensor frequency response is obtained by combining the frequency response curve of the at least one SWIR band with the transmittance of the emitted gas (here, methane gas), as illustrated in FIG. 2. In other words, the effective sensor frequency response of a given SWIR band may be calculated as the product of the emitted gas transmittance function and sensor frequency response over the given SWIR band. For a given SWIR band, the area under the frequency response curve and under the emitted gas transmittance function is a measure of the amount of infrared light that is collected by the sensor in the presence of methane.

[00049] The method may further comprise calculating (e.g., determining) a dimming function for at least one SWIR band captured by the sensor. More particularly, the method may comprise calculating a dimming function for the active band and for the reference band of the infrared sensor. [00050] The dimming function is a measure of a percentage of infrared light that is not absorbed by the emitted gas in an environment. The dimming function, a, may be associated to a gas concentration for a given band B. Here, the initial infrared light of a given band, Bo , and the infrared light of the same given band as effected by the emitted gas, B _t. The presence of the emitted gas can be explained by the dimming according to the formula:

B _t = aB ₀

[00051] The dimming function is the ratio of the integral of the effective sensor frequency response to the integral of reference effective sensor frequency response for a given SWIR band. As a ratio, the dimming function has a value in a range extending from 0 to 1, which values respectively correspond to full absorption of infrared light by the emitted gas and no absorption of infrared light by the emitted gas (e.g., full transmission through the emitted gas). Accordingly, by calculating the dimming function, the method comprises determining a percentage of infrared light that was absorbed by and that was not absorbed by (e.g., transmitted through) the emitted gas before reaching the sensor.

[00052] FIGS. 3A-3C are graphs plotting the dimming function as a function of the respective SWIR bands of the WV3 satellite sensor, wherein the optical path lengths are respectively 1 cm, 2 cm, and 3 cm.

[00053] In each of FIGS. 3A-3C, for the sensor bands SWIR1 and SWIR2, 100% of infrared light is transmitted through methane gas; therefore, the dimming function is 1. Infrared light starts to be absorbed by methane gas at SWIR3, as the dimming function is less than 1. The presence of methane is best detected in SWIR8. As shown in each of FIGS. 3A-3C, the lowest dimming function value is obtained in sensor band SWIR8, which indicates the greatest absorption of infrared light by methane gas in this band. As further shown in each of FIGS. 3 A-3C, the greatest dimming function value having a value less than 1 is obtained in sensor band SWIR5. Accordingly, among SWIR bands that absorb infrared light, the least amount of infrared light is absorbed by methane gas in SWIR5. From a comparison of FIGS. 3A-3C, the dimming function can be seen to decrease as the optical path length increases.

[00054] FIGS. 4A-4C illustrate the dimming function as a function of the optical path length through methane gas for SWIR bands 5, 7 and 8, respectively, of the WV3 satellite sensor. As can be seen from FIGS. 4A-4C, the dimming function value (indicated at SWIR8) is strictly decreasing (as expected from the Beer-Lambert law) with increasing optical path length. Accordingly, as the optical path length increases, the amount of infrared light absorbed by the methane gas increases. As illustrated from the FIGS. 4A-4C, the dimming function of SWIR8 has the largest sensitivity to the optical path length.

[00055] Finally, the dimming function maps the optical path length through methane gas and some infrared light that has been absorbed by some methane gas. Conversely, the bijectivity evidenced implies the existence of an inverse dimming function and a 1-1 correspondence between a percentage of infrared light absorbed at a pixel and an equivalent methane optical path length.

[00056] The calculated dimming function for at least one SWIR band of the sensor may be used in order to quantify some emitted gas concentration given some percentage of infrared light is absorbed by the emitted gas: depending on the extent of variation in infrared light intensity within the image (e.g., variation in infrared light intensity between pixels of the image), an optical path length of the emitted gas may be computed and, from the optical path length, a concentration of the emitted gas may be quantified.

[00057] Accordingly, the method of the present disclosure may further comprise processing of the obtained image to determining an amount of attenuation of infrared light therein. As previously stated, an amount of attenuation of infrared light may be correlated to a concentration of the emitted gas in the obtained image using the previously calculated dimming function.

[00058] A schematic infrared image that may be obtained in the method of the present disclosure is provided as FIG. 5. FIG. 5 depicts an area of interest (AOI) including a site where the gas is expected to be emitted.

[00059] In order to determine an amount of infrared light absorption in each pixel of the image, the method may further comprise calculating a linear combination of logarithms of each SWIR band of the infrared sensor. Coefficients of the logarithms of the SWIR bands that are more affected by the gas emission (e.g., the active band(s)) are positive, while coefficient of the logarithms of the SWIR bands that are less affected by the gas emission (e.g., the reference band(s)) are negative. The coefficients of the logarithms of the SWIR bands are computed by an optimization algorithm based on a correlations of the SWIR bands and on the gas absorption spectra so as to maximize the signal -to-noise ratio of the resulting image.

[00060] Using the Beer-Lambert law, the dimming of a light from a source caused by the atmosphere for a frequency,/ as measured at a sensor is: where 1(f) is the light intensity reduced as a function of an atmosphere, /, through which the light passes,

Io is the initial light intensity,

A, is the absorption of the atmosphere traversed by the light,

L, is the length of the atmosphere traversed by the light, and lf is the reflection coefficient, or albedo.

[00061] The signal, w, seen by the sensor having a sensitivity, 5, on the overhead image acquisition device is then:

[00062] Assuming that the initial light intensity does not vary as a function of the frequency,/ assuming that the reflection coefficient, l, is constant, and assuming that a gas having a length, L _S0Urce , traversed by the light has been emitted into the atmosphere and is the source of the dimming of the light intensity, the signal, w, seen by the sensor on the overhead image acquisition device becomes:

[00063] A ratio of the signal with the gas emission, u( x, y, t), and the signal without the gas emission, u(x, y, t), is:

[00064] By way of example of the foregoing steps with reference to SWIR bands of the WV3 satellite, the images for bands SWIR8, which has the greatest absorption by methane gas, and SWIR5, which has relatively low absorption by methane gas, are selected. SWIR5 is further selected as the closest spectral band to SWIR8 with low absorption by methane. For both SWIR8 and SWIR5, the absorption behavior is expected to be the same in the background of the image including the reference area.

[00065] The method may comprise determining a ratio of the infrared light intensity of at least one active band image to the infrared light intensity of at least one reference band image. By dividing at least one active band by at least one reference band, objects that absorb infrared light in the active band(s), including methane gas, are highlighted while the remaining elements in the obtained image are smoothed (e.g., muted). [00066] FIG. 6 illustrates the AOI of FIG. 5 for which an active band/reference band image has been obtained.

[00067] The method of the present disclosure may comprise normalizing SWIR images of at least two SWIR bands of the sensor. In some embodiments, the method may comprise normalizing SWIR images for the active band and the reference band. In other embodiments in which there exists more than one active band and/or more than one reference band for a given sensor, the bands that are normalized may be a first band in which absorption by the emitted gas is the greatest and a second band selected may be a band in which there is no absorption or another band in which absorption by the emitted gas occurs but at the lowest level relative to other bands.

[00068] The method may further comprise normalizing the active band(s)/reference band(s) ratio and fitting the ratio into the previously determined dimming function. The active band/reference band ratio may be normalized as the active band/reference band ratio should be equal to 1 in areas of the image in which the emitted gas is not present. The active band/reference band ratio is normalized against an active band/reference band ratio of the reference area (hereinafter “the reference active band/reference band ratio”). The reference active band/reference band ratio is the ratio that would have been obtained without the emitted gas absorbing infrared light. The ratio of (active band/reference band)/(reference active band/reference band) can be used to fit the dimming function: by construction, if (active band/reference band)/(reference active band/reference band) is 1, there is no emitted gas; if (active band/reference band)/(reference active band/reference band) is less than 1, the emitted gas is present.

[00069] The method may further comprise applying a denoising algorithm to simulate a background of the image of FIG. 6. In other words, the denoising algorithm simulates an environment behind the detected gas emission shown in the image.

[00070] By way of example, denoising may comprise applying a non-local means denoising algorithm to the obtained image in order to obtain the background of the image. The non-local means denoising algorithm may be the image denoising method as described in “A non local algorithm for image denoising,” by A. Buades, et ak, published in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Volume 2, pages 60-65, the entire disclosure of which is incorporated herein by this reference.

[00071] The denoising algorithm is based on using the reference band for which the emitted gas does not absorb to determine regions of the image with similar intensity. These regions are then used to extrapolate to the area hidden by the absorbing emitted gas in the active band/reference band ratio image.

[00072] Other methods than the non-local means denoising algorithm may be used to obtain (e.g., estimate) the background of the image.

[00073] For example, SWIR images of the AOI taken at other times (e.g., a time series of images) may be used to obtain the background of the gas emission. Alternatively or additionally, similar pixels can be combined within the same image or from different images obtained at different times. Further, the combinations can be obtained by averaging the N nearest neighbors (with or without weights) or by projecting the current pixel on the vector space spanned by the neighbors.

[00074] By way of brief description for each pixel of the obtained image to be denoised, a surrounding region of pixels of fixed size is compared to all other regions of the same size within some delimited zone far from the region including the emitted gas. For each of these delimited zones, a distance with the region to be denoised together with a gaussian coefficient which exponent depends on that distance are computed. A value of the gaussian coefficient is relatively high if the region to be denoised is similar to the surrounding region of pixels, and the value of the gaussian coefficient is relatively low if the region to be denoised is dissimilar from the surrounding region of pixels. The surrounding region of pixels is finally replaced by the coefficients-weighted average of all regions. Gaussian coefficients are computed for spectral bands for which methane gas does not absorb the infrared light, then applied to spectral bands where absorption occurs (e.g., the active band). This has the effect of bringing back pixels out of the region in which the emitted gas is depicted. In other words, an appearance of the region behind the emitted gas may be simulated (e.g., estimated).

[00075] Yet a further method for obtaining the background of the obtained image using SWIR images of the AOI taken at other times includes using a pixel-wise regularized linear regression. Given a time series of images split into a batch of T dates, for each date t, the method includes projecting the image on date t on the t-1 images. Put differently, the method comprises training a linear regression model with intercept on X and Y, where X is a matrix of size (n, t-1), where n is the number of pixels in the image and Y is a (n, 1) vector corresponding to the date t. The residual of the projection y yields the gas emission mask for the date t. A theoretical threshold may be set on this residual to detect a given gas emission concentration so as to eliminate false gas emission concentrations. [00076] The method may also optionally comprise applying a cloud cover detection algorithm to the time series of SWIR images of the AOI so as to remove images from any of the foregoing analyses that are comprised of, for example more than 15% of cloudy pixels (e.g., pixels in which gas emissions are undetectable due to cloud cover). The denoised image of the AOI that illustrates the simulated background obtained by the method is illustrated in FIG. 7. In FIG. 7, the emitted methane gas plume is no longer visible while any other background in the originally obtained image are substantially the same. Accordingly, the denoising step reveals the background behind the emitted gas.

[00077] The simulated background image is then used to generate the normalized image of FIG. 8. After normalization with the simulated background, regions where the emitted gas plume is present are highlighted, and the remainder of the image has significantly reduced.

[00078] The method of the present disclosure may further comprise applying an automatic thresholding procedure to the normalized image of FIG. 8 to obtain FIG. 9. Applying the automatic thresholding procedure to the normalized image to isolate the gas emission relative to a remainder of the image not including the gas emission. The automatic thresholding procedure may be the Multi-Otsu thresholding method as described in “A fast algorithm for multilevel thresholding,” by Liao, P-S., et al., published in Journal of Information Science and Engineering 17(5), pages 713-727 (2001). the entire disclosure of which is incorporated herein by this reference. Using this method, a histogram of pixel intensity is automatically split into three parts. The lower bound is taken as a threshold. Pixels with a normalized ratio below the threshold are considered in the method as being methane, leading to the generation of a methane mask. The methane mask refers to a depiction of the methane gas in the obtained image.

[00079] Any portion of the denoised image of FIG. 8 that is not considered to be the emitted gas is removed from the analysis and FIG. 9 is produced. FIG. 9 illustrates the gas emission mask. The areas removed from the analysis are the areas outside of the gas emission mask. The remaining image is normalized against the reference active band/reference band ratio. The normalized active band/reference band ratio may have a value extending in a range from a value of 0 to a value of 1, where a value of 0 represents total absorption and a value of 1 represents total transmission. As some, but not all, of the infrared light is absorbed in the emitted gas, the normalized active band/reference band ratio generally falls in a narrower range between 0 and 1 such that the normalized active band/reference band ratio approaches a value of 0 but does not reach 0. Accordingly, in some embodiments, the active band/reference band ratio has a value that is greater than 0 and less than or equal to 1. [00080] The method further comprises calculating (e.g., determining) the normalized active band/reference band ratio for each pixel of the image. Each normalized active band/reference band ratio value is fit to the dimming function in a 1 : 1 relationship as previously discussed. Accordingly, from the normalized active band/reference band ratio, the concentration of the emitted gas in a given pixel can be determined using the dimming function.

[00081] FIG. 10 illustrates the concentration of the emitted gas in each pixel of the image of FIG. 5.

[00082] As previously described with reference to and illustrated in FIGS. 4A-4C, the dimming function is related to the optical path length through the emitted gas. Accordingly, from the normalized active band/reference band ratio, the optical path length through the emitted gas can be determined for each pixel pursuant to Beer-Lambert’s law. Subsequently, the concentration of the emitted gas for each pixel of the image can be obtained from the calculated optical path length as explained below.

[00083] With respect to the exemplary case discussed herein, if the infrared light detected by the sensor has traveled through a volume consisting of pure methane gas, the intensity of the infrared light will decrease by a factor equal to some normalized intensity measured at that pixel (e.g., the active band/reference band ratio having a value less than 1). The volume consisting of pure methane gas is determined from the cross-sectional area of the pixel and a height equal to that of the calculated optical path length through the methane gas. The total amount of molecules of gas contained in that volume of pure methane can then be derived from the ideal gas law. Accordingly, in the present method, a decrease in infrared light intensity should be determined for regions of the image within the methane mask.

[00084] If infrared light travels through a volume consisting of pure air (e.g., no methane gas is present or assuming that a volume of methane is negligible), the infrared light intensity will not be decreased (e.g., the normalized active band/reference band ratio having a value of 1). The volume consisting of pure air is determined from the cross-sectional area of the pixel and a height equal to the entire atmospheric column. The total amount of molecules of air contained in that volume of pure air can then be derived from the ideal gas law. Accordingly, in the present method, a decrease in infrared light intensity should not be calculated for regions of the image outside the methane mask.

[00085] A final concentration (in ppb) of the emitted gas shown in the image may be calculated from the concentration of the emitted gas and concentration of pure air calculated for each pixel of a plurality of pixels of the image. The final concentration is computed as the ratio between the total number of molecules in the volume of pure emitted gas and the total number of air molecules contained the volume of pure air.

[0001] Another method for determining a concentration of the emitted gas shown in the image may be calculated using an integrated mass enhancement (IME) method. The IME method used according to embodiments of the present disclosure may be the IME method as described in D. J. Varon, et al., “Quantifying methane point sources from fine-scale satellite observations of atmospheric methane plumes,” Atmospheric Measurement Techniques , vol. 11, no. 10, pp. 5673-5686, 2018, which is hereby incorporated in its entirety by this reference.

[0002] The IME method relates an emission rate Q to a total detected plume mass IME. The IME method uses plume enhancements D.i\mol x m2] observed over i E [1, TV] plume pixels with area A^m ² ], an effective wind speed U _e ^[m x s ^_1 ] , and a plume length scale L[m\ to estimate the emission rate Q:

[00086] Here, N is obtained with a Boolean plume mask that distinguishes plume pixels from background pixels in the gas concentration image, and L is defined as the square root of a total plume area of the gas emission mask identified from the obtained image. According to methods of the present disclosure, plumes may be mixed over the entire Planetary Boundary Layer (PBL), such that U _e ^ is taken to be equal to a local wind speed, which is representative of the average PBL wind.

[00087] To obtain the effective wind speed necessary for the IME method, the method of the present disclosure may further comprise a step of obtaining meteorological data for the area of interest in which the gas emission is detected. Such meteorological data may include wind velocity and/or wind direction for a given time period. The meteorological data may be obtained from one or more meteorological databases. The meteorological databases may obtain meteorological data from one or more remote sensing devices including overhead image acquisition devices. Such databases may include at least one of the National Oceanic and Atmospheric Administration (NOAA), European Centre for Medium-Range Weather Forecasts (ECMWF), the National Center for Environmental Prediction (NCEP), and Global Forecast System (GFS).

[00088] While the method has been described with reference to processing of a single image and quantification of the emitted gas in the image, the method is not so limited. The method may be applied to a plurality of images. In some embodiments, the method may be applied to a time series of images such that a variation in the quantity of and/or total quantity of the emitted gas may be determined over a period of time.

[00089] The method of the present disclosure may further comprise determining confidence intervals for the aforementioned calculation of the final concentration of methane present in the image to account for errors in the methodology, if any. For example, when the calculation of (active band/reference band)/(reference active band/reference band) results in values not equal to 1, the expected value outside of absorbing parts of the image (out of the methane mask computed above) leads to the determination of the standard deviation of the noise (sigma) considered as gaussian. Some range of dimming is obtained by fitting the dimming function into some desired range of ppb. Upper and lower endpoints of the confidence intervals may be obtained by performing the inverse dimming function to that dimming range increased and decreased by standard deviation of the noise, respectively.

[00090] A posteriori, pixels with (active band/reference band)/(reference active band/reference band) lower than sigma are not taken into account.

[00091] Typical confidence intervals are shown in FIG. 11. Confidence intervals may strongly vary depending on the overall quality of the image).

[00092] In the examples provided herein, small gas emission detections are obscured by the noise (about 250 ppb in the example case of detection of methane gas in the WV3 image).

[00093] Average error (average between upper and lower confidence limits) as a function of the concentration is illustrated in FIG. 12.

[00094] Errors are very small for high concentration releases (lower than 5% with the example case of detection of methane gas in the WV3 image).

[00095] Embodiments of the present disclosure may comprise or utilize a special purpose or general-purpose computer including computer hardware, such as, for example, one or more processors and system memory, as discussed in greater detail below. Embodiments within the scope of the present disclosure also include physical and other computer-readable media for carrying or storing computer-executable instructions and/or data structures. In particular, one or more of the processes described herein may be implemented at least in part as instructions embodied in a non-transitory computer-readable medium and executable by one or more computing devices (e.g., any of the media content access devices described herein). In general, a processor (e.g., a microprocessor) receives instructions, from a non-transitory computer-readable medium, (e.g., a memory, etc.), and executes those instructions, thereby performing one or more processes, including one or more of the processes described herein. [00096] Computer-readable media can be any available media that can be accessed by a general purpose or special purpose computer system. Computer-readable media that store computer-executable instructions are non-transitory computer-readable storage media (devices). Computer-readable media that carry computer-executable instructions are transmission media. Thus, by way of example, and not limitation, embodiments of the disclosure can comprise at least two distinctly different kinds of computer-readable media: non-transitory computer-readable storage media (devices) and transmission media.

[00097] Non-transitory computer-readable storage media (devices) includes RAM, ROM, EEPROM, CD-ROM, solid state drives (“SSDs”) (e.g., based on RAM), Flash memory, phase-change memory (“PCM”), other types of memory, other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store desired program code means in the form of computer-executable instructions or data structures and which can be accessed by a general purpose or special purpose computer.

[00098] Further, upon reaching various computer system components, program code means in the form of computer-executable instructions or data structures can be transferred automatically from transmission media to non-transitory computer-readable storage media (devices) (or vice versa). For example, computer-executable instructions or data structures received over a network or data link can be buffered in RAM within a network interface module (e.g., a “MC”), and then eventually transferred to computer system RAM and/or to less volatile computer storage media (devices) at a computer system. Thus, it should be understood that non-transitory computer-readable storage media (devices) can be included in computer system components that also (or even primarily) utilize transmission media.

[00099] Computer-executable instructions comprise, for example, instructions and data which, when executed at a processor, cause a general purpose computer, special purpose computer, or special purpose processing device to perform a certain function or group of functions. In some embodiments, computer-executable instructions are executed on a general- purpose computer to turn the general-purpose computer into a special purpose computer implementing elements of the disclosure. The computer executable instructions may be, for example, binaries, intermediate format instructions such as assembly language, or even source code. Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the described features or acts described above. Rather, the described features and acts are disclosed as example forms of implementing the claims.

[000100] Those skilled in the art will appreciate that the disclosure may be practiced in network computing environments with many types of computer system configurations, including, personal computers, desktop computers, laptop computers, message processors, hand-held devices, multi-processor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, mobile telephones, PDAs, tablets, pagers, routers, switches, and the like. The disclosure may also be practiced in distributed system environments where local and remote computer systems, which are linked (either by hardwired data links, wireless data links, or by a combination of hardwired and wireless data links) through a network, both perform tasks. In a distributed system environment, program modules may be located in both local and remote memory storage devices.

[000101] Embodiments of the present disclosure can also be implemented in cloud computing environments. In this description, “cloud computing” is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources. For example, cloud computing can be employed in the marketplace to offer ubiquitous and convenient on-demand access to the shared pool of configurable computing resources. The shared pool of configurable computing resources can be rapidly provisioned via virtualization and released with low management effort or service provider interaction, and then scaled accordingly.

[000102] A cloud-computing model can be composed of various characteristics such as, for example, on-demand self-service, broad network access, resource pooling, rapid elasticity, measured service, and so forth. A cloud-computing model can also expose various service models, such as, for example, Software as a Service (“SaaS”), Platform as a Service (“PaaS”), and Infrastructure as a Service (“IaaS”). A cloud-computing model can also be deployed using different deployment models such as private cloud, community cloud, public cloud, hybrid cloud, and so forth. In this description and in the claims, a “cloud-computing environment” is an environment in which cloud computing is employed.

[000103] FIG. 13 illustrates a block diagram of an example system 1300 that may be configured to perform one or more of the processes described above. The system 1300 is communicatively coupled to a plurality of databases. The plurality of databases may include at least one database 1304 from which SWIR images from an overhead image acquisition device 1302 are obtained, at least one database 1306 from which at least one relative radiometric response curved for the at least one infrared band captured by the infrared sensor of the overhead imaging device is obtained, and at least one database 1308 from which meteorological data is obtained. The plurality of databases may obtain images and other data from overhead image acquisition devices. In some embodiments, the overhead image acquisition devices may be different satellites comprising different sensors and image acquisitions devices.

[000104] The databases 1304, 1306, and 1308 may be stored on a non-transitory computer-readable storage media (device) as previously described herein. As shown by FIG. 13, the system 1300 may comprise a computing device 1326. The computing device 1326 may comprise a processor 1328, a memory 1330, a storage device 1332, an I/O interface 1334, and a communication interface 1336, which may be communicatively coupled by way of a communication infrastructure 1338. While an example computing device 1326 is shown in FIG. 13, the components illustrated in FIG. 13 are not intended to be limiting. Additional or alternative components may be used in other embodiments. Furthermore, in certain embodiments, the computing device 1326 can include fewer components than those shown in FIG. 13. Components of the computing device 1326 shown in FIG. 13 will now be described in additional detail.

[000105] In one or more embodiments, the processor 1328 includes hardware for executing instructions, such as those making up a computer program. By way of non-limiting example, to execute instructions, the processor 1328 may retrieve (or fetch) the instructions from an internal register, an internal cache, the memory 1330, or the storage device 1332 and decode and execute the instructions. In one or more embodiments, the computing device 1326 may include one or more internal caches for data, instructions, or addresses. By way of non limiting example, the computing device 1326 may include one or more instruction caches, one or more data caches, and one or more translation look aside buffers (TLBs). Instructions in the instruction caches may be copies of instructions in the memory 1330 or the storage 1332.

[000106] The computing device 1326 includes memory 1330, which is coupled to the processor 1328. The memory 1330 may be used for storing data, metadata, and programs for execution by the processor(s) 1328. The memory 1330 may include one or more of volatile and non-volatile memories, such as Random Access Memory (“RAM”), Read Only Memory (“ROM”), a solid state disk (“SSD”), Flash, Phase Change Memory (“PCM”), or other types of data storage. The memory 1330 may be internal or distributed memory. [000107] The computing device 1326 includes the storage device 1332 that includes storage for storing data or instructions. By way of non-limiting example, storage device 1332 can comprise a non-transitory storage medium described above. The storage device 1332 may include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disc, a magneto optical disc, magnetic tape, or a Universal Serial Bus (USB) drive or a combination of two or more of these. The storage device 1332 may include removable or non-removable (or fixed) media, where appropriate. The storage device 1332 may be internal or external to the computing device 1326. In one or more embodiments, the storage device 1332 is non-volatile, solid-state memory. In other embodiments, the storage device 1332 includes read-only memory (ROM). Where appropriate, this ROM may be mask programmed ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), electrically alterable ROM (EAROM), or flash memory or a combination of two or more of these.

[000108] The computing device 1326 also includes one or more input or output (“I/O”) devices/interfaces 1334, which are provided to allow a user to provide input to, receive output from, and otherwise transfer data to and receive data from computing device 1326. The EO devices/interfaces 1334 may include a mouse, a keypad or a keyboard, a touch screen, a camera, an optical scanner, network interface, modem, other known I/O devices or a combination of such I/O device/interfaces. The touch screen may be activated with a stylus or a finger.

[000109] The I/O devices/interfaces 1334 may include one or more devices for presenting output to a user, including, but not limited to, a graphics engine, a display (e.g., a display screen), one or more output drivers (e.g., display drivers), one or more audio speakers, and one or more audio drivers. In certain embodiments, the I/O interface 1334 is configured to provide graphical data to a display for presentation to a user. The graphical data may be representative of one or more graphical user interfaces and/or any other graphical content as may serve a particular implementation.

[000110] The computing device 1326 can further include a communication interface 328. The communication interface 1336 can include hardware, software, or both. The communication interface 1336 can provide one or more interfaces for communication (such as, for example, packet-based communication) between the computing device 1326 and one or more other computing devices or networks. As an example and not by way of limitation, the communication interface 1336 may include a network interface controller (NIC) or network adapter for communicating with an Ethernet or other wire-based network or a wireless MC (WNIC) or wireless adapter for communicating with a wireless network, such as a WI-FI. The computing device 1326 can further include a bus. The bus can comprise hardware, software, or both that couples components of computing device 1326 to each other.

[000111] While the present disclosure has been described herein with respect to certain illustrated embodiments, those of ordinary skill in the art will recognize and appreciate that it is not so limited. Rather, many additions, deletions, and modifications to the illustrated embodiments may be made without departing from the scope of the invention as hereinafter claimed, including legal equivalents thereof. In addition, features from one embodiment may be combined with features of another embodiment while still being encompassed within the scope of the invention as contemplated by the inventors.