A Joint Inversion Method and System for Methane Emissions at the Industrial Park Scale Based on Point-Source and Area-Source Collaborative Analysis

By using a joint inversion method based on observation data from multiple platforms, platform-specific observation operators and error covariance matrices are constructed to adaptively optimize methane emissions in the industrial park. This solves the problem of overlapping attributions between point source and area source fluxes, and improves the interpretability and stability of the inversion results.

CN123117638APending Publication Date: 2026-11-17WUHAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611297658.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-25
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing methane remote sensing monitoring technologies are difficult to effectively coordinate the analysis of point source and area source emissions at the park scale, resulting in overlapping attributions of point source and area source fluxes. Existing methods fail to fully utilize the complementarity of multi-platform observations and lack physical mapping relationships.

Method used

A platform-specific observation operator is constructed using data from multiple observation platforms. Joint inversion is performed using Bayes' theorem. By combining the observation error covariance matrix and the prior error covariance matrix, the inversion results are adaptively optimized to construct the posterior emission field, thereby reducing system errors and simultaneously optimizing point source and area source emissions.

Benefits of technology

This study improved the interpretability and stability of methane emissions at the park scale, reduced the attribution aliasing between point source and area source fluxes, and enhanced the objectivity and reliability of the inversion results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117638A_ABST
    Figure CN123117638A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for joint inversion of methane emissions at the industrial park scale, oriented towards point-source and area-source collaborative analysis. It belongs to the technical fields of greenhouse gas remote sensing monitoring, emission flux inversion, atmospheric transport simulation, and environmental monitoring verification. The method constructs a candidate library of point sources and a rasterized prior list of area sources based on observation data from multiple observation platforms within the industrial park. It then constructs observation operators and observation vectors corresponding to each observation platform, stacking them to form a joint observation operator and a joint state vector before performing joint inversion to construct the posterior emission field. Finally, iterative optimization is performed by calculating the observation residuals to obtain the final inversion result. This invention is suitable for providing data support for greenhouse gas emission accounting and emission reduction decisions in industrial parks, and can reduce attribution aliasing between point-source and area-source emission fluxes, improving the interpretability and stability of methane emission inversion results at the industrial park scale.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the technical fields of greenhouse gas remote sensing monitoring, emission flux inversion, atmospheric transport simulation and environmental monitoring verification, and in particular relates to a joint inversion method and system for methane emissions at the park scale oriented towards point source-area source collaborative analysis. Background Technology

[0002] Methane is a significant greenhouse gas, with widespread emission sources in energy production, petrochemicals, coal chemicals, waste treatment, wastewater treatment, and urban functional parks. Unlike regional-scale emissions, industrial park-scale methane emissions typically exhibit both point source and area source characteristics. Storage tanks, well sites, valve chambers, pipeline nodes, flares, vents, and suspected leak facilities exhibit localized high emissions and plume characteristics, classifying them as point source emissions. Background releases from roads, plant areas, wastewater treatment units, storage yards, and functional areas exhibit spatial continuity and gradually varying intensity characteristics, classifying them as area source emissions.

[0003] Existing methane remote sensing technologies can acquire various observational data from different platforms. Satellite observations typically provide column concentrations or concentration enhancements at a large pixel scale; UAV observations can obtain local concentration profiles at different altitudes along the flight path; mobile observations can obtain near-ground trajectory concentration sequences; and ground station observations can provide continuous background concentrations and local meteorological information. These observations differ significantly in terms of observation altitude, spatial resolution, time window, sensor response, representativeness scale, and error structure, and cannot be simply pieced together into the same type of observational data.

[0004] Current common methods include using observations from a single platform to estimate plume flux, or using empirical weighted fusion of observations from different platforms before inverting emission flux. The former makes it difficult to fully utilize the complementarity of multi-platform observations, while the latter usually does not clearly establish the physical mapping relationship between different observations and emission states. When point source plume signals and area source background signals coexist, simple fusion methods are prone to misassigning point source anomalies to area source fluxes, or misjudging continuous area source backgrounds as point source anomalies, thus causing aliasing of point source and area source flux attributions.

[0005] Therefore, there is an urgent need for a joint inversion method and system for methane emissions at the industrial park scale, which is oriented towards point source-area source collaborative analysis, to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to address the problems existing in the prior art by providing a joint inversion method and system for methane emissions at the campus scale, oriented towards point source-area source collaborative analysis.

[0007] To achieve the above objectives, the invention employs the following technical solution:

[0008] In a first aspect, the present invention provides a joint inversion method for methane emissions at the industrial park scale based on point source-area source collaborative analysis, the method comprising the following steps:

[0009] Acquire observation data from various observation platforms in the park, and construct a candidate library of point sources and a rasterized prior list of area sources;

[0010] Based on the point source candidate library and the area source rasterization prior list, platform observation operators and platform observation vectors corresponding to multiple types of the observation platforms are constructed respectively.

[0011] The platform observation operator and the platform observation vector are stacked to form a joint observation operator and a joint state vector, and the joint state vector is jointly inverted to construct the posterior emission field.

[0012] The observation residuals of the posterior emission field are calculated, and the input parameters of the joint inversion are adaptively updated to iteratively optimize the joint inversion results.

[0013] By adopting the above technical solution, platform-specific observation operators are constructed for multiple observation platforms, enabling unified joint inversion of different observation data without the need for forced resampling to the same data form. This achieves native collaborative utilization of multi-source heterogeneous data and significantly reduces the systematic errors introduced by data preprocessing.

[0014] As a further technical solution, the method for constructing the posterior emission field is as follows:

[0015] All the platform observation operators are stacked row by row to form a joint observation operator, and the corresponding platform observation vectors are stacked vertically to form a joint state vector;

[0016] The observation error covariance matrix is ​​determined based on the observation error characteristics of the various observation platforms, and the prior error covariance matrix is ​​determined based on the uncertainty of the point source candidate library and the area source rasterized prior list.

[0017] Using the joint observation operator, the observation error covariance matrix, and the prior error covariance matrix, the joint state vector is jointly inverted based on Bayes' theorem to obtain the posterior state vector;

[0018] The point source components in the posterior state vector are mapped to the emission rates at each point source location, and the area source components are mapped to the emission fluxes in each grid region. The posterior emission field is constructed by combining these components according to their spatial locations.

[0019] By adopting the above technical solution, the point source emission intensity sub-vector and the area source grid flux quantum vector are incorporated into the same joint state vector for joint inversion. The two types of emission sources are simultaneously optimized in the same inversion framework, thereby reducing the attribution aliasing between point source and area source emission fluxes and improving the interpretability and stability of methane emission inversion results at the park scale.

[0020] As a further technical solution, the observation data of the various observation platforms includes methane observation data and auxiliary data. The methane observation data includes at least two of the following: satellite observation data, UAV observation data, mobile observation data, and ground station observation data. The auxiliary data includes meteorological data, background concentration data, park functional zoning data, and emission source prior data.

[0021] By adopting the above technical solution, the input data boundary is defined for the joint inversion method, and the data source and content on which the method depends are clarified.

[0022] As a further technical solution, the method for constructing the point source candidate library and the area source rasterized prior list is as follows: generate corresponding observation descriptors through the methane observation data of each observation platform, and construct the point source candidate library and the area source rasterized prior list based on the observation descriptors and the auxiliary data.

[0023] By adopting the above technical solution, corresponding observation descriptors are generated for the methane observation data of each observation platform, transforming the original, heterogeneous observation data into standardized data units that can be uniformly processed by subsequent algorithms.

[0024] As a further technical solution, the platform observation operator includes an emission field mapping operator, a meteorological transport response operator, a sensor response operator, and a spatiotemporal sampling operator. The emission field mapping operator is used to map the point source emission intensity of each candidate facility in the point source candidate library and the area source grid flux of each grid in the area source gridded prior list to a unified spatial emission field. The meteorological transport response operator is used to convert the unified spatial emission field into a concentration response field. The sensor response operator includes any one of the following: satellite pixel average kernel, UAV sampling volume, mobile instrument response time, and ground station representative volume. The spatiotemporal sampling operator is used to extract simulated observation values ​​corresponding to the actual observation time and observation location from the concentration response field based on the actual observation time and observation location of each observation platform.

[0025] By adopting the above technical solution and through the division of labor and cooperation of four sub-operators, the organic integration of multi-source heterogeneous data at the physical model level is achieved.

[0026] As a further technical solution, the joint state vector includes a point source emission intensity sub-vector, a surface source grid quantum vector, a background concentration correction sub-vector, and a model bias correction sub-vector. The background concentration correction sub-vector is used to characterize the background concentration field correction amount in the observation area, and the model bias correction vector is used to characterize the systematic bias correction amount of each observation operator of the platform. The point source emission intensity sub-vector is subject to the position constraints, non-negativity constraints, and sparsity constraints of each candidate facility in the point source candidate library, and the surface source grid quantum vector is subject to spatial grid constraints, functional area prior constraints, non-negativity constraints, and spatial smoothness constraints.

[0027] By adopting the above technical solution, through the division of four sub-vectors and the application of differentiated physical constraints, a framework with complete structure, clear physical constraints, and mathematical solvability is provided for joint inversion.

[0028] As a further technical solution, the joint observation equation includes a multi-source observation residual term, a point source sparse regularization term, a surface source spatial smoothing regularization term, and a priori deviation constraint term. The multi-source observation residual term is determined by the joint observation operator and the observation error covariance matrix. The observation error covariance matrix is ​​used to characterize the sensor errors, representativeness errors, spatiotemporal registration errors, and meteorological transmission errors of various observation platforms.

[0029] By adopting the above technical solution and through the synergistic optimization of four independent action terms, the optimal balance point is found among "fitting observation data", "maintaining prior rationality", "promoting point source sparsity" and "ensuring surface source smoothness", thereby ensuring that the inversion results conform to both observation facts and physical laws.

[0030] As a further technical solution, the spatial resolution of the posterior emission field is 10~1000m and the assimilation time window is 5~1440min.

[0031] Secondly, this invention provides a joint inversion system for methane emissions at the industrial park scale based on point-source and area-source collaborative analysis, applicable to methods such as the joint inversion of methane emissions at the industrial park scale based on point-source and area-source collaborative analysis, including:

[0032] The data acquisition and prior construction unit is used to acquire observation data from various observation platforms in the park and construct a candidate library of point sources and a rasterized prior list of area sources.

[0033] The platform observation operator and vector construction unit is used to construct platform observation operators and platform observation vectors corresponding to multiple types of the observation platforms based on the point source candidate library and the area source rasterization prior list.

[0034] The joint inversion solution unit is used to stack the platform observation operator and the platform observation vector to form a joint observation operator and a joint state vector, respectively, and to perform joint inversion on the joint state vector to construct the posterior emission field;

[0035] An adaptive optimization unit is used to calculate the observation residuals of the posterior emission field and adaptively update them according to the spatial distribution characteristics of the residuals to iteratively optimize the inversion result.

[0036] Thirdly, the present invention provides a non-transitory computer-readable storage medium storing computer instructions that cause the computer to execute the steps of a campus-scale methane emission joint inversion method oriented towards point-source-area-source collaborative analysis.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] 1. This application incorporates the point source emission intensity sub-vector and the area source grid flux quantum vector into the same joint state vector for joint inversion, and simultaneously optimizes the two types of emission sources in the same inversion framework, thereby reducing the attribution aliasing between point source and area source emission fluxes and improving the interpretability and stability of methane emission inversion results at the park scale.

[0039] 2. This application constructs platform-specific observation operators for multiple observation platforms, enabling unified joint inversion of different observation data without the need for forced resampling to the same data form, realizing the native collaborative utilization of multi-source heterogeneous data, and significantly reducing the systematic errors introduced by data preprocessing;

[0040] 3. This application utilizes the observation error covariance matrix to automatically allocate observation weights based on the observation accuracy and error characteristics of each observation platform, enabling high-precision observation platforms to exert a greater constraint on the inversion results, thereby reducing the influence of low-precision observation platforms, avoiding the subjectivity and blindness of manual weighting, and improving the objectivity and credibility of the joint inversion results.

[0041] 4. This application calculates the observation residuals of the posterior emission field and adaptively updates them according to the spatial distribution characteristics of the residuals, forming a closed-loop iterative optimization mechanism. This mechanism can gradually approach the true emission distribution without human intervention, ensuring the convergence and stability of the final inversion results. Attached Figure Description

[0042] Figure 1 This is a technical roadmap for the joint inversion method of methane emissions at the industrial park scale for point source-area source collaborative analysis, as described in this invention.

[0043] Figure 2 This is a conceptual rendering of the point source and area source flux collaborative analysis in the campus-scale methane emission joint inversion method for point source-area source collaborative analysis of the present invention. Detailed Implementation

[0044] It should be noted that: The terms “comprising” and “having”, and any variations thereof, in the specification, claims, and accompanying drawings of this invention are intended to cover a non-exclusive inclusion, such as a process, method, system, product, or apparatus that includes a series of steps or units, not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0045] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices. The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be decomposed, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0046] 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. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined to form new technical solutions. Such combinations are not bound by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0047] The overall system architecture diagram in this embodiment of the invention is as follows: Figure 1 As shown, this invention provides a joint inversion method for methane emissions at the industrial park scale based on point source-area source collaborative analysis. This method includes the following steps: Step 1: Obtain observation data from various observation platforms in the park, and construct a candidate library of point sources and a gridded prior list of area sources; The method for constructing the point source candidate library and the area source rasterized prior list is as follows: generate corresponding observation descriptors through methane observation data from each observation platform, and construct the point source candidate library and the area source rasterized prior list based on the observation descriptors and auxiliary data.

[0048] Step 2: Based on the point source candidate library and the area source rasterization prior list, construct platform observation operators and platform observation vectors corresponding to multiple types of observation platforms.

[0049] Step 3: Stack the platform observation operator and the platform observation vector to form a joint observation operator and a joint state vector, and perform joint inversion on the joint state vector to construct the posterior emission field; Specifically, all platform observation operators are stacked row-wise to form a joint observation operator, and the corresponding platform observation vectors are stacked vertically to form a joint state vector. The observation error covariance matrix is ​​determined based on the observation error characteristics of multiple observation platforms, and the prior error covariance matrix is ​​determined based on the uncertainties of the point source candidate library and the area source rasterized prior list. Using the joint observation operator, the observation error covariance matrix, and the prior error covariance matrix, the joint state vector is jointly inverted based on Bayes' theorem to obtain the posterior state vector. The point source components in the posterior state vector are mapped to the emission rates at each point source location, and the area source components are mapped to the emission fluxes in each grid region. The posterior emission field is constructed by combining these components according to their spatial locations.

[0050] Step four: Calculate the observation residuals of the posterior emission field and adaptively update the input parameters of the joint inversion to iteratively optimize the joint inversion results.

[0051] It should be noted that the above method constructs a candidate library of point sources and a gridded prior list of area sources based on observation data from multiple observation platforms in the park. It also constructs observation operators and observation vectors for each observation platform, stacks them to form a joint observation operator and a joint state vector, and performs joint inversion to construct the posterior emission field. Then, iterative optimization is performed by calculating the observation residuals to obtain the final inversion result. By constructing platform-specific observation operators for each type of observation platform, different observation data can be uniformly jointly inverted without forced resampling to the same data format. This achieves native collaborative utilization of multi-source heterogeneous data, significantly reducing systematic errors introduced by data preprocessing. Furthermore, it incorporates the point source emission intensity sub-vector and the area source grid flux quantum vector into the same joint state vector for joint inversion, simultaneously optimizing the two types of emission sources within the same inversion framework. This reduces attribution aliasing between point source and area source emission fluxes, improving the interpretability and stability of the park-scale methane emission inversion results.

[0052] In this embodiment, the observation data from multiple observation platforms includes methane observation data and auxiliary data. The methane observation data includes at least two of the following: satellite observation data, UAV observation data, mobile observation data, and ground station observation data. The auxiliary data includes meteorological data, background concentration data, park functional zoning data, and emission source prior data.

[0053] It is understandable that satellite observation data, UAV observation data, mobile observation data, and ground station observation data are respectively the system access satellite methane column concentration or concentration enhancement product, UAV methane concentration profile, mobile near-ground methane concentration sequence, and ground station continuous concentration observation; the auxiliary data include meteorological data such as wind speed, wind direction, boundary layer height, temperature, air pressure, and humidity within the assimilation time window, park functional zoning data such as park boundaries and functional zones (industrial area / residential area / agricultural area, etc.), and emission source prior data such as facility vectors, roads, storage tanks, pipelines, well sites, sewage treatment units, and historical emission inventories.

[0054] Specifically, to acquire observation data from multiple observation platforms, firstly, satellite observation platforms are screened for cloud contamination, solar zenith angle, inversion quality indicators, and outliers; UAV observation platforms are screened for flight attitude, sampling altitude, sensor response time, and wind field matching; mobile observation platforms are screened for vehicle speed, instrument lag, road obstruction, and near-ground background; and ground station observation platforms are screened for continuity, drift, and local anomalies. Then, according to the preset spatial resolution Δs and assimilation time window Δt for the posterior emission field, the spatial resolution Δs ranges from 10 to 1000 m (preferably 100 m) and the assimilation time window Δt ranges from 5 to 1440 min (preferably 30 to 180 min).

[0055] Furthermore, for the r observation platform, its l-th observation record defines an observation descriptor. Represented as: , , .

[0056] in, For observation time, For spatial location, For observation altitude, For pixel footprints or sampling trajectories, For the sensor response function, Here, r represents the observation error parameter, R represents the observation platform type, and R represents the set of observation platform types. r=sat, where r represents a satellite observation platform; r=uav, where r represents a UAV observation platform; r=car, where r represents a mobile observation platform; and r=gnd, where r represents a ground station observation platform. This refers to the observation record number within the r observation platform. The number of valid observation records for the r observation platform; This generates corresponding observation descriptors for methane observation data from each observation platform, transforming the raw, heterogeneous observation data into standardized data units that can be uniformly processed by subsequent algorithms.

[0057] In this embodiment, the platform observation operators include an emission field mapping operator, a meteorological transport response operator, a sensor response operator, and a spatiotemporal sampling operator. The emission field mapping operator is used to map the point source emission intensity of each candidate facility in the point source candidate pool and the area source grid flux of each grid in the area source gridded prior list to a unified spatial emission field. The meteorological transport response operator is used to convert the unified spatial emission field into a concentration response field. The sensor response operator includes any one of the following: satellite pixel average kernel, UAV sampling volume, mobile instrument response time, and ground station representative volume. The spatiotemporal sampling operator is used to extract simulated observation values ​​corresponding to the actual observation time and observation location from the concentration response field based on the actual observation time and observation location of each observation platform.

[0058] Understandably, the division of labor and cooperation among the four sub-operators of the platform's observation operator enables the organic integration of multi-source heterogeneous data at the physical model level.

[0059] Specifically, to construct the platform observation operator, firstly, using the observation data from the aforementioned various observation platforms, the observation equation and operator decomposition of the r observation platform are expressed as follows: , .

[0060] in, For the platform observation vector, Let x be the platform observation operator, and x be the joint state vector. For the background or systematic bias term corresponding to the r observation platform, Let r be the observation error vector corresponding to the observation platform; P is the emission field mapping operator, and T is the observation error vector corresponding to the observation platform. r Meteorological transport response operator, A r Sensor response operator and S r Spatiotemporal sampling operator; meteorological transport response operator takes wind field, boundary layer height, atmospheric stability, topography and atmospheric transport model as input, and converts the emission intensity of each emission source in a unified spatial emission field into the concentration response field at each location in the park.

[0061] Based on the above observation equations and operator decomposition representations, further solutions are obtained for the l-th platform observation vector of the r observation platform. Discretize to obtain the discrete platform observation vector. expression: .

[0062] Where, q i Let e ​​be the emission intensity of the i-th candidate point source. j Let A be the flux of the j-th area source grid. j Let j be the area of ​​the j-th grid cell. Let i be the sensitivity of the i-th point source to the l-th observation on the r-th observation platform. Let be the sensitivity of the j-th area source grid to the l-th observation of the r-th observation platform. This is a background concentration correction term. is the model bias correction term for the r observation platform, m is the total number of candidate point sources, and n is the total number of area source grids.

[0063] Furthermore, the sensor response operators can be understood as follows: the satellite pixel average kernel is used to describe the weighted average contribution of the spatial emission distribution within the satellite pixel coverage area to the pixel observation value; the UAV sampling volume is used to describe the concentration average effect within the spatial volume swept by the UAV flight path; the mobile instrument response time is used to describe the smoothing effect of the mobile sensor on mobile observation due to the response time characteristics; and the ground station representative volume is used to describe the spatial average concentration range represented by the observation value of the fixed point of the ground station.

[0064] In this embodiment, the joint state vector includes a point source emission intensity sub-vector, a surface source grid quantum vector, a background concentration correction sub-vector, and a model bias correction sub-vector. The background concentration correction sub-vector is used to characterize the background concentration field correction amount in the observation area, and the model bias correction sub-vector is used to characterize the systematic bias correction amount of the observation operators of each platform. The point source emission intensity sub-vector is subject to the position constraints, non-negativity constraints, and sparsity constraints of each candidate facility in the point source candidate pool. The surface source grid quantum vector is subject to spatial grid constraints, functional area prior constraints, non-negativity constraints, and spatial smoothing constraints. The joint observation equation includes a multi-source observation residual term, a point source sparse regularization term, a surface source spatial smoothing regularization term, and a priori deviation constraint term. The multi-source observation residual term is determined by the joint observation operator and the observation error covariance matrix. The observation error covariance matrix is ​​used to characterize the sensor error, representativeness error, spatiotemporal registration error, and meteorological transmission error of multiple types of observation platforms.

[0065] Understandably, the joint state vector, through the division of four sub-vectors and the application of differentiated physical constraints, provides a structurally complete, physically constrained, and mathematically solvable framework for joint inversion. Meanwhile, the joint observation equation, through the synergistic optimization of four independent action terms, seeks the optimal balance between "fitting observation data," "maintaining prior rationality," "promoting point source sparsity," and "ensuring surface source smoothness," thereby ensuring that the inversion results conform to both observational facts and physical laws.

[0066] Specifically, to construct the joint state vector and covariance, firstly, based on the aforementioned discrete platform observation vector... The expression defines the joint state vector x, the observation error covariance matrix R, and the prior error covariance matrix B as follows: , , .

[0067] Where T represents the transpose of a vector or matrix, q1~q m Let m be the emission intensities of candidate point sources, e1~e n For n area source grid fluxes, This is a correction term for the bias of the satellite observation platform. This is a deviation correction term for the UAV observation platform. For the deviation correction term of the mobile observation platform and G represents the ground station observation platform bias correction term; G is the observation error covariance matrix. The satellite observation error covariance submatrix, The covariance submatrix of UAV observation errors. The submatrix of covariance of underway observation error and B is the ground station observation error covariance submatrix; B is the prior error covariance matrix. Used to characterize the point source prior uncertainty in the prior error covariance matrix. Used to characterize the spatial correlation of surface sources and the prior uncertainty of functional regions in the prior error covariance matrix. Used to characterize the background concentration error in the prior error covariance matrix. It is used to characterize the platform bias and transmission model error in the prior error covariance matrix.

[0068] Furthermore, such as Figure 2 As shown, for the joint inversion of point source and area source, the observation equations and operator decomposition representation based on the r-observation platform are used to represent the platform observation operators of various observation platforms. The joint observation operator is formed by stacking rows. And the platform observation vector of the corresponding observation platform Vertical stacking forms a joint state vector Then, the background or systematic bias term of the corresponding observation platform is... Vertical stacking The observation error vector Stacked by row The joint observation equation can be expressed as: , , .

[0069] in, Observation vectors for satellite observation platforms transpose, Observation vectors for UAV observation platform transpose, Observation vectors for mobile observation platform transpose, Observation vectors for ground station observation platform transpose; For satellite platform observation operators transpose, For observation operators of unmanned aerial vehicle platforms transpose, For mobile platform observation operators transpose, For ground station platform observation operators The transpose of .

[0070] Furthermore, based on the joint observation equation and Bayes' theorem, a four-dimensional variational assimilation method is employed to perform joint inversion by minimizing the objective function. Obtain the posterior state vector x a Represented as: , Add constraints , .

[0071] Where, x b Let the prior state vector be... For point source emission intensity sub-vectors, Here is the point source candidate weight matrix. Let L be the quantum vector of the surface source grid, L be the spatial smoothing operator, and Le be the spatial difference term after the spatial smoothing operator is applied to the surface source flux. , and , where is the weighting coefficient; the first to fourth terms of the joint observation equation correspond to the multi-source observation residual term, the point source sparse regularization term, the surface source spatial smoothing regularization term, and the prior deviation constraint term, respectively.

[0072] Understandably, the point source sparsity constraint preferentially attributes local high-value plumes to a small number of candidate facilities, while the area source spatial smoothness constraint preferentially attributes continuous weak anomalies to the area source background.

[0073] In addition, joint inversion can also be achieved using Bayesian update methods or ensemble Kalman filtering methods based on Bayes' theorem. The Bayesian update method can employ batch update or sequential update strategies; the ensemble Kalman filtering method can use a sequential update strategy. When using a sequential update strategy, multi-source observation data are absorbed batch by batch.

[0074] Specifically, for example, for the first Batch observation, using the current prior state vector and the current prior error covariance matrix Combined with the current joint observation operators of this batch and the current observation error covariance matrix The current posterior state vector is calculated using either Bayesian update or ensemble Kalman filter update formulas. and the current posterior state vector The corresponding posterior error covariance matrix is ​​used as prior information for the next batch of observations; then, the process is repeated batch by batch until all observations have been processed, finally obtaining the jointly inverted posterior state vector. ; For Bayesian update methods employing a batch update strategy, multi-source observation data within the assimilation time window are unified into a joint observation vector. The joint observation operator, observation error covariance matrix, and prior error covariance matrix are then used to jointly update all observation information, yielding the jointly inverted posterior state vector. .

[0075] Furthermore, through the posterior state vector To construct the posterior emission field, firstly by analyzing the posterior state vector... The emission rates of all point source components are summed directly, and the flux density of all area source grids is multiplied by the corresponding grid area and then summed again to obtain the total emission flux of the point sources. Total non-point source emission flux Total emission flux of the park Sample variance of total emission flux in the park The representation is: , , , .

[0076] in, Let be the posterior emission intensity of the i-th point source. Let be the posterior flux of the j-th area source grid. Let k be the total emission flux of the park corresponding to the k-th set member or disturbance sample. Its set average, This refers to the number of set members or the number of perturbed samples. The sample variance of the total emission flux in the park. The corresponding standard deviation is used as a characterization of the uncertainty of total emission flux, thereby constructing a posterior emission field using these data.

[0077] Furthermore, after constructing the posterior emission field, it is necessary to apply the posterior emission field to the joint observation operator H. R The observation residual vector v is then remapped to the observation space of each platform, and thus the observation residual vector v is calculated, which is expressed as: , If local positive residuals continuously appear near candidate facilities in multiple time windows among multiple types of observation platforms, then update the prior weights of the candidate point source or add a new candidate point source; if the residuals are spatially continuous, then update the prior flux or covariance of the surface source in the corresponding functional area; if there is a systematic bias in the residuals of a certain type of observation platform, then update the observation weights or model bias terms of that type of platform.

[0078] The implementation of the various embodiments of this invention is based on programmed processing through a device with processor functionality. Therefore, in practical engineering, the technical solutions and functions of the various embodiments of this invention are encapsulated into various units. Based on this reality, and building upon the above embodiments, the embodiments of this invention provide a joint inversion system for methane emissions at the park scale, oriented towards point-source and area-source collaborative analysis, applicable to methods such as the joint inversion of methane emissions at the park scale, oriented towards point-source and area-source collaborative analysis.

[0079] The system includes: a data acquisition and prior construction unit, used to acquire observation data from multiple observation platforms in the park and construct a point source candidate library and a rasterized area source prior list; a platform observation operator and vector construction unit, used to construct platform observation operators and platform observation vectors corresponding to multiple observation platforms based on the point source candidate library and the rasterized area source prior list; a joint inversion solution unit, used to stack the platform observation operators and platform observation vectors to form a joint observation operator and a joint state vector, and perform joint inversion on the joint state vector to construct the posterior emission field; and an adaptive optimization unit, used to calculate the observation residuals of the posterior emission field and adaptively update them according to the spatial distribution characteristics of the residuals to iteratively optimize the inversion results.

[0080] The specific implementation methods of each unit are the same as those of each step, and will not be described in this invention.

[0081] Based on the above embodiments, the present invention provides a non-transitory computer-readable storage medium that stores computer instructions, which instruct the computer to execute the steps of a joint inversion method for methane emissions at the park scale oriented towards point-source and area-source collaborative analysis.

[0082] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A joint inversion method for methane emissions at the industrial park scale, oriented towards point-source and area-source coordinated analysis, characterized in that, The method includes the following steps: Acquire observation data from various observation platforms in the park, and construct a candidate library of point sources and a rasterized prior list of area sources; Based on the point source candidate library and the area source rasterization prior list, platform observation operators and platform observation vectors corresponding to multiple types of the observation platforms are constructed respectively. The platform observation operator and the platform observation vector are stacked to form a joint observation operator and a joint state vector, and the joint state vector is jointly inverted to construct the posterior emission field. The observation residuals of the posterior emission field are calculated, and the input parameters of the joint inversion are adaptively updated to iteratively optimize the joint inversion results.

2. The method for joint inversion of methane emissions at the industrial park scale based on point source-area source collaborative analysis as described in claim 1, characterized in that, The method for constructing the posterior emission field is as follows: All the platform observation operators are stacked row by row to form a joint observation operator, and the corresponding platform observation vectors are stacked vertically to form a joint state vector; The observation error covariance matrix is ​​determined based on the observation error characteristics of the various observation platforms, and the prior error covariance matrix is ​​determined based on the uncertainty of the point source candidate library and the area source rasterized prior list. Using the joint observation operator, the observation error covariance matrix, and the prior error covariance matrix, the joint state vector is jointly inverted based on Bayes' theorem to obtain the posterior state vector; The point source components in the posterior state vector are mapped to the emission rates at each point source location, and the area source components are mapped to the emission fluxes in each grid region. The posterior emission field is constructed by combining these components according to their spatial locations.

3. The method for joint inversion of methane emissions at the industrial park scale based on point source-area source collaborative analysis as described in claim 1, characterized in that, The observation data from the various observation platforms include methane observation data and auxiliary data. The methane observation data includes at least two of the following: satellite observation data, UAV observation data, mobile observation data, and ground station observation data. The auxiliary data includes meteorological data, background concentration data, park functional zoning data, and emission source prior data.

4. The method for joint inversion of methane emissions at the industrial park scale based on point source-area source collaborative analysis as described in claim 3, characterized in that, The method for constructing the point source candidate library and the area source rasterized prior list is as follows: generate corresponding observation descriptors through methane observation data of each observation platform, and construct the point source candidate library and the area source rasterized prior list based on the observation descriptors and the auxiliary data.

5. The method for joint inversion of methane emissions at the industrial park scale based on point source-area source collaborative analysis as described in claim 1, characterized in that, The platform observation operators include an emission field mapping operator, a meteorological transport response operator, a sensor response operator, and a spatiotemporal sampling operator. The emission field mapping operator is used to map the point source emission intensity of each candidate facility in the point source candidate library and the area source grid flux of each grid in the area source gridded prior list to a unified spatial emission field. The meteorological transport response operator is used to convert the unified spatial emission field into a concentration response field. The sensor response operator includes any one of the following: satellite pixel average kernel, UAV sampling volume, mobile instrument response time, and ground station representative volume. The spatiotemporal sampling operator is used to extract simulated observation values ​​corresponding to the actual observation time and observation location from the concentration response field based on the actual observation time and observation location of each observation platform.

6. The method for joint inversion of methane emissions at the industrial park scale based on point source-area source collaborative analysis as described in claim 1, characterized in that, The joint state vector includes a point source emission intensity sub-vector, a surface source grid quantum vector, a background concentration correction sub-vector, and a model bias correction sub-vector. The background concentration correction sub-vector is used to characterize the background concentration field correction amount in the observation area, and the model bias correction sub-vector is used to characterize the systematic bias correction amount of each observation operator of the platform. The point source emission intensity sub-vector is subject to the position constraints, non-negativity constraints, and sparsity constraints of each candidate facility in the point source candidate library, and the surface source grid quantum vector is subject to spatial grid constraints, functional area prior constraints, non-negativity constraints, and spatial smoothness constraints.

7. The method for joint inversion of methane emissions at the industrial park scale based on point source-area source collaborative analysis as described in claim 1, characterized in that, The joint observation equation includes a multi-source observation residual term, a point source sparse regularization term, a surface source spatial smoothing regularization term, and a priori deviation constraint term. The multi-source observation residual term is determined by the joint observation operator and the observation error covariance matrix. The observation error covariance matrix is ​​used to characterize the sensor errors, representativeness errors, spatiotemporal registration errors, and meteorological transmission errors of various observation platforms.

8. The method for joint inversion of methane emissions at the industrial park scale based on point source-area source collaborative analysis as described in claim 1, characterized in that, The spatial resolution of the posterior emission field is 10–1000 m, and the assimilation time window is 5–1440 min.

9. A joint inversion system for methane emissions at the industrial park scale based on point-source and area-source collaborative analysis, applied to the joint inversion method for methane emissions at the industrial park scale based on point-source and area-source collaborative analysis as described in any one of claims 1 to 8, characterized in that, include: The data acquisition and prior construction unit is used to acquire observation data from various observation platforms in the park and construct a candidate library of point sources and a rasterized prior list of area sources. The platform observation operator and vector construction unit is used to construct platform observation operators and platform observation vectors corresponding to multiple types of the observation platforms based on the point source candidate library and the area source rasterization prior list. The joint inversion solution unit is used to stack the platform observation operator and the platform observation vector to form a joint observation operator and a joint state vector, respectively, and to perform joint inversion on the joint state vector to construct the posterior emission field; An adaptive optimization unit is used to calculate the observation residuals of the posterior emission field and adaptively update them according to the spatial distribution characteristics of the residuals to iteratively optimize the inversion result.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions that cause the computer to perform the steps of the campus-scale methane emission joint inversion method oriented towards point-source-area-source collaborative analysis as described in any one of claims 1 to 8.