Ecological and environmental monitoring systems and methods

Abstract

This application provides an ecological environment monitoring system and method. By jointly detecting jumps in each wavelength channel of the raw spectral data, it identifies and removes anomalous spectra caused by cloud cover and electronic noise, obtaining effective spectral data. Based on atmospheric vertical profile data and all ground pollutant concentrations, it performs source intensity inversion assimilation on all ground pollutant concentrations, and performs slant path optical path segment integration at each observation path to obtain multiple path equivalent pollution levels. Based on the effective spectral data, it determines the initial pollution concentration for each path, cross-validates all path equivalent pollution levels and all initial pollution concentrations, and corrects all initial pollution concentrations to obtain the atmospheric pollution concentration of the target city. Using the scheme of this application, the atmospheric pollution concentration of a city can be accurately monitored based on the anomaly detection of spectral data and the precise conversion from ground point data to path equivalent pollution levels.

Description

Technical Field

[0001] This application relates to the field of environmental monitoring technology, and more specifically, to ecological environment monitoring systems and methods. Background Technology

[0002] Monitoring air pollution concentration is a core aspect of urban ecological environment governance and management. Differential optical absorption spectroscopy has become one of the mainstream technologies for urban air pollution monitoring due to its advantages of non-contact, large-scale, and continuous monitoring. The construction and application of multi-point monitoring networks in cities has also become an industry development trend.

[0003] Currently, most industry solutions acquire urban air pollution concentrations by deploying optical sensors to collect solar scattering spectral data and combining this with data such as wind speed and ground pollutant concentrations. Some solutions only perform simple anomaly screening on the spectral data and conduct basic diffusion analysis to achieve a preliminary conversion between ground data and path data. However, these solutions typically lack multi-dimensional joint judgment logic for anomaly detection in spectral data, making it difficult to accurately identify and eliminate abnormal cloud spectra, and ensuring the acquisition of effective spectral data. Secondly, they fail to effectively adapt ground point data to path data or cross-validate and correct the initial pollution concentration obtained from spectral inversion, resulting in insufficient accuracy of the inverted concentration data and an inability to accurately reflect the actual air pollution status of the target city. Therefore, how to accurately monitor urban air pollution concentrations based on anomaly detection in spectral data and accurate conversion from ground point data to path-equivalent pollution levels has become a challenge for the industry. Summary of the Invention

[0004] This application provides an ecological environment monitoring system and method that can accurately monitor the concentration of air pollution in cities based on the detection of anomalies in spectral data and the precise conversion of ground point data to path equivalent pollution levels.

[0005] Firstly, this application provides a method for monitoring air pollution concentrations at multiple locations in an urban area, applied to an ecological environment monitoring system. The method includes the following steps: Optical sensors were deployed at various monitoring points in the target city, and solar scattering spectra along the observation paths of each monitoring point in the target city were collected synchronously and continuously to obtain raw spectral data. Joint jump detection is performed on each wavelength channel in the original spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining effective spectral data; Meteorological wind field data of the target city was obtained through an ultrasonic anemometer network, along with atmospheric vertical profile data and ground pollutant concentrations at multiple locations. Based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations, source intensity inversion and assimilation are performed, and then slant path optical path segment integration is performed at each observation path to obtain multiple path equivalent pollution levels. Based on the effective spectral data, the initial pollution concentration of each path is determined. Cross-validation is performed based on the equivalent pollution levels of all paths and all initial pollution concentrations. Then, all initial pollution concentrations are corrected to obtain the atmospheric pollution concentration of the target city.

[0006] In some embodiments, the solar scattering spectra along the observation paths of various monitoring points in the target city are collected synchronously and continuously to obtain raw spectral data, specifically including: Passive differential absorption spectrometers were installed at each monitoring point; Through an integrated satellite positioning and timing module, a unified time reference is provided for all monitoring points; The observation path for each monitoring point is determined based on the zenith angle and azimuth sequence in the zenith observation mode. All passive differential absorption spectrometers are controlled to synchronously and continuously acquire solar scattering spectra along various observation paths under a unified time reference, thereby obtaining raw spectral data.

[0007] In some embodiments, joint transition detection is performed on each wavelength channel in the original spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining effective spectral data. Specifically, this includes: Determine the relative jump rate of the light intensity time series of each wavelength channel in the original spectral data at adjacent time points; The sliding window for the data points of each wavelength channel is determined, and then the median absolute deviation of the relative jump rate of all wavelength channels within the sliding window is obtained. By jointly judging all the median absolute deviations and all the relative jump rates and calculating the rate of change of absorption depth of the oxygen A-band channel, multiple abnormal spectral points caused by cloud cover and electronic noise are obtained, resulting in abnormal spectra. Based on the temporal trend of the adjacent valid data points of each abnormal spectral point in the abnormal spectrum, each abnormal spectral point is reconstructed to obtain valid spectral data.

[0008] In some embodiments, acquiring meteorological wind field data of a target city through an ultrasonic anemometer network, and acquiring atmospheric vertical profile data and ground pollutant concentrations at multiple locations in the target city specifically includes: Near-surface meteorological wind field data of the target city are obtained through a network of ultrasonic anemometers distributed in the target city; Vertical stratification data of atmospheric aerosols, temperature, humidity and pressure were obtained by using microwave radiometer and micropulse lidar to obtain atmospheric vertical profile data. Ground concentration data of various pollutants in the target city were obtained using a high-precision point gas analyzer.

[0009] In some embodiments, source intensity inversion and assimilation based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations specifically includes: The atmospheric boundary layer height and vertical diffusion parameters are determined based on the atmospheric vertical profile data. Based on all ground pollutant concentrations, the meteorological wind field data, the boundary layer height, and the vertical diffusion parameters, a four-dimensional variational method considering four types of emission sources—industrial, transportation, residential, and dust—is used to perform source strength inversion assimilation, resulting in the pollutant concentration distribution field.

[0010] In some embodiments, performing slant path optical path segment integration at each observation path to obtain multiple path equivalent contamination degrees specifically includes: Extract discrete grid points from the concentration distribution field obtained by source strength inversion and assimilation for each observation path; Determine the contribution weight of each discrete grid point in each observation path to the total optical path integral of the path. The concentration values ​​of all grid points along each observation path in the concentration distribution field are weighted and integrated according to their contribution weights to obtain the path equivalent pollution degree of the corresponding observation path.

[0011] In some embodiments, determining the initial contamination concentration of each path based on the effective spectral data specifically includes: The standard absorption cross section of the target pollutant gas in the effective spectral data is selected for differential absorption spectral fitting; By minimizing the fitting residuals, the initial pollution concentration along each observation path can be inverted.

[0012] Secondly, this application provides an ecological environment monitoring system, which includes an air pollution concentration monitoring unit, the air pollution concentration monitoring unit comprising: The acquisition module is used to deploy optical sensors at various monitoring points in the target city and synchronously and continuously acquire the solar scattering spectrum of the observation path of each monitoring point in the target city to obtain raw spectral data. The processing module is used to perform joint jump detection on each wavelength channel in the original spectral data, identify and remove abnormal spectra caused by cloud cover and electronic noise in the original spectral data, and thus obtain effective spectral data. The processing module is also used to acquire meteorological wind field data of the target city through the ultrasonic anemometer network, and to acquire atmospheric vertical profile data and ground pollutant concentrations at multiple locations in the target city. The processing module is also used to perform source strength inversion and assimilation based on the atmospheric vertical profile data, meteorological wind field data and all ground pollutant concentrations, and then perform slant path optical path segment integration at each observation path to obtain multiple path equivalent pollution degrees. The execution module is used to determine the initial pollution concentration of each path based on the effective spectral data, perform cross-validation based on the equivalent pollution degree of all paths and all initial pollution concentrations, and then correct all initial pollution concentrations to obtain the air pollution concentration of the target city.

[0013] Thirdly, this application provides a computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described method for monitoring air pollution concentrations at multiple locations in urban areas.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for monitoring air pollution concentrations at multiple locations in a city.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: The ecological environment monitoring system and method provided in this application firstly deploy optical sensors at various monitoring points in the target city and simultaneously and continuously collect solar scattering spectra along the observation paths of each monitoring point in the target city to obtain raw spectral data; jointly detect jumps in each wavelength channel of the raw spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining effective spectral data; acquire meteorological wind field data of the target city through an ultrasonic anemometer network, and acquire atmospheric vertical profile data and ground pollutant concentrations at multiple locations in the target city; perform source intensity inversion assimilation based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations, and then perform slant path optical path segment integration at each observation path to obtain multiple path equivalent pollution levels; determine the initial pollution concentration of each path based on the effective spectral data, cross-validate all path equivalent pollution levels and all initial pollution concentrations, and then correct all initial pollution concentrations to obtain the atmospheric pollution concentration of the target city.

[0016] Therefore, in the process of this application's method for monitoring air pollution concentration at multiple locations in a city, firstly, optical sensors are deployed at various monitoring points in the target city, and the solar scattering spectrum along the observation path of each monitoring point is collected synchronously and continuously to obtain raw spectral data. Then, joint jump detection is performed on each wavelength channel in the raw spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining effective spectral data. Through multi-dimensional joint jump detection, abnormal spectra affected by external interference are accurately identified and removed, effectively eliminating the influence of factors such as cloud cover and electronic noise on the spectral data. The obtained effective spectral data avoids abnormal... The monitoring bias introduced by the data provides a reliable and accurate spectral basis for subsequent pollution concentration inversion. Secondly, source intensity inversion and assimilation are performed based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations. Then, slant path optical path segment integration is performed at each observation path to obtain multiple path equivalent pollution levels. The source intensity inversion and assimilation combined with atmospheric vertical profile data can accurately restore the distribution characteristics of pollutants in three-dimensional space. Slant path optical path segment integration realizes the effective conversion of ground point concentration data to path integrated concentration data. The above scheme can accurately monitor the atmospheric pollution concentration in cities based on the anomaly detection of spectral data and the accurate conversion of ground point data to path equivalent pollution levels. Attached Figure Description

[0017] Figure 1 This is an exemplary flowchart of a method for monitoring air pollution concentration at multiple locations in an urban area, according to some embodiments of this application. Figure 2 This is an exemplary flowchart illustrating the determination of valid spectral data according to some embodiments of this application; Figure 3 This is an exemplary flowchart illustrating the determination of path equivalent contamination according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of an air pollution concentration monitoring unit according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing a method for monitoring air pollution concentration at multiple locations in an urban area, according to some embodiments of this application. Detailed Implementation

[0018] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0019] refer to Figure 1 The figure is an exemplary flowchart of a method for monitoring air pollution concentration at multiple locations in an urban area according to some embodiments of this application. The method for monitoring air pollution concentration at multiple locations in an urban area mainly includes the following steps: In step 101, optical sensors are deployed at various monitoring points in the target city, and solar scattering spectra along the observation paths of each monitoring point in the target city are collected synchronously and continuously to obtain raw spectral data.

[0020] In some embodiments, the solar scattering spectra along the observation paths of various monitoring points in the target city are synchronously and continuously acquired to obtain the raw spectral data, which can be achieved through the following steps: Passive differential absorption spectrometers were installed at each monitoring point; Through an integrated satellite positioning and timing module, a unified time reference is provided for all monitoring points; The observation path for each monitoring point is determined based on the zenith angle and azimuth sequence in the zenith observation mode. All passive differential absorption spectrometers are controlled to synchronously and continuously acquire solar scattering spectra along various observation paths under a unified time reference, thereby obtaining raw spectral data.

[0021] In specific implementation, the observation path of each monitoring point can be determined based on the sequence of zenith angle and azimuth angle in the zenith observation mode in the following way: a set of combination sequences of zenith angle and azimuth angle is preset, where the zenith angle defines the angle between the line of sight and the vertical direction, and the azimuth angle defines the projection direction of the line of sight on the horizontal plane; the telescope is driven to point according to each preset set of angles, and the straight line swept by the telescope's line of sight in three-dimensional space is the observation path at that moment. The observation path determines the scattered sunlight received by the sensor from different regions of the atmosphere and different optical paths; other methods can also be used in other embodiments, which are not limited here.

[0022] In specific implementation, controlling all passive differential absorption spectrometers to synchronously and continuously acquire solar scattering spectra along various observation paths under a unified time reference can be achieved in the following way: A central server or a local timer triggered by a timing module sends synchronous acquisition commands to the passive differential absorption spectrometers at all monitoring points under a unified time reference. Then, each passive differential absorption spectrometer, according to the commands, sequentially drives its gimbal to a preset observation path angle. After each angle stabilizes, the passive differential absorption spectrometer performs spectral dispersion and photoelectric conversion on the sky scattering light collected by the telescope, converting the light signal into a digital spectrum containing wavelength and intensity information. Finally, the spectral data with time, location, and path labels is transmitted back as the raw spectral data. Other embodiments may also employ other methods, which are not limited here.

[0023] It should be noted that the original spectral data in this application were synchronously acquired by passive differential absorption spectrometers at each monitoring point. Essentially, it is a set of spectra arranged in chronological order. Each spectrum contains the wavelength channels acquired at each sampling time and their corresponding light intensity measurements, along with the timestamp when the spectrum was acquired, the location of the monitoring point, and the observation path information.

[0024] In step 102, joint jump detection is performed on each wavelength channel in the original spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining effective spectral data.

[0025] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart for determining valid spectral data in some embodiments of this application. In this embodiment, joint transition detection is performed on each wavelength channel in the original spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining valid spectral data. This can be achieved by the following steps: First, in step 1021, the relative jump rate of the light intensity time series of each wavelength channel in the original spectral data at adjacent time points is determined; Secondly, in step 1022, the sliding window of the data points of each wavelength channel is determined, and then the median absolute deviation of the relative jump rate of all wavelength channels within the sliding window is obtained. Furthermore, in step 1023, all median absolute deviations and all relative jump rates are jointly judged and the absorption depth change rate of the oxygen A-band channel is calculated to obtain multiple abnormal spectral points caused by cloud cover and electronic noise, thus obtaining abnormal spectra. Finally, in step 1024, each abnormal spectral point is reconstructed based on the temporal trend of its adjacent valid data points in the abnormal spectrum to obtain valid spectral data.

[0026] In specific implementation, determining the relative jump rate of the light intensity time series of each wavelength channel in the original spectral data at adjacent moments can be achieved in the following way: For each wavelength channel in the original spectral data, the light intensity values ​​measured at any two consecutive sampling moments are compared, the change amplitude of the light intensity at the later moment relative to the light intensity at the previous moment is calculated, and the change amplitude is divided by the light intensity value at the previous moment to obtain the dimensionless relative jump rate for the corresponding two consecutive sampling moments. The relative jump rate is used to quantify the degree of abrupt change of the signal at each wavelength point at adjacent moments, thereby obtaining multiple relative jump rates for the corresponding wavelength channel. The multiple relative jump rates for the remaining wavelength channels are then determined. Other embodiments may also use other methods, which are not limited here.

[0027] In specific implementation, determining the sliding window of data points for each wavelength channel and then obtaining the median absolute deviation of the relative jump rate of all wavelength channels within the sliding window can be achieved in the following way: Select a data point as the selected data point, and take the sampling time of the selected data point as the center, select data from several consecutive moments before and after the selected data point (e.g., 5 minutes before and after each moment) to form a time sliding window; calculate the median absolute deviation of the relative jump rate of all wavelength channels at each moment within the above time sliding window. The calculation process of the median absolute deviation is as follows: first, calculate the median of all jump rates within the window, then calculate the absolute value of the difference between each relative jump rate and the median, and finally take the median of these absolute values; other embodiments may also use other methods to achieve this, which are not limited here.

[0028] In practice, all median absolute deviations and all relative jump rates are jointly judged, and the rate of change of absorption depth of the oxygen A band channel is calculated to obtain multiple abnormal spectral points caused by cloud cover and electronic noise. The abnormal spectra can be obtained in the following way: First, three judgment thresholds are set: jump threshold, deviation threshold, and oxygen absorption change threshold. The high percentile value (e.g., the 99th percentile) of the statistical distribution of relative jump rates for all wavelength channels during historical normal monitoring periods is calculated. This value is used as an empirical jump threshold for relative jump rates to determine whether the passive differential absorption spectrum signal has changed drastically. The relative jump rate of each data point is divided by the median absolute deviation calculated by the sliding window of the corresponding data point to obtain a standardized deviation. This threshold is usually set to a constant greater than 1 to determine whether the jump is significantly abnormal in a local time. The oxygen A band channel absorption depth change rate is calculated. The absorption depth of the channel (760-770nm) is the ratio of the maximum light intensity to the minimum light intensity in the channel. The rate of change of absorption depth between adjacent time points is calculated, and an oxygen absorption change threshold is set, for example, 10%, to distinguish between the overall light intensity decrease caused by cloud cover and the actual change in pollutant concentration. A data point is identified and marked as an abnormal spectral point when any of the following conditions are met: 1. The sag rate of a data point in any wavelength channel exceeds the sag threshold, and its standardized deviation exceeds the deviation threshold (electronic noise determination); or 2. The sag rate of all wavelength channels exceeds the sag threshold, and the standardized deviation exceeds the deviation threshold, while the rate of change of oxygen A-band absorption depth exceeds the set threshold (cloud obstruction determination). The remaining data points are further evaluated and labeled. Multiple anomalous spectral points caused by cloud cover and electronic noise are grouped together as anomalous spectra. Based on the dual criteria of statistical distribution and local consistency test, combined with the physical constraint of the oxygen A band reference channel, it is ensured that only those transient interference points that have both global drastic changes and significant outliers in the local time series are accurately identified, thereby effectively distinguishing anomalous spectral points caused by rapid cloud cover or electronic noise from the actual pollutant concentration change signal. In other embodiments, adaptive thresholds or other statistical criteria can also be used, which are not limited here.

[0029] In specific implementation, the effective spectral data can be obtained by reconstructing each abnormal spectral point based on the temporal trend of its adjacent effective data points in the abnormal spectrum. This can be achieved in the following way: for each spectral data point marked as abnormal (i.e., an abnormal spectral point), it is removed; then, the light intensity value of the nearest effective (non-abnormal) data point before and after the abnormal spectral point is used to perform linear interpolation calculation, and the calculated estimated value is used to replace the original abnormal value, thereby filling the gaps and repairing the curve in the time series, and finally obtaining a temporally continuous effective spectral data sequence that has eliminated instantaneous interference, i.e., effective spectral data; other embodiments may also use other methods, which are not limited here.

[0030] It should be noted that the valid spectral data in this application refers to a continuous and reliable spectral data sequence that has completed outlier removal and repair. This data is used as direct input to the differential absorption spectroscopy inversion algorithm to obtain a more stable and accurate initial pollutant column concentration, thus laying a reliable data foundation for subsequent high-precision calibration.

[0031] In step 103, meteorological wind field data of the target city is acquired through an ultrasonic anemometer network, and atmospheric vertical profile data and ground pollutant concentrations at multiple locations in the target city are also acquired.

[0032] In some embodiments, acquiring meteorological wind field data of a target city through an ultrasonic anemometer network, and acquiring atmospheric vertical profile data and ground pollutant concentrations at multiple locations in the target city can be achieved through the following steps: Near-surface meteorological wind field data of the target city are obtained through a network of ultrasonic anemometers distributed in the target city; Vertical stratification data of atmospheric aerosols, temperature, humidity and pressure were obtained by using microwave radiometer and micropulse lidar to obtain atmospheric vertical profile data. Ground concentration data of various pollutants in the target city were obtained using a high-precision point gas analyzer.

[0033] In specific implementation, obtaining near-surface meteorological wind field data of the target city through a network of ultrasonic anemometers distributed in the target city can be achieved in the following way: continuously measuring the horizontal wind speed and direction at the corresponding location using the existing environmental monitoring network or multiple specially deployed ultrasonic anemometers in the target city, and collecting the real-time data through the network to form meteorological wind field data describing the horizontal air flow near the target city. The meteorological wind field data refers to a vector dataset with spatiotemporal identification used to characterize the horizontal motion state of the atmosphere near the ground, including at least the horizontal wind speed and horizontal wind direction of each monitoring point; other methods can also be used in other embodiments, which are not limited here.

[0034] In specific implementation, atmospheric vertical profile data can be obtained by acquiring vertical stratification data of atmospheric aerosols, temperature, humidity, and pressure based on microwave radiometers and micropulse lidars. This can be achieved in the following way: One microwave radiometer and one micropulse lidar are set up within the monitoring area of ​​each monitoring point in the target city. The microwave radiometer receives microwave radiation emitted by the atmosphere itself and inverts the vertical profiles of temperature and humidity. The micropulse lidar emits a laser beam vertically into the atmosphere and receives backscattered light signals from aerosols at different altitudes. Next, by analyzing the time delay and intensity of the received backscattered light signals, the distribution of aerosol extinction coefficient with altitude is inverted. Finally, the temperature and humidity data inverted by the microwave radiometer and the ground pressure measurement values ​​are combined to form the atmospheric vertical profile data of the vertical stratification structure of atmospheric physical properties. Other embodiments may also use other methods, which are not limited here.

[0035] In specific implementation, obtaining ground-level concentration data of multiple pollutants in a target city using a high-precision point gas analyzer can be achieved in the following way: Data from automatic ambient air quality monitoring stations that have been constructed and operated by the target city's ecological and environmental departments and meet national air quality monitoring standards can be retrieved. These monitoring stations are equipped with high-precision point gas analyzers based on principles such as ultraviolet fluorescence and chemiluminescence, thereby continuously and accurately measuring the concentrations of pollutants such as sulfur dioxide and nitrogen dioxide at each monitoring point. The set of all the above parameters is used as the ground-level concentration data of multiple pollutants in the target city. Other methods can also be used in other embodiments, which are not limited here.

[0036] It should be noted that the meteorological wind field data in this application are vector data describing the direction and speed of horizontal atmospheric movement; the atmospheric vertical profile data are vertical distribution data characterizing the relationship between atmospheric physicochemical properties (such as aerosol density, temperature, and humidity) and altitude; and the ground pollutant concentration data refer to discrete point measurements that provide a benchmark for the actual near-ground pollutant concentration.

[0037] In step 104, source intensity inversion and assimilation are performed based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations. Then, slant path optical path segment integration is performed at each observation path to obtain multiple path equivalent pollution levels.

[0038] In some embodiments, source intensity inversion and assimilation based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations can be achieved using the following steps: The atmospheric boundary layer height and vertical diffusion parameters are determined based on the atmospheric vertical profile data. Based on all ground pollutant concentrations, the meteorological wind field data, the boundary layer height, and the vertical diffusion parameters, a four-dimensional variational method considering four types of emission sources—industrial, transportation, residential, and dust—is used to perform source strength inversion assimilation, resulting in the pollutant concentration distribution field.

[0039] In practice, determining the atmospheric boundary layer height and vertical diffusion parameters based on the atmospheric vertical profile data can be achieved as follows: Extract the vertical distribution data of temperature (or calculate the potential temperature) from the atmospheric vertical profile data. Start searching upwards from near the ground to find the height at which the potential temperature gradient changes significantly. Typically, within the boundary layer, the potential temperature changes little with altitude (uniform mixing); above the top of the boundary layer, the potential temperature increases significantly with altitude (stable stratification). The starting height at which the vertical potential temperature gradient first shows a sustained and significant increase is determined as the top height of the atmospheric boundary layer. Simultaneously, the vertical profile of the extinction coefficient obtained from aerosol lidar inversion can be analyzed, showing a rapid and continuous decrease in aerosol concentration from the high value region near the ground upwards. The height of the clean background value is used as another criterion for the boundary layer height. Below the determined boundary layer height, key meteorological elements such as the average wind speed within the boundary layer in the profile data, the frictional velocity measured or estimated at the ground, and the surface heat flux (which can be derived from the temperature profile and wind speed profile) are used. Then, a parameterization scheme based on the Monin-Obukhov similarity theory is applied to calculate the key parameter characterizing atmospheric stability—the Monin-Obukhov length. Based on this length value and the observation height, the corresponding dimensionless wind speed and temperature gradient universal functions are selected. Finally, the vertical turbulent diffusion coefficient, i.e., the vertical diffusion parameter, which describes the vertical transport capacity of turbulent motion to pollutants, is calculated. Other methods can also be used in other embodiments, which are not limited here.

[0040] In specific implementation, based on all ground pollutant concentrations, the meteorological wind field data, the boundary layer height, and the vertical diffusion parameters, a four-dimensional variational method considering four types of emission sources—industrial, traffic, residential, and dust—is used for source strength inversion and assimilation. The resulting pollutant concentration distribution field can be achieved as follows: a four-dimensional variational assimilation system based on a Gaussian plume diffusion model is constructed, dividing the target city into several emission source grids, with the emission source strength of each grid serving as the control variable to be inverted; the observed pollutant concentration values ​​from all ground monitoring points are used as observation constraints for the assimilation system; the meteorological wind field data, boundary layer height, and vertical diffusion parameters are used as input parameters for the assimilation system; the control variables are iteratively optimized to minimize the sum of squared residuals between the ground concentration simulated by the diffusion model and the observed concentration, thereby obtaining the optimal emission source strength distribution; finally, using the optimal emission source strength distribution and real-time meteorological conditions, the diffusion model is run to obtain a three-dimensional pollutant concentration distribution field covering the entire target city; other embodiments may also use ensemble Kalman filtering or other assimilation methods, which are not limited here.

[0041] It should be noted that the Gaussian plume diffusion model is based on the aforementioned input data and operates within a defined three-dimensional spatial computational grid. The core computational steps of this model are: for each ground-based pollution source, simulating the concentration contribution of the pollution plume released under the current wind field and turbulence conditions in each grid cell of the three-dimensional space; finally, linearly superimposing the concentration contribution values ​​of all pollution sources to the same grid cell to generate a three-dimensional pollutant concentration distribution field covering the entire target urban area with a specified spatiotemporal resolution; wherein, the concentration distribution field is a digital matrix containing spatial coordinates and concentration values, with its spatial dimension covering the horizontal area and vertical height, and its temporal dimension aligned with the timestamp of the input data.

[0042] In some embodiments, reference Figure 3 As shown, this figure is an exemplary flowchart for determining the equivalent contamination degree of a path in some embodiments of this application. In this embodiment, the slant path optical path segment integration is performed at each observation path to obtain multiple equivalent path contamination degrees, which can be achieved by the following steps: First, in step 1041, discrete grid points in the concentration distribution field obtained by source strength inversion assimilation are extracted for each observation path; Secondly, in step 1042, the contribution weight of all discrete grid points in each observation path to the total optical path integral of the path is determined. Finally, in step 1043, the concentration values ​​of all grid points along each observation path in the concentration distribution field are weighted and integrated according to their contribution weights to obtain the path equivalent pollution degree of the corresponding observation path.

[0043] In specific implementation, extracting discrete grid points from the concentration distribution field obtained by source intensity inversion and assimilation for each observation path can be achieved in the following way: First, for each passive differential absorption spectrometer at a monitoring point, obtain its precise geographical location at a specific sampling time and the instrument's observation pointing parameters at that time, defined by the zenith angle and azimuth angle; based on the above parameters, calculate the mathematical expression in three-dimensional space of the observation path starting from each instrument position and pointing to a specific sky direction, i.e., a ray defined by the starting point and direction vector; then, for each calculated observation path ray, perform geometric optical path tracing in the concentration distribution field. This tracing process is completed through the following steps: viewing the three-dimensional space of the concentration distribution field... The path is a set of regular cubic grid cells. Starting from the origin of the path ray, it moves forward step by step along the ray direction. In each step, the three-dimensional spatial coordinates of the current point are calculated, and it is determined which grid cell's boundary range the coordinates fall into. When the path first enters a new grid cell, the grid cell is marked as passed, and the three-dimensional coordinates of its center point are recorded. This step-by-step moving and judgment process continues until the path ray completely passes out of the spatial range defined by the concentration distribution site. Finally, a list containing the coordinates of the center points of all the grid cells passed is generated for the observation path, which is the discrete grid point set of the spatial sampling points traversed by the observation path. Other embodiments may also use other methods to implement this, which are not limited here.

[0044] In specific implementation, the contribution weight of all discrete grid points in each observation path to the total optical path integral can be determined in the following way: For each discrete grid point traversed on an observation path, the distance from the center of the grid point to the instrument position on the observation path is calculated. The core idea of ​​inverse distance weighting is then adopted, that is, the closer the grid point is, the greater its contribution to the total absorption on the entire optical path. Thus, a weight is assigned to each grid point, and the weight value is inversely proportional to the square of the distance from the grid point to the instrument. The weights of all grid points on the corresponding observation path are then normalized to ensure that the sum of all weights is one, thereby obtaining the contribution weight of each grid point on each observation path. Other embodiments may also use other methods, which are not limited here.

[0045] In practice, the path-equivalent contamination degree of the corresponding observation path is obtained by weighted integration of the concentration values ​​of all grid points along each observation path in the concentration distribution field according to their contribution weights. This can be achieved as follows: For an observation path, the predicted pollutant concentration value corresponding to each grid point it passes through in the concentration distribution field is extracted and multiplied by the corresponding contribution weight (i.e., weighting). Then, all these weighted concentration values ​​are summed, and the summation result is used as the integral average of the pollutant concentration along the actual observation path optical path, which is the path-equivalent contamination degree that can be directly compared with the column concentration obtained from the spectral inversion. Here, the light signal received by the observation instrument is the integral result of the pollutant absorption along the entire inclined observation path (optical path), and its physical essence is... It is a continuous integral of the concentration at each point along the path; however, the concentration distribution field obtained through source intensity inversion assimilation is composed of a discrete three-dimensional grid. Therefore, to obtain a numerical benchmark that can be directly compared with the instrument measurement (optical path integral concentration), it is necessary to perform an integration operation on this discrete three-dimensional concentration field along a specific spatial oblique line (i.e., the observation path). In this way, by using the three-dimensional concentration information provided by the diffusion model and the geometric optical path tracing technology, the concentration data of points that originally only represented the ground situation can be transformed into a path integral concentration reference value that is completely consistent with the physical meaning of the optical sensor observation and has the same dimensions. This solves the industry problem of the lack of reliable true value calibration for spectral inversion results. Other embodiments can also be implemented in other ways, which are not limited here.

[0046] It should be noted that the concentration distribution field in this application is a digital model used to describe the predicted concentration of pollutants at each point in three-dimensional space; the contribution weight to the total optical path integral of the path is a proportionality coefficient dynamically calculated based on the spatial geometric relationship (such as inverse distance) between discrete grid points and the observation instrument, used to quantify the contribution share of the concentration in each grid cell to the total optical path absorption of the entire path; the path equivalent pollution degree of the observation path is the optical path integral concentration obtained by weighted integration of the concentration values ​​of all grid cells traversed by the path, used as a physical reference benchmark for direct comparison and calibration with the inversion results of the passive differential absorption spectrometer, which facilitates the provision of a supervision signal with clear physical meaning and spatiotemporal matching for the deep learning model, thereby realizing a reliable mapping from ground point concentration to air path concentration and system error correction.

[0047] In step 105, the initial pollution concentration of each path is determined based on the effective spectral data. Cross-validation is performed based on the equivalent pollution levels of all paths and all initial pollution concentrations, and then all initial pollution concentrations are corrected to obtain the air pollution concentration of the target city.

[0048] In some embodiments, determining the initial contamination concentration for each path based on the effective spectral data can be achieved using the following steps: The standard absorption cross section of the target pollutant gas in the effective spectral data is selected for differential absorption spectral fitting; By minimizing the fitting residuals, the initial pollution concentration along each observation path can be inverted.

[0049] In practice, selecting the standard absorption cross section of the target pollutant gas from the effective spectral data for differential absorption spectral fitting can be achieved as follows: Read the standard absorption cross section data of the target pollutant (e.g., nitrogen dioxide) within the instrument's monitoring wavelength range from a standard atmospheric molecular spectral database (such as HITRAN). This standard absorption cross section data describes the gas's absorption capacity for different wavelengths of light. Simultaneously input the preprocessed effective spectral data and these standard absorption cross section data into the differential absorption spectral inversion algorithm for differential absorption spectral fitting. By minimizing the fitting residual, the initial absorption cross section along each observation path can be inverted. The initial pollution concentration can be achieved in the following way: In the inversion algorithm, the concentration of the pollutant column to be determined is used as the key adjustable parameter. The algorithm is based on the Beer-Lambert law and uses the concentration parameter and standard absorption cross section to calculate a theoretical spectrum. By continuously adjusting the concentration parameter, the difference between the calculated theoretical spectrum and the measured effective spectral data (i.e., the fitting residual) is minimized. When the residual is minimized, the concentration parameter value used is determined as the initial pollution concentration obtained by inversion along the observation path. The initial pollution concentration of the remaining observation paths is then determined. Other methods can also be used in other embodiments, which are not limited here.

[0050] It should be noted that the initial pollution concentration in this application refers to the pollutant column concentration directly obtained by performing differential absorption spectral fitting and residual inversion calculation on the effective spectral data sequence, without undergoing systematic error calibration. This concentration is used as a direct input feature of the deep learning calibration model, which facilitates the model to learn the nonlinear error relationship between the concentration and the high-precision path reference concentration, thereby realizing batch, automatic, and intelligent correction of the original data of the entire monitoring network.

[0051] In some embodiments, the atmospheric pollution concentration of the target city can be obtained by cross-validating all path-equivalent pollution levels and all initial pollution concentrations, and then correcting all initial pollution concentrations using the following steps: All path equivalent pollution levels are used as monitoring signals, and training sample pairs are formed with the initial pollution concentrations at the corresponding times and monitoring points. A deep neural network model integrating residual connections and attention mechanisms is constructed and trained with initial pollution concentration and synchronous meteorological and solar geometric parameters as inputs and path equivalent pollution degree as the output target. Using the trained model, the initial pollution concentration of the observation path at each monitoring point is automatically corrected; The atmospheric pollution concentration of the target city is generated based on all corrected pollution concentrations.

[0052] In specific implementation, a deep neural network model integrating residual connections and attention mechanisms is constructed. The model is trained using initial pollution concentration and synchronous meteorological and solar geometric parameters as inputs, with path-equivalent pollution level as the output target. This can be achieved as follows: A deep neural network is designed, containing residual blocks to alleviate the training difficulties of deep networks; an attention layer is embedded in the network to enable the model to automatically focus on input features that have a greater impact on calibration; parameters such as initial pollution concentration, synchronous temperature, humidity, wind speed, and solar zenith angle from the training samples are input into the network. The training objective of the network is to make its output value as close as possible to the corresponding path-equivalent pollution level in the samples. The internal parameters of the network are iteratively optimized through a large number of samples to complete model training. Finally, a trained model that automatically corrects the initial pollution concentration of the observation path at each monitoring point is obtained. Other implementation methods can also be used in other embodiments, which are not limited here.

[0053] Furthermore, in another aspect of this application, in some embodiments, this application provides an ecological environment monitoring system, which includes an air pollution concentration monitoring unit, with reference to... Figure 4 The figure is a schematic diagram of the structure of an air pollution concentration monitoring unit according to some embodiments of this application. The air pollution concentration monitoring unit includes: a data acquisition module 401, a processing module 402, and an execution module 403, which are described below: The acquisition module 401 in this application is mainly used to deploy optical sensors at various monitoring points in the target city and synchronously and continuously acquire the solar scattering spectrum of the observation path of each monitoring point in the target city to obtain raw spectral data. Processing module 402, in this application, is used to perform joint jump detection on each wavelength channel in the original spectral data, identify and remove abnormal spectra caused by cloud cover and electronic noise in the original spectral data, and thus obtain effective spectral data; It should be noted that the processing module 402 in this application is also used to acquire meteorological wind field data of the target city through the ultrasonic anemometer network, and to acquire atmospheric vertical profile data and ground pollutant concentrations at multiple locations in the target city. In addition, it should be noted that the processing module 402 in this application is also used to perform source intensity inversion assimilation based on the atmospheric vertical profile data, meteorological wind field data and all ground pollutant concentrations, and then perform slant path optical path segment integration at each observation path to obtain multiple path equivalent pollution degrees. The execution module 403 in this application is mainly used to determine the initial pollution concentration of each path based on the effective spectral data, perform cross-validation based on the equivalent pollution degree of all paths and all initial pollution concentrations, and then correct all initial pollution concentrations to obtain the air pollution concentration of the target city.

[0054] In addition, this application also provides a computer device, the computer device including a memory and a processor, the memory storing code, the processor being configured to acquire the code and execute the above-described method for monitoring air pollution concentration at multiple locations in a city.

[0055] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device for implementing a method for monitoring air pollution concentration at multiple locations in an urban area, according to some embodiments of this application. The method for monitoring air pollution concentration at multiple locations in an urban area as described above can... Figure 5 The computer device shown is used to implement this, and the computer device includes at least one processor 501, a communication bus 502, a memory 503, and at least one communication interface 504.

[0056] Processor 501 can be a general-purpose central processing unit (CPU) or an application-specific integrated circuit (ASIC).

[0057] The communication bus 502 can be used to transmit information between the aforementioned components.

[0058] Memory 503 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CDROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. Memory 503 may exist independently and be connected to processor 501 via communication bus 502. Memory 503 may also be integrated with processor 501.

[0059] The memory 503 stores program code for executing the scheme of this application, and its execution is controlled by the processor 501. The processor 501 executes the program code stored in the memory 503. The program code may include one or more software modules. The method used in the above embodiments can be implemented by the processor 501 and one or more software modules in the program code in the memory 503.

[0060] Communication interface 504 uses any transceiver-like device to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0061] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single CPU) processor or a multi-core (multi CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0062] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0063] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for monitoring air pollution concentrations at multiple locations in a city.

[0064] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0065] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for monitoring air pollution concentrations at multiple locations in an urban area, applied to an ecological environment monitoring system, characterized in that, The method includes the following steps: Optical sensors were deployed at various monitoring points in the target city, and solar scattering spectra along the observation paths of each monitoring point in the target city were collected synchronously and continuously to obtain raw spectral data. Joint jump detection is performed on each wavelength channel in the original spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining effective spectral data; Meteorological wind field data of the target city was obtained through an ultrasonic anemometer network, along with atmospheric vertical profile data and ground pollutant concentrations at multiple locations. Based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations, source intensity inversion and assimilation are performed, and then slant path optical path segment integration is performed at each observation path to obtain multiple path equivalent pollution levels. Based on the effective spectral data, the initial pollution concentration of each path is determined. Cross-validation is performed based on the equivalent pollution levels of all paths and all initial pollution concentrations. Then, all initial pollution concentrations are corrected to obtain the atmospheric pollution concentration of the target city.

2. The method as described in claim 1, characterized in that, The solar scattering spectra along the observation paths of various monitoring points in the target city were collected synchronously and continuously to obtain the raw spectral data, which specifically included: Passive differential absorption spectrometers were installed at each monitoring point; Through an integrated satellite positioning and timing module, a unified time reference is provided for all monitoring points; The observation path for each monitoring point is determined based on the zenith angle and azimuth sequence in the zenith observation mode. All passive differential absorption spectrometers are controlled to synchronously and continuously acquire solar scattering spectra along various observation paths under a unified time reference, thereby obtaining raw spectral data.

3. The method as described in claim 1, characterized in that, Joint transition detection is performed on each wavelength channel of the original spectral data to identify and remove abnormal spectra caused by cloud cover and electronic noise, thereby obtaining effective spectral data, specifically including: Determine the relative jump rate of the light intensity time series of each wavelength channel in the original spectral data at adjacent time points; The sliding window for the data points of each wavelength channel is determined, and then the median absolute deviation of the relative jump rate of all wavelength channels within the sliding window is obtained. By jointly judging all the median absolute deviations and all the relative jump rates and calculating the rate of change of absorption depth of the oxygen A-band channel, multiple abnormal spectral points caused by cloud cover and electronic noise are obtained, resulting in abnormal spectra. Based on the temporal trend of the adjacent valid data points of each abnormal spectral point in the abnormal spectrum, each abnormal spectral point is reconstructed to obtain valid spectral data.

4. The method as described in claim 1, characterized in that, Meteorological wind field data for the target city was acquired through an ultrasonic anemometer network, along with atmospheric vertical profile data and ground-level pollutant concentrations at multiple locations. Near-surface meteorological wind field data of the target city are obtained through a network of ultrasonic anemometers distributed in the target city; Vertical stratification data of atmospheric aerosols, temperature, humidity and pressure were obtained by using microwave radiometer and micropulse lidar to obtain atmospheric vertical profile data. Ground concentration data of various pollutants in the target city were obtained using a high-precision point gas analyzer.

5. The method as described in claim 1, characterized in that, The source intensity inversion and assimilation based on the atmospheric vertical profile data, meteorological wind field data, and all ground pollutant concentrations specifically includes: The atmospheric boundary layer height and vertical diffusion parameters are determined based on the atmospheric vertical profile data. Based on all ground pollutant concentrations, the meteorological wind field data, the boundary layer height, and the vertical diffusion parameters, a four-dimensional variational method considering four types of emission sources—industrial, transportation, residential, and dust—is used to perform source strength inversion assimilation, resulting in the pollutant concentration distribution field.

6. The method as described in claim 1, characterized in that, Slant path optical path segment integration is performed at each observation path to obtain the equivalent contamination degree of multiple paths, specifically including: Extract discrete grid points from the concentration distribution field obtained by source strength inversion and assimilation for each observation path; Determine the contribution weight of each discrete grid point in each observation path to the total optical path integral of the path. The concentration values ​​of all grid points along each observation path in the concentration distribution field are weighted and integrated according to their contribution weights to obtain the path equivalent pollution degree of the corresponding observation path.

7. The method as described in claim 1, characterized in that, Determining the initial pollution concentration for each path based on the effective spectral data specifically includes: The standard absorption cross section of the target pollutant gas in the effective spectral data is selected for differential absorption spectral fitting; By minimizing the fitting residuals, the initial pollution concentration along each observation path can be inverted.

8. An ecological environment monitoring system, comprising an air pollution concentration monitoring unit, characterized in that, The air pollution concentration monitoring unit includes: The acquisition module is used to deploy optical sensors at various monitoring points in the target city and synchronously and continuously acquire the solar scattering spectrum of the observation path of each monitoring point in the target city to obtain raw spectral data. The processing module is used to perform joint jump detection on each wavelength channel in the original spectral data, identify and remove abnormal spectra caused by cloud cover and electronic noise in the original spectral data, and thus obtain effective spectral data. The processing module is also used to acquire meteorological wind field data of the target city through the ultrasonic anemometer network, and to acquire atmospheric vertical profile data and ground pollutant concentrations at multiple locations in the target city. The processing module is also used to perform source strength inversion and assimilation based on the atmospheric vertical profile data, meteorological wind field data and all ground pollutant concentrations, and then perform slant path optical path segment integration at each observation path to obtain multiple path equivalent pollution degrees. The execution module is used to determine the initial pollution concentration of each path based on the effective spectral data, perform cross-validation based on the equivalent pollution degree of all paths and all initial pollution concentrations, and then correct all initial pollution concentrations to obtain the air pollution concentration of the target city.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the urban multi-point air pollution concentration monitoring method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for monitoring air pollution concentration at multiple locations in the city as described in any one of claims 1 to 7.

Images