RU2463630C1 - Method to detect sources of emissions into atmosphere by images of megacities - Google Patents

Abstract

FIELD: ecology.

SUBSTANCE: method is proposed to detect sources of emissions into atmosphere by images of megacities. A multispectral image of a region is produced remotely with high resolution (≤2 m/pixel), containing control industrial sites, in the form of digital values of spectral brightness of the image I(x,y) in the red stripe of the visible spectrum (570…670 nm). Contours are identified on images by methods of spatial differentiation with established brightness gradations, lineation analysis is carried out on gradient fields inside identified contours, and graphical images are produced for sources of emissions in the form of aligned lineaments. The point of lineaments crossing is identified with location of a source of emission. Image pixel brightness values are normalised relative to maximum brightness, and pixel brightness is calibrated in PLC values by their values for control sites.

EFFECT: higher statistical stability, efficiency and validity of megacity atmosphere environmental condition assessment results.

7 dwg, 1 tbl

Description

The invention relates to the field of ecology, in particular to remote methods for monitoring natural environments, and can find application in systems of sanitary and epidemiological control of industrial regions.

Industrial progress is inevitably associated with an increase in emissions of so-called "greenhouse" gases into the atmosphere, which are one of the causes of global climate change on the planet. Monitoring the state of air pollution is an integral part of the obligations of the states that have signed the Kyoto Protocol for environmental monitoring of natural environments.

A methodological problem in environmental assessment is the documentary control of the compliance of the actual emissions with each enterprise in the region with the established limit value, according to the environmental passport, for which the company pays fines to the budget.

A known method for assessing the state of the atmosphere by calculating the total index q _Σ [see, for example, “Methodology for calculating the concentrations of harmful substances in atmospheric emissions in enterprises”, All-Union Normative Document, OND-86, USSR, Hydrometeoizdat, Len, 1987 , p. 4 ... 5 - analogue]:

Usually q _Σ (calculated for five components that determine the main contribution to air pollution, provided that q ₁ > q ₂ > q ₃ > q ₄ > q ₅ [see, for example, the Yearbook of the State of Atmospheric Pollution in Cities in Russia, under Edited by E.Yu. Bezuglova, State Technical University named after A.I. Voeykov, St. Petersburg, 1994 ... 1996 - analogue]:

where m _i [mg / m ³ ] - the average annual concentration of the i-th substance in the atmosphere, determined by local measurements on the ground;

CH _i - the maximum permissible sanitary norm for the concentration of the i-th substance in atmospheric air according to GOST;

j is an indicator of the degree of isoefficiency of the harmful substance equal to 0.85; one; 1.3; 1.5 for substances of IV, III, II and I hazard classes.

The disadvantages of the known analogues are:

- statistical instability of the method of single local measurements on the ground at control points as such;

- the uncertainty of the choice of the sampling control points themselves and the significant influence on the result of measurements of the wind rose and random atmospheric turbulence at the sampling points;

- the inoperability of analytical chemistry methods by sampling, ranging from 6 to 12 hours per sample, the high cost of chemical analysis of the sample;

- the inability to simultaneously control the entire set of ingredients (up to 600) due to inoperability and high cost;

- the impossibility of documentary verification of subjectively compiled protocols based on the results of local measurements.

The closest analogue to the claimed technical solution is the "Method for assessing atmospheric pollution" Patent RU No. 2117286, 1998

The closest analogue method involves remotely obtaining a spectrozonal image of a region, including control industrial sites, in the form of digital values of spectral brightness I (x, y) of visible G, R images, dimension | m × n | each element, with a linear resolution of each element larger than the fractal portion of the underlying surface, elementwise logical sorting of pixels in both matrices in accordance with the algorithm, if R≥G, then R, if R <G, then R = R _max - | k | · G , where k is the correlation coefficient of the chromatic coefficients r, g, the resulting matrix of the same size is obtained, the numerical characteristics of the resulting matrix are calculated - mathematical expectation, variance, envelope of the spatial spectrum, autocorrelation function, the histogram of the races is calculated definiteness pixel brightness, binds the obtained relative to the law of distribution of absolute values of the state of the atmosphere in the region of its index values and area control sites.

The disadvantages of the closest analogue include:

- the inadequacy of the algorithm of logical sorting of pixels of the resulting matrix to the measured physical process, and as a result, the discrepancy between the obtained histograms and the physical process;

- the non-obviousness of the calibration of the relative law of the distribution of pixels in brightness by measurements of control sites.

The problem solved by the invention is to create a document from a spectrozonal image of an industrial region from the location of sources of local emissions into the atmosphere with an estimate of their quantitative contribution to the total pollution.

The technical result is achieved in that a method for determining sources of atmospheric emissions from images of megacities, including remote sensing of a high-resolution spectrozonal image of a region (≤2 m / pixel) containing industrial control sites, in the form of digital values of the spectral brightness of the image I (x, y) in the red band of the visible range (570 ... 670 nm), the allocation by the methods of spatial differentiation of the contours in the image with established gradations of brightness, linear analysis of gradient along inside the selected contours, obtaining graphical images of the sources of emissions in the form of oriented lineaments, identifying the point of intersection of the lineaments with the coordinates of the sources of emission, normalizing the brightness of the image pixels relative to the maximum brightness, calibrating the pixel brightness in the MPC values for their reference areas, recalculating the image brightness pixels in the contour of the MPC values by the area of megacities.

The invention is illustrated by drawings, where:

figure 1 is a fragment of the original spectrozonal image of the metropolis;

figure 2 - the interaction of the quanta of visible light with gas molecules during fluorescence;

figure 3 - contours of the established gradations of brightness on the processed image;

figure 4 - pattern of the lineament field inside the selected path;

5 is a histogram of the distribution of image pixels by brightness: a) absolute values; b) normalized values;

6 is a contour of the values of the MPC of the contours, calibrated by the values of the MPC of control sites;

7 is a functional diagram of a device that implements the method.

The technical essence of the invention is as follows.

The surface concentration of harmful substances in the atmosphere depends on many factors: the number of sources, emissions, wind speed, stratification of the temperature gradient, time of year, day, landscape and infrastructure of the region. Therefore, the result of any local measurement on the ground is a random variable. In this regard, the assessment of the state of the air by spectrozonal images, with the number of pixels in the measurement matrix, for example, | 512 × 512 | digital readings provides statistical stability of the result.

Atmospheric pollution in megacities directly above industrial sites reaches units and even tens of MPC [see - analogue]. Physically, this is manifested in an increase in the number of particles of anthropogenic origin in the air, an increase in the turbidity of the atmosphere, and the formation of smogs. When shooting from space, the turbidity of the atmosphere determines the interval of variation of the coefficient of spectral brightness (CSN) of the "atmosphere-underlying surface" system. The interaction of solar radiation with anthropogenic particles occurs at the molecular level. When photons of the light flux collide with gas molecules, energy quanta (hυ ^* ) are transferred to the molecules, which transform into an excited state. For all types of possible interaction of the light flux with smog molecules over megacities, such as absorption, scattering, and fluorescence re-emission, the integral effect is to shift the spectrum of the visible range to its long-wavelength part (red region), [see, for example, R. Mežeris, Laser remote sensing, translation from English, Moscow: Mir, 1987, p. 124, Tables 3, 4. Wave numbers of the Raman shift at a wavelength of 337.1 nm]. Below are some extracts from this Table for some greenhouse smog molecules.

Type of substance molecule NO ₂ SO ₂ CO ₂ NH ₃ C ₂ H ₂ H ₂ s CO NO H ₂ O The wavelength of scattered radiation, nm 345.7 350.8 352.5 378.8 380.3 369 363.9 365.9 384.4 The absolute value of the displacement Δλ, nm 8.6 13.7 15.4 42.7 43,2 32 16.8 18.8 47.3

As a result of the Raman scattering of sunlight, energy is redistributed between the spectral components of the visible range, and the recorded spectral image of anthropogenically contaminated sites acquires a predominantly reddish or dark cherry hue. In addition, in cases of powerful anthropogenic haze of organic and photochemical origin, there is an additional intense absorption (red and near infrared ranges), accompanied by “blurring” of the pixels of the image of the underlying surface [see, for example, “Spectral brightness coefficients of airborne haze”, L. I. Chapursky “Reflective properties of natural objects in the range 400 ... 2500 nm”, part 1, Min. Defense of the USSR, 1986, pp. 103-107, 134-137, tab. P. 10].

Visual selectable signs of contaminated atmospheric portions during spectrozonal imaging from space are the yellow-orange shift of the spectrum of scattered re-radiation and the dark cherry hue of the areas of intense absorption. Obviously, the greater the shift in the spectrum and the absorption of energy from the incident light flux, the greater the pollution of the atmosphere. In the digitized spectrozonal image, in the standard quantization scale of 0 ... 255 levels, the difference in pixel brightness in the red band of the visible spectrum reaches I _min ≈92, I _max = 248. (see figure 1).

The irregularity of the spectrozonal image contains hidden information about the location of local sources of emissions.

The ultimate goal of the proposed method is to create a document with the coordinates of the sources of emissions in the analyzed image and the MPC values of the adjacent territory. The latter is achieved by isolating the contours (contour drawing) in the images, carried out by methods of spatial differentiation (see, for example, Duda PO, Hart P.E. “pattern recognition and scene analysis”, translated from English, ed. Mir, M, 1976 d. "Spatial differentiation", pp. 287-288). There are several well-known standard operators (Roberts, Laplace, Sobel) that allow you to calculate the contour in two-dimensional images. In particular, for example, the Roberts cross operator is calculated in a window (mask) of 2 × 2 elements, for each discrete image point I (x, y) from the relation:

R (i, j) = | I (i, j) -I (i + 1, j + 1) | - | I (i + 1, j) -I (i, j + 1) |

The calculated value of the operator is compared with the set threshold R (i, j)> П, points for which R (i, j) exceed the threshold are displayed. The selection of the contours in the image using masks of various operators seems to be a standard mathematical operation included in the specialized software package (see, for example, MATH SAD. 6.0. PLVS, second edition, stereotyped, information and publishing house Filin, 1997, p. 50-68). The highlighted contours of the established gradations of brightness in the processed image of FIG. 1 are illustrated in FIG.

The next task of the analysis is to extract hidden information from the image. Note that the amplitudes of the image pixels are inversely proportional to the degree of atmospheric pollution of the respective sections. Therefore, the gradient of the brightness of the pixels contains information about the location of the emission sources. By definition, the vector field of the gradients of the scalar image function I (x, y) for any point is calculated as (see, for example, “Derivative in direction” in the book of N. S. Piskunov “Differential and Integral Calculus”, textbook for technical schools, 5 -th ed., vol. 1, Moscow: Nauka, 1964, pp. 264-266)

Near each pixel in the image, there are 8 adjacent pixels. Therefore, in principle, the derivatives of each image point can be calculated in at least 8 directions. The calculation of the directional derivatives is carried out in the “scanning point” mode. The length of highlighted lineaments is set in the input parameters of the processing program, which depend on a number of conditions: the image scale, the resolution of one pixel, the scale of the quantization levels of the signal in amplitude. With the known coordinates of the upper and lower ends of the lineament (x ₁ , y ₁ ) and (x ₂ , y ₂ ), the length of the lineament is determined by the formula:

and azimuth:

при x_i<x_j; y_i<y_j;

for x _i <x _j ; y _i <y _j ;

, при x_i<x_j; y_i>y_j;

, for x _i <x _j ; y _i > y _j ;

A rose chart is defined as a function:

where l _i (A) is the length of the i-th lineament oriented in azimuth A; k is the number of lineaments oriented in azimuth A. Generalization of the distribution of lineaments in azimuthal directions is carried out by the method of “scanning window”. The total length or number of lineaments in the scanning window at this step is calculated. The result is entered into the cell of the output array, the index of which strictly corresponds to the coordinates of the “scanning window”. After the scan is completed, the formed array is processed and the lineament density isolines are built. The program for generalization and construction of the density of lineament isolines is presented in the implementation example. The result of the software calculation is illustrated in figure 4

As a result of the generalization of the azimuths of the lineament rose-diagrams, a vector field is obtained of the preferred directions of the image brightness changes inside the contours. The intersection point of lineaments within the contours is identified with the location of the source of emissions. The shape of the resulting drawings (patterns) gives an idea of the "image" of emission sources.

To quantify the air pollution of a megalopolis, the image pixels are sorted by brightness and their distribution histogram is built. Since the absolute value of the brightness of pixels depends on the shooting conditions (the height of the Sun, viewing angle, time of day), they are brought to a single scale by normalizing with respect to the maximum brightness. This operation allows you to calibrate the relative brightness values in the MPC values (maximum permissible concentrations) against the relative brightness values of the control sites. Histograms of the distribution of pixels of the processed fragment of the original image (figure 1) are illustrated in figure 5. The result is a document (Fig.6) in the form of contours of the MPC values by the area of megacities and the location of emission sources for environmental decisions.

An example implementation of the method.

The inventive method can be implemented on the basis of the device according to the scheme of Fig.7. The functional diagram of the device of Fig. 7 contains an orbital observation complex of the Resource type 1 spacecraft (SC) with a high-resolution digital camera 2 (type DCS760) installed on its board for receiving a light flux signal reflected from the underlying surface 3 through a light filter 4 in red band of visible range. Operational shooting of the planned objects and turning on the camera 2 is carried out by the onboard control complex (BCC) 5 according to the commands transmitted from the flight control center (MCC) 6 via the command control radio 7. The captured frames are recorded in the buffer memory 8 and in the visibility areas of the spacecraft from ground points through an autonomous high-speed data transmission channel 9, they are dumped to information reception points (PPI) 10, where they are recorded on the means 11. Information from the PPI is distilled via terrestrial communication channels 12 to the environmentally friendly Room 13, where the selection is carried out of captured frames on official grounds. Arranged arrays of images of megacities are placed in the database 14 of the Internet 15. (Public access site for users "Resource").

Software image processing for the operations of the method is carried out on a PC 16 in a standard set of elements: a processor 17, a hard drive 18, a random access memory 19, a display 20, a printer 21, a keyboard 22. The contouring and linear analysis programs are recorded on the hard drive 18. As an example A spectrozonal image of Mytishchi, Moscow Region was processed. The picture was ordered at the State Center "Nature", Order No. 11 / 93-42. The amplitude quantization level of the signal corresponded to a scale (0 ... 255), the spatial resolution per pixel was 1.56 m. A fragment of the processed image (Fig. 1) corresponded to the | m × n | matrix of elements [512 × 512]. The maximum brightness I _max = 248, the minimum brightness I _min = 92, the average brightness value I _cf = 190. When programmatically selecting contours, the gradation of brightness was ΔI ~ 20. The highlighted contours in the image are illustrated in FIG. For one of the contours, a gradient analysis of the directions of the lineament field was carried out. The operation of obtaining a gradient field of lineaments is implemented by the following program:

The atmosphere is most polluted in the image above the territory of the Mytishchi Machine-Building Plant (MMZ), chosen as a reference industrial site. The average MPC level in the territory of the MMZ as of the date of receipt of the signal was 4.7 MPC. A histogram of the distribution of pixels by brightness of the analyzed image is illustrated in FIG. 5, where a) a histogram of absolute brightness values; b) a histogram of values normalized with respect to Imax. The minimum brightness in the image corresponds to the maximum MAC values (maximum signal attenuation). For a reference industrial site (MPC = 4.7), the relative value of the pixel brightness corresponds to: I _{industrial area.} / I _max ≈k = 0.37. Whence the calibration coefficient inversely proportional for any point of the relative brightness of the histogram corresponds to:

Figure 6 presents the working paper in the form of MPC isolines and locations of emission sources in the image for environmental decision-making.

The effectiveness of the method is characterized by statistical stability, efficiency, reliability, low cost and documentary results of the assessment of the ecological state of the atmosphere of megacities in comparison with well-known analogues.

Claims (1)

A method for determining sources of atmospheric emissions from images of megacities, including remote sensing of a high-resolution spectrozonal image of a region (≤2 m / pixel) containing industrial control sites in the form of digital values of the spectral brightness of the image I (x, y) in the red band of the visible range ( 570 ... 670 nm), the allocation by methods of spatial differentiation of the contours in the image with established gradations of brightness, lineament analysis of gradient fields inside the selected contours, obtaining g afichesky images of sources of emissions in the form of oriented lineaments, identification of the point of intersection of lineaments with the coordinates of the sources of emissions, normalization of the brightness of the image pixels relative to the maximum brightness, calibration of the brightness of pixels in the MPC values for their reference areas, recalculation of the image brightness pixels in the contour of the MPC values area of megacities.

Images