Method for determining sources of pollution

Abstract

Disclosed is a method for determining pollutant-emitting sources comprising, in an initial phase, collecting information on the study area and determining locations suitable for sample collectors using chemical and meteorological models such as CHIMERE and WRF; in an intermediate phase, analysing the samples using techniques adapted to the pollutant, focusing on particle concentration; and, in a final phase, analysing the data obtained and the information collected in the preceding phases in order to identify and locate the polluting sources. This analysis includes a statistical study of percentiles and meteorological conditions in order to detect significant exceedances in particle concentration, and an analysis using a positive matrix factorization model, also referred to as a PMF model, to break down the concentrations according to the pollution factors. This comprehensive approach makes it possible to accurately determine the sources of pollution and their specific contribution in the study area.

Description

[0001] METHOD FOR DETERMINING SOURCES OF POLLUTION

[0002] DESCRIPTION

[0003] OBJECT OF THE INVENTION

[0004] The technical field of the present invention is situated at the intersection of environmental engineering with a particular focus on the assessment and management of air quality in areas contaminated with suspended particulate matter of anthropogenic or natural origin.

[0005] Specifically, the present invention relates to an automated method for determining point or dispersed sources of pollutants, which allows for the identification, characterization, localization, and estimation of the contribution of such pollutant sources, with the objective of evaluating the impact of particulate or gaseous emission sources on nearby urban centers. The method is especially focused on the study of dispersed and heterogeneous sources, such as open mining, landfills, agricultural operations, etc., located in an area of ​​varying size on the outskirts of the study site.

[0006] BACKGROUND OF THE INVENTION

[0007] Currently, there are several studies related to the impact of particle-emitting activity areas. These activities can be diverse: they include industrial, agricultural, road traffic, natural sources, or activities that contaminate the soil. Various methodologies have been used to assess the extent and magnitude of the pollution. These include techniques to determine the best location for air, water, and soil sampling and analysis devices to identify and quantify pollutants. These techniques actually have two subphases: in the first, the locations with the greatest potential or gradient of "positive" or significant data are studied. In the second, collectors or analyzers are placed in these areas, sampling is performed, and non-zero data are obtained, allowing pollutant modeling and impact estimation.Epidemiological studies have also been used to assess the health effects of the exposed population, as well as computational modeling to predict the dispersion of contaminants and their reach in surrounding urban areas.

[0008] On the one hand, the CHIMERE chemical transport model is well-known and is accepted in the air pollutant modeling sector. CHIMERE is an open-source physical / chemical model designed to perform accurate analyses of atmospheric pollutant propagation, pollution episodes, daily forecasts of ozone, aerosols, and other pollutants, as well as long-term simulations (entire seasons or years) under emission control scenarios. In technical terms, CHIMERE is a multi-scale chemical transport model, from urban to continental, that allows the quantification of the evolution of air masses and pollution plumes (dispersions of pollutants in the air originating from a specific source) over time.It is known that CHIMERE can be used as an air quality forecasting model, and can also be integrated with meteorological models, which increases the accuracy of the results.

[0009] On the other hand, despite the progress, there are challenges and drawbacks to this type of research. One of the main limitations is the difficulty in establishing a clear line of causality between the emitting activities and pollution at the study site, given the presence of other polluting factors and the variability in environmental and exposure conditions. Likewise, the lack of adequate historical data and the complexity of quantifying long-term effects on human health and the environment represent significant obstacles to fully understanding the impacts. Furthermore, active participation and collaboration between public entities, industries, local communities, and scientists are often challenging, which can limit data availability and the implementation of effective mitigation measures.In this sense, transparency and collaboration are essential to comprehensively address the impacts of pollution from various activities in urban centers or sensitive areas and work toward more effective and sustainable solutions. State-of-the-art solutions are generally limited to retrospective analysis of historical data, without offering a detailed methodology for determining optimal sampling locations or accurately identifying and characterizing air pollutant emission sources.

[0010] Therefore, the objective technical problem that arises is the lack of precision and specificity in the study of the implementation of sampling points and the identification of sources of contaminants in the target areas, such as in the case of an applied use related to mining activities.

[0011] DESCRIPTION OF THE INVENTION

[0012] The present invention addresses the aforementioned problem by means of a computer-implemented methodology that allows for a more accurate and adaptive determination of critical areas for the placement of air quality monitoring devices and a more precise characterization of pollution sources, including their type and contribution to the overall pollutant profile in the atmosphere. This method overcomes the challenges of conventional approaches by integrating advanced numerical simulation models and adapting existing sampling criteria, thereby improving the effectiveness of mitigating the environmental impact of mining activities and protecting public health.

[0013] An aspect of the invention relates to a method implemented to determine pollutant emitting sources that is defined according to claim 1. The dependent claims define advantageous embodiments.

[0014] Another aspect of the invention relates to a computer program product comprising instructions that, when the program is executed by a computer, cause it to carry out the method defined above.

[0015] Another aspect of the invention relates to a computer-readable medium comprising instructions that, when executed by the computer, cause it to execute the method defined above.

[0016] The advantages of the present invention compared to the prior art are fundamentally: Through a preliminary study of macro- and micro-implementation of the proposed methodology, the shortcomings present in other existing studies are covered, where, although a background study is carried out to establish the arrangement of the sampling points, all are limited to an analysis of historical data or databases of the locations where there is a probability of being a potential source of emission.

[0017] The present invention's use of numerical simulation technologies and a physical / chemical model of atmospheric pollutant propagation (e.g., the CHIMERE model), with specific model configuration, allows for the optimal area for the placement of air quality devices to be defined and determined, assessing the areas most affected by the sources considered. In terms of micro-implementation, criteria applied in sampling campaigns for other pollutants are adapted to adaptively combine them with the target pollutant of the study.

[0018] The proposed methodology allows for the identification of sources and the determination of not only the sources of emissions but also the characterization of the sector from which they originate. In contrast, other source identification methods assess pollutant contributions from emitters only approximately, identifying the most likely sources based on the nature of the pollutant and its concentration, or by breaking down the emitting sources from the point of emission, that is, determining the dispersion of the pollutant, but with a prior knowledge of the source of emission. Existing identification methods use different approaches: (i) either the quantities and nature of the pollutants can be determined, but without specifically knowing the source of emission; (ii) or the nature of the pollutant and therefore its dispersion can be determined, but assuming the point of emission is known.

[0019] These and other advantages can be derived in light of the description of the invention presented in detail below.

[0020] DESCRIPTION OF THE DRAWINGS

[0021] To complement the description being made and in order to help better understand the characteristics of the invention, in accordance with a preferred example of practical implementation thereof, a set of drawings is attached as an integral part of said description, in which, for illustrative and non-limiting purposes, the following has been represented:

[0022] Figure 1.- Shows an example of a diagram for the sectorization by quadrants used in the identification of the origin of the pollutant emitting source.

[0023] Figure 2.- Shows a histogram of wind directions for the analysis of environmental data performed using positive matrix factorization.

[0024] Figure 3.- Shows trends in frequency and intensity of wind directions in the quadrant sectorization diagram.

[0025] Figure 4.- Shows an image of a study area with the overprint of the quadrant diagram and wind directions and the graphic marking of various metal species zones for different wind speeds.

[0026] Figure 5.- Shows a histogram of the concentration percentages of each metal species obtained using a PMF model for a prevailing wind direction.

[0027] Figure 6.- Shows a heat map of different concentration levels over a study domain resulting from a simulation of the CHIMERE model.

[0028] PREFERRED EMBODIMENT OF THE INVENTION

[0029] A detailed explanation of a preferred embodiment of the object of the present invention is provided below, with the aid of the aforementioned figures, referring to a methodology that uses different processes by which the emission sources of the target pollutant are determined. These processes are divided into different stages that range from the background study to the selection of sampling points up to the final phase where the emission sources are determined. These main stages are: • Selection of sampling points

[0030] • Measurement and data extraction campaigns

[0031] • Determination and evaluation of sources

[0032] Below, the aforementioned stages of the method are detailed according to a possible implementation, which are differentiated into three main phases:

[0033] - A first phase of preliminary study and macro-implementation of sample collectors (or analyzers), for example, airborne metal collectors.

[0034] - A second phase involves designing campaigns capable of covering a study area to collect as much data as possible on the collected samples.

[0035] - A final phase corresponding to a post-processing of these collected data and a study of contribution and determination of sources from the collected samples.

[0036] I. First phase: Selection of sampling points a) Previous studies

[0037] Background information is collected from the entire sampled area, as well as information on potential types of pollution, industry or other sources of emissions, geographic location, meteorological conditions, etc., in that area. This first step is important because, depending on the potential sources of pollution found, they will be introduced into the CHIMERE pollutant propagation model described below. b) Macro-implementation

[0038] Macro-deployment refers to the study of the best locations for large-scale device deployment, that is, determining the location of sample collectors (usually within a few kilometers; this could be a town or similar area). To do this, it is first necessary to determine which areas are emitting metals into the atmosphere. A physical / chemical pollutant propagation model (in the example described here, the CHIMERE model) coupled with a meteorological model is used to simulate the dispersion of airborne particles, thus determining the most affected areas.In other words, "virtual" sources of contamination are created, previously detected in section a) and used to study the behavior and propagation of metals in the air. The areas most prone to contamination are determined, and therefore defined as the study area. This macro-implementation study ensures that the samples subsequently collected by the sampling equipment are not biased by any contaminating activity or situation that is not significant.

[0039] CHIMERE offers numerical simulation and prediction of the dispersion of particles and other pollutants using initial emissions input data and subsequently performs zoning to detect regions ("clusters") with similar particle concentration values. This allows determining the best location for sample collectors using principal component analysis (PCA) correlations.

[0040] CHIMERE works in conjunction with various meteorological simulation models. In this case, the WRF (Weather Research and Forecasting) model was used, which represents the state-of-the-art in numerical weather prediction (NWP). The WRF model is capable of forecasting meteorological variables such as wind, temperature, and precipitation based on global meteorological files from NOAA (National Oceanic and Atmospheric Administration). In this phase, the emission from a potential mining source can be simulated and its propagation observed in space, thus defining the areas receiving the most particulate matter. Figure 6, described below, shows an example of how the defined zones are represented.

[0041] In one possible implementation, a simulation is run for the hours specified in a file ("script" in "bash") containing a series of commands and control structures executed by a Unix-based operating system using a command interpreter that provides an interface between the user and the operating system. Depending on the study to be conducted, a domain is defined, which is the geographic area of ​​study to be covered along with the emissions scenario in that geographic area. The study domain, and the domain within which CHIMERE works, is defined in different cells, units of minimum space over which the results are obtained. This determines the model's resolution; in this case, for example, 1x1 km cells were selected.

[0042] In order to simulate particle emissions from sources or focal points, "artificial" emissions are generated. To do this, all gas-related emissions are first eliminated to prevent the model from considering unrelated chemical reactions in the atmosphere. In this way, only particle emissions that do not react chemically in the atmosphere are generated; that is, they are direct emissions, not immissions (generated by gases such as volatile organic compounds or VOCs). The most important direct emissions are black carbon (BC) and organic carbon (OC). Therefore, these particles behave similarly to metallic particles since they are only affected by diffusion and transport processes.If potential emission points from BC and OC mining activities are generated in the studied locations, these emissions move like metal particles, so the transport of particulate matter is only correlated with the distribution of metal particles. Artificial emissions are generated based on the position, size, activity, metals present in the studied areas, as well as any other relevant prior information. Artificial emissions are generated based on a multiplier of actual emissions, extracted from available inventoried or satellite databases, in this case from the CAMS (Copernicus Atmosphere Monitoring Service) inventory, and increased in increments in those cells or positions that coincide with the emissions, improving the overall gradients for better identification. The area of ​​the mining operation is proportional to the increased emissions.This involves creating a specific and extraordinary configuration of the model that allows you to get the most out of it.

[0043] In one possible simulation example, the emission type is PM10 particles (with an aerodynamic diameter less than 10 pm - micrometers), using BC particulate pollutants as a basis. These particles are defined as ultrafine particles, smaller than PM1 (polluting particles smaller than 1 micron), but the simulation is forced to generate particles between 0 and 10 pm, based on the composition and stability of BC particles, in order to model transport.

[0044] The next step is to simulate particle emissions from the study areas. These emissions are assigned to each of the cells (e.g., 10x10 km) taking into account the study area within the cell so that the emissions are proportional to the size of the study area. To do this, the locations of the shape files for the study areas defined in the previous section are taken into account to assign them to a cell and observe the temporal evolution of the concentration on the grid. c) Micro-implementation

[0045] The micro-implantation study aims to pinpoint the exact location from which the contaminant will be sampled, that is, to determine the suitability of the location for the sampling device. In the case of particles, this refers to sample collection, and collectors can be divided into High Volume (HV), Low Volume (LV), and settleable collectors. In the case of gases, sampling devices can include gas analyzers, portable cabinets, solid or gel media collection, or Tediar bags for subsequent laboratory analysis.

[0046] For example, the following criteria are used when selecting the locations of the collection devices:

[0047] • Flow restriction: The accumulation of contaminants due to environmental characteristics (buildings, topography, etc.) that hinder air circulation and thus prevent the measurement of low-dispersion microenvironments must be avoided.

[0048] • Height: The building has a suitable height to install the device, between 1.5 and 4 meters to improve air quality.

[0049] • Free arc 270° / 180° (<3m): At the installation site, the device has a clear angle where it can collect data in the most efficient way possible.

[0050] • Obstacles (<6m): There are no obstacles in the immediate vicinity of 6 meters, thus avoiding obstruction by trees, public lighting, structures, etc., to the extent possible.

[0051] • Proximity to highways and heavy traffic: Particle emissions from road traffic can affect the capture of target particles, so locations away from roads with heavy traffic are chosen.

[0052] • Contaminant trajectories: These simulations performed with CHIMERE are used as a determining factor when rejecting or prioritizing a study area.

[0053] • Equipment security conditions: Locations are sought where the equipment has a secure environment that guarantees the integrity of the devices against vandalism, taking into account the availability of fencing, security cameras, etc.

[0054] • Access for inspections and sampling: It is highly valued that the environment where the device is located is accessible to the technician and facilitates sampling as much as possible.

[0055] Based on the macro- and micro-implantation criteria, a numerical system is established to evaluate the locations of collectors for particle measurement, with the aim of selecting the best location for the installation of high and low volume collectors (AV, BV) and sedimentation tanks.

[0056] II. Second phase: Measurement campaigns according to reference technique

[0057] Data collection, an intermediate phase of the methodology, based on measurements made by collectors (e.g., settleable particle collectors), can be implemented using different techniques depending on the nature of the air pollutant whose source needs to be determined. For example, for a project to determine particle emission sources, measurement campaigns can be divided into main groups, one for each type of collector: low-volume (LV) campaigns, high-volume (AV) campaigns, and settleable particle campaigns. Due to the nature of the samples to be identified, for example, metals, sample analyses are performed with a pretreatment of the collector filters, consisting of washing with distilled water, stabilization, weighing, and identification, which ensures sample traceability.For example, for the determination of soluble metals, samples must be acidified upon arrival at the analytical laboratory, while for the determination of insoluble metals, the entire sample volume is filtered through a cellulose acetate filter. The analytical results are the input to the final phase of the method, which determines the sources of the target pollutant.

[0058] III. Final phase: Determining sources

[0059] The identification and contribution of sources is the ultimate goal of processing laboratory results, based on the data obtained from sampling using the different techniques in the previous phase (analytical results). Two methodologies are applied to analyze the analytical results: the first is the statistical study of percentiles and baseline meteorological conditions; the second consists of simulations using the statistical software Positive Matrix Factorization (PMF), a tool designed to analyze and decompose environmental data, especially in the context of air quality. Finally, both data are combined to discern between sources contributing to the samples. i) Statistical analysis of percentiles and baseline meteorological conditions

[0060] For the first method of analyzing analytical results, exceedances of the 90th percentile for each metal species are identified and grouped based on established criteria in the quadrant division by wind direction. Using this, along with wind speed, backtracking can be traced to potential metal emitting sources.

[0061] To identify by quadrant the exceedances of the 90th percentile (anomalous emissions) for each metal species for each location, exclusively data collected in high-volume measurement campaigns are used, at the locations where they have been carried out. In locations where such campaigns do not exist, low-volume campaigns with a more limited spectrum are used. Once these percentiles have been identified, they are broken down using the following criterion: they are divided into four quadrants (four for each location), each quadrant corresponding to a range of degrees referring to the directional angles of the wind downwind of the collector. These quadrants, illustrated in Figure 1, are named as follows:

[0062] Q1 : >0° to <90°, corresponds to a northeast NE direction

[0063] Q2: >90° to <180°, corresponds to a southeast SE direction Q3: >180° to <270°, corresponds to a southwest SW direction Q4: >270° to <360° (0o ), corresponds to a northwest NW direction

[0064] The quadrant diagram example shown in Figure 1 is based on a compass rose indicating the cardinal points and the diagram implementable in a graphical interface shows:

[0065] • Degrees: Indications around the entire circumference near the outer edge, indicating the degrees belonging to the intervals into which the quadrants are divided.

[0066] • Quadrants Q1, Q2, Q3, and Q4. In the example, quadrant Q1, downwind of the collector, is the direction in which the pollutants are detected to be dispersed (represented by arrows pointing outwards); while quadrant Q3 indicates the direction in which the emitting sources are located, in this case, upwind of the collector.

[0067] • Arrows: Example of information representation to visually appreciate the direction of particle dispersion according to the prevailing winds.

[0068] In this way, exceedances of the 90th percentile in a specific direction are associated to trace a back trajectory and identify the origin of the emitting source. In addition, a wind speed criterion is added to further disaggregate the samples. This criterion separates the exceedances into two groups within the quadrant: the first group as exceedances that only occur when there is a wind speed greater than or equal to 3 m / s, and the second group, which includes exceedances that occur only where the wind is less than 3 m / s. The resulting difference places the emitting sources at a greater or lesser degree of proximity, with distant sources being those that give rise to percentile exceedances where the wind speed is >3 m / s, and close sources where it is <3 m / s. In some cases, correlation matrices are used to find groups of metals with the same origin.

[0069] i) Positive Matrix Factorization Analysis The simulation part using the PMF model collects the metal concentration signals from each of the collectors and separates them into different factors that contribute to these results through statistical methods and linear regression adjustments. This model is fed, in addition to the pollutant concentrations collected by the collectors, by the uncertainties of the measurements taken.

[0070] Conducting a source contribution analysis for air quality is essential for understanding and addressing environmental pollution problems, especially in regions near industrial and mining activities. In this context, the EPA PMF 5.0 (Environmental Protection Agency - Positive Matrix Factorization) model is specifically used as a tool to estimate the contribution of different sources to the concentration of air pollutants. This model unravels complex mixtures of atmospheric particles and provides quantitative source identification, providing insight into the impacts of mining activities on air quality and, ultimately, on human health and the environment.

[0071] Particulate collectors, in this case high- and low-volume collectors and settling particle filter collectors, provide accurate data on metal concentrations in the air. However, directly interpreting these data can be complicated due to the complex mix of air pollution sources in urban and industrial areas. Therefore, the purpose of using this model is to discern which of the samples collected by the collectors originate from mining operations and which are due to other factors. This includes differentiating between direct emissions from mines, natural sources, or industrial sources. To achieve this, the PMF model is fed with two types of input data: measurements of the sample concentrations (metals) collected by the collectors for each location and the uncertainties associated with each of these measurements.The PMF model assumes that there are p sources, source types, or source regions, called factors, that influence the receptor and that, in addition, the observed concentrations for the different sources have been produced by a linear combination of the impacts of the p factors. The mathematical expression describing the final concentration is as follows: where x¡j is the concentration in the receptor for the j species on day i, g¡k is the contribution of the k factors on day i, f, is the fraction of the k factors that is species j, and finally e is the residual of species j on day i.

[0072] That is, this model assumes that only the concentrations x are known, and the objective is to find the contributions g¡k and the fractions f¡¡. To do this, the PMF model uses the least squares technique to minimize the difference between the concentration data and those obtained through the weights g and f. The objective is to minimize the following equation: where s is the uncertainty of the j species for day i.

[0073] As a final result, the PMF model returns the percentage contribution of each of the factors to the provided samples.

[0074] Before running the model, the model configuration is set to indicate the number of factors into which the concentration of the samples will be broken down, which in a possible case could be broken down into the following factors:

[0075] • Factor A: Industrial

[0076] • Factor B: Combustion gases (road traffic, biomass combustion...)

[0077] • Factor C: Potential sources of mining origin

[0078] • Factor D: Others, such as background (urban, agricultural or whatever is relevant to the area studied)

[0079] • Factor E: Calima or other extraordinary natural sources

[0080] After compiling all the historical data collected by collectors in each campaign, the PMF model is run for each site, thus obtaining the contribution of each factor to each species. Furthermore, the PMF model simulations are distinguished by sampling technique, i.e., high-volume and low-volume. For example:

[0081] - First case: using the entire historical sample collection. The PMF model is run using the entire historical sample collection for each location, separating between high and low volumes. The PMF model returns the percentage contributions of each source to each metal.

[0082] - Second case: data filtered by wind direction. Additionally, and since wind is one of the variables that plays an important role in the transport of pollutants in the atmosphere, the samples are divided according to the prevailing wind direction. At each location, the samples are divided 50%-50%. A meteorological study of the two most predominant wind directions is performed to subsequently divide the data set of species concentrations into two equal subsets, assigning each value to one wind direction or the other depending on the day. This analysis helps determine the type of each of the factors that PMF displays in the simulation. Figure 2 shows, in a histogram, the separation of the wind directions into two based on the prevailing directions for each of the locations, representing the frequency and wind angle for each of the two prevailing wind directions.

[0083] The following Figures 3-5 show results according to an example of the implemented methodology that uses data resulting from the measurement campaigns that are identified in the Figures with the following terminology:

[0084] • AV: Data collected through high-volume campaigns

[0085] • BV: Data collected through low-volume campaigns

[0086] • MIXED: Data collected through low-volume campaigns, which were considered insufficient for PMF executions, so they were completed (same location and not coinciding in time) with data from high-volume campaigns carried out in the same location and date.

[0087] The methodology, according to an example, is based on the following steps to determine the factors:

[0088] 1. The data corresponding to each location are divided: data from the AV, BV, and MIXED collectors, meteorology, including wind speed and direction. This information is then divided into four quadrants (Q1 of 1 o at 90°, Q2 from 91° to 180°, Q3 from 181° to 270° and Q4 from 271° to 360°). The position from which the directionality data were taken with respect to metal capture is always taken into account to determine the quadrant.

[0089] For example, Figure 3 represents the frequency and intensity of the wind directions that occur in the North (N), South (S), East (E) and West (W), or in degrees of North azimuth (0-360°) for a given specific area:

[0090] • 0° = North (North wind)

[0091] • 45° = Northeast (Northeast wind)

[0092] • 90° = East (East wind)

[0093] • 135° = Southeast (Southeast wind)

[0094] • 180° = South (South wind)

[0095] • 225 ° = Southwest (Southwest wind)

[0096] • 270 ° = West (West wind)

[0097] • 315° = Northwest (Northwest wind)

[0098] • 360° = North

[0099] Wind data uses a different angle projection than the algebraic (traditional) one, with North always being the origin of the coordinates, so it is important to take this into account based on wind projections.

[0100] 2. For each quadrant, the metals in the exceedances are grouped (90th percentile) into two large groups: metals transported at wind speeds below 3 m / s and metals transported by wind speeds greater than 3 m / s (This value of >3 or <3 m / s is an example of a value in which a difference has been observed between the metal groupings). This separation determines the approximate proximity of the emitting source, that is, particles transported by winds below 3 m / s are found at a shorter distance from the collection site than particles transported by higher speed winds.

[0101] For example, Figure 4 shows an image of a given study area with the wind rose superimposed on the image representing the following parameters and results:

[0102] Predominant wind direction: SE, quadrant Q2. Possible sources from direction: NW, quadrant Q4.

[0103] Species that only occur with wind speeds greater than 3 m / s, represented in the areas outlined in black: Aluminum, Barium, Phosphorus, Iron, Potassium, Titanium, Zinc and Sodium.

[0104] Species that only occur at wind speeds below 3 m / s, represented in the areas outlined in white: Lead and Sulfur.

[0105] In this example, it can be deduced that, using also the collection of information from the background of the sampled area, since lead and sulfur occur in nearby transport winds, it is possible that the majority of these two metals are due to an existing mine in the area and that the remaining metals are of industrial origin due to the companies located in the area and the agricultural sector.

[0106] 3. Using the results obtained from the PMF model, the percentages of each metal associated with each factor are verified, analyzing the correlation between metals and factors to determine which metals have a common source. These analyzed results have been previously disaggregated into two predominant directions for each location, using 50% of the samples in each. For quadrants where prevailing wind results disaggregated from the PMF are not available, PMF results without disaggregation by prevailing direction can be used (all samples). In addition, raw data on possible wind sources and directions are used.

[0107] 4. To determine the potential sources, the factors are divided into PMF and, if the quadrants do not coincide, the sources present in the area. For this determination, some of the possible factors are the following: a. Composition of potential sources due to mining activity that falls within the possible wind paths (upwind of the collector and downwind of said source). b. Composition of industrial emissions, extracted from the PRTR (Pollutant Release and Transfer Registers). c. Episodes of Saharan dust intrusion or other natural sources. d. Emissions from agricultural and livestock sources. e. Other background emissions from combustion gases (traffic, boilers, etc.)

[0108] For example, Figure 5 represents the results of the PMF model in a histogram, considering one of the two predominant wind directions and the sample data from high volume (HV) collectors, obtaining the concentration percentages of the possible species for four factors: Factor 1 - Agricultural

[0109] Factor 2 - Industry - There is a grouping of metals of industrial origin

[0110] Factor 3 - Potential sources of mining activity

[0111] Factor 4 - Undetermined

[0112] Finally, Figure 6 shows a heat map of the emissions simulated with the CHIMERE model; in this example, referring to lead concentrations. Heat maps can be obtained for each mineral mined, with kilometer resolution, forming the main results of the macro-implantation study. This information is key in determining the "target," reference point, or background contamination of the study area.

[0113] The method described is equally applicable to gaseous emissions, modifying only some data sources and adapting the field sampling methods.

Claims

CLAIMS 1. A computer-implemented method for determining emission sources of a target pollutant, comprising the following steps executed by one or more processors: in a first phase, collecting background information from one or more areas to be sampled within a study area, the background information comprising pollution types, industry, geographical location and given meteorological conditions in each area to be sampled; determining at least one location of one or more sample collectors in each area to be sampled to cover the study area according to macro-implementation criteria, where the at least one location is determined using a combination of chemical and meteorological models configured to predict a dispersion of particles of the target pollutant in the study area;determining, according to micro-implementation criteria, a suitability value for each location determined in the previous step for each sample collector and choosing the location with the highest suitability value; in a second phase, obtaining data from the samples collected at the location chosen by the one or more sample collectors in each area to be sampled, where the data are obtained by means of analytical techniques adapted to the target pollutant applied to the collected samples, the data obtained comprising at least one measurement of the concentration of particles of the target pollutant;and in a third phase, analyzing the data obtained in the previous phase to determine an identification and location of sources emitting the target pollutant using the background information collected in the first phase and estimate a contribution of each source to the measurement of concentration of particles of the target pollutant obtained in the second phase, where the data are analyzed by two procedures: i) a statistical analysis of percentiles and conditions; meteorological to identify exceedances of the 90th percentile in the prediction of particle concentration, and (i) a positive matrix factorization analysis to separate the predictions of particle concentration according to factors contributing to pollution in the study area.

2. The method according to claim 1, wherein the chemical model used to determine the location of the one or more sample collectors is the CHIMERE model configured to identify regions according to particle concentration.

3. The method according to claim 2, wherein the CHIMERE model is coupled with an atmospheric simulation climate model which is the WRF model.

4. The method according to any of the preceding claims, wherein high volume, low volume and settleable sample collectors are used.

5. The method according to any of the preceding claims, wherein the statistical analysis of percentiles and meteorological conditions comprises a quadrant analysis based on wind directionality to obtain particle backtracking that determines the location of emitting sources.

6. The method according to any of the preceding claims, wherein the analysis by positive matrix factorization uses the measurement of particle concentration of the target pollutant and an uncertainty value of the measurement obtained in the second phase.

7. The method according to any of the preceding claims, wherein the positive matrix factorization analysis uses the measurement of particle concentration of the target pollutant filtered by wind direction.

8. A computer program product comprising instructions that, when the program is executed by a computer, cause the computer to perform the method of claims 1-7.

9. A computer-readable medium comprising instructions that, when executed by a computer, cause the computer to carry out the method of claims 1-7.

Images