Ministry of Natural Resources, Ecology and Environmental Protection of the Republic of Mari El
Monitoring of environmental impact during mining operations and development of common minerals in the territory of the Republic of Mari El. Detection of illegal subsoil use accompanied by environmental pollution, namely, detection and updating of information on illegal quarries (outside the boundaries of licensed areas) and associated solid waste dumps.
In the framework of the customer’s technical assignment, cataloging and preliminary photogrammetric processing of archival images, automatic and manual interpretation, data analysis and GIS update, created in 2013-2016, field work were carried out.
An archival survey was ordered for the areas of work – satellite images from the Kanopus-V and Resource-P satellites for 2017–2018. Then, based on the metadata, the experts of Innoter LLC cataloged the available remote sensing data, which made it possible to efficiently select suitable to decrypt images and weed out data with a high percentage of clouds, the presence of snow cover and other defects.
The next stage of work was orthorectification (by RPC-polygons and SRTM-1) of the selected images, their geometric and radiometric correction, including for accurate binding to the images used earlier in the project for 2013-2016.
After that, it was necessary to assess the spectral separability of classes in order to identify objects of subsurface use. The main purpose of the assessment of spectral separability is a visual or automatic comparison of the spectral brightness curves that determine the reflectivity of various ground cover objects. Were selected test plots in the most characteristic areas for this landscape, as well as subsoil use objects (quarries, tree stands, man-made objects, lands not covered with vegetation, water bodies).
Test sites for assessing the spectral characteristics of landscape objects. (a - sand, b - tree-shrub vegetation, c – asphalt)
Based on this assessment, the specialists of Innoter LLC were able to determine which objects would be well classified in the photographs, having different spectral characteristics (such as sand and forest), and which objects are similar in parameters and can be misleading as ("sand" and "soil") and, therefore, it will be necessary to more carefully select the reference areas and carry out expert control of automatic interpretation. But even objects that have good standards can often be classified falsely in automatic mode, so visual interpretation of specialists is necessary.
Update and supplement the data of the previously created GIS
Through the NextGIS QGIS software product we updated the data using new information obtained from 2017-2018 snapshots, as well as from the register of quarries in the Republic of Mari El, (this was provided by the Ministry of Natural Resources, Ecology and Environment of the Republic of Mari El).
The resulting map in the GIS format contained geo-information layers that reflected preliminary information about all the objects of the “subsoil use objects” class, open-cast areas, information about mineral resources, landfills, etc.
An example of one of the map layers
As already mentioned above, at first the selection of remote sensing data was carried out. Of the 65 archived images included in the first delivery, only 40 satisfied the time range. Further selection took place according to the criteria of reliability as a source of information regarding the interpretation of the signs. Were excluded images with snow and a continuous cloud cover. Snow greatly complicates the detection of quarries, for example. So do the clouds. Thus, priority was given to images that were taken in the summer-autumn period.
The discovered quarry (polygon of red color) is reliably interpreted by the summer (left) image, but not readable in the winter (right) image
In preparation for the main stage of work, a database cluster was launched – a software and hardware complex that allows employees of Innoter LLC to work simultaneously on a single project using the SQL query language. The advantage of using such a cluster over working in a conventional desktop GIS is that a lot of decoder professionals work simultaneously with the same data: you don’t need to duplicate data, work on a summary of the decryption results from several performers, etc.
At the first stage, the interpretation was carried out visually by specialists. The following interpretive features were used:
- spectral brightness (photo tone);
- the size;
- the form;
- context (proximity of roads, settlements);
As a result, preliminary data were obtained.
Potential objects of mining, identified by the results of the interpretation of the satellite imagery 2017-2018.
Next, the objects of interest were digitized, and then it was decided to perform an automatic uncontrolled classification of images (clustering) within the designated quarries. For example, in this way it became possible to isolate the developed part of the quarries. For this, several models were created that processed data in a special way (based on the ISODATA method). As a result, several classes were identified that were automatically decrypted by the program.
A fragment of the image. Classification by 6 grades.
At the next stage, the data also in the automatic mode using a specially created model were vectorized and merged into one layer.
The next significant stage of work was the analysis of the objects obtained in conjunction with the existing data of the subsoil fund and preparation for field research. In addition to data integration, a special “confidence” scale was developed:
- 25% – the presence of a career is unlikely;
- 50% – the presence of a pit is determined with an average degree of probability;
- 75% – the presence of a career is determined with a high degree of probability;
- 100% – the object is a career.
The data were evaluated in accordance with it and all other relevant attribute data were entered (the state of the quarry, the presence of the landfill, the area of production, the type of mineral used for interpretation of the satellite image, the year of the survey).
In preparation for the field work, several road graphs were created and detour routes for the objects of interest were planned.
Field route №1
The final stage of the project was in the field survey of objects of interest. As a result of field work, 43 potential subsoil use objects were surveyed; 21 of the 43 visited sites turned out to be a quarry.
In addition, the experience of the field survey allowed us to specify the templates and standards for office interpretation, which will help the staff of Innoter LLC in similar projects in the future. Now Innoter LLC has its own bank of interpretation standards and the accuracy of the identification of quarries reaches 70% or more.
Monitoring of the environmental impact during mining operations on the territory of the Republic of Mari El was carried out using Resurs-P and Kanopus-V satellite imagery for 2017–2018. Considering the available coverage of the territory of the Republic with satellite images suitable for decoding, according to the results of the work carried out, more than 100 objects of subsoil use were identified.
The results of the work showed that remote sensing data can, and should, be used to monitor the environment, for example, in this case to detect illegal subsoil use and mining, as well as to detect unauthorized landfills.
In addition to the identified subsoil use objects, additional interpretive signs of quarries were formed, as well as other objects that can be confused with quarries with automated and manual interpretation.
A GIS project for the maintenance and monitoring of subsoil use objects in the NextGIS QGIS format, containing information on the contours of legal and illegal workings ever discovered in the territory of the Republic, has been compiled; on field survey data and unconfirmed cases of illegal activities (including photo centroids); about the pictures used in the work. In addition, GIS contains topographic and administrative information about the territory.
The final look of the project