Detecting Environmental Impacts of Uranium Mining with the use of Remotely Sensed and Geophysical Data
Petr Dobrovolný1), Lubomil Pospíšil2), Otakar Pazdírek2)
1) Department of Geography, Faculty of Science, Masaryk University, Kotlářská 2, 611 37 Brno, Czech Republic
2) GEOFYZIKA a.s., Ječná 29a, P.O. Box 62, 612 46 Brno, Czech Republic
SUMMARY
The presented paper summarizes the first results of the project which tries to find a method for detection of contamination due to uranium mining and connected activities. The study area is situated in the western part of Bohemian Moravian Uplands (village Dolní Rožínka and surroundings, area approx. 80 sq. km). At this time the decline of mining activity is typical for the area. In the project, remotely sensed data (LANDSAT TM and aerial photos) and geophysical data (aerial gamma-ray spectrometry measurements) have been used. It was proved, that higher concentrations of radionuclides 40K, 238U and 232Th are restricted above all only to the mines, mine tailings, and tailing ponds and that the concentrations fall down sharply with the increasing distance. Using spectral enhancements (PCA) or ISODATA clustering it is possible to find a good correlation between geophysical data and LANDSAT TM multispectral imagery.
KEY WORDS
Uranium mining, contamination, Bohemian Moravian Highland, LANDSAT TM, gamma-ray spectrometry, image processing
INTRODUCTION
An enormous development of geoinformatics in the 1990s is connected above all with the GIS and Remote Sensing technologies. These technologies give us a powerful tools for effective analysis, control and decision making of various processes. The mentioned methods of investigation have a great significance in environmental applications - especially in the processes where conflicts between the nature and antropogenic activity should be solved (LILLESAND and KIEFER 1994, LEGG 1995). Uranium mining and related activities can certainly be mentioned among such processes.
The objective of this paper is to show the potential of various kinds of remotely sensed and geophysical data for monitoring of contamination of the environment due to uranium mining. The intent was also to suggest a method how to study the dynamics of contamination processes with the help of image processing of remotely sensed data, geophysical data and GIS methods of analysis.
Comparing to "classical" methods of environmental data collection, remote sensing methods offer several advantages. First, advantages of economic character can be mentioned. Second, through remote sensing synoptic view of various processes can be obtained. Moreover, with the help of remote sensing data collection and interpretation, spatial patterns of various processes can be studied. Many processes of such character like pollution or contamination can be normally investigated through point measurements only. Third, remote sensing data are important from point of view of their good temporal resolution. In case of constant parameters of a measuring system, the variability of measured data gives us the evidence on process dynamics.
STUDY AREA
The area of uranium mining is situated in the western part of Bohemian Moravian Uplands (Fig. 1), area approximately 80 sq. km. Originally this area was mainly used for extensive agricultural activities. Discovery of uranium deposits in the second part of the 1950s had a strong impact both on environment and lives of people in this region. The area is characterized by a great diversity from geological and geomorphological point of view. It comprises from narrow and deep valleys in the south and east part and from plateau in the northwestern part. Steep slopes of the valleys are covered with forest mostly. The plateau and narrow flood plains along the rivers are used for agriculture (fields and pastures mostly). The height above the sea level changes in very broad interval (from 300 m to more than 600 m a.s.l.). The occurrence of the uranium ore in this region is tied to the contact of two geological units - Moldanubicum and the Svratka River crystalline unit. The morphogenetic type of uranium deposits in the study area is characterized as a graphitized zones of crushing in rocks of crystallinic fundament (Československá ložiska uranu 1984).
Fig. 1. Location of the study area
The study of the possible impacts of mining on environment was significant not only in the time of a full development of these activities, but the research is even more important in the present time of decline of mining when reclamation of the area and mitigation of possible harmful consequences should be solved.
DATA
Different kinds of data have been used for the project solving. The first group of data is represented with a topographical map 1 : 10 000 in a digital format. The content of the topographical map was digitized to several coverages (digital terrain model, communication network, streams, settlements, etc.). Besides topography there are also several coverages which inform on spatial distribution on mining (active and re-cultivated mines) and related activities (management in tailing ponds) etc. The digital map of the study area was also used for rectification of the second group of data, which comprised from a subscene of multispectral imagery acquired from LANDSAT Thematic Mapper in May 1986. The period of 1980s can be denoted as a peak of the mining activity. After that there was a decline connected with the change of political situation after the year 1989. At this moment only a few mines are active, some areas are under a process of reclamation (see Fig. 1). For detail mapping of land cover and land use changes aerial photographs have also been used. Two sets of photographs from two time slices have been available - from the first part of the 1950s (before the beginning of mining activities) and also from the beginning of 1990s (decline of mining activities in some localities).
The last group of data was represented with a geophysical data - mostly the results of gamma-ray spectrometry. Aerial survey was performed with the line spacing of 250 m. Original measured data were interpolated into a regular 80 m grid. Data from the gamma-ray spectrometry allow to map a concentration of 40K, 238U, and 232Th and also the total activity of the mentioned radionuclides.
METHODS
Digital processing of remotely sensed data is based above all on multispectral approach. It means that with the help of several images of the same area, acquired in different but precisely defined intervals of electromagnetic spectrum, it is possible to recognize specific processes, even such ones, which can not be directly seen in the environment. This is based on the assumption that the searched processes are connected with the characteristic features in their spectral response. The typical example of such features can be the change of spectral response of vegetation cover due to various kinds of stress.
Possible approaches to the contamination detection connected with uranium mining with the help of remotely sensed data can be summarized briefly. Is some circumstances direct observation of rocks and minerals is possible. This approach seems to be very promising, especially with the help of hyperspectral data. Contaminated areas can show specific spectral response. On the other hand rocks and minerals are covered with the vegetation mostly.
The so called geobotany can be mentioned as the second possible approach. Geobotany is based on correlation between the occurrence of some species of vegetation and deposits of rocks and minerals (CAMPBELL 1996). Through recognition of vegetation species identification of rocks and minerals is possible to some extent. The main drawback of use of remotely sensed data in this approach is their limited spatial resolution and the fact that not "pure" but "mixed" spectral responses are recorded mostly.
In some cases, processes of contamination can be enhanced with the help of various kinds of spectral enhancements. This group of image processing techniques transforms original multispectral data to a new set of images. Changes in vegetation condition can be stressed using e.g. principal component analysis, color transformations or multispectral classification. Some methods of enhancement and classification allow to combine imagery with a different kinds of data - DEM ( contamination can be spread in the area through drainage network), or geological map (contamination can be controlled by faults of specific direction). Principal component analysis for the detection of polluted areas due to uranium mining has been successfully used by NEVILLE (1996). The author proved, that some transformed images - principal components (PCs) - correlate with the occurrence of specific rocks and minerals or even with the occurrence of contaminated areas.
Fig. 2. Concentrations of 238U (in ppm) in the study area
Radioactive pollution due to uranium mining can be studied directly with the help of geophysical remote sensing. Concentrations of the most radionuclides can be measured using airborne gamma-ray spectrometry. This kind of investigation has been used in Czech Republic for aerial monitoring of the environment in the uranium mining area Stráž pod Ralskem - Hamr (JURZA 1994)
OVERVIEW OF THE MAIN RESULTS
For identification of contamination connected with the uranium mining activities in the area of Dolní Rožínka and surroundings, geophysical data were used along with the correlation with LANDSAT TM data. From gamma-ray spectrometry records it is evident, that higher concentrations of 40K, 238U, and 232Th are more or less restricted on the mine shafts, mine tailings and tailing ponds. These concentrations of the radioactive isotopes sharply decrease with the increasing distance from the mentioned spots. It is clearly evident especially in the case of uranium 238U (see Fig. 2). Average natural concentration of 238U is about 2-4 ppm. The highest concentrations of this isotope in the processed area are more than 30 ppm. In case of thorium the average natural concentration is about 8-12 ppm. In the study area the concentrations of 232Th reach more than 24 ppm. At this time it is not possible to prove some dynamics in polluted areas. Temporal changes in spatial distribution of contaminated areas can be defined with the help of series of spectrometry measurements. On the other hand, in case of good correlation with gamma-ray spectrometry this dynamic aspect can be traced using multispectral satellite imagery. Relations between the two types of data are discussed further on.
Using LANDSAT Thematic Mapper imagery the mining areas are clearly visible even on a suitable enhanced RGB color composites, especially on those ones where channels from the middle infrared part of the electromagnetic spectrum are included. Further identification of mining areas is possible using spectral enhancements. As a suitable method principal component analysis can be designated. In this project several combinations of input original channels was tested. It was possible to delineate specific features in a transformed output channels (PCs). Some of them were restricted on mining areas only. So that they correlate well with the contaminated areas, which were defined using geophysical data. Other specific classes are not connected neither with mining areas nor with the specific land cover features. This holds especially about higher order PCs (see FIG. 3). Bright spots on the image show good spatial correlation with the higher concentration of uranium and thorium and therefore with the areas of mining activity (compare with Fig. 1 and 2.)
The second approach how to localize contaminated areas on LANDSAT TM imagery was ISODATA classification. This widely used clustering algorithm was used for definition as many as 50 fine spectral classes. Through subsequent aggregation of the spectral classes only those were preserved which classified pixels in the areas of mining activities. Such classes again show good correlation with the polluted areas denoted from gamma-ray spectrometry.
Fig. 3. Example of the 3rd principle component (PC3) transformed from the original LANDSAT imagery (TM1-TM5, TM7)
CONCLUSIONS
Uranium mining and related activities had a strong impact both on environment and also on life of people in the study area. Moreover, substantial changes had both positive and negative meaning and they hit such a different areas like land cover or land use of the region of e.g. an employment rate of inhabitants.
At this time higher concentrations of radioisotopes are restricted only to the places directly connected with the mining activities and are not spreading to the surroundings. This is very important for controlling decontamination and reclamation processes in the study area.
Using spectral enhancement and ISODATA classification, good agreement was proved between LANDSAT TM imagery and aerial gamma-ray spectrometry for the area contaminated on account of uranium mining. Due to temporal resolution of satellite imagery this method allows to study also the dynamics of the processes in contaminated areas. The success of environmental monitoring of similarly stressed areas with the help of remotely sensed data can be even more promising with better spatial and spectral resolution of used imagery.
REFERENCES
ACKNOWLEDGEMENTS
Authors would like to thank mining company DIAMO s.p. o.z. GEAM for providing of digital topographic maps and also GEOFYZIKA a.s. for gamma-ray spectrometry data.