Araştırılması, İTÜ FBE Yüksek Lisans Tezi,. İstanbul. Önal, G., 1981. Kilyos Bölgesi Kumlarının. Değerlendirme Olanaklarının Araştırılması, Türkiye. Madencilik ...
Açık Ocak Madenciliğinin Çevresel Etkilerinin Belirlenmesinde ve Rekultivasyon Çalışmalarının Takibinde Uzaktan Algılamanın Kullanımı Use of Remote Sensing in Determining the Environmental Effects of Open Pit Mining and Monitoring the Recultivation Process Zehra Damla UÇA AVCI
İstanbul Teknik Üniversitesi, Uydu Haberleşmesi ve Uzaktan Algılama Merkezi, İSTANBUL
Muhittin KARAMAN
İstanbul Teknik Üniversitesi, Uydu Haberleşmesi ve Uzaktan Algılama Merkezi, İSTANBUL
Emre ÖZELKAN
İstanbul Teknik Üniversitesi, Uydu Haberleşmesi ve Uzaktan Algılama Merkezi, İSTANBUL ÖZET Uzaktan algılama, madencilik faaliyetlerinden dolayı oluşan morfoloji değişimlerinin belirlenmesinde kullanılmaktadır. Buna ek olarak, madencilik faaliyetleri sonrası gerçekleştirilmesi gereken ağaçlandırma, yeniden düzenleme ve iyileştirme gibi rehabilitasyon çalışmalarının planlanması ve takibi aşamasında da ekonomik, hızlı ve doğru çözümler sunmaktadır. Kilyos ve Karaburun (İstanbul) arasındaki bölgede, yaklaşık 50 sene boyunca kömür çıkartma amaçlı açık ocak madenciliği yapılmıştır. Bu çalışmada, Ekim 2009 tarihli SPOT 5 uydu görüntüsü kullanılarak, bölgenin arazi örtüsü haritası çıkartılmıştır. Öncelikle bölgedeki doğal yüzey örtüsü sınıfları ile madencilik çalışmaları sonrasında oluşan yeni yüzey örtüsü sınıfları (bitki örtüsü tahrip edilmiş bölgeler, yapay gölet alanları gibi) belirlenmiş ve daha sonra segmentasyon ve sınıflandırma olmak üzere iki aşamalı görüntü analizi gerçekleştirilmiştir. Elde edilen tematik sınıflar konum, alansal büyüklük ve dağılım açısından değerlendirilmiştir. Ayrıca, uzaktan algılamanın bölgenin rehabilitasyonu sürecinde sağlayacağı faydalar da ortaya konmuştur. Sınıflandırılmış uydu görüntüleri, bölgedeki kömür madenciliği aktivitelerinin çevresel boyutunu analiz etmenin yanı sıra, ileriye yönelik olarak mekansal planlama için de iyi bir altlık olmaktadır. ABSTRACT Remote Sensing is conventionally used for detection of land cover change caused by open-pit mining activities. It also provides an economic, fast and accurate solution for planning and monitoring of rehabilitation processes. Open pit coal mining has been maintained in Kilyos - Karaburun region (Istanbul) for about 50 years. In this study, a SPOT 5 satellite image acquired on October, 2009 has been used for land cover mapping. First, the natural and the post-formed land cover types (such as vegetation removed surface and artificial pit lakes) were defined as thematic classes. Then, image analyzing methods were applied in two steps as segmentation and classification. The classes were evaluated with respect to their location, area and distribution. Besides, use and advantages of remote sensing for rehabilitation process were also presented. Classified satellite images can be used as base maps, both for analyzing the environmental effects and spatial planning and management of the region for the future use.
1 INTRODUCTION Mining is the extraction of valuable minerals or other geological materials from the earth, generally with an operation that involves the physical removal of earth surface. Mining activity includes excavations in underground mines, surface excavations in open-pits, and also the mining from the seafloor. The process of obtaining useful minerals, using the raw material in industry, and furthermore exportation of them, are very important for the development of a country. However, the mining activity has to be organized not only in the aim of extracting the material but also saving the natural surrounding environment. The process has to involve the protection of environment and sustainability principles, which require social, ecological, economic, spatial and cultural dimensions as well. Environment-friendly mining will save the human-nature balance, meet needs of the day in consideration with the plan for future generations. All these require some operations that have to be done all before the mining activity, during the mining operations and after the mining activities are completed. Respecting it in all process steps is important not to cause permanent harm to the ecology of region. Generally, the pre-mining activities are categorized as precaution / protection operations, and post-mining activities are categorized as retrieval / recruitment applications, which may be enhanced to reformation, nature restoration and rehabilitation. Post-operations may include the conversion of the mining area to the initial or to a better state; or changing it for a totally different functionality / usage. All these processes including the pre- and postoperations can be defined as recultivation (or reclamation). (Simsir vd., 2007) The planning, monitoring and controlling of mining and recultivation operations can be done by remote sensing as well as ground surveying. RS technology provides the
possibility of collecting data of wide areas in a short time. The data acquired is in a digital form, can be derived and analyzed in various ways and integrated to other information systems. In this study, satellite data is analyzed to determine the present status of the region and to show the potential of optical data use in mining for monitoring the recultivation process in scope of environmental conscience. 2 STUDY AREA AND DATA USED One of the important underground resources of Istanbul is the coal mines occurring between Kilyos and Karaburun areas. Coal formation of Trakya Tersiyer basin is hosted by Danismen Formation of Oligocene age. The Danismen Formation is consisted of grey-colored mica sand; and grey-colored marly, clay stone, silt stone and lignite layers. In Agacli region, where clay-coal layers are observed, the clay seams exist between medium and upper coal seams as 11.5 m width. In Bolluca region it occurs as 2.5 m thick layer alternating with upper medium and lower seams. (Koroglu) The lignite formation is observed in fringe of Istranca massifs. The Agacli lignite basin covers approximately 25 km. squares extending from the Bosphorus entrance at north, to Terkos Lake at west. (Sengüler, 2007) The Ciftalan, Ihsaniye, Bolluca, Akpinar, Yenikoy, Agacli coal basins account for Agacli basin lignites. (Koroglu) The recent studies show that at the depths of 350-550m and 550-700 m, existing coal seams with thickness between 0.5-4 m have calorific values between 2000-3000kcal/kg. The total reserve was determined as more than 520 million tons. (MTA report, 2005) As for potential, the sand reserves of the Kilyos region were found to be 55 million tons of quartz in sand dunes of Gumusdere and Agacli, 7 million tons for feldspar and 3.8 million tons for the heavy minerals. (Onal G. 1981)
The coal deposits have been mined out since 1919 using open-pit techniques. All the mining activities have been done by open-pit mining operation. In this coal basin, the licensees started the production from the coast side through the insides of land and they poured the extracted land\rock into the closest seaside. In some regions, this poured rock has been used for operating the coal veins that lie through the sea bottom. The land is enlarged and reformed by each filling activity and the coast is shifted through the sea. As confirmed by studies in literature, coastline changes are very important for environment and the living beings of the habitat. This region’s coastline has been in change since mining in this region has been started. (Uça, Z. D. et al., 2006) Pouring land into the sea to form pools and then pumping water out has been applied to reach the coal veins and mine extraction in the sea part. Coast environment has been destroyed by these activities. The land part has also been destroyed by the formation of wide pits which turn into artificial lakes after abandoned by the operators. In this region many artificial lakes formed after mining activity are observed. (Turnacigil, 2008) In many ways, the lignite coal mining has caused distortions on the morphology of the region. 2.1 Study Area In this study, to represent the influence area of mining -which is affected in positive or negative ways, before, during or after the operations-, focus area is selected as seen in Figure 1. The region is in Sariyer, Eyup and Arnavutkoy country boundaries. Gradient of the slopes in the region are between 20-30 %; the highest elevations on the west is about 120 m. and the lowest is beyond negatives by being inside the natural sea level on the north. (Turnacigil, 2008)
Figure 1. Study Area*
2.2 Data Used In this study, multispectral SPOT 5 data acquired on October2nd, 2009 was used. SPOT 5 is an optical satellite that has sensors to detect and record the reflectance of land surfaces on visible and infrared regions of the electromagnetic spectrum. SPOT 5 has multispectral and monospectral data acquisition modes. In this study, multispectral SPOT 5 image was used which has been recorded in four bands. The spectral intervals of SPOT 5 bands are shown in Table 1. The spatial resolution of SPOT 5 images in multispectral mode is 10 m. Table 1. Spectral Properties of SPOT 5 Data Band Number Band 1 Band 2 Band 3 Band 4
Spectral Region 0.50 - 0.59 µm 0.61 - 0.68 µm 0.78 - 0.89 µm 1.58 - 1.75 µm
3 METHOD The present situation of a mining region and its environment may well be evaluated by using a segmentation and image classification method on satellite data. This kind of situation analysis can also be used as a base and data source for spatial planning process before starting the mining activities.
In addition to the production of a status map, remote sensing may also be used for monitoring the region and quantifying some of the recultivation activities. Revegetation, which is one of the rehabilitation steps, can be followed up by using satellite data and image analyzing methods. One of the most practical methods, which can be used for identifying the vegetation presence, status and distribution, is using normalized difference vegetation index (NDVI) images. In this study, as a first step of image processing, segmentation was applied to form image objects. Then image classification was performed to determine the thematic classes. As a second step, the NDVI image was produced to obtain the vegetation status in the region. 3.1 Segmentation Segmentation process is the division of whole image into separate image parts according to a homogeneity criterion, which is controlled by color, shape, compactness and smoothness parameters. The convenient values for segmentation parameters and the scale parameter, which is also an initial input, were assessed by the analyst. 3.2 Classification After segmentation process was completed, classification was performed on image objects. Class descriptions were defined and each image object was assigned to a suitable class. In this study, both of the nearest neighbor and condition-based classification algorithms were used. 3.3 NDVI Image NDVI is a commonly used vegetation index in remote sensing analysis to distinguish the vegetation from non-vegetation features and also identify the vegetation condition. In an NDVI image, pixels take values between -1 and +1. Higher NDVI values indicate a greater level of photosynthetic activity. (Sellers, 1985; Tucker et al., 1991) NDVI image is produced with a band math equation as: NDVI = ( NIR - R ) / ( NIR + R ) where NIR: Near infrared band, R:Red band.
4 APPLICATION 4.1 Segmentation Three scale parameters were tested to decide the suitable segmentation layer, which will provide a convenient base for the discrimination of thematic classes. The resultant segmentation image produced for three different scale parameters (sp) is given in Figure 2.
(a)
(b)
(c)
Figure 2. Segmented Image with sp (a) 200, (b) 100, (c) 50
Sp = 50 was selected as the scale parameter to fit the requirements, by using visual analysis. The other parameters of the segmentation were determined as given in Table 2. Table 2. Segmentation Parameters Used Parameter Color Shape Compactness Smoothness
Value 0.9 0.1 0.5 0.5
4.2 Classification Thematic classes for this region were determined as: ‘Mine Type 1, Mine Type 2, Mine Type 3, Vegetation Type 1, Vegetation Type 2, Vegetation Type 3, Destroyed Land Surface Type 1 (DLS 1), Destroyed Land Surface Type 2 (DLS 2), Destroyed Land Surface Type 3 (DLS 3), Water, Road. The samples and the related surface views in (RGB: 4,1,2) are given in Table 3.
Table 3. Surface Views of Training Samples Class Name
The resultant classification image and the
Sample Region
Vegetation Type 3 Vegetation Type 2 Vegetation Type 1 DLS Type 3 DLS Type 2 DLS Type 1
legend of the final classification are given in Figure 4 and 5, respectively. Figure 4. Classification Image
Water Mine Type 3 Mine Type 2 Mine Type 1 Figure 5. Legend of the Classification Image
The set of sample objects, which were used as training data, is visualized in Figure 3.
After performing classification, the areal percentages of the classes were obtained as given in Table 4. Table 4. Areal Percentages of the Classes Class Name Vegetation DLS Mine Lake Road
Figure 3. Locations of Training Samples Used
For all of the classes except ‘road’ class, standard nearest neighbor classifier was used to obtain class distribution functions. For ‘road’ class, the feature for class description was defined by length/width proportion criterion.
Area % 53 29 7 6 5
The classified image and areal sizes of the classes represent the present situation of the region by the date satellite image was acquired. 4.3 NDVI Image The NDVI image produced is given in Figure 6.
Figure 6. NDVI Image
The NDVI values of pixels were found to be between [-0.63, 0.45] interval. Using the NDVI image, it can be said that the surrounding of the mining areas and especially the coastal region doesn’t have dense vegetation when compared to regions relatively far from mining areas. This may probably be an indicator of naturally overrunning weeds, or sodding, whereas high NDVI values may indicate a more dense and/or aged vegetation cover or woodlands (such as in the southeast of the region). Obtaining satellite images periodically and applying NDVI analysis may be used in recultivation operations to observe the change detection in vegetated areas. 5 RESULTS & CONCLUSION One of the most common methods of expressing the classification accuracy is using error matrix. Based on error matrix, the overall accuracy of the classification was found to be 83%. It can be thought that mining activity is relational to ‘mine’, ‘DLS’ and ‘lake’ classes. These classes cover 42 % of the total region, showing that the action has a wide affection area. Land cover/land use maps derived from satellite data, provide information on both the areal sizes of the thematic classes and spatial distribution of them. This allows the use of results for spatial analysis with RS and/or GIS tools.
In addition to determination of the present situation of the region, the management after completing mining activities and planning the future use of the region may well be organized by using satellite data. Use of satellite data and image analyzing methods may have a significant part on the control mechanism of some recultivation steps. RS will also be an economic, fast and accurate way of planning and checking on after-use processes. Using satellite data over the region of interest periodically will allow multitemporal processing for change detection analysis. Sustainable development primarily requires environment-friendly arrangements on the management of sectors which are directly involved with environment. RS can be a useful tool to observe and plan the human activity and its effects on Earth. Authors would like to thank to Istanbul Technical University - CSCRS, for SPOT 5 data.
REFERENCES Köroğlu, Ç., 2007. Ağaçlı-Bolluca (İstanbul) Yöresi Seramik Killerinin Malzeme Özelliklerinin Araştırılması, İTÜ FBE Yüksek Lisans Tezi, İstanbul. Önal, G., 1981. Kilyos Bölgesi Kumlarının Değerlendirme Olanaklarının Araştırılması, Türkiye Madencilik Bilimsel ve Teknik 7. Kongresi, MMO, Ankara. Sellers, P.J., 1985. Canopy Reflectance, Photosynthesis and Transpiration, International Journal of Remote Sensing, 6:1335-1372. Şengüler, İ., 2007. Ülkemiz Enerji Bütünlemesinde Marmara ve Trakya Bölgesi Kömürlerinin Yeri, MTA Genel Müdürlüğü Enerji Dairesi Başkanlığı, Ankara. Şimşir, F., Pamukçu, Ç., Özfırat, M. K., 2007. Madencilikte Rekültivasyon ve Doğa Onarımı (Mine Reclamation and Restoration of Nature), DEÜ Mühendislik Fakültesi, Fen ve Mühendislik Dergisi, Cilt: 9 Sayı: 2 sh. 39-49. Turnacigil, A., 2008. Yeniköy Ağaçlı Civarındaki Maden Ocaklarının Rehabilitasyonu, İTÜ FBE Yüksek Lisans Tezi, İstanbul. Tucker, C.J., Dregne H.E., and Newcomb W.W., 1991. Expansion and Contraction of the Sahara Desert from 1980 to 1990, Science, 253:299-301. Uça, Z. D., Sunar Erbek, F., Kuşak, L., Yaşa, F., and Özden, G., 2006. The Use of Optic and Radar Satellite Data for Coastal Environments,
International Journal of Remote Sensing, Vol. 27, No. 17. *Figure 1. was adapted from maps.yahoo.com
