Environ Monit Assess (2010) 163:397–410 DOI 10.1007/s10661-009-0843-7

Monitoring deterioration of vegetation cover in the vicinity of smelting industry, using statistical methods and TM and ETM+ imageries, Sarcheshmeh copper complex, Central Iran F. Rastmanesh · F. Moore · M. Kharrati-Kopaei · M. Behrouz

Received: 25 June 2008 / Accepted: 16 February 2009 / Published online: 19 March 2009 © Springer Science + Business Media B.V. 2009

Abstract Simple statistical methods on Normalized Difference Vegetation Index (NDVI) and bands 3 and 4 data of relatively coarse resolution Thematic Mapper (TM) and Enhanced Thematic Mapper plus (ETM+ ) imageries were used to investigate the impacts of air pollution on the deterioration of the vegetation cover in the Sarcheshmeh copper complex of central Iran. Descriptive statistics and k-means cluster analysis indicated that vegetation deterioration had already started in the prevailing wind directions. The results show that combination of simple statistical methods and satellite imageries can be used as effective monitoring tools to indicate vegetation stress even in regions of sparse vegetation. Despite various possible perturbing factors upon NDVI, this index remains to be a valuable quantitative vegetation monitoring tool.

F. Rastmanesh (B) · F. Moore Department of the Earth Sciences, College of Sciences, Shiraz University, Shiraz, 71454, Iran e-mail: [email protected] M. Kharrati-Kopaei Department of Statistics, College of Sciences, Shiraz University, Shiraz, 71454, Iran M. Behrouz Sarcheshmeh Copper Complex, Office of Research and Development, Kerman, Iran

Keywords Air pollution · NDVI · Statistical evaluation · Remote sensing · Vegetation cover deterioration · Sarcheshmeh copper complex · Iran

Introduction A major cause of vegetation cover deterioration is air pollution. The smelting industry sector is a well-known air pollutant source worldwide. In the vicinity of nonferrous metal smelters, high concentrations of toxic compounds have been detected in soils and vegetation (e.g. Ettler et al. 2005). Smelting industries have markedly degraded the natural environment around the world. In areas like Kola Peninsula, the reported total industrial barrens around Severonickel Smelter are as much as 200 km2 (Kalabin and Evdokimova 1993). The main types of air pollutants caused by smelting industries are noxious gases such as SO2 and heavy metal particulates (e.g. Ress and Williams 1997; Tommervik et al. 2003). Inflicted damage to vegetation cover is usually monitored using satellite imagery (Rigina et al. 1999). The reason is the synoptic view and multitemporal sensing of satellite imagery making it suitable for monitoring vegetation health, that is, condition and change through time. According to Mikkola (1996), the application of remote sensing methods can produce a more accurate spatial view

398

Environ Monit Assess (2010) 163:397–410

of the phenomenon. Multitemporal image analysis and change detection methods can also give some historical insight on vegetation deterioration. While it is necessary to characterize the nature of the problem and the ranges of magnitude of effects on vegetation through field sampling, it is difficult to determine the geographic extent and locations of the areas affected using conventional methods (Woodcock et al. 2002). It has been demonstrated that stress and reduced vitality have some effect on the spectral signature of green plants, generally manifested as a decrease in near infrared (NIR) and increase in the visible wavelengths (Hame 1991; Jensen 1996). According to Gupta (1991), healthy vegetation normally has a strong reflection in the nearinfrared (NIR). Vegetation indices are known to correlate with such related physicochemical properties as chlorophyll concentration, leaf area index, foliar loss and damage, photosynthetic activity, and carbon fluxes (Tucker 1979; Asrar et al. 1984; Sellers 1985; Tucker et al. 1986; Vogelmann 1990; Buschmann and Nagel 1993). The most widely used vegetation index for ecological applications is the Normalized Difference Vegetation Index (NDVI) (Garty et al. 2001; Toutoubalina and Ress 1999; Mikkola 1996; Bala et al. 2000; Giannico 2007). Rouse et al. (1974) have defined NDVI as: NDVI = (Near−Infrared−Red) /(Near Infrared+Red) This index as well as its less popular modifications (e.g. Soil-Adjusted Vegetation Index (SAVI), Transformed Vegetation Index (TVI)) is based on the difference between the maximum absorption of radiation in the red band due to chlorophyll pigments and the maximum absorption radiation in the NIR band due to leaf cellular structure, and the fact that soil spectra, lacking these properties, do not show a dramatic spectral difference (Garty et al. 2001). NDVI is the most commonly used vegetation index as it retains the ability to minimize topographic effects while producing a linear measurement scale. The higher the NDVI values the higher the probability that the corresponding area

on the ground has a dense coverage of healthy green vegetation (Campbell 1996). In most of the studies which use satellite imageries for assessing the impact of air pollution on vegetation, NDVI evaluations (e.g. Giannico 2007; Garty et al. 2001), image classification techniques (e.g. Mikkola 1996; Ress and Williams 1997), land cover maps with ancillary data (e.g. Tommervik et al. 1998, 2003), or multitemporal change detection technique (e.g. Hanger and Rigina 1998) are used. However, most of these studies deal with dense vegetation cover such as forest and very little research is done on sparsely vegetated areas. The main purpose of this paper is to present a method based on simple statistical approaches on NDVI values and bands 3 and 4 data, for investigating the effect of air pollution on vegetation in Sarcheshmeh, a sparsely vegetated area. In order to serve this purpose, image descriptive statistics and modified soil brightness line plot (Richardson and Wiegand 1977) are used.

Study area Iran ranks 16th among the world’s major copper producers (Edelstein 2003). Sarcheshmeh copper deposit, the largest porphyry copper deposit in Iran is located 160 km SW of Kerman city in Kerman Province (Fig. 1). The Sarcheshmeh orebody along with a number of other porphyry copper deposits occur in the so called central Iranian volcanic belt (Waterman and Hamilton 1975; Forster 1978; Shahabpour and Kramers 1987). Reverb and converter stacks of the smelting plant release 136,000 and 163,000 m3 /h of gas into the atmosphere, respectively. SO2 gas constitutes 2.6% and 4.8% of the emissions, respectively (Ebrahimi and Hakimi 2002). The region’s temperature varies between −20◦ C in the winter to +32◦ C in the summer (Shirashiani 2004). Average rainfall is 440 mm. Prevailing wind directions based on available meteorological data in the period 1987 to 2000 are NE and N (Sarcheshmeh meteorological station). Vegetation cover reflects the climatic conditions of the region, i.e., in low-lying plains vegetation cover is mainly bush and desert plants, while

Environ Monit Assess (2010) 163:397–410

399

Fig. 1 Study area location and selected windows on NDVI image for the year 1987

wild almond and pistachio flourish in mountainous areas (Shafiei 2000). According to Shahidi et al. (2007), the region’s vegetation density varies between 58.4 and 126.6 bushes per 25 m2 . Although the overall natural vegetation is relatively sparse, it is fairly concentrated along stream courses and orchards. A residential complex built for the personnel and known as “The copper city” (Fig. 1) is the nearest population center to Sarcheshmeh copper complex and is located 5 km W-NW of the complex. The population of this city in 2001 was reported to be 9,018 (Shirashiani 2004). There are also several villages in the vicinity of the copper complex.

Satellite data A cloud-free Landsat ETM+ subscene from Aug. 7, 2000 and a somewhat cloudy subscene of TM Landsat from Aug. 28, 1987 over the study area were chosen for detection of vegetation cover changes. The partial cloudiness of TM data did not have any effect on the image processing, as the clouds were covering less than 10% of the subscene, mostly south of the open pit. The reason for choosing August imageries was to

avoid snow cover and represent the peak of the growing season. Prior to statistical analysis, satellite data were geometrically and atmospherically corrected. The Landsat images were rectified to UTM projection system (ellipsoid WGS-1984; Zone, 40N) using detailed topographic maps at scale of 1:20,000 and choosing ground control points. This resulted in an estimated registration error of less than one pixel. A nearest neighbor algorithm was chosen in order to avoid spectral mixing between adjacent pixels (e.g., Ress and Williams 1997). Since NDVI is sensitive to atmospheric influence (Holben 1986) and the purpose of this study is detection changes on vegetation cover, the simple dark object subtraction was used for atmospheric correction, as suggested by Song et al. (2001). This atmospheric correction method has already been used by other authors (e.g. Franklin et al. 1997). However, NDVI partially compensates changes in the illumination conditions, surface slope, aspect, and other illumination factors (Lillesand and Kiefer 1987; Giannico 2007). Also NDVI like any other ratio indices substantially reduces topographic effects (Walsh et al. 1997). The impacts of sun angle differences are also partially reduced if the data belong to the similar dates (Singh 1989; Jensen 1996).

400

Environ Monit Assess (2010) 163:397–410

Table 1 Selected windows and their distances from the stacks

Selected windows

Distance from stacks (Km)

N NE1 NE2 E SW W

3.8 5.2 11.5 1.2 13.7 5.2

Methodology Window selection In order to evaluate air pollution effects on vegetation cover in a 14-years period (1987–2000), six windows (50 × 50 pixels each) were selected in different geographical directions. The chosen sizes of the windows are proportionate with vegetation cover in the study area. Table 1 shows the geographical directions and distances of windows from the smelter stacks. In selecting these windows the following consideration were taken into account: 1. Despite the scarcity of the vegetation cover, relatively dense vegetation occurs in main drainage courses and orchards. 2. The windows must be selected in such a way to best reveal the effect of pollutant emissions on vegetation, and indicate subtle vegetation stress. Hence the windows were selected in the upwind and downwind directions to ease comparison.

Three of the selected windows with varying distances from the smelting plant stacks were chosen in the direction of the prevailing wind, i.e., North and Northeast. Only one window was selected in the North direction. The reason is acid drainage of the mine which also flows in this direction, making it impossible to differentiate acid mine drainage (AMD) and polluting emission effects on vegetation. However, the single selected window in this direction will reveal the overall effect on vegetation. The fourth window was also selected west of the smelting plant, where the copper city is located. The remaining two windows are selected in SW and E directions. Figure 1 shows the windows locations. Descriptive statistics In order to show NDVI variations in the selected windows, statistical parameters, i.e., maximum, minimum, mean, median, standard deviation, first (Q1 ), and third (Q3 ) quartiles, were determined. All statistical calculations were carried out by Splus 6.1.2, and SPSS 10 softwares. Table 2 presents the calculated statistical parameters for the years 1987 and 2000. All values (except maximums) are negative and close to zero indicating that most pixels display unvegetated surface. The data for the year 2000 are more negative compared with 1987 data, probably reflecting vegetation loss in the 14 years interval. Standard deviation (SD) is generally used as a criterion for homogeneity; in this case the smaller the SD, the more homogenous the NDVI values.

Table 2 Descriptive statistics for selected windows Window Year Mean N1 NE1 NE2 W SW E

1987 2000 1987 2000 1987 2000 1987 2000 1987 2000 1987 2000

−0.054 −0.116 −0.58 −0.97 −0.083 −0.117 −0.053 −0.071 −0.064 −0.084 −0.062 −0.098

First quartile (Q1 ) Median (Q2 ) Third quartile (Q3 ) Standard deviation (SD) Max. Min. −0.098 −0.135 −0.097 −0.116 −0.111 −0.141 −0.092 −0.119 −0.107 −0.123 −0.085 −0.121

−0.077 −0.12 −0.062 −0.104 −0.102 −0.132 −0.076 −0.103 −0.095 −0.112 −0.064 −0.103

−0.049 −0.104 −0.044 −0.088 −0.088 −0.119 −0.044 −0.070 −0.078 −0.097 −0.048 −0.087

0.089 0.029 0.035 0.034 0.064 0.052 0.080 0.10 0.101 0.095 0.040 0.048

0.503 0.090 0.348 0.261 0.369 0.265 0.638 0.658 0.463 0.503 0.489 0.428

−0.163 −0.190 −0.12 −0.176 −0.151 −0.178 −0.191 −0.218 −0.142 −0.176 −0.164 −0.232

Environ Monit Assess (2010) 163:397–410

If the SD is small and mean pixel values are more positive, the homogeneity implies more vegetation cover; inversely, if SD is small and the mean of pixel values is more negative, the homogeneity indicates barren soil or rock. Comparison of windows SD and mean values for the year 1987 and 2000 (Table 2) shows that the greatest difference in SD values occurs in N window, where the SD in 2000 is smaller and the mean value is more negative. Therefore, pixel

Fig. 2 Box plots for the seleted windows. a N-Window; b NE1-Window; c NE2-Window; d SW-Window; e W-Window; f E-Window. In a to f, the left and right panels show the box plot for NDVI in 1987 and 2000 respectively. The y-axis is the value of NDVI

401

cover in this window is more homogeneous and there is less vegetation than in 1987. A similar trend (although less pronounced) is seen in other windows. First and third quartiles (Q1 and Q3) may also be used to compare NDVI values in 1987 and 2000 to avoid comparison of the extreme values, i.e. maximum and minimum. It must be noted that 25% and 75% of the dataset fall at or below the first and third quartiles, respectively. Obviously,

402

the more negative the Q1 and Q3 , the less is the vegetation cover. Table 2 shows that all Q1 and Q3 values in 1987 and 2000 are negative and near zero indicating that most NDVI values represent barren soil or rock. However, Q1 and Q3 in 2000 have lower values. The more negative Q3 values in the year 2000 compared with Q1 in 1987 emphasizes the fact that vegetation cover (positive NDVI values) is much reduced in the 14 years interval. The

Fig. 2 (continued)

Environ Monit Assess (2010) 163:397–410

reduced vegetation is more prominent in N, NE1 , and NE2 windows. To provide a visual comparison between the NDVI values of the two studied years on the basis of Q1 , Q2 (median), Q3 , and also the range of NDVI values in the two studied years, box plot (Mc Clave and Sincich 2000; Upton and Cook 2001) is used. Box plots generally provide useful information on the variation and central point of a data set. Note that the median (Q2 ) shows the

Environ Monit Assess (2010) 163:397–410

central point of a data set much better than the mean if there are outliers. Results for the six selected windows are given in Fig. 2a–f. It is clear that the NDVI values for the year 2000 are dramatically reduced in N, NE1, and NE2 windows, i.e, in the prevailing wind directions (compare Q1 and Q2 and NDVI ranges in Fig. 2a–f).

Fig. 2 (continued)

403

Q1 and Q3 are also more negative in W, SW, and E windows in 2000 (Fig. 2 and Table 2), indicating an increasing trend in the number of more negative pixels. However the reduction is not very prominent. Figure 3 is a surface diagram for N window. Note that two units are added to each NDVI values (Z axis) to enhance clarity. X and Y axes

404

Environ Monit Assess (2010) 163:397–410

Fig. 3 Surface diagram of the N window. a 1987 and b 2000. Note that two units are added to NDVI values (Z -axis) to enhance clarity

Fig. 4 Soil brightness line (adopted from Jensen 1996). After Richardson and Wiegand (1977)

Fig. 5 Results of k-means cluster analysis on N window of 1987 using two clusters

Environ Monit Assess (2010) 163:397–410

are the number of pixels in the X and Y directions, respectively. The reduction of vegetation cover in the year 2000 is obvious.

405

Note that no statistical tests (e.g. t-test) are used to compare the windows of the studied years, since these windows were not selected at random

Fig. 6 Results of k-means cluster analysis using three clusters on N window a (1987), b (2000); SW window c (1987), d (2000); and NE1 window e (1987), f (2000)

406

and all pixels in these windows are taken into account. k-means cluster analysis Measured reflectance by a sensor often contains soil and vegetation components (Gibson and Power 2000). On a scatter diagram of nearinfrared and red band values, soil response will form a more or less straight line. This is the soil brightness line which was first devised by Richardson and Wiegand (1977). The perpendicular distance to the soil line is an indicator of plant development (Jensen 1996). Generally, the farther away from the soil line to the left, the greater the amount of vegetation present (Fig. 4, Curran 1983), since healthy vegetation has low value in the visible red and high value in the near-infrared (Gibson and Power 2000). The goal in cluster analysis is to find an optimal grouping for which the observations within each cluster are similar but between the clusters are dissimilar (Rencher 2002). Plotting bands 3 and 4 data for N-1987 window on a scatter diagram reveals two separate clusters of soil and vegetation (Fig. 5). If soil is further divided into wet and

Fig. 6 (continued)

Environ Monit Assess (2010) 163:397–410

dry, then a third cluster will also appear on the scatter plot. Therefore, k-means cluster analysis with 3 clusters on bands 3 and 4 values of selected windows were carried out. Cluster analysis for N-1987 window, produced three separate clusters of which, vegetation cluster tends to the left (Fig. 6a). Clustering of N-2000 data also produced three clusters on the soil line (Fig. 6b). k-means cluster analysis with SW (upwind) and NE1 (downwind) produced interesting results. In the case of SW (1987 and 2000), cluster analysis provides three separate clusters, with a distinct vegetation cluster (Fig. 6c, d); while for NE1 (1987 and 2000) there are three mixed clusters of vegetation and soil data (Fig. 6e, f). The reason is probably lack of dense vegetation in these 2 years to produce distinct clusters. The mixed clusters seem to delineate partially vegetated pixels with varying vegetation to soil ratios (Gibson and Power 2000). Apparently, whenever the vegetation cover is healthy and dense, cluster analysis divides the pixel values into three clusters with a distinct vegetation cluster. Hence, in our experience, pixels of healthy vegetation with high spectral reflectance in near-infrared band (high value) and low value

Environ Monit Assess (2010) 163:397–410

in red, and high perpendicular distance to the soil line, form a separate cluster. Note that when there are three distinct clusters (in the case of healthy vegetation cover), the value of NDVI increases and hence it is expected that Q1 , Q2 , and Q3 (in Section “Descriptive statistics”) have higher

Fig. 7 NDVI image of N window a (1987), b (2000); SW window c (1987), d (2000); and NE1 window e (1987), f (2000)

407

values. For instance, in the left panel of Fig. 2a the values of Q1 , Q2 , and Q3 are −0.098, −0.077, and −0.049, respectively, resulting in three separate clusters in Fig. 6a. While more negative values for Q1 , Q2 , and Q3 in the right panel of Fig. 2a (see Table 2), resulted in three mixed clusters in

408

Fig. 6b. Thus k-means cluster analysis is a good means of monitoring overall vegetation status. It should be noted that, increasing the number of clusters will not affect the vegetation cluster; and merely increases the number of clusters on the soil line, probably as a function of wetness. Since in Fig. 4, the lower end of the soil line represents the wet soil and the upper end represents the dry soil.

Concluding remarks Figure 7 represents NDVI images for the selected windows. Vegetation deterioration is evident especially in N, NE1, and NE2 windows in the year 2000. This is in agreement with the statistical data and indicates the impact of prevailing wind direction. The coincidence of the inflicted vegetation damage and the prevailing wind direction emphasizes the role played by air pollution than other causes such as draught conditions. A recent study carried out by Lotfian (2007) on lichens in the vicinity of Sarcheshmeh copper complex indicate reduced lichen chlorophyll contents due to air pollution. The reduction is also more severe northeast of the complex (Lotfian 2007). Ebrahimi and Hakimi (2002) showed that SO2 concentration is higher north and northeast of the complex. Close investigation of Fig. 7 shows that vegetation deterioration has resulted in reduced image contrast in NDVI images in 2000. If NDVI value is considered as a brightness measure of pixels then, whenever vegetation is healthy and dense, i.e., higher difference NDVI value among pixels, higher image contrast exists. Conversely, whenever there is stressed or sparse vegetation cover (as in the case of NE1 (1987, 2000) and N (2000), there is no significant difference among adjacent pixels and NDVI values. According to Hanger and Rigina (1998), topography also plays an important role in vegetation sensitivity to air pollution. In general elevated and windward sites appear to be more affected than depressed and leeward locations. Window locations in this study also indicate that

Environ Monit Assess (2010) 163:397–410

wind direction is apparently more important in Sarcheshmeh than topography. This study showed that simple statistical methods can be used as a monitoring tool to indicate vegetation cover deterioration even in a region of sparse vegetation, using relatively coarse resolution TM and ETM data. Finally, despite many possible perturbing factors upon the NDVI, it remains a valuable quantitative vegetation monitoring tool. Acknowledgements The authors would like to thank R&D department of the Sarcheshmeh copper complex for financially supporting this research. Thanks are extended to the research committee of Shiraz University for logistical support.

References Asrar, G., Fuchs, M., Kanemasu, E. T., & Hatfield, J. L. (1984). Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance in wheat. Agronomy Journal, 76, 300–306. Bala, A. E., Batelaan, O., & Desmed, F. (2000). Appraisal of remote sensed natural resources around Dustin MA, northwestern Nigeria. Journal of Science Engineering and Technology, 7(4), 2860–2869. Buschmann, C., & Nagel, E. (1993). In vivo spectroscopy and internal optics of leaves as basis for remote sensing of vegetation. International Journal of Remote Sensing, 14, 711–722. doi:10.1080/01431169308904370. Campbell, J. B. (1996). Introduction to remote sensing. London: Taylor and Francis. Curran, P. Z. (1983). Estimating green LAI from multispectral aerial photography. Photogrammetric Engineering and Remote Sensing, 49, 1709–1720. Ebrahimi, H., & Hakimi, H. (2002). Unpublished internal report, Investigating SO2 emissions in Sarcheshmeh copper complex and Copper City. R&D Center, Sarcheshmeh Copper Complex (in Farsi). Edelstein, D. L. (2003). Copper, Minerals Yearbook 2001, accessed March 10, 2003, at URL & http:// minerals.usgs.gov/minerals/pubs/commodity/copper/ coppmyb01.pdf. Ettler, V., Vanke, A., Minaljevic, M., & Bezdickal, P. (2005). Contrasting lead speciation in forest and tilled soils heavily polluted by lead smelting. Chemosphere, 58, 1449–1459. doi:10.1016/j.chemosphere.2004.09.084. Forster, H. (1978). Mesozoic-Cenozoic metallogenesis in Iran. Journal of Geological Society of London, 135, 443–455. Franklin, S. E., Lavigne, M. B., Deuling, M. J., Wulder, M. A., & Hunt, R. E. (1997). Estimation of forest leaf area index using remote sensing and GIS data for modeling

Environ Monit Assess (2010) 163:397–410 net primary production. International Journal of Remote Sensing, 16, 3459–3471. Garty, J., Tamir, O., Hassid, I., Eshel, A., Cohen, Y., Karnieli, A., & Orlovsky, L. (2001). Photosynthesis, Chlorophyll integrity and spectral reflectanece in lichens exposed to air pollution. Journal of Environmental Quality, 30, 884–893. Giannico, C. (2007). Remote sensing of vegetation in the Calabrian region. Acta Astronautica, 60, 119–131. doi:10.1016/j.actaastro.2006.07.003. Gibson, P. J., & Power, C. H. (2000). Introductory remote sensing digital image processing and applications. London: Taylor and Francis. Gupta, R. P. (1991). Remote sensing geology. New York: Springer. Hanger, O., & Rigina, O. (1998). Detection of forest decline in Monchegorsk area. Remote Sensing of Environment, 63, 11–23. doi:10.1016/S0034-4257(97) 00113-2. Hame, T. (1991). Spectral interpretation of changes in forest using satellite scanner images. Acta Forestalia Fennica, 222, 1220. Holben, B. (1986). Soil influence in remote sensed vegetation-canopy spectra. In C. Elachi (Ed.), Introduction to the physics and techniques of remote sensing. New York: Wiley-Interscience. Jensen, J. R. (1996). Introductory digital image processing a remote sensing perspective. Upper Saddle River: Prentice Hall. Kalabin, G., & Evdokimova, G. (1993). Ecology and natural protection of the Kola North. Educational text book (252 pp). Apatity: Kola Scientific Center. Lillesand, T. M., & Kiefer, R. W. (1987). Remote sensing and image interpretation. Canada: Wiley. Lotfian, S. (2007). Detemination of lichens around Sarcheshmen copper complex and estimation the effect of pollutants from complex on them with special reference to the bioindicator species Rhizoplaca melanophthalma. MSC thesis, University of Kerman, Iran. Mc Clave, J. & Sincich, T. (2000). Statisticsm (8th ed.). Upper Saddle River: Prentice Hall. Mikkola, K. (1996). A remote sensing analysis of vegetation damage around metal smelters in the Kola Peninsula, Russia. International Journal of Remote Sensing, 17(18), 3675–3690. doi:10.1080/01431169608949177. Ress, W. G., & Williams, M. (1997). Monitoring changes in land cover induced by atmospheric pollution in the Kola Peninsula, Russia, using Landsat-Mss data. International Journal of Remote Sensing, 18(8), 1703–1723. doi:10.1080/014311697218061. Rencher, A. C. (2002). Methods of multivariate analysis (2nd ed.). New York: Wiley. Richardson, A. J., & Wiegand, G. L. (1977). Distinguishing vegetation form soil background. Photogarmmetric Engineering and Remote Sensing, 43, 1541– 1552. Rigina, O., Baklanov, A., Hanger, O., & Olsson, H. (1999). Monitoring of forest damage in the Kola Penninsula, northern Russia, due to smelting industry. Journal

409 of Science of the Total Environment, 229(3), 147– 163. Rouse, J. W., Haas, R. H., Schell, J. A., & Deering, D. W. (1974). Monitoring vegetation systems in the Great Plains with ERTS. In Proceedings 3rd. ERTS symposium, Greenbelt: NASA SP-351, pp. 3010–317. Sellers, P. J. (1985). Canopy reflectance, photosynthesis and transpiration. International Journal of Remote Sensing, 6, 1335–1372. doi:10.1080/01431168508948283. Shafiei, B. (2000). The study of geochemical distribution, occurrence and metallogenic model of silver and gold in Sarcheshmen porphyry copper deposit. MSC thesis, university of Kerman, Iran. Shahabpour, J., & Kramers, J. D. (1987). Lead istope data from the Sarcheshmeh porphyry copper deposit, kerman, Iran. Mineralium Deposita, 22, 278–281. doi:10.1007/BF00204520. Shahidi, G. h., Yeganeh, M., Rezaiezadeh, M., Zahmatkesh, H., Jalali, F., Daei, H., et al. (2007). Natural vegetation cover assessment in Sarcheshmeh Copper Complex and adjacent areas. In Proceedings of 7th sym. on safety, health and environment in mines and related industries, Kerman, Iran. Shirashiani, D. (2004). Copper City, limitations, faculities and current status R&D center, Sarcheshmeh Copper Complex, Unpublished internal report (in Farsi). Singh, A. (1989). Digital change detection techniques using remotely-sensed data. International Journal of Remote Sensing, 6, 989–1003. doi:10.1080/01431168908903939. Song, C., Woodcock, C. E., Seto, K. C., Lenney, M. P., & Macomber, S. A. (2001). Classification and change detection using Landsat data: When and how to correct atmospheric effects. Remote Sensing of Environment, 75, 230–244. doi:10.1016/S0034-4257(00)00169-3. Tommervik, H., Johansen, M. E., Pedersen, J. P., & Guneriussen, T. (1998). Integration of remote sensed and in-situ data in an analysis of the air pollution effects on terrestrial ecosystems in the border areas between Norway and Russia. Environmental Monitoring and Assessment, 49, 51–85. doi:10.1023/A: 1005755706302. Tommervik, H., Hogda, K. A., & Solheim, I. (2003). Monitoring vegetation changes in Pasvik (Norway) and pechenga in Kola Peninsula (Russia) using multitemporal landsat MSS/TM data. Remote Sensing of Environment, 85, 370–388. doi:10.1016/S00344257(03)00014-2. Toutoubalina, O. V., & Ress, W. G. (1999). Remote sensing of industrial impact on Arctic vegetation around Noril’sk, northern Siberia: Preliminary results. International Journal of Remote Sensing, 20(15, 16), 2979– 2990. Tucker, C. J. (1979). Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing of Environment, 8, 127–150. doi:10.1016/ 0034-4257(79)90013-0. Tucker, C. J., Fung, I. Y., Keeling, C. D., & Gammon, R. H. (1986). Relationship between atmospheric CO2 variations and a satellite-derived vegetation index. Nature, 319, 195–199. doi:10.1038/319195a0.

410 Upton, G., & Cook, I. (2001). Introducing statistics (2nd ed.). Oxford: Oxford University Press. Vogelmann, J. E. (1990). Comparison between two vegetation indices for measuring different types of forest damage in the north-eastern united states. International Journal of Remote Sensing, 11, 2281–2297. doi:10.1080/01431169008955175. Walsh, S. J., Moody, A., Allen, T. R., & Brown, D. C. (1997). Scales dependence of NDVI and its relationship to mountainous terrain. In D. A. Quattrochi, &

Environ Monit Assess (2010) 163:397–410 M. F. Goodchild (Eds.), Scale in remote sensing and GIS (pp. 27–55). Boca Raton: Lewis Publishers. Waterman, G. C., & Hamilton, R. L. (1975). The Sarcheshmeh porphyry copper deposit. Economic Geology and the Bulletin of the Society of Economic Geologists, 70, 568–578. Woodcock, C. E., Macomber, S. A., & Kumar, H. (2002). Vegetation mapping and monitoring. In A. Skidmore (Ed.), Environmental modeling with GIS and remote sensing (pp. 97–120). UK: Taylor and Francis.