Forest losses and gains in Kurdistan province, western Iran: Where do we stand?

The Egyptian Journal of Remote Sensing and Space Sciences (2017) 20, 51–59

H O S T E D BY

National Authority for Remote Sensing and Space Sciences

The Egyptian Journal of Remote Sensing and Space Sciences www.elsevier.com/locate/ejrs www.sciencedirect.com

RESEARCH PAPER

Forest losses and gains in Kurdistan province, western Iran: Where do we stand? Maedeh Sadeghi a,*, Mansoureh Malekian a, Loghman Khodakarami b a b

Department of Natural Resources, Isfahan University of Technology, Isfahan, Iran Department of Petroleum Engineering, Koya University, Koya, Kurdistan Region, Iraq

Received 30 December 2015; revised 22 June 2016; accepted 12 July 2016 Available online 26 April 2017

KEYWORDS Forest mapping; Change detection; Zagros forest; Persian squirrel

Abstract Zagros forests in Kurdistan province of Iran are important habitats for many species including Persian squirrel. In the past few decades, human activities such as clear cutting and agriculture have reduced the quality and quantity of these forests. The current study used Landsat Thematic Mapper images from years 1987, 2000, 2010 and 2015 to map and monitor changes in the forest extent and density in the region. Radiometric, geometric and topographic corrections were applied to the satellite images. For supervised classification training signatures for different classes were established. Training sites were evaluated for possible discrimination of each class using transformed divergence method. The maximum likelihood algorithm was used for supervised classification and post-classification method was used to detect changes over a period of 28 years. The final map of land covers consisted of seven classes including barren land, agriculture, dense forest, semi-dense forest, mixed of sparse forest and rangeland, wetland plant, and water bodies. The results of change detection showed an increase in the amount of forests in this period of time, however, 3083.8 hectares of dense forests, which is one of the most important areas for wildlife such as the Persian squirrel, has reduced. This information can be used by natural resource managers and it is prerequisite for further ecological studies on wildlife species in the region. Ó 2016 National Authority for Remote Sensing and Space Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/ by-nc-nd/4.0/).

1. Introduction The protection of all things connected with nature has gained increasing attention because of the massive rate of decline and * Corresponding author at: Department of Natural Resources, Isfahan University of Technology, Isfahan 84156-83111, Iran. Fax: +98 33912840. E-mail address: [email protected] (M. Sadeghi). Peer review under responsibility of National Authority for Remote Sensing and Space Sciences.

loss of habitats and the natural environments over the past century. Forest degradation is a major issue worldwide, resulting in temporary or permanent deterioration of density and structure of vegetation which can affect species composition and diversity (Grainger, 2013). Monitoring changes in forest cover and canopy structure through the time is important for many applications, such as forest planning and management (Sironen et al., 2001; Zimble et al., 2003), climate change studies (Nuutinen and Kellomaki, 2001; Zimble et al., 2003; Matala, 2005), and wildlife conservation (Coops and Catling, 1997).

http://dx.doi.org/10.1016/j.ejrs.2016.07.001 1110-9823 Ó 2016 National Authority for Remote Sensing and Space Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

52

M. Sadeghi et al.

Traditionally, forest inventories using aerial photography interpretation and ground-based sampling have been used for forest mapping. These methods are rather time consuming and costly, especially in mountainous areas. Development of remote sensing technologies over the past few decades has provided new tools for advanced ecosystem management (Ayodeji Opeyemi, 2006) and a more automatic and efficient way for collecting data and measuring forests in terms of both cost and time (Yu, 2007). Since the 1970s, satellite data have been contributing to forest mapping (Peterson et al., 1987; Martin et al., 1998; Tokola et al., 2001; Ahmadi Sani et al., 2009; Broich et al., 2011). Several approaches such as post-characterization comparison of disturbance index (Masek et al., 2008), time-series and classification of forest cover loss analysis (Broich et al., 2011; Potapov et al., 2011), time-series of forest classifications using joint probabilities (Caccetta et al., 2007) and change vector thresholds (Xian et al., 2009; Xian and Homer, 2010) have been developed for forest change detection and monitoring. Among different satellite images used for forest mapping, Landsat Thematic Mapper (TM) imagery is generally used for mapping over larger areas (Dorren et al., 2003). Zagros forests with an area of about 6 million hectares (3.5 percent of Iran), are located in the west of the country. These forests have also been called western oak forests, due to the dominancy of oak species (Quercus spp.). Western oak forests are home to many species including, the Persian squirrel (Sciurus anomalus) which is the indicator species of this region (Ziaaee, 2010). Persian squirrels and oak trees have symbiotic relationships, in which forests provide ecological requirements of Persian squirrels such as food and shelter and, in return, the Persian squirrel contributes in seed germination and forests‘ regeneration (Ziaaee, 2010). Deforestation and poaching are the major threats to wildlife in the region (Yigit et al., 2012). The lack of accurate knowledge about these forest habitats also hinders the effective planning and

Figure 1

management of these forests and the wildlife within them. Preparing update maps and awareness of past to present changes would be useful for wildlife conservation planning. Change detection studies in parts of Zagros forests including Arasbaran (Ranjbar and Mesgari, 2002; Rasuly et al., 2010; Rezaee Moghadam et al., 2010), Bane (Amini et al., 2009), Ilam dam catchment (Shahkooeei et al., 2014), Saman forests, Chaharzebar forests (Khan Hasani et al., 2008), (Susani et al., 2010), Ilam province forests (Mahdavi, 2010) and Marivan forests (Yusefi et al., 2012), showed considerable degradation in these forests. However, considerable parts of Zagros including forests of Kurdistan province which are important for conservation of Persian squirrel are yet to be investigated. The current study aimed to map a large area of Zagros forests in Kurdistan province, detect changes over a period of 28 years and discuss the importance of forest change for wildlife conservation. This information can be used by natural resource managers to adopt conservation and restoration strategies. In addition, such studies are prerequisite for further ecological studies on wildlife species such as the Persian squirrel in the region. 2. Material and methods 2.1. Study area The study area is a mountainous forest located in the west of Kurdistan province, Iran (35°060 2500 N–36°130 2200 N and 45°330 2200 E–46°430 2700 E) (Fig. 1). The region covers more than 466,902 hectares of Zagros forests and is one of the most biologically diverse landscapes. The region is typically characterized by a semi-humid climate with extremely cold winters and annual precipitation of about 800 mm. Increasing populations, low level of awareness and high dependence of local communities on forests for their primary livelihood

Study area located in Kurdistan province, western Iran.

Forest losses and gains in Kurdistan province

53

needs, are the main reasons of forest destruction (Fattahi, 1996). Many parts of the region are now semi-degraded and coppiced (Jazirehi and Rostaghi, 2002).

2.5. Classification

In order to extract and quantify meaningful information from remotely sensed data, pre-prepossessing is required (Ghebrezgabher et al., 2016). Pre-processing on Landsat data includes radiometric calibration, radiometric normalization and geometric correction (AbdelRahman et al., 2016). These required pre-processing steps help to ensure that detected land cover changes correspond to changes of surface conditions and not artifact of atmospheric conditions, imaging and viewing conditions, sensor deficiency, or pixel misalignment (Schott et al., 1988; Furby and Campbell, 2001; Coops et al., 2006). All images were rectified to UTM zone 38, WGS84 using at least 25 well distributed ground control points and nearest neighbor re-sampling. The root mean square errors for each of the four images were less than 0.5 pixels. Radiometric normalization was then applied to account for differences in atmospheric conditions, solar angle, and satellite sensor characteristics. In order to account for radiometric distortions, an atmospheric correction was carried out for all images using ERDAS imagine 9.2 software. Each image was enhanced using linear contrast stretching and histogram equalization to improve the image and help identify ground control points. After corrections, the study area from the whole scene images was extracted using the ArcGIS 10.1.

The two common pixel-based classification methods, unsupervised and supervised classifications, were carried out on each of the images. Optimum Index Factor (OIF) method was used for selection of the best combination of ETM bands. As a result, False Color Composite (FCC) image using bands 2, 3 and 4 was used for classification. For supervised classification training signatures for different classes were established (20–30 training samples for each class) based on ground truth data collected during field surveys, spectrum features, topographic maps, Google earth images and FCC and NDVI images. For better and more accurate classification, every class was divided into minor subclasses, i.e., barren land (4 classes), agriculture (3 subclasses), dense forests (canopy density 50–80%, 2 subclasses), semi-dense forests (canopy density 25–50%, 1subclass), sparse forests (canopy density 5–25%, 1subclass), rangelands (1subclass), water (lake and ponds) and wetland plants. Training sites were evaluated for possible discrimination of each class using transformed divergence (TD) method available within ERDAS imagine 9.2 processing software. Transformed divergence calculates separability on a scale ranging between 0 and 2000. A minimum TD value of 1700 is considered good for separability (Bharti et al., 2012). A satisfactory spectral signature is the one that there is minimum confusion among the land covers to be mapped (Butt et al., 2015) and hence training samples with less than this value were removed. In average, a minimum separability above 1700 was achieved for all training samples of classes except water (lake and ponds) and wetland plants. TD of water and wetland plant classes was low and unacceptable. Also, unsupervised classification could not separate these two classes from other green areas such as forests. Therefore, these areas (water and wetland plants) were masked from images and classified separately using maximum likelihood algorithm. The TDs of rangelands and sparse forests were also low and algorithm was unable to separate these two classes. Therefore, these two classes merged together. After obtaining a satisfactory discrimination between classes, a final classification was conducted to produce the land cover map.

2.4. Topographic normalization

2.6. Accuracy assessment

Topographic normalization is generally used for correcting the so called slope and aspect effect, which causes misclassification and hinders multi-temporal analysis (Pellikkaa et al., 2005). Use of satellite data from mountainous areas generally requires additional pre-processing, including corrections for relief displacement and solar illumination differences (Meyer et al., 1993; Riano et al., 2003). The sun’s elevation and azimuth angles were obtained from the image header file. Illumination (IL) ranges between 1 (minimum) and +1 (maximum), and is computed as in Eq. (1).

When the supervised classification was completed, the precision of classified images was assessed, using control points. The control points were selected based on the topographic maps, Google earth images, FCC and during the field survey. Data from 310 ground survey sites were used in accuracy assessment. Each control point contains pixels of different land use. Each sample point was compared to the reference data to determine whether the Landsat classified change had actually performed. Overall accuracy, user’s and producer’s accuracies, and the Kappa statistic were then derived from the error matrices.

2.2. Data acquisition Four Landsat TM images were used from a proximate time (August 20, 1987; September 8, 2000, August 22, 2011 and August 17, 2015) with a general resolution of about 30 m. In addition, 1:25,000 topographic maps, road maps and the catchment maps were used for the accuracy assessment of the images and geometric correction. Field survey was conducted for selecting control points. ERDAS imagine 9.2 and ArcGIS 10.1 were used for image processing and forest change detection. 2.3. Geometric and radiometric correction

IL ¼ cos ci ¼ cos hp cos hz þ sin hp sin hz cosðua  uoÞ where: ci = incidence angle hp = slope angle hz = solar zenith angle = (90 – sun’s elevation angle) ua = solar azimuth angle uo = aspect angle (topographic-normalization).

ð1Þ

2.7. Change detection We used post-classification method for change detection to improve classification accuracy and reduce misclassifications (Butt et al., 2015). Post classification compares the classifications maps obtained independently from different images of

54

M. Sadeghi et al.

the same area acquired at different times (Jayanth et al., 2015). Based on these comparisons, we calculated the quantity of changes in land use types. 3. Results and discussion We aimed to detect forest changes in an interval of 28 years from 1987 to 2015; therefore, land cover was classified to separate forests from other land covers. Overall seven classes, including barren land, agriculture, dense forest, semi-dense forest, mixed of sparse forest and rangeland, wetland plant, and water bodies were obtained (Fig. 2 and 3). Maximum likelihood algorithm was unable to separate the wetland plants and water (lake and ponds) from other vegetation covers because of spectral interferences. Therefore, water and wetland plants were classified separately and joined to the final map. The forests were classified into three classes based on the canopy density. Overall classification accuracies of thematic maps derived from error matrices for 1987, 2000, 2011 and 2015 were 85.16%, 87.42%, 90.00% and 89.10% respectively (Table 1). Low producer’s accuracy of semi dense forest class (60.42%) for both the period of 1987 and 2000 may be due to spectral interference of this class with dense forest, while high producer’s accuracy of barren land, water, wetland plant classes (above 90%) indicates that almost all of its pixels have been classified successfully. Kappa coefficients for the above years were 0.81, 0.84, 0.87 and 0.87 respectively. Accuracy of semi dense forest and agriculture classes was low due to low image resolution and spectral interference of forests, rangelands and agriculture classes. Overall, Kappa coefficient, user and producer accuracy of agriculture, semi dense forests and mixed of sparse forests and rangelands classes were lower due to spectral interference between these classes and higher for barren land, water and wetland plants.

Figure 2

Figure 3 Land cover classification of Zagros forests in Kurdistan province for (A) 1978, (B) 2000, (C) 2011 and (D) 2015. Seven classes including barren land, dense forests, semi dense forests, mixed of sparse forests and rangelands, agriculture, wetland plants and water bodies are shown.

Landsat classification area statistics (ha) for 1978, 2000, 2011 and 2015.

90.00 100.00 – 85.16 0.81

90.00 90.00 – 0.90 0.90 – 90.00 100.00 – 87.42 0.84

90.00 90.00 – 0.90 0.90 – 90.00 100.00 – 90.00 0.87

90.00 90.00 – 0.90 0.90 – 95.71 96.67 100.00 0.95 0.96 1.00

89.10 0.87 Overall accuracy Total kappa

98.53 93.55 100.00

94.87 96.00 82.05 60.42 80.65 92.50 80.00 80.00 72.50 83.33 0.88 0.78 0.77 0.67 0.79 98.26 88.00 94.59 71.74 73.53 94.17 73.33 87.50 82.50 83.33 0.91 0.71 0.86 0.71 0.79 94.21 80.65 92.50 85.00 85.00 95.00 83.33 94.87 85.00 85.00 0.92 0.81 0.91 0.83 0.81 94.44 72.86 92.86 86.15 84.29 0.92 0.67 0.88 0.82 0.82

Barren land Agriculture Dense forests Semi dense forests Mixed of sparse forests and range lands Wetland plants Lake Water

88.54 91.07 85.53 84.85 81.94

Kappa Kappa User accuracy (%) Kappa Kappa Class name

User accuracy (%)

Producer accuracy (%)

2011 2015

Table 1

Year

Summary of Landsat classification accuracies for 1987, 2000, 2011 and 2015.

Producer accuracy (%)

2000

User accuracy (%)

Producer accuracy (%)

1987

User accuracy (%)

Producer accuracy (%)

Forest losses and gains in Kurdistan province

55 Our findings showed that from 1987 to 2000, the area of barren land, agriculture, dense forest, mixed of sparse forest and rangeland and wetland plant classes has reduced (Fig. 2). The water area had almost a fixed size and semidense forests have increased. From 2000 to 2011 the barren land, agriculture and dense forests have reduced in size while other land covers have increased (Fig. 2). At the first glance, the amount of dense forests has reduced but by a closer look to change detection results it is evident that a significant part of this part of Zagros forests has changed to semi dense forests. Based on the change detection results, the amount of semi-dense forests has increased by 20409.8 ha from 1987 to 2000 (Table 2a), while from 2000 to 2011 and from 2011 to 2015 the area of dense forests has reduced by 3356.7 ha and 11034.45 ha respectively and added to the semi-dense forest class (Table 2b and Fig. 4.). Also, between 2011 and 2015 the extent of semi dense forests has reduced and changed to barren land. During this period of time, 1907.01 ha of dense forests have changed to semi dense forest (Table 2f). Overall, results showed an increase in the extent of forests in the region over a 28-year period (from 1987 to 2015), however, that density of these forests has changed negatively. This highlights the importance of scale and selecting appropriate categories in change detection studies which can otherwise be misleading. A large amount of natural and dense forests of the region has converted to semi-dense forests or other classes which may not be favorable for wildlife species, especially, the Persian squirrel. Harvesting trees without replacing reduces the density of forest and its quality as species habitats. Plowing the forest soil, livestock feeding on leaves, fruits and young branches, collier, and forest fire are the main causes of forest destruction in this area. Plowing the soil under trees does not permit sprouting the trees, resulting in gradual death of Zagros forests. Degraded forests have lost much of their productivity and biodiversity as well as many of the ecological goods and services they once provided (Lamb and Gilmour, 2003). Although climate change and drought have affected the region, human intervention, deforestation and nonsystematic use of these forests have caused the degradation of Zagros forests. In specific, dense forests are important habitats for wildlife species, especially, the Persian squirrel. Squirrels prefer dense habitat where they can move between trees without descending to the ground (Summers and Proctor, 1999). Tall and large trees within dense canopy cover provide more food and protection from predator and, therefore, are preferred by squirrels as nesting sites (Edelman and Koprowski, 2005, 2007). Decreasing the quality of canopy cover might affect squirrel populations in the region. We collected the Persian squirrel presence localities in the region and produced a layer of presence locations. This layer was overlaid with the vegetation map, showing that Persian squirrels are present in dense and semi-dense forests. However, the species doesn’t present in all dense and semi-dense forests but in specific areas. Therefore, we merged the dense and semi-dense forest classes as the potential habitats of the species for further investigation of habitat selection by the Persian squirrel in the region (Sadeghi and Malekian, 2016). The most land cover change refers to mixed of sparse forest and rangeland class that has decreased 53827.2 ha from 1987

56 Table 2

M. Sadeghi et al. Transition area matrices between 1987–2000 (a), 2000–2011 (b), 1987–2011 (c), 1987–2015 (d) 2000–2015 (e) and 2011–2015 (f).

Class name

Barren land

Agriculture

Dense forests

Semi dense forests

Mixed of sparse forests and rangelands

Wetland plant

Lake

(a) Barren land Agriculture Dense forests Semi dense forests Mixed of sparse forests and rangelands Wetland plants Lake and ponds

224408.1 1687.68 1349.55 4597.2 46802.79 240.3 6.39

2395.26 2837.07 744.93 1108.8 13289.13 0 0

689.13 986.67 11649.6 3207.96 892.98 0 0

20121.57 1010.34 2214.99 28943.37 19544.94 1.08 0

9504 2429.37 1194.66 13568.76 43497.36 2.34 0

25.92 0 0 0 0 766.07 1.17

5.13 0 0 0 0 0.09 795.96

(b) Barren land Agriculture Dense forests Semi dense forests Mixed of sparse forests and rangelands Wetland plants Lake and ponds

225787.8 3086.01 251.01 10412.73 5188.86 34.2 8.28

7069.14 6204.69 268.02 1607.67 4030.47 0 0

1511.28 373.41 10811.25 589.41 785.25 0 0

21502.89 2163.69 6074.55 42948.36 27427.32 0.09 0

22861.8 5847.39 21.51 16277.4 32762.61 00 0

319.41 0 0 0.72 1.98 754.38 2.34

39.69 0 0 0 0 14.49 790.56

(c) Barren land Agriculture Dense forests Semi dense forests Mixed of sparse forests and rangelands Wetland plants Lake and ponds

204164.2 1301.28 564.21 3529.62 35113.32 86.4 8.82

3875.49 2896.29 843.03 1625.79 99.38.34 1.08 0

1120.05 866.97 9349.11 2146.77 587.43 0.27 0

31101.03 2566.35 6119.82 33070.49 25277.67 0.54 0

16693.29 1317.15 277.38 11052.27 51130.44 0.18 0

59.12 0.09 0.18 0.18 0 916.92 2.34

37.89 0 0 0 0 14.49 792.36

Class name

(d) Barren land Agriculture Dense forests Semi dense forests Mixed of sparse forests and rangelands Wetland plants Lake and ponds (e) Barren land Agriculture Dense forests Semi dense forests Mixed of sparse forests and rangelands Wetland plants Lake and ponds (f) Barren land Agriculture Dense forests Semi dense forests Mixed of sparse forests and rangelands Wetland plants Lake and ponds

Barren land

Agriculture

Dense forests

Semi dense forests

Mixed of sparse forests and rangelands

Wetland plant

Lake

Water

176522.13 1589.22 1108.71 8921.43 28266.3

1572.84 534.15 210.42 702.63 1622.52

2034.9 1440.81 10482.57 5344.74 2018.34

40303.98 4591.26 5116.95 28275.12 48010.86

33675.75 669.33 173.25 8040.33 43434.99

207.99 39.96 9.63 3.51 5.94

35.1 0 0 0 0

634.95 41.04 16.38 34.47 168.39

9 0

2.25 0

4.86 0

12.15 0

0.99 0

943.38 8.1

43.56 795.15

0 0

187547.85 2320.2 680.94 17905.95 7961.58

1247.4 488.34 157.05 1022.49 1729.53

2201.31 814.05 12308.31 3403.35 2597.58

47418.84 10595.25 4205.07 31220.19 32870.7

37244.16 6008.49 30.69 18007.92 24703.38

390.78 14.31 3.42 17.46 29.43

36 0 0 0 0

724.68 36.99 5.94 61.47 66.15

0.27 0

0 0

1.62 0

0.27 0

0 0

755.19 7.92

44.64 793.17

0 -

181269.45 2538.72 953.91 21001.86 10648.17

501.48 577.89 99.81 3337.11 122.22

468.18 1031.58 11034.45 8726.22 49.95

37197.81 11453.04 1907.01 40254.66 35487.9

22316.13 3467.7 11.7 26383.59 33815.16

129.24 15.12 10.53 34.92 3.15

46.08 0 0 0 0

605.88 38.52 0 125.82 90.81

4.5 0

6.3 0

15.84 0

9.9 0

0.36 0

1007.37 18.18

27 800.73

0 25.65

to 2000 (Table 2a) and increased 10280.9 ha between 2000 and 2011 (Table 2b) because some areas of barren lands or forest have changed to rangelands in this period. Other studies in

Zagros showed an increase in the amount of forests (Arekhi, 2013; Shahkooeei et al., 2014) because they generally considered all forest areas with different canopy density as one class.

Forest losses and gains in Kurdistan province

Figure 4

57

Example of some pixels of dense forests in 1987 (A) changed into semi dense forest in 2011 (B).

In the current study, however, by investigating the forest density as three subclasses this misconception was removed. Although the total forest extent shows apparently an increase, our findings showed that a large extent of natural and dense forests in Kurdistan region has reduced from 1987 to 2015. A reduction in the forest and rangeland extent was evident in some studies in the region (Karami, 2004; Khan Hasani et al., 2008; Amini et al., 2009; Rezaee Moghadam et al., 2010; Susani et al., 2010). The discrepancy of the findings might be the result of scale difference, different images used, place or the timing of the studies. In addition, in most of these studies the canopy density didn’t take into account. Forest degradation can be prevented through correct management and general public awareness. Lack of information in many parts of Zagros forests also hampers the appropriate management. There is no harvesting plan for Zagros forest, because the native woody species are not commercially harvestable and the region is ecologically sensitive and prone to soil erosion. These forests have potential for non-woody forest product extraction such as natural gum, herbs and oak galls. Local communities, however, use forests to collect firewood and livestock feeding. Over a three month grazing period, the number of livestock is several times the pasture capacity. Low pasture capacity results in the incursion of a large number of the livestock into the forests. Forest ground forage constitutes a negligible proportion of the total feedstock and the livestock generally feed on leaves, fruits and young branches. This has led to gradual death of the wildlife due to the food supply restrictions. Non-irrigated farming is another major ecosystem degradation factor. Thousands of hectares dry-land farms under the Zagros forests are main centers of soil erosion. Presently, only 7 percent of Zagros forests have seed origin and the majority of them (93%) are coppice. These trees have

same age and height less than 5 meters and may not provide suitable habitats for the Persian squirrel. To conserve wildlife in the region it is vital to protect their habitats. Consequently suitable habitats should be declared protected areas. In the southern parts of the study area two protection areas exist, however, none has been established to protect Persian squirrels against illegal hunting. There is also Zarivar lake wildlife sanctuary, which has established for waterfowl protection. Establishing protected areas in the region could be through educational efforts and awareness among locals. The current study provided a foundation for further research in the region, illustrating the importance of satellite imagery, with up to 90 percent accuracy. To obtain more precise environmental information, an object-based image analysis by evaluating high resolution images is recommended which can be applied for monitoring of exact changes over the time and space. Conflict of interest There is no conflict of interest. References AbdelRahman, M.A., Natarajan, A., Hegde, R., 2016. Assessment of land suitability and capability by integrating remote sensing and GIS for agriculture in Chamarajanagar district, Karnataka, India. EJRS 19 (1), 125–141. Ahmadi Sani, N., Darvish Sefat, A.A., Zobeyri, M., Farzane, A., 2009. Ability of ASTER images for mapping of Zagros forests density (case study: Marivan forests). J. Nat. Res. 61 (3), 603–614.

58 Amini, M.R., Shetaee, S., Ghazanfari, H., Moayeri, M.H., 2009. Change detection of Zagros forests extent by using airborne photo satellite images (case study: Bane forests). JASNR 15 (2), 10–20. Arekhi, M., 2013. Modeling spatial pattern of deforestation using GIS and logistic regression: a case study of northern Ilam forests, Ilam Province, Iran. AJB 10 (72), 16236–16249. Ayodeji Opeyemi, Z., 2006. Change detection in land use and land cover using remote sensing data and GIS (A case study of Ilorin and its environs Kwara state) (M.Sc. thesis). Department of Geography, Ibadan University, Nigeria. Bharti, R.R., Bhupendra, S.A., Rawat, G.S., 2012. Assessing vegetation changes in timberline ecotone of Nanda Devi National Park, Uttarakhand. Int. J. Appl. Earth Obs. Geoinf. 18, 472–479. Broich, M., Hansen, M.C., Potapov, P., Adusei, B., Lindquist, E., Stehman, S.V., 2011. Time-series analysis of multi-resolution optical imagery for quantifying forest cover loss in Sumatra and Kalimantan, Indonesia. Int. J Appl. Earth. Obs. Geoinf. 13 (2), 277–291. Butt, A., Shabbir, R., Ahmad, S.S., Aziz, N., 2015. Land use change mapping and analysis using Remote Sensing and GIS: a case study of Simly watershed, Islamabad, Pakistan. Int. J Appl. Earth. Obs. Geoinf. 18 (2), 251–259. Caccetta, P., Furby, S., O’Connell, J., Wallace, J., Wu, X., 2007. Continental monitoring: 34 years of land cover change using Landsat imagery. In: 32nd International symposium on Remote Sensing of Environment, June 25–29, 2007, Costa Rica. Coops, N.C., Catling, P.C., 1997. Predicting the complexity of habitat in forests from airborne videography for wildlife management. Int. J. Remote Sens. 18, 2677–2686. Coops, N.C., Wulder, M.A., White, J.C., 2006. Identifying and Describing Forest Disturbance and Spatial Pattern: Data Selection Issues and Methodological Implications, Florida, USA. Dorren, L.K.A., Maierb, B., Seijmonsbergena, A.C., 2003. Improved Landsat-based forest mapping in steep mountainous terrain using object-based classification. For. Ecol. Manage. 183, 31–46. Edelman, A.J., Koprowski, J.L., 2005. Selection of drey sites by Abert’s squirrels in an introduced population. J. Mammal 86, 1220–1226. Edelman, A.J., Koprowski, J.L., 2007. Communal nesting in asocial Abert’s squirrels: the role of social thermoregulation and breeding strategy. Ethology 113, 147–154. Fattahi, M., 1996. Investigation on the Zagros Quercus Forests and the Important Deforestation Parameters. Iranian Forest and Rangelands Research Institute Press, Tehran, Iran. Furby, S.L., Campbell, N.A., 2001. Calibrating images from different dates to ’like value’ digital counts. Remote Sens. Environ. 77, 186– 196. Ghebrezgabher, M.G., Yang, T., Yang, X., Wang, X., Khan, M., 2016. Extracting and analyzing forest and woodland cover change in Eritrea based on landsat data using supervised classification. EJRS 19 (1), 37–47. Grainger, A., 2013. Controlling Tropical Deforestation. Earthscan Pubns Ltd, London, UK. Jayanth, J., Kumar, T.A., Koliwad, S., Krishnashastry, S., 2015. Identification of land cover changes in the coastal area of Dakshina Kannada district, South India during the year 2004–2008. EJRS 19 (3), 73–93. Jazirehi, M., Rostaghi, E., 2002. Silviculture in Zagros. University of Tehran Press, Tehran, Iran. Karami, F., 2004. Application of Landsat Satellite Data for Studying Zagros Forest Destruction (M.Sc. thesis). Natural Resources College, Mazandaran University, Iran. Khan Hasani, M., Sharifi, M., Tavakoli, A., 2008. Using aerial photos in estimation of destruction rate of Chahrzobor forests in Kermanshah province. IJFPR 15 (2), 105–114. Lamb, D., Gilmour, D., 2003. Rehabilitation and restoration of degraded forests. IUCN, Gland, Switzerland and Cambridge, UK and WWF, Gland.

M. Sadeghi et al. Mahdavi, A., 2010. IRS-1C image data applications for land use/land cover mapping in Zagros region, case study: Ilam watershed, West of Iran. CJES 8 (1), 35–41. Martin, M.E., Newman, S.D., Aber, J.D., Congalton, R.G., 1998. Determining forest species using high spectral resolution remote sensing data. Remote Sens. Environ. 65, 249–254. Masek, J.G., Huang, C., Wolfe, R., Cohen, W., Hall, F., Kutler, J., Nelson, P., 2008. North American forest disturbance mapped from a decadal Landsat record. Remote Sens. Environ. 112 (6), 2914– 2926. Matala, J., 2005. Impacts of climate change on forest growth: a modelling approach with application to management Ph.D thesis. Faculty of Forestry. University of Joensuu, Finland. Meyer, P., Itten, K.I., Kellenberger, T., Sandmeier, S., Sandmeier, R., 1993. Radiometric corrections of topographically induced effects on Landsat TM data in an alpine environment. ISPRS 48 (4), 17– 28. Nuutinen, T., Kellomaki, S., 2001. A comparison of three modelling approaches for large-scale forest scenario analysis in Finland. Silva Fenn. 35 (3), 299–308. Pellikkaa, P.K.E., Clarka, B.J.F., Sirvioa, T., Masalina, K., 2005. Environmental change monitoring applying satellite and airborne remote sensing data in the Taita Hills, Kenya. In: 1st International Conference on Remote Sensing and Geoinformation Processing in the Assessment and Monitoring of Land Degradation and Desertification, Trier, Germany. Peterson, D.L., Spanner, M.A., Running, S.W., Teuber, K.B., 1987. Relationship of thematic mapper simulator data to leaf area index of temperate coniferous forest. Remote Sens. Environ. 22, 322–341. Potapov, P., Turubanova, S., Hansen, M.C., 2011. Regional-scale boreal forest cover and change mapping using Landsat data composites for European Russia. Remote Sens. Environ. 115 (2), 548–561. Ranjbar, A., Mesgari, M.S., 2002. Investigation and Estimating Forest destruction using RS & GIS M.Sc. thesis of remote sensing. Khajeh Nasir Toosi University, Tehran, Iran. Rasuly, A.A., Naghdifar, R., Rasuly, M., 2010. Detecting of Arasbaran forest changes applying image processing procedures and GIS techniques. Procedia. Environ. Sci. 2, 454–464. Rezaee Moghadam, M., Rezaee Banafshe, M., Feyzi Zade, B., Nazm Far, H., 2010. Classification of land use/cover based on objectoriented techniques and satellite images (case study: Azarbayjan-egharbi province). JWMR 87, 20–35. Riano, D.E., Chuvieco, J., Salas, I., Aguado, I., 2003. Assessment of different topographic corrections in Landsat-TM data for mapping vegetation types. IEEE TGRS 41 (5), 1056–1061. Sadeghi, M., Malekian, M., 2016. The Persian squirrel of Kurdistan Province, western Iran: what determines its geographic distribution? Mammalia. http://dx.doi.org/10.1515/mammalia-2015-0166 (in press). Schott, J.R., Salvaggio, C., Volchok, W.J., 1988. Radiometric scene normalization using pseudoinvariant features. Remote Sens. Environ. 26, 1–16. Shahkooeei, E., Arekhi, S., Najafi Kani, A., 2014. Remote sensing and GIS for mapping and monitoring land cover and land use changes using support vector machine algorithm (case study: Ilam dam watershed). Res. J. Appl. Basic. Sci. 8 (4), 464–473. Sironen, S., Kangas, A., Maltoma, M., Kangas, J., 2001. Estimating individual tree growth with the k-nearest neighbour and k-most similar neighbour methods. Silva Fenn. 35 (4), 453–467. Summers, R.W., Proctor, R., 1999. Tree and cone selection by Crossbills Loxia sp. and red squirrels Sciurus vulgaris at Abernethy Forest, Strathspey. For. Ecol. Manage. 118, 173–182. Susani, J., Zobeyri, M., Feghhi, J., 2010. application of aerial photos and satelitte images in viewing Zagros forest cover (case study: forests of Kaka Reza area of Lorestan province). IJFPR 17 (2), 237–249.

Forest losses and gains in Kurdistan province Tokola, T., Sarkeala, J., Linden, M.V.D., 2001. Use of topographic correction in Landsat TM-based forest interpretation in Nepal. Int. J. Remote Sens. 22 (4), 551–563. Xian, G., Homer, C., 2010. Updating the 2001 national land cover database impervious surface products to 2006 using Landsat imagery change detection methods. Remote Sens. Environ. 114 (8), 1676–1686. Xian, G., Homer, C., Fry, J., 2009. Updating the 2001 national land cover database land cover classification to 2006 by using Landsat imagery change detection methods. Remote Sens. Environ. 113 (6), 1133–1147. Yigit, N., Krystufek, B., Sozen, M., Bukhnikashvili, A., Shenbrot, G., 2012. Sciurus anomalus. In: IUCN 2008: IUCN Red List of Threatened Species. Available at: .

59 Yu, X., 2007. Methods and techniques for forest change detection and growth estimation using airborne laser scanning data (Ph.D. thesis). Helsinki University of Technology, Finland. Yusefi, S., Moradi, H.R., Hoseini, S.H., Mirzaee, S., 2012. Monitoring of land use change of Marivan by using TM and ETM+ sensor of Landsat satellite. J. Appl. RS and GIS Tech. Nat. Resour. 3, 97– 105. Ziaaee, H., 2010. A field Guide to the Mammals of Iran. Iranian Wildlife Center, Tehran, Iran. Zimble, D.A., Evans, D.L., Carlson, G.C., Parker, R.C., Grado, S.C., Gerard, P.D., 2003. Characterizing vertical forest structure using small-footprint airborne LiDAR. Remote Sens. Environ. 87 (2–3), 171–182.