Spatial and temporal dynamics of Pai forest vegetation in

Preparing to load PDF file. please wait...

0 of 0
Spatial and temporal dynamics of Pai forest vegetation in

Transcript Of Spatial and temporal dynamics of Pai forest vegetation in

J. For. Res. DOI 10.1007/s11676-016-0327-x
Spatial and temporal dynamics of Pai forest vegetation in Pakistan assessed by RS and GIS
A. A. Siyal1,2 • A. G. Siyal2 • R. B. Mahar1

Received: 18 September 2014 / Accepted: 6 February 2016 Ó Northeast Forestry University and Springer-Verlag Berlin Heidelberg 2016

Abstract Pai, an arid forest in Sindh Province of Pakistan, is important for the environmental, social, economic development and conservation of ecosystems of the province. Considering the significance of the forest for Sindh and the calls from the local population for its deforestation, we quantified the spatial and temporal variation in the vegetation of the forest and land surface temperature (LST) using optical and thermal Landsat satellite data. Our analysis of temporal (1987–2014) images with ArcGIS 10.1 revealed that the dense forest area was greatest at 725 ha (37 % of the total forest area) during 2013 while it was smallest at 217 ha (11 %) in 1992. The sparse forest area peaked during 1987 at 1115 ha (58 %) under shrubs whereas it was smallest at 840 ha (43 %) in 1992, and the maximum deforestation of Pai forest occurred during 1992. Spatial change in vegetation over a period of about 27 years (1987–2014) revealed that vegetation increased on an area of 735 ha (37 %), decreased on 427 ha (22 %), and there was no change on 808 ha (41 %) of the forest. Variation in temperature between shaded (dense forest) and unshaded areas (bare land) of the forest was from 6 to 10 °C. While the temperature difference between areas with sparse forest and bare land ranged from 4 to 6 °C. An
The online version is available at
Corresponding editor: Yu Lei.
& A. A. Siyal [email protected]
1 U.S.-Pakistan Center for Advanced Studies in Water (USPCAS-W), Mehran University of Engineering and Technology, Jamshoro, Pakistan
2 Department of Land and Water Management, Sindh Agriculture University, Tandojam, Pakistan

inverse relationship between LST and NDVI of Pai forest with coefficients of determination of 0.944 and 0.917 was observed when NDVI was plotted against minimum and maximum LST, respectively. The vegetation in the forest increased with time and the areas of more dense Pai forest supported lower surface temperature and thus air temperature.
Keywords Landsat Á Arid forest Á NDVI Á Deforestation Á Ecosystem Á Land surface temperature (LST)
Forests not only provide livelihood to about one billion people (UNEP 2011; Sunderland et al. 2013) but also provide employment directly or indirectly to more than one hundred million people around the globe (UNEP 2011). Forests protect environment through soil conservation, watershed management, and provide protection against floods and landslides. Forests have an important impact on local as well as global environment and climate change (Xiao et al. 2004). Temporal changes in forest land cover can result in a wide variety of ecological impacts, including changes in productivity, composition, nutrient dynamics, species diversity, and increased atmospheric carbon dioxide (Braswell et al. 2003).
Measuring, mapping and classifying forests is important for monitoring spatial and temporal variation in vegetation and for planning and implementation of forest restoration programs (Rikimaru 1999). Traditional methods of mapping and categorizing forests through field surveys, literature reviews, map interpretation and secondary data analysis are time consuming, laborious, and uneconomical. Therefore, Remote Sensing and GIS tools are often used


A. A. Siyal et al.

for mapping, monitoring and analyzing forest vegetation as these tools are economical and rapid (Kumar 2010). Acquiring high spatial resolution satellite imagery is often expensive and cost-prohibitive. In contrast, Google EarthÒ (GE) is a free source of high resolution satellite imagery (Madin et al. 2011; Hughes et al. 2011; Pringle 2010) but its applications are limited by inadequate temporal availability of imagery and low spectral resolution and number of spectral bands (Potere 2008; McInnes et al. 2011). Hence, Landsat imagery is widely used for monitoring global surface change as it is a free multispectral and medium spatial resolution resource of imagery (Goward et al. 2006; Masek et al. 2008; Wulder et al. 2008; Xie et al. 2008; Bhandari et al. 2012). The diurnal land surface temperature (LST) derived from thermal infrared (TIR) bands is reported to be closely related to the density of the forest canopy (Nemani and Running 1997). Pongratz et al. (2006) confirmed that increase in LST occurs in areas with lower forest cover compared with dense forest. Yang et al. (2007) observed a good agreement with coefficient of determination of R2 = 0.80 and RMSE = 0.7 °C between predicted and observed air-surface temperature profile in mixed forest. Wickham et al. (2012) reported that forest surface temperatures were cooler than cropland surface temperatures. Thus, day LST can be used to detect changes in the forest cover (van Leeuwen et al. 2011).
Pai forest, one of the major forests of Sindh, Pakistan is reported to be endangered due to natural factors such as chronic water shortage, soil salinization, threats from land grabbers and lumber mafia, and deforestation due to cutting of trees by nearby villagers (WWF 2008). It is claimed by local people that trees are cut from an area of about 200 ha and the land is then brought under cultivation for crops. These threats are countered by calls from civil society to prevent deforestation and conserve the forest. There is a need to study the temporal and spatial change in the vegetation of the forest in order to analyze threats to the forest as claimed by people. Our study was thus conducted using temporal Landsat optical data to quantify the spatial and temporal variation in the vegetation of the forest. Spatial variation in LST during winter (January), spring (March), summer (June) and autumn (October) was determined from Landsat TIR data for the year 2013–2014 only.
Materials and methods
Study area
Pakistan has about 4.57 million hectares of forests, which is equivalent to 5.2 % of the total geographical area (GoP 2003). This is low compared to the global average of 30 % (FAO 2001) but the climate of Pakistan is predominantly

arid and largely unsuited to forest development. The total riverine area in Sindh, Pakistan is 0.855 million ha which is confined between flood protective embankments on both sides of the main stream of the river Indus (Panhwar 2002). The riverine forests in Sindh are mostly on rich alluvial soils that support trees such as Prosopis juliflora (Babol), Acacia nilotica (Kikar), Populus euphratica (Bahan), Tamarix aphylla and Tamarix dioica (Lai), Dalbergia sissoo (Shesham), Prosopis cineraria (Kandi), and Azadirachta indica (Neem). These forests are the major sources of timber, firewood and fodder for livestock (Mughal 2010).
Pai forest is considered to be one of the important forest of Sindh province and covers an area of 1977 ha. It is surrounded by vast fertile cultivated lands (Qureshi and Bhatti 2010; Faruqi 2011). WWF (2008) reported that during 2008, about 1502 ha (76 %) were under tree cover while the remaining 468 ha (23.7 %) were either cultivated or were bare land with salty patches. The forest is located in arid area with mean annual rainfall of about 150 mm and maximum surface evaporation rates of about 9–10 mm day-1 during May and June.
The forest is situated in Shaheed Benazirabad, a central district of Sindh, near Sakrand town on the left bank of the river Indus between longitudes 68°1202000 and 68°1700400E and latitudes 26°0405000 and 26°0704000N at 30 m above mean sea level (Fig. 1). The Pai forest in true color, pseudo natural color, and false color is shown in Fig. 2. The study area is mainly covered by vegetation (Fig. 2c). Areas covered by isolated bushes are shown in purple color in Fig. 2b, agricultural land is shown in magenta color in Fig. 2b, and bare land is shown in white color in Fig. 2a.
Pai forest was a riverine forest until its source of irrigation was cut off from the river Indus due to construction of flood protection bunds along the Indus river banks during the era of British rule. It is now a canal irrigated forest, being irrigated by the Rahib Shah distributary, a branch of the Rohri Canal and by groundwater through tube-wells installed for this purpose. The forest provides habitat for flora and fauna including the indigenous Mesquite (P. juliflora), Arabic gumtree (A. nilotica), Spunge tree (P. cineraria), Indian rosewood or Bombay Blackwood (D. sissoo) and Hog deer, Partridges, Asiatic jackals, Jungle cat, Porcupine, reptiles. Qureshi and Bhatti (2010) reported total of 93 plant species in the forest. The forest also provides livelihood to 21 villages located on the periphery of the forest. They are dependent upon the natural products of the forest to meet their daily requirement of food, fuel wood and earnings.
Satellite imagery
Temporal and spatial variation in vegetative cover of Pai forest was quantified using Landsat imagery. Landsat


Spatial and temporal dynamics of Pai forest vegetation in Pakistan assessed by RS and GIS
Fig. 1 Location of Pai forest in Sind, Pakistan

imagery of WRS-2 path 152, row 42, datum WGS84, processing L1T from November, 1987 to November, 2014 was downloaded from the United States Geological Survey portal (GloVis) website. Details of the downloaded imagery are listed in Table 1. Only two images of the site were available from the GloVis archive for the year 1992, hence, for that year we used the image made in April.
Image processing
Extraction of Pai forest and image classification
The acquired Landsat imagery was processed in ArcGIS 10.1. The image of only Pai forest was extracted from the entire scene using ‘Extract by Mask’ tool, by adding the shapefile of the forest as a mask. The extracted images of Pai forest of all acquired imagery were classified using a maximum likelihood algorithm. For this, the training samples were produced on the raster image of the entire area of Pai forest using Training Sample Manager. Training samples were created by drawing polygons at different locations in the forest area and then all samples were merged to a single class. The area was thus trained for three classes viz. dense forest (dense trees), sparse forest (scrub and stunted bushes) and bare land. From the training area and classes, a signature file was prepared which was then used as an input file during running of the maximum likelihood algorithm. The area under each class was

computed by converting the raster data into polygons and then summing the area of all polygons of the same class.

Temporal and spatial variations in crop cover and NDVI

The temporal and spatial vegetation change in forest between November 1987 and November 2014 was determined by comparing the near infrared (NIR) bands (band 4 of Landsat 5 and band 5 of Landsat 8) of both years and running the image difference algorithm in the Image Analysis Window of ArcGIS 10.1.
Normalized difference vegetation index (NDVI) is usually used to estimate vegetation biomass, greenness and production of the vegetative cover (Kawamura et al. 2005; Telesca et al. 2006). It is generally calculated from the visible red and near-infrared light reflected by vegetation. The NDVI of a given pixel always results in a number that ranges from -1 to ?1. A value between -1 and 0 shows a water body in the area, while NDVI between 0 and 0.2 represents bare land and 0.2 and 1.0 indicates the vegetative cover. The NDVI of all the extracted images of Pai forest was calculated using following relation (Rouse et al. 1974):

NDVI ¼ qnir À qred


qnir þ qred

where, qnir is the near infrared band, and qred is the red band.


A. A. Siyal et al.

Table 1 Landsat optical images used in the study with years of acquisition of the data, row, path, the sensors, and day of the year (DOY)

Year Acquisition date Satellite

Sensor ID


1987 1992 2000 2010 2013 2014

November 08 April 27 November 19 November 07 November 15 November 18

Landsat 5 TM


Landsat 5 TM


Landsat 7 ETM


Landsat 5 TM


Landsat 8 OLI and TIRS 319

Landsat 8 OLI and TIRS 322

can be derived using Plank’s law (Dash et al. 2002). Temporal and spatial variation in temperature of Pai forest was also monitored throughout the year i.e. winter, spring, summer and autumn. Landsat 8 imagery (Table 2) was also processed for this purpose. A newly launched Landsat 8 satellite is equipped with TIRS, a new instrument for capturing surface temperature information. Two electromagnetic spectral bands are delivered by this sensor with 100 m resolution to estimate the temperature at the Earth’s surface. To calculate the actual surface temperature of the area under study, digital numbers (DNs) of thermal bands (band 10 or band 11) were first converted into top-of-theatmosphere (ToA) radiance values and then ToA radiance values to ToA brightness temperature in Kelvin and then to centigrade using Eqs. 2 and 3.

L ¼ ðML Â QcalÞ þ AL


T ¼  K2  À 273:0


ln KL1 þ 1

where, Lk TOA spectral radiance, ML Band-specific multiplicative rescaling factor, AL Band-specific additive rescaling factor, Qcal Quantized and calibrated standard product pixel values (DN), T ToA brightness temperature (°C), K1 Band-specific thermal conversion constant, and K2 Band-specific thermal conversion constant.
The input parameters of Eqs. 2 and 3 (ML, AL, K1 and K2) for band 10 and band 11 of Landsat 8 obtained from metafile of images are given in Table 3.

Fig. 2 Pai forest in true (a), pseudo natural (b) and false colors (c). The band combinations in RGB were a B4, B3, B2 b B6, B5, B4 and c B5, B4, B3
Land surface temperature (LST)
Satellite thermal infrared sensor (TIR) is usually used to quantify the top of the atmosphere (TOA) radiances, from which brightness temperatures (blackbody temperatures)

Table 2 Landsat OLI and TIRS images used for determination of temporal and spatial variation in temperature


Acquisition date



2013 2014 2014 2014

October 13 January 18 March 07 June 11










Spatial and temporal dynamics of Pai forest vegetation in Pakistan assessed by RS and GIS

Table 3 Landsat 8-TIRS constants for bands 10 and 11





Band 10




Band 11




1321.08 1201.14

By extracting the LST values from randomly selected points of ToA brightness temperature raster image, the spatial variation of LST under different land use classes (bare land, sparse forest and dense forest) was determined.
A study conducted by Li et al. (2008) to investigate the relationship between LST and air temperature revealed that

Fig. 3 Landsat images (pseudo natural color) and classified images of Pai forest for the years of 1987, 1992, 2000, 2010, 2013 and 2014

Table 4 Temporal variation in vegetative cover of Pai forest from 1987 to 2014


Dense forest (ha)

Sparse forest (ha)

Forest total (ha)

Bare land (ha)













1115 840
1010 1042
894 948













Total (ha)
1977 1977 1977 1977 1977 1977

A. A. Siyal et al.
Vegetative cover ratio (%) 78.5 53.46 65.25 77.08 81.89 82.8

Fig. 4 Temporal variation (%) in vegetative cover of Pai forest
differences are relatively small in areas of high temperature but are larger in low temperature areas. They also reported moderate to high correlations ranging from 0.73 to 0.79 between air temperature and LST. For the present study difference between air temperature and LST was assumed very small as Pai forest is located in hot and dry climate region.

Results and discussion
Temporal and spatial variations in vegetative cover
Figure 3 shows the Landsat images in pseudo natural color and classified images of Pai forest for the years 1987, 1992, 2000, 2010, 2013 and 2014. All the Landsat images of Pai forest were classified for dense forest (pixels represent only trees), sparse forest (mixed pixels representing both sparse shrubs and bare land) and bare land (pixels represent only bare land). Variation in vegetative cover of Pai forest from 1987 to 2014 showed that about 1552 ha (79 %) of forest were covered with vegetation (trees and shrubs) in 1987. This declined to 1057 ha (53 %) in 1992, a 26 % reduction in vegetative cover in 5 years (Table 4). This might have been due to illegal harvest of trees or drying of trees due to shortage of irrigation water during that period as reported

Fig. 5 Spatial change in vegetation of Pai forest occurred in 27 years (a) and pie graph of variation in vegetation (% area) of forest in 27 years (b)
by Siddiqui (2009). We were unable to confirm the cause of the decrease in vegetative area in 1992 through consultation with the Forest Department Sindh. Another reason for reduction in vegetation in 1992 might have been bias of the satellite image as it was captured during April while the rest of images were captured during winter. Our analysis of an image captured during June 1992 revealed that there


Spatial and temporal dynamics of Pai forest vegetation in Pakistan assessed by RS and GIS

Fig. 6 Temporal variation in NDVI of Pai forest from 1987 to 2014

was a non-significant difference in vegetative area represented by both images (April and June). Thus, we conclude that the decrease in vegetation from 1987 to 1992 was not due to the influence of the difference in season.
After 1992, the vegetated area increased to 1290 ha in 2000 (65 %), 1524 ha in 2010 (77 %), 1619 ha in 2013 (82 %), and 1637 ha in 2014 (83 %). Vegetation gradually

increased on about 467 ha (24 %) over a period of 18 years (1992–2010) with an average annual increase in area of 26 ha (Table 4). While from 2010 to 2014, vegetation increased on an area of about 113 ha (6 %) with an average annual increase of 28 ha. Rapid increase in vegetation from 2010 to 2014 might be due to favorable soil moisture conditions in forest due to above average rainfall and heavy


Table 5 Spatio-temporal


distribution of vegetation cover

of Pai forest, Sindh, Pakistan

Nov. 08, 1987

Apr. 27, 1992

Nov. 19, 2000

Nov. 07, 2010

Nov. 15, 2013

Nov. 18, 2014

A. A. Siyal et al.

NDVI Area (ha) Area (%) NDVI Area (ha) Area (%) NDVI Area (ha) Area (%) NDVI Area (ha) Area (%) NDVI Area (ha) Area (%) NDVI Area (ha) Area (%)

Vegetative cover Dense forest Sparse forest

0.35–0.58 437 22.2 0.35–0.46 217 11.0 0.35–0.46 280 14.1 0.35–0.41 482 24.3 0.35–0.44 725 36.6 0.35–0.43 698 34.8

0.20–0.35 1115 56.3 0.20–0.35 840 42.5 0.20–0.35 1010 51.1 0.20–0.35 1042 52.7 0.2–0.35 894 45.2 0.2–0.35 948 48.0

Bare land and water body
-0.03–0.20 425 21.5 -0.03–0.20 920 46.5 0.06–0.20 687 34.8 -0.04–0.2 453 23.0 0.01–0.2 358 18.2 -0.01–0.2 340 17.2

-0.03–0.58 1977 100 -0.03–0.46 1977 100 0.06–0.46 1977 100 -0.04–0.41 1977 100 0.01–0.44 1977 100 -0.01–0.43 1977 100

Fig. 7 Spatial and temporal variation in temperature of Pai forest (a) October, 2013 (b) January, 2014 (c) March, 2014, and (d) June, 2014

Spatial and temporal dynamics of Pai forest vegetation in Pakistan assessed by RS and GIS

Table 6 Forest area under different temperature ranges during winter, spring, summer and winter

Temperature (°C)
16–20 20–24 24–28 28–32 32–36 36–40 40–46 Total area (ha)

Area under different temperatures

Oct. 14, 2013

Jan. 18, 2014

– – – 1788 189 – – 1977

1821.4 155.6 – – – – – 1977

March 07, 2014
– 1305.2 671.8 – – – – 1977

June 11, 2014
– – – – – 975 1002 1977

Temporal and spatial variations in NDVI

The temporal and spatial variation in NDVI of Pai forest from November, 1987 to November, 2014 showed values between -0.045 and 0.58 (Fig. 6). The NDVI value above 0 is categorized into three groups viz. bare land (0–0.2), sparse forest with scrub and stunted bushes (0.2–0.35) and dense forest ([0.35). Thus, for all evaluated years NDVI was higher at those places where dense forest was located, whereas it was minimum for bare land. It varied from year to year such that maximum NDVI of vegetation was recorded in 1987 and minimum NDVI in 2010.
Based on NDVI, the area classification of Pai forest from 1987 to 2014 is given in Table 5. The area under dense forest was greatest at 725 ha (37 %) during 2013 and least at 217 ha (11 %) in 1992. While sparse forest covered the greatest area during 1987 (56 %) and the smallest area in 1992 (43 %). Thus, the smallest areas of both dense and sparse vegetative cover in Pai forest occurred in 1992.

Fig. 8 Relationship between NDVI and a minimum LST and b maximum LST
floods during 2010 and 2011 in Sindh. The temporal variation in vegetative cover of the forest is graphically presented in Fig. 4. Spatial variation in vegetation cover of the forest shows that reduction in vegetation before 2010 mainly occurred at the central and southern parts along the boundary of Pai forest but after 2010, the northeast part of the forest was affected (Fig. 3).
The spatial change in vegetation of Pai forest occurred over nearly 27 years from 1987 to 2014 and shows that there was more forestation than deforestation (Fig. 5a). This change was reflected by an increase in vegetation on an area of 735 ha (37 %), a decline on 427 ha (22 %), and no change on 808 ha (41 %) of the forest (Fig. 5b). The same qualitative and quantitative conclusions can be derived from Fig. 4 and Table 4.

Temporal and spatial variation in Land Surface Temperature (LST)
Temporal and spatial variation in LST of forest for the years 2013–2014 shows that the LST of the forest during autumn, winter, spring and summer varied between 27 and 35 °C, 16 and 23 °C, 22 and 28 °C, and 35 and 45 °C, respectively (Fig. 7). The bare land area had high temperature whereas area under vegetation had low temperature. A variation in LST from 6 to 10 °C between shaded (dense forest) and unshaded areas (bare land) of the forest was observed. Similarly, the temperature difference between areas with sparse forest and bare land ranged between 4 and 6 °C. LST was higher for the bare land and lower for areas with vegetative cover. The forest areas and their temperature ranges during winter, spring, summer and winter are listed in Table 6. Forest, thus decrease surface and air temperatures by providing shade and through evapotranspiration. Akbari and Kurn (1997) reported that temperature under tree canopy can be 11–25 °C cooler than


A. A. Siyal et al.

on bare lands. A study conducted in Spain showed that temperature differences between barren and forested lands ranged up to 13 °C (Gomez et al. 2004). Regression analysis of both minimum and maximum LST and NDVI was carried out to assess the impact of extreme temperatures on NDVI of the forest. An inverse relationship was documented between LST and NDVI of Pai forest with coefficients of determination of 0.944 and 0.917 when NDVI was plotted against minimum and maximum LST, respectively (Fig. 8). This might have been due to latent heat transfer through evapotranspiration, lower heat capacity, or thermal inertia of vegetation compared to soil (Goward and Hope 1989). Similar results were reported by Wilson et al. (2003), Small (2006), Yue and Tan (2007), and Singh (2015).
Because increasing vegetative cover brings down the LST, hence a healthy Pai forest will lower the surface and air temperature, increase the humidity and thus decrease the intensity of scorching heat in the area. Because increasing vegetative cover reduces LST hence greater tree density and canopy closure in Pai forest would be expected to lower LST and thus air temperature as there is generally a close correlation between LST and surface air temperature (Farina 2012).
Maximum reduction in the vegetation of Pai forest was observed during 1992. Spatial change in vegetation of Pai forest over about 27 years (1987–2014) revealed that there was more increase in vegetation than reduction of vegetation. Thus, over 27 years, vegetation increased on 37.31 % of the forest, decreased on 21.68 %, and there was no change on 41.02 % of the forest. Variation of LST between shaded (dense forest) and unshaded areas (bare land) of the forest ranged from 6 to 10 °C. While the LST difference between areas with sparse forest and bare land ranged between 4 and 6 °C. Regression analysis of both minimum and maximum LST and NDVI showed an inverse relationship between LST and NDVI of Pai forest with coefficients of determination of 0.94 and 0.92, when NVI was plotted against minimum and maximum LST, respectively.
Our results how that overall vegetation in forest increased from 1552 ha (78.5 %) in 1987 to 1637 ha (83 %) in 2014 though there was a significant decrease in vegetation 1992. GIS and remote sensing tools can be used for monitoring of spatial and temporal change in vegetation as well as spatial change in LST of forests in arid regions. These results can also be helpful for decision and policy makers for developing an effective forest management system and viable policies for the Pai and other arid forests.

Akbari H, Kurn D (1997) Peak power and cooling energy savings of shade trees. Energy Build 25:139–148
Bhandari S, Phinn S, Gill T (2012) Preparing Landsat image time series (LITS) for monitoring changes in vegetation phenology in Queensland, Australia. Remote Sens 4:1856–1886
Braswell BH, Hagen SC, Frolking SE, Salas WA (2003) A multivariable approach for mapping sub-pixel land cover distributions using MISR and MODIS: application in the Brazilian Amazon region. Remote Sens Environ 87(2–3):243–256
Dash P, Gottsche FM, Olesen FS, Fischer H (2002) Land surface temperature and emissivity estimation from passive sensor data: theory and practice-current trends. Int J Remote Sens 23(13):2563–2594
FAO (2001) State of World’s Forests 2001. Food and Agriculture Organization of The United Nations, Rome
Farina A (2012) Exploring the relationship between land surface temperature and vegetation abundance for urban heat island mitigation in Seville, Spain. Thesis submitted to Department of Earth and Ecosystem Sciences, Lund University
Faruqi S (2011) Into the Pai Forest. The Daily Dawn, December 16, 2011. ( Accessed 5 Aug 2014
Gomez F, Jabaloyes J, Van˜o E (2004) Green zones in the future or urban planning. J Urban Plan Dev 130(2):94–100
GoP (2003) Economic survey of Pakistan. Ministry of Finance, Islamabad
Goward SN, Hope AS (1989) Evapotranspiration from combined reflected solar and emitted terrestrial radiation: Preliminary FIFE results from AVHRR data. Adv Space Res 9:239–249
Goward S, Irons J, Franks S, Arvidson T, Williams D, Faundeen J (2006) Historical record of Landsat global coverage: mission operations, NSLRSDA, and international cooperator stations. Photogramm Eng Remote Sens 72:1155–1169
Hughes BJ, Martin GR, Reynolds SJ (2011) The use of Google Earth (TM) satellite imagery to detect the nests of masked boobies Sula dactylatra. Wildl Biol 17:210–216
Kawamura K, Akiyama T, Yokota H et al (2005) Quantifying grazing intensities using geographic information systems and satellite remote sensing in the Xilingol steppe region, Inner Mongolia, China. Agric Ecosyst Environ 107(1):83–93
Kumar P, Meenu R, Pandey PC, Majumdar A, Nathawat MS (2010) Monitoring of deforestation and forest degradation using remote sensing and GIS: a case study of Ranchi in Jharkhand (India). Report Opin 2(4):55–67
Li Z, Gu X, Dixon P, He Y (2008) Applicability of land surface temperature (LST) estimates from AVHRR satellite image composites in northern Canada. Prairie Perspect 11:119–130
Madin EMP, Madin JS, Booth DJ (2011) Landscape of fear visible from space. Nat Sci Rep 1:14. doi:10.1038/srep00014
Masek JG, Vermote 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 112:2914–2926
McInnes J, Vigiak O, Roberts AM (2011) Using Google Earth to map gully extent in the West Gippsland region (Victoria, Australia). 19th International Congress on Modelling and Simulation, Perth, Australia, 12–16 December. Accessed 5 Aug 2014
Mughal MS (2010) State of forests in Sindh as well as the economic benefits flowing out of it. World Environment Day. Pakistan Energy Congress.

ForestPai ForestLstVegetationVariation