Pakistan being one of the eight member countries of the International Centre for Integrated Mountain Development (ICIMOD), is actively engaged in its macro level land cover assessment in collaboration with UNEP Environment Assessment Programme for Asia and the Pacific (UNEP/EAP.AP). With the limited resources and time frame, however, it will not be possible to bring about a detailed micro-level assessment of the land cover. Still, it has been endeavored to add as many parameters to the report as possible. Some of the data included in the report is as old as 10 years. However, this data could be used as a baseline for a comprehensive report to be prepared in the future. That would greatly facilitate the comparison between the old and the new data and might be helpful in ascertaining the physical changes occurring particularly in the land use pattern and look for possible corrective measures if the situation so warranted.
1.2 Pakistan: Study Area
1.2.1 Location and Physical Characteristics
Pakistan has a great variety of landscapes with a diversified relief. It has all the majestic high mountain ranges of the sub-Continental north: the Himalayas, the
Karakorams and the Hindu Kush. The vast and rich irrigated plains of the Indus Basin covering vast tracts of the Panjab and Sindh, the stark deserts of Cholistan (Punjab!) and Thar (Sindh!), the inter-montane valleys of NWFP (North West Frontier Province) and the awe-inspiring rugged plateaus of Balochistan and the meeting point of the Himalayas, the Hindu Kush, and the Karakorams in the Northern areas are some of the most varied features of the country’s landscape.
Geographically, Pakistan lies between 24o and 37o N latitude and 61o and 75o E longitude. It is bordered by China in the north, Arabian Sea in the south, Iran in the west, Afghanistan in the north-west and India in the east (Fig. 1). It is a federation of four provinces: NWFP, Panjab, Sindh and Balochistan with its capital at Islamabad. The land area of Pakistan is about 796,000 sq. km and an estimated population of 134 million (June 1996), having a population density of 168 persons per sq. km.
Of the 79.6 million hectares land area, only about 22 million hectares (23%) are available for cultivation: 18 million ha irrigated and 4 million ha rain-fed. Forests, both natural and man-made, cover about 4% of its land area.
1.2.2 Present State of Land Cover
The climate of Pakistan varies with altitude, which in turn affects the type of vegetation. It has some of the world’s highest cold areas that occur above 5,175 masl in the Himalayas and the hottest low areas in the Indus Plains with many intermediate ecological zones.
Pakistan has nine major Ecological Zones with the main ecological determinants as Arid and Semi-Arid conditions. These conditions prevail over most part of the Indus Plains and the Balochistan Plateau. The Humid conditions exist over the hills and mountains in the north. In the Arid and Semi-Arid areas, most parts are bare of vegetation. At riverbanks and deltas, Riverain and Mangrove forests have emerged. On humid hills and mountains, pines and coniferous forests occur that change with altitude. The Dry sub-Tropical forests dominate up to an altitude of 1,000 masl, the Coniferous forests from 1,000 to 4,000 m. Above tree line, Dwarf Alpine forests followed by the Alpine Pastures occur up to snow line.
Based on the above classification, the following seven forest types are recognized:
The Coniferous Forests occur from 1,000 to 4,000 m altitudes. Swat, Dir, Malakand, Mansehra and Abbottabad districts of NWFP, and Rawalpindi districts of the Panjab are the main areas covered with coniferous forests. Fir (Abies spp.) and Spruce (Picea smithiana) occupy the highest altitudes, Deodar (Cedrus deodara) and Blue Pine (Pinus wallichiana), the intermediate heights, and Chir Pine (Pinus roxburghii), the lower areas. The Coniferous forests also occur in Balochistan hills. Chilghoza Pine (Pinus gerardiana) and Juniper (Juniperous macropoda) are the two most common species of Balochistan.
The sub-Tropical Dry Forests are found in the Attock, Rawalpindi, Jhelum and Gujrat districts of the Panjab, and in the Mansehra, Abbottabad, Mardan, Peshawar and Kohat districts of NWFP up to a height of 1,000 m. In Balochistan, they are confined to the Sulaiman mountains and other hilly areas. Dominant tree species are Phulai (Acacia modesta), Kau (Olea cuspidata) and Donoenia viscosa.
The Tropical Thorn Forests are dominated by Xerophytic Scrubs. They are most widespread in the Panjab plains. They also occupy small areas in southern Sindh and western Balochistan. Common species are Acacia spp., Salvadora oleodes, Prosopis cineraria, Capparis aphylla, etc.
The Irrigated Plantations were first developed in 1866 at Changa Manga (Panjab!). Today they occupy about 226,000 ha. Shisham (Dalbergia sissoo), Mulberry (Morus alba), Babul (Acacia nilotica), Eucalyptus and Populus spp. are the common tree species grown in the Irrigated Plantations.
The Riverain Forests grow in narrow belts along the banks of Indus and its tributaries. They are more commonly found in Sindh and to some extent in the Panjab. Babul (Acacia nilotica), Shisham (Dalbergia sissoo) and Tamarax dioica are the most common species. Prosopis cineraria, Tamarax spp. and Populus euphratica are some other species found in these Forests.
The Mangrove Forests are located in the Indus delta. However, lack of fresh water has resulted in their stunted growth. Avecennia officanilis is the main species. Ceriops and Rhizophoras are the other tree species but are fast disappearing because of human pressure.
1.2.3 Areas of Major land cover Transformations (Hot Spots)
As a consequence to deforestation and changing land use patterns, the most critically affected ecosystems of Pakistan are:
Juniper Forests of northern Balochistan, essentially categorized as Protected Forests, are heavily harvested for timber and fuelwood. The unrestricted grazing has further hampered the natural regeneration of trees.
Indus River Zone is the other such area where ecological changes have drastically affected the Riverain Forests as also the Coastal Mangrove Forests. Large tracts of riverain forests have been cleared for agriculture. The river Indus has been dammed and dyked and thus confined. The resulting drop in river level has left extensive areas of the riverain forests high and dry.
As a further consequence of the construction of upstream dams (in the northern parts) and barrages in the plains of the Panjab and Sindh, there has been considerable drop in the Indus water. This has caused reduction of fresh water flow in the delta (at the confluence of the Arabian Sea) resulting in increased salinity which in turn has damaged the mangrove forests and thus eliminated several tree species. The uncontrolled cutting of trees for firewood and fodder needs has further accelerated the reduction and degradation of swamp forests.
The Himalayan Temperate Forests are also under severe pressure from logging for timber and firewood and making clearings for agriculture; for the ever-increasing population pressure.
1.2.4 Need for Monitoring
Satellite imageries at varying spectral, spatial, and temporal resolutions are useful to map natural vegetation types and to detect and delineate major changes over time.
To carry out the inventory of the vast and scattered areas at macro scale, coarse spatial resolution scanners data with high temporal resolution are needed to reduce data volume and increase the probability of cloud-free data. The source of information used in this report is the digital data of NOAA AVHRR LAC format node having 1.1 kilometre spatial resolution acquired during ascending node (1330 LST). Pakistan is covered within one LAC format NOAA AVHRR imagery.
1.3 NOAA Satellite Series
The NOAA satellite series commenced with TIROS-N (Television and InfraRed Observation Satellite) (launched in October 1978) and continued with NOAA-A (launched in June 1979 and renamed NOAA-6) to NOAA-J (launched in 30 December 1994 and renamed NOAA-14). At present two satellites, NOAA-12 and NOAA-14, are operational. NOAA satellite series are polar orbiting sunsynchronoumus satellites orbiting at a height of 830 - 870 km above the Earth’s surface. The data can be acquired twice daily, one in ascending mode and the other one in descending mode. The even numbered satellites have daylight (0730 LST) north-south equatorial crossing times and the odd-numbered satellites have nighttime (0230 LST) north-south equatorial crossing times. Alternatively, even numbered satellites have evening (1930 LST) south-north crossing times and odd numbered satellites have afternoon (1330 LST) south-north equatorial crossings. The crossing time of ascending and descending nodes of NOAA-12 is 1915 LST and 0715 LST respectively, and that of NOAA-14 is 1330 and 0130 LST respectively. The satellite has an orbital period of 102 minutes covering the Earth’s surface 14.1 times daily. The inclination of the satellite is 112 degree with the scanning angle of 99 degrees. The sensors onboard the NOAA Satellite series are AVHRR (Advanced Very High Resolution Radiometer), TOVS-SSU (TIROS Operational Vertical Sounder Microwave Sounding Unit), TOVS-MSU (TIROS Operational Vertical Sounder Stratospheric Sounding Unit), TOVS-HIRS/2 (TIROS Operational Vertical Sounder High resolution Infrared Radiation Sounder / 2), SAR (Search and Rescue), SBUV/2 (Solar Backscatter Ultraviolet Radiometer/2), and ERB (Earth Radiation Budget Experiment).
1.4 NOAA AVHRR
The AVHRR is a broad-band scanner that consists of four to five bands depending on the models of the sensor on board NOAA Satellites. AVHRR data are archived in three formats: HRPT (1.1 km spatial resolution at satellite nadir), LAC (1.1 km spatial resolution at satellite nadir), and GAC ( 4 km spatial resolution) formats. The AVHRR sensor provides global, pole to pole, data from all spectral channels. The swath width is 2399 km with 2048 pixels per scan line. The entire Earth can be covered in just 14.5 days. LAC data are also full resolution satellite images but are recorded on an on-board digital tape recorded for subsequent transmission. Though the spatial resolution of the AVHRR HRPT and LAC data is 1.1 km at satellite nadir, the resolution decreases with the increase in the view angle off-nadir. The maximum off-nadir resolution along track is 2.4 km and across track is 6.9 km. The advantage of the NOAA AVHRR LAC data are: synoptic coverage and hence low data volume (swath width 2700 km), high radiometric resolution (10 bit), relatively low cost (Free!, only handling cost), twice daily coverage and hence high possibilities of having cloud free data. The major disadvantages are: coarse spatial resolution (1.1 km at the nadir), pre-processing is time consuming, the methodology is not well developed, and LAC data has limited capability to record on-board. Designated originally for meteorological studies, AVHRR data can be used for various land applications, such as land cover assessment and monitoring. Due to the coarse spatial resolution (1 km), the pixel might represent different land cover types on the ground, but the spectral characteristic will be representation of the predominant cover type within that pixel. One should be aware that the minimum mapping unit for the NOAA AVHRR data is one square kilometer. That is, land cover features smaller than one square kilometer is not distinctive.
Maps 1 to 8 have been added to explain AVHRR spectral ranges and their combinations and applications as listed in Table 1.
Table 1. AVHRR spectral ranges, their combinations and applications
|Map No.||Channel Number
Or Channel combination
|1||1||0.58 - 0.68||Reflected Visible||Discriminating clouds, Daytime cloud and surface features mapping.|
|2||2||0.72 – 1.10||Visible / Near Infrared||Mapping land / water discrimination (water has lesser reflectance than other land uses), discriminating daytime cloud.|
|3||3||3.55 - 3.93||Emitted Thermal Infra- Red / Reflected Solar InfraRed||Determining temperature of radiating surface, night cloud mapping|
|4||4||10.50 -11.50||Thermal IR / Emmited Thermal Infra Red||Determining sea surface temperature, day/night cloud mapping|
|5||11.50 - 12.50||Thermal InfraRed / Emmited Thermal InfraRed||Determining sea surface temperature, soil moisture, day/night cloud mapping|
|NDVI||Normalized Difference Vegetation Index||Vegetation Index is ratio or difference of reflectance value in the visible (Red) and Near Infra Red region of the spectrum|
|FCC||False Color Composite||Generated by compositing three multi-band images with the use of three primary colours: by assigning blue to Reflected Visible band, green to Reflected Visible, and red to Visible/Near-Infrared band. Green vegetations appears in different tones of red color. Snow & ice appears in white.|
|FCC||False Color Composite||Generated by compositing three multi-band images with the use of three primary colours: by assigning blue to Reflected Visible band, green to Visible/Near-Infrared band., and red to Reflected Visible. Green vegetations appears in different tones of green color.|
1.5 Methods Used
Ten sets of the NOAA AVHRR data covering Pakistan were supplied by UNEP to ICIMOD for processing. The digtial data of NOAA AVHRR in LAC formats of the following dates were used in the present study: 14 Feb 1993, 20 Feb 1993, 28 Feb 1993, 17 March 1993, 23 Apr 1993, 1 Dec 1992, 14 Oct 1992, 23 Oct 1992, 13 Apr 1993. Pre-processing was carried out for all the datasets. The NOAA AVHRR data Pre-processing consists of data extraction and noise removal, radiometric calibration, geometric correction, and cloud masking procedures. AVHRR imagery of 14 October 1992 was used as the base image owing to its least cloud coverage. The data set are resampled to one square kilometer spatial resolution (i.e. one pixel represents one square kilometer on ground) after geometric correction. The country mask is generated using the country boundary available in the Digital Chart of the World (DCW) vector map. Spectral characteristics of the individual bands, NDVI, and color composites were studied for land cover mapping. Unsupervised digital land cover classification was performed using
spectral signature definition by iterative clustering technique and
maximum likelihood classification method. Interactive labeling of this
signature information into major land cover categories requires substantial
field information, forest seasonality/ phenology and ancillary data like
topography and climate. Minor decision rules based on GIS overlay operations
were performed for the finalization of the classification generalization.
Accuracy assessment of the classification result is done using other available