GISdevelopment.net ---> AARS ---> ACRS 2002 ---> Earth Observation from Space

Accuracy and Application of Remote Sensing in Nepal

Punya Prasad Oli
P.O.Box. 9447
Handigau, Kathmandu-5
Tel. +977 1 438869
E-mail: p_oli@htp.com.np
Nepal



Abstract :
The aerial photography started in 1933 by photographing Mt. Everest areas. The accuracy of topographical and resources maps and information were improved using satellite remote sensing during 1970-1980s. The Remote Sensing Centre was establish in 1970s and disappeared in 1990s. Presently, satellite sensing is being used to study and management of land resources like forestry. In this article, is briefly described the ups and downs of remote sensing, applications, accuracy achieved and recommendation for the development of remote sensing in Nepal.

Introduction:
Remote Sensing is the science and technology of acquiring, processing, analysis of the object or area without touching or from remote and use of such data for decision making. The term was first used in 1960s in USA for photogrammetry and used extensively after launching the ERTS satellites in 1972.

It has three components of emitting or radiating object of the earth surface, the receiving sensor on platform in the aircraft or satellite, and receiving and processing station on the ground. Some cases the platform itself emits and receives the reflected or scattered ray back and forms image or receive data on the sensor.

The sensors may be active or passive and imaging or no imaging. The photography of Shuttle is passive, non-scanning and imaging type and Real Aperture Radar Imagery is the example of active, scanning and imaging sensors. They may be sensitised films, receiving lens, lenses or mirror consisting of series of detectors in arrays.

The detectors convert the received electromagnetic energy into electrical signals. They are amplified and transmitted to the receiving station directly or through satellite to the ground stations in binary signals of pixels. Both cases, the signal passes through atmosphere, which effects the signals/ images. Attitudes of sensors, detectors, orbit of satellite and duration of day also affect the received images or signals.

The received signals of images processed and by sampling and quantisation where number of pixels will be reduced and noises are eliminated. They will be calibrated with ground data, inter band sensitivity, band to band sensitivity ratio and absolute sensitivity. The atmospheric correction and validation will also be conducted for instrumental errors, physical data and use of data. As different channels/bands have different set of detectors and will be registered slightly in different positions. The images are registered each other pixel by pixel for each image. After correction, the images of each scene will be registered and stored by band sequential (BSQ) or BIL (band interleaved by line) format along with auxiliary data. The scene part of scene will be registered to UTM or other projection.

The topographical map data at the scale of 1:25,000 or 1:50, 000 are required for spatial rectification, thematic map data for training data to classify, and socio economic map data for the application and GIS. The digital terrain model data will be used to generate the orthophoto and 3 dimensional view. Generally, remotely sensed data are stored in digital tapes and distributed on CD-ROM or floppy diskettes.

As the resolution of images or pixel size are improved and are getting better than 1 m. It is required to access accuracy of spatially and semantically.

Past and Present of Remote Sensing
In 1972, Earth Resources Technology Satellite and later satellites took imagery of Nepal, which was extensively used as map substitute. The Shuttle Mission took photograph of Nepal, which was provided better understanding of snow covered area of Himalayas.

Remote Sensing Centre established in Dept. of Forest and produced imageries and Imagery Maps of Nepal at the scale of 1: 250, 000. Population distribution or demographic maps of census in 1991 was also produced on the basis of Landsat data.

ICIMOD, Dept of forestry also produced forest inventory maps of Nepal using Remote Sensing data Landsat MSS and data with assistance of Govt. of Finland. T. U., Geography Dept. digitised some maps of Nepal. Survey Dept. started digitisation in 1980s and converted new topographical maps into digital form accurately and available to all projects. Latest aerial photographs are being converted into digital orthophoto maps based on aerial photos with resolution better than 0.5-1 m and high-resolution imageries with resolution up to 1m. Both of them are used in Nepal for planning, monitoring and evaluation of development projects. The orthophoto maps of Mahakali or Sikta Irrigation and Orthophoto Maps of municipal areas of Survey Departments are the examples of digital orthophoto maps used in 2001.

Introduction to Modified UTM Projection of Nepal
The shape of the earth (Geoid) is not exactly a spheroid. The shape varies place to place. There are more than 15 reference spheroids, the shape formed by the rotation of ellipse, used in the modern world. The Everest Spheroid 1830, which is thought to be best fitting for this region and used by Nepal, India, Pakistan, Bangladesh, Myanmar and Sri Lanka. Nepal is continued to use the spheroid. The Indian foot was shorter (1 foot = 0.3047995m) as compared by A. R. Clarke in 1865.

The Universal Transverse Macerator Projection (U T M) with 3° belts is used to reduce the scale errors of computation. The central scale factor (k o) is 0.9999 and origin is equator and 500,000 m at central meridians (81°, 84 and 87°E). The grids are divided in 001 to 180 sheets between 350km-650km at each zone in the y direction and 2900 km to 3400km in x (North) direction. The sheet covers 50km x 50 km area and numbered from NW corner. About 86 sheets of 50 km x 50 km grid sheets will cover the whole country. Topographic maps, 1:25,000 and smaller are numbered by graticules (latitude and longitude) number and their sub divisions.

  1. False Origin
    The origin of easting (y) is central meridian, which is also called false easting, and is 500,000 m at 81° for the area between 79° 30’-82° 30’, 84° for the area between 82° 30’-85° 30’ and 87° for the area between 85° 30’-88° 30’ east longitude.

  2. Figures of the Earth
    The co-ordinates /grid conversion is carried out using of the following figures of the earth: -
    Spheroid - the Everest Spheroid (1830)
    a = 6,377, 276.345m,
    b = 6,356,075.413 m
    1/f =300.80

    The height is used on the basis of BMs established by Survey of India on the points near the border, which are based on mean sea level of 1930 of Bay of Bengal. The fundamental station at Nagarkot will be used as basis of future elevation measurement of Nepal.

  3. Co-ordinates Transformation
    Co-ordinates of pixels / map data required to transfer to fit the map compiled in one systems into another system, grid, projection or spheroid or scale factors. Some times, it is also required to transfer pixels/maps from unknown system to required system. Such as 1''= 1 mile map or land resources maps were compiled in Lambert Conformal Polyconic Projection. A portion or whole map sheet required to transfer into UTM or another projection. The formula may be different depending upon the size and projection system.

    The co-ordinates of small area less than 1° square (111.3 km)2 and of the same projection system will be transformed using linear transformation equations with 3 common / known points.

    The co-ordinates of larger area and different projection systems will be transformed using binomial transformation equations and 6 known or common points.

  4. Crustal Movement:
    The annual movement of southern part of Nepal or collision of Indian plates to Tibet below Himalayan about 20 km north of main central thrust, is generating 20mm north east movement in horizontal direction and 2,5,10 mm rise of terai, hill and Himalayan regions respectively. It is also creating problems on accuracy detection.
Currently available Cartographic Imageries :
The satellite with remote sensors to observe the earth is called a remote sensing satellite. The meteorological satellites, which are placed at higher altitude to provide the meteorological data, are excluded here for comment.

Most of the remote sensing satellites have the sun synchronise and polar orbit with altitude around 800 Km. above the earth surface and they use ultra-violet to infrared radiation and microwave for radiometric purposes.

Presently, countries like USA, France, Russia, EU, Japan, India, China have their remote sensing satellites. Some private companies are also deploying satellites for remote sensing works. The IKONOS, EOS and Quick Bird imageries are providing 1 m or better resolution with stereo pair images.

Some of the popular satellite imageries are Lands at, IRS and IKONOS that are presently used in Nepal.

Sources and Correction of Errors :
There are some uncertainties on the accuracy of images due to its passage through atmosphere, lenses and various electronic devices which may swift, displace, distort the actual position of pixels. The radiometric characteristics of image also change during the detection and transmission processes. The ground relief, curvatures of the earth and map projections are also effecting the position of pixels.

The correction of photographic/analogue images i.e. lens and film errors will be corrected as interior orientation using the correcting lenses or factor. The image is further rectified for geometrical errors using ground control points or large scale topographical and thematic maps. Map elevations are used to rectify the errors due to displacement of images and lack of light on shadow areas due to relief. The satellite imageries are acquired very high altitude and heightening accuracy is inversely proportional to altitude of satellite, hence the imageries will not be used for heightening purpose except the stereo imageries.

The digital images are quantised and sampled for each detector and each pixel. Different sensors and multi-temporal images are registered to correct for the geometric errors.

The observed data of test sites are calibrated and corrected with reference data to correct the sensitivity of the detector, atmospheric condition, alignment of detector and so on. It is, the ground calibration done before the launch of satellite and on board calibration data like sunlight, shadow, black body radiation will be transmitted to the control station along with remote sensing data. The remote sensing data also validated with the ground data as well as specification of instruments and physical situation of ground which are collected from the test sites. The correction of vignatting, sunspot and shades due to relief on the images called the radiometric correction will be carried out.

For accurate geometric correction or transformation, ground control points or large-scale topographical database along with elevation data will be used. It will correct the errors caused by tilt, tips, scale error, radial and tangential distortions, projection, and curvature of the earth and relief displacement. Number of ground control points or map points and their accuracy depend open the number of unknowns (4-10 unknowns) and the method of transformation. The accuracy of control points should be better than one pixel and generally accuracy of one pixel is achieved during correction process.

In practise large number of control points are needed to transform an imagery including matching the side imageries.

Institutional and other local Expertise:
In order to carry out the remote sensing works, distribution of imageries and provision of weather satellite data, the Remote Sensing Centre was established in Forestry Dept. in 1981 and was converted into Remote Sensing Unit during the reorganisation of HMG/Nepal agencies in November 1991.

Survey Department established National Geographical Information Infrastructures (NGII), which needs to co-operate with other organisations and to update topographic and land resources data. It is high time to develop remote sensing capabilities in the department to update these Land other data. The departments of Survey, Forest, Mines and Geology, Irrigation are using and have acquired hardware and software to use remote sensing data. The Landsat MSS, IRS Pan and LISS and other data have been used for some areas and aerial photographs is extensively used for engineering survey works than the satellite imagery due to high cost of imagery. The quality control and assessment is not properly documented.

The Remote Sensing and Photogrammetric society was also established in March 1992. Like other professional societies and organisations, it is also in active in the development of Remote Sensing technology.

Applications and Accuracy of Remote Sensing Data :
Remote Sensing has been used for many applications specially for the study of the natural resources, orthophoto preparation of the terrain and over view of latest ground situation of the larger area as well as updating data of various inventory and data base. The aerial photography is still used to compile detail information in Nepal, which is being replaced by high-resolution satellite imageries.

The remote sensing is used in Nepal mainly for the following purposes: -
  1. Base Mapping at the Large Scale
    As the availability of aerial photographs is restricted in this region, the use of high-resolution satellite imageries is essential for the planning and development of this region. The only restricting factors are high cost of corrected imageries and software, and necessitate of higher skills.

    High-resolution (1m-pixel) imageries are used to prepare digital and hard copy orthophoto maps at the scale of 1:2,500 to 1:10,000 of flat areas. The contours or DTM is generated either by field survey or from the existing large-scale topographical maps. There is not much experience on the elevation accuracy. The accuracy of Ikonos Panchromatic images were ± 0.1 m and standard deviation was ± 0.6m in terai areas and hill areas. Private firms mostly generated the imagery data with various methods and purposes.

    The semantic accuracy of image recognition may be around 80% or better (Scott 2002 3).

    The high-resolution imageries are being used in planning and monitoring of irrigation, drinking water supply, rural roads, small town and other small development projects.

  2. Revision of Map: -
    Survey Depart is also working with 5.5m resolution IRS Imageries for the revision of medium scale maps. The irrigation projects are updating the existing 1:5,000 scale maps using high-resolution imagery as orthophoto maps.

    Other imageries like Landsat, SPOT also have stereo capabilities and high-resolution panchromatic band with 15m and 10m resolutions respectively. Three seasons data will improve the semantic accuracy and will be about.99% correct (Mercier 20024).

    The remote sensing image data and other analysed data are used to obtain a higher accuracy of classification. The digital elevation model data is used with remote sensing for correction of error due to shadow.

  3. Land Cover Classification and change detection
    The study of land cover or physical condition of any part of world will be carried out extensively using satellite imagery. Geo-coded remote sensing data will be superimposed on the topographical map data pixel by pixel. The maximum likelihood method of classification is used for classification. The textual information could be used for detail change detection. The multi-band and multi-temporal (data of different dates) are used to detect the changes and classes of data. The forest-mapping programme in Nepal used these techniques.

    The semantic accuracy of cropland interpretation will be 90% and pasture 55% and it will improve to 99.5% by three season observations( Mercier 20024). and which is also varies 63% to 96 % from crop to crop (Craig 2002 5).

  4. Generation of D E M and up dating maps
    The remote sensing data like Landsat, Spot, IRS have stereovision of panchromatic image and has higher resolution up to 10m or so and may organisation used them for generation of digital elevation model (D E M). It is claimed that the theoretical heighting accuracy of Landsat TM has 14.5m and SPOT has 14.8 m (Ching2). Similarly, the panchromatic images have resolution about 5-10m and may be used to up date topographical maps at the scale of 1:50,000 and in some cases to revise th e maps at the scale of 1:25,000.

  5. Geological Study
    The remote sensing data will also be useful to detect the lineament and rock types and the study of geological structures and minerals.

    Landsat 5 TM MSS data were used to prepare engineering and environmental Geological Study of municipal areas and deposits of Siwalik areas (Duvadi 19986). Presently IRS LISS and panchromatic imageries are mostly used for their works. The SPOT, Landsat imageries are also available for geological works.

    The GIS is established on the basis digital topographical data provided by Survey Department. The features recognition of high Himalayan areas are quit good and more field work needed in hills and terai areas of thicker soil deposit / vegetated areas.

  6. Remote sensing and GIS
    The following three necessary functions are required to fulfil by remote sensing to be used in GIS: -
    1. To store and manage the geographic information comprehensively and effectively,
    2. To display the geographic information depending on the purposes of use
    3. To execute, query analysis and evaluation of geographic information effectively.
    The remote sensing data after geometric correction are overlaid with other data in vector/raster form. It has two uses in GIS i.e. classified data and us ed as image data.

    Use of classified data- the remote sensing data is classified and converted vectors data. These data along with other field data are used as GIS purposes.

    Use of Image data- the remote sensing data may overlay onto other geographical data, which enable to analysis for environmental monitoring and change detection.

    The accuracy part of imagery and spatial data are poorly understood by surveyors as well as by experts. The large-scale spatial data are used planning and cost estimation of project. In Nepal the in accurate information means black listing of experts and lost of income or jobs. Hence, lot of ground control data and trial and error methods are used before really producing the usable information.
Conclusion
As conventional mapping and data acquiring process are times consuming and impractical to use for periodic monitoring of the resources of the country. The satellite remote sensing has proven its capability to provide the service effectively. The national mapping organisation should use this technology and assist the planners and decision-makers for their judicious decision. The accuracy of the remote sensing data are reaching to about 1-0.5 m resolution in plannimetric and 2 -4 m in elevation information (Scott 2002 3). They are quite useful to compile quickly the recent orthophoto data/maps of any area and to update maps, G I S and data regularly.

Acknowledgement
I would like to express my appreciation to Department of Survey, Mine and Geology, Forest and Irrigation as well as Welink, ERDAS Nepal and other friends for providing valuable information about the activities of their organisations and A C R S for providing me opportunity to present the paper.

Reference
  • Murai, Prof. S Edited by 1991 Application of Remote Sensing in Asia and Oceania, AARS
  • Ching N.P. Mapping from SPOT Stereo Imagery, Application of Remote Sensing in Asia and Oceania, A A R S. PP 197-198
  • Scott Lee D., Shan J. and Bethel J, 2002, , ACSM- APRS, Annual Conference Washington May 2002
  • Mercier B.M and other 2002, Remote Sensing Tools for state wide Management; Developing a Landsat 7 Imagery Date base of Kansas 2002, ACSM- APRS, Annual Conference Washington May 2002
  • Gaig M.E. 2002, comparing 30 meter Imagery from Landsat sand 7 for Crop Area Estimation, 2002, , ACSM- ASPRS, Annual Conference Washington May 2002
  • Duvadi A.K. and other 1998, use of Landsat TM Data in Mapping of Siwalik Deposits, Nepal, Journal of Nepal Geological Society, 1998 vol. 18, pp, 59-69