Data Acquisition through
Remote Sensing for Management Planning of National Parks and Protected
Areas Somasiri, S*, and M.
Herath# Abstract*Dr .S, Somasirim managing Director, AGRIDEV Consultants, Ltd. Kandy, Sir Lanka #Dr. Manthrithilake Herath, Director, Upper Mahawali Catchment Management Division MASL, Sri Lanka. Sri Lank is endowed with a rick bio-diversity and endemictity, but the species are under severe threat from a rapidly increasing population. the government has established a large number of protected areas (Pas), which constitute 12 percent of the total area of the biological resources of the Pas, which are scattered over the country have been a major constraint for management planning. The severe competition for resources in Pas and escalating human-animal conflicts, rapid destruction of the key animals, and loss of property strategies for which obtaining resources information in a very short time is essential. Application of remote sensing information seems the quickest method of resources inventory. This study attempts to identify, classify habitat types and map them; identify available remote sensing data. Subsequently, field checking was used to verify the interpretations. Paper discusses the advantages and the limitations of the remote sensing methodology in resource data acquisition. Introduction Sri Lanka is endowed with a rick bio-diversity and endemicity, but the species are under severe threat of loss and in some cases extinction due to lack of proper and adequate by the government with a view to conserve and maintain the biodiversity, no management resources. There is an urgent need to inventorize the physical and biological resources, identify the cultural resources and identify the cultural resources and evaluate the external impacts on the protected areas. Remote sensing has been applied for earths resource inventory form the middle of this century (Hoffer, 1975). Use of satellite MSS data began in late 60s (Floyd F. Sabins. Jr 1977) In Sri Lanka aerial photography was used for resources inventory in the early 60s (Nanayakkara, 1983) and other high altitude remote sensing data came into use in the early eighties (Geiser and Sommer, 1984; Molegoda, Kotagama & Sommer, 1984).The emphasis of previous studies in Sri Lanka has been on the vegetative cover and it changes. In this paper attempt is to apply the remote sensing methodology for a rapid appraisal of the physical and some biological resources and to inventorize them for the purpose of management planning and development of strategies for the conservation and maintenance of biodiversity. Also identify the extent of encroachments by the settlements in to the protected areas. However, the wild life sector is not equipped with computer hardware and software as well as the required expertise for sophisticated processing of that the resources inventory for wildlife management has to rely more on visual interpretations of remote sensing data and low altitude aerial photography. Objectives: This research activity is designed to test the applicability of the remote sensing methodologies for rapid appraisal of the resources and the inventorization of physical, to species conservation. Following are the specific objectives. *Inventory of the soil and the land form/geological features *Inventory of the habitats and characterize them in the field *Inventory of the geo-hydrological features: surface and sub surface water resources. *Inventory of the land use changes over time. The overall objective is to provide the essential resources information in a short period for management planning and development of management strategies. Study Area Two sample areas were selected from the list of protected areas, both in the dry to semi-dry zones of Sri Lanka. the area (1) northern part of Wasgamuwa National Park, E longitudes and (2) Kahalla-Pallekelle sanctuary situated in the districts of Kurunegala and Anuradhapura between 285 and 307 N latitudes and 163 and 168 longitudes. Methods: The study methods included the visual interpretation of Landsat imagery, spot imagery the scale 1:50,000 of HRV 2 XS (1996) interpretation of aerial photo of 1:20,000 of the same areas, and computer-aided IRS data analysis. The digital computer-aided analysis involved machine -aided "training field selection" or "non supervised spectral response resorted to. A limited field surveys and other secondary data were used to verify the computer aided classification and visual interpretations. Results and Discussion Geomorphology and Landform The visual interpretation of Landsat imagery and false colour composite maps provided useful information on landform for both study areas. However, only the very distinct geomorphic units such as high relief ridges and hills were easily identified. The more subdued landform such as river levees, rock knob plains and inland valleys were not conspicuous in the images use, therefore, such features are not identified. The aerial photo interpretation provided more detailed information on important landform features. In the study area (1) the river levees, rock knob plains and ridges which are very important feature in distinguishing wild life habitat types were conspicuous and were easily mapped. Soils Digital processing of IRS data or visual interpretation of spot imagery could not identify the soil variations on land slopes. Also the residual from alluvial soils could not be separated by this methods. for semi-detailed mapping of soil variations in small areas such as few thousand hectares of protected area, low altitude aerial photography proved to be the most useful. Hydrology Except for the linearity of the geologic features, the structural information on the geologic formations could not be inferred. There are not distinct lineaments present in the study areas. The high relief of ridges and low density of dissection suggest that the ridges are formed of rocks resistant weathering. The MSS band 7 image showed very low stream density. These geologic features in sample areas are very well covered with vegetation. Therefore, it is possible that the rainfall runoff is low and much of the rainfall is infiltrated through the ridge surface. These geologic features should, therefore act as a ground water reservoir, and that water should emerge at the base of the ridges. During the field surveys the existence of a series of springs along the foot of the ridges were observed and confirmed the inferences drawn from the visual interpretation of the remote sensing data. Vegetative Cover types The computer-aided analysis of the IRS data by the unsupervised method of training field selection in the first instant classified the spectral response into six ecological classes. This resulted in the division of the dense forest vegetation into two different spectral signature classes. The dense forest vegetation on the west face of the ridge, which had the shadows of the ridges at the time of the data acquisition had the same spectral class as the agriculture field in the adjacent areas. The other spectral class included all the forest types, i.e. dense forest, open forest and grasslands with sparse tree cover (Damana or dry savanna). Subsequently the analysis and classification of the same data into 10 spectral classes differentiated the dense forest from other vegetation types and also into 10 spectral differentiated the dense forest from other vegetation types and also overcame the problem of the division of the dense forest into two distinct spectral classes. The second analysis gave six different spectral classes for the study area (2). They are: Dense forest, open forest, bare rocks/bare soil, grass cover with sparse tree vegetation, paddy and water bodies. The false colour composite (4,3,1) of IRS of 1992 March 10 confirmed the spectral identifiable on low altitude aerial photographs and found that open forest including the secondary forest growth in abandoned chena lands. The visual interpretation of the imagery HRV I XS of 1:50,000 scale of the spot data of 1986 recognized eight different ecological units for the same study area one (1).They are (1) Dense forest, (2) Open forest, (3) Secondary vegetation, (4) Scrub on river levee, (5) Chena/abandoned chena, (6) Scrub, (7) Homesteads (8) Paddy, (9) Rock out corps and (10) identified on the 1:50,000 image used. Further forest plantations could be separated from other types of forest. the ecological features clearly identified in both the computer-aided analyses and the visual interpretations were, dense forest, paddy, bare rock/bare soil and water bodies. In the unsupervised spectral analysis, the open forest, secondary vegetation, secondary vegetation in river levee, chena/abandoned chena, homesteads, upland croplands were classified in one spectral class. Further, study is required to separate these units by computer-aided analysis. The manual training field selection, that is "supervised training field selection" with ground observations would perhaps be able to separate these ecological units. The ecological unit, grassland with spare tree vegetation appears to include more area than the damana lands that is found in the study area. Further, field work is necessary to confirm this observation. A major problem for the conservation of biodiversity in the protected areas is the encroachment of agricultural uses. Even the unsupervised classification of the spectral signatures was able to show the extent of encroachment. The irrigation structure (tanks), paddy cultivation, upland croplands and settlements showed distinct spectral signatures, and such activities were readily identified. These observations were confirmed by the false colour composite image of IRS data. Unless the low altitude photography is very recent, the extent of encroachments into the protected areas can not be accurately mapped. The digital processing of IRS data appears to be very useful in the inventory of resources for protected area management. It is very rapid method, Major ecological differences could be easily distinguished and mapped. However, for wildlife management mere inventory of the natural resources is not adequate. The identification and characterization of habitat types are very necessary. The interpretation of low altitude photographs of a suitable scale provided a very good basis for the identification of habitat types. Such habitat types included more than one spectral class, but in different proportions. A method involving a combination of the interpretation of low altitude photographs, visual interpretation of spot imagery and computer-aided analysis of IRS data together is most fruitful in the identification and characterization of the wildlife habitat type. The multi-temporal satellite data is useful for assessment of stress conditions, arising from droughts, fires etc. what would to develop management plans for protection, and of vulnerable areas. To monitor deforestation and detect areas vulnerable to encroachment can be made with the Acknowledgements I am very grateful to Mr. Sarath Jayatilake and his staff for inviting me and giving me the opportunity to use their data. Mr. Ranjith Premalal of the Faculty of Agricultural Engineering readily assistant me in the computer - aided classification of the IRS data. I wish to thank Mr. T.M.J. Bandara of Natural Resources Management Centre for the useful suggestion made in the preparation of this paper. References
|