Detection and mapping of vegetal cover depletion of a catchment areas using Remote Sensing

Detection and mapping of vegetal cover depletion of a catchment areas using Remote Sensing

Mustapha K. Idris & Hamirdin I.
Department of Geography University Malaya 59100 Kuala Lumpur, Malaysia

1. Introduction
Vegetal cover , especially types of vegetation, plays a crucial role in the hydrological cycle of an areas where water resources are being exploited. Durian Tungal dam, in the state of Melaka, Malaysia became completely dried up and rendered inoperative in January 1991. The inclement dry spells in the state and errors in calculating the depth-level of the reservoir were tendered as major causes but the development of events was far too gradual as to ignore the importance of soil canopy cover over the dam watershed area. Recent studies showed that the watershed areas have been gradually depleted of forest cover over a period of years in lieu of developmental projects. This paper focuses on mapping the depletion of vegetation cover and the consequences that resulted in the inability of the dam to widstand the prolonged dry spells.

2. Study Area
The study area is located in the north-eastern part of the state of Melaka. Being one of the smaller states in Malaysia, Melaka is experiencing an expanding population and rapid industrialization. The dam, located near the town of Simpang Gading, is an earth-filled impounding dm. It was envisaged that water from the dam is discharged regularly into the intake plant on sungei Melaka before being processed for domestic consumption.

The watershed covers an area of about 41 km2 and is drained by Sungai Machap Umbor, Sungei Ayer Pasir, and other tributaries. In 1974 the vegetation cover comprised mainly of rubber plantations and smallholdings, some patches of secondary forests and belukar, while the upper reaches was mainly under the Bukit Sedanan forest reserve. The topography is generally low and undulating in the lower regions while the headwaters consist of hilly and dissected terrain. The highest point, Bt. Sedanan, at 1093 metres also marked the northern boundary of the watershed.

3. Materials and Methods
Temporal detection of thematic features based solely on remote sensing data, especially in humid tropical region, has been established. Problems relating to cloud cover and non-availability of time-specific images, amongst others, have been recognized. This study is no exception . A two-stage procedure was followed. Firstly, topographic maps of 1959 and 1974 were combinely utilized to derived a vegetation map of the study area prior to the construction of the dam as shown in Fig. 1. The watershed boundary was also delineated based on contours present on the same maps. Reference to aerial photographs taken around this period was made to ascertain the land cover then present.

Figure 1. Land cover of Durian Tunggal Catchment Area. 1974
Tow Landsat TM images of the area were available for this study. The 1990 image was found unsuitable as most of the watershed areas was under heavy cloud cover. Another Landsat TM image of the same area, path/row 126/58, taken on 02.07.89 was selected, despite the presence of clouds, for the second stage. A 512 x 512 pixel subset of the image was inputted for processing using IDRISI, a grid-based geographic analysis system, which also incorporates an image processing module. A variety of modules was used to generate a landcover classification of the area using normalized ratio, principal component and clustering. Visual interpretation of the grey-scale display of TM4 was also pursued for mapping purposes. The same band was also used to digitize the boundaries of the watershed and the impounded lake.

A field study to update land cover types was undertaken in late 1990 and early 1991 and the observations were incorporated with those derived from the satellite data, as in figure 2. Measurements relating to stream characteristics within the watershed area were also made for further analysis.

Figure 2. Land cover of Durian Tunggal Catchment Area. 1989
4. Results
The results of the mapping exercise of the study area, Figure.1 and Figure 2, showed that there clearly was a gross depletion of primary and secondary forest within the watershed. The change in land cover was particularly discernibal in the upper stretches of the main river system. Major parts of the forest reserve have been devegetated for FELCRA’s Penempatan Penyusunaan Semula kampong Kaum Nelayan ( settlement For image shows this particular areas to be intensively berrain except for small strips of vegetation within the gullies. Further west of this area, the forest have been coverted to rubber plantation while the remaining stands of forest cover, in the northeastern corner of the watershed, was observed to be heavily logged.

In the north-west and southern parts of the watershed, rubber and oil palm have replace secondary jungle and bushes as the dominant vegetation cover. About 10% of the area was inundated by the impounding reservoir which is clearly visible in the 1989 infrared image. The clearing of some rubber smallholdings, persuadably for replanting purposes, within the watershed area can also be interpreted from TM4 image. Such affected areas were most often left unvegetated for one to three months before the tree crops mature. Thus, Fig. 2 deplorable phenomena that, in the north-eastern corner of proportions of the Durian Tungal Dam watershed area is fully under various man-induced landuses.

5. Discussion
The extent of the reservoir, as shown in the July 1989 image, catchment area, is remarkable. Yet, less than 18 months after the image was taken the reservoir was completely drained of its content. It was a known fact that water was intermittently released, to beef up the level of Sungai Melaka, a few kilometers south of the dam, and to mitigate the paralyzing water supply deficit experienced by the state during the dry spells of 1990. The decision to release the water from this reservoir was in anticipation that rain would fall and the reservoir is capable to replenish itself. Both did not happen.

It has been expressed earlier that the watershed is devoid of forest cover. The protection of mineral soil from the full impact of falling rain afforded by forest canopy cover is well established. In a tropical rainforest , the canopy is placed in storeys, the floor is covered with litter and lush undergrowth. Operating in tandem these elements provide increased interception, reduced raindrop impact effects, and enable the rain reaching the soil to be infiltrated efficiently and become part of the soil and ground water. The clearing, of forest and smallholdings in the Durian Tunggal catchment area, as evident from the 1989 image, gave rise to a consolidating opposite effects. Subsequently large proportion of the rainfall that occurred prior to 1990 was translated into surface runoff. Very little, if at all, moisture is thus retained in the soil.

Random sampling of water pixel values taken within the reservoir returned a range between 34 to 60 in band TM4. Taking into account the irradiance from the cloudy atmosphere, the readings are still considered too high for water bodies. It can only mean that the water in the reservoir is heavily laden with suspended materials brought down by streams draining the devegetated areas. Thus, the main reason for the inability of main reason for the inability of the dam to recharge itself quickly, after being drained out, can be conjectured.

About 80% of the catchment area have been alienated to major tree crops namely rubber, oil palm and orchards. Planted trees not only lack canopy storeys, contain variable amount of litter and sparse undergrowth but also have less canopy density except where intercropping is practiced. The importance of planted-tree cover in terms of water resource is that hey intercept much less amount of rainfall such that more water is available for the soil to absorb. The lack of interceptive properties, however, will result in the rate of rainfall intensity to exceed the rate of infiltration capacity of the soil. Thus, much of the rainfall is again translated into surface runoff and eventually lost through stream flow and evaporation during the dry season. This affects the subsurface flow of the area with planted – trees canopy, such as the Durian Tunggal dam, can be considered to be an inferior water resource production unit.

Water resource production of the dam is derived mainly from surface water which flow in the streams found in the watershed as illustrated in Figure 2. Prior to 1990, equatorial rainfall, which contributed to the surface flow, was able to sustain the needs of the dam, such that spillway level of 25. 9 meters was always attained. The degenerative repercussion of vegetal cover removal within the watershed was not realized until the areas experienced a period of incessant dry spell. With little or no surface runoff in the streams the dam had to rely on subsurface flow and ground water, by then already lacking in the soil, to replenish its fast dwindling level. The cumulative effects of soil canopy removal; and, the continual release of much-needed water delivered the death blow to the Durian Tunggal dam. On 1st January 1991, the water depth middle of January the dam was completely dry and rendered inoperative.

6. Conclusion
It can be safely stated that the rainwater absorbing and retaining capability of the Durian Tunggal watershed has been damaged by deforestation . The planted-tree canopy cover, the predominant landuse in the basin but apparently does not retain much of the rainwater, also failed to act as a buffer against an inclement dry condition. It may said that the debacle of Durian Tunggal dam could even be taken to be a “blessing in disguise” in that it would provide lessons in future attempts at managing the environment for water supply project.

Landsat TM data and image processing techniques were utilized to update the vegetation maps for hydrological analyses of a catchment area. The study’s initial objective of obtaining a time series data to observe shorter temporal dynamics of the Durian Tunggal watershed was aborted due to lack of acceptable images. Nonetheless, satellite data when combined with published maps, aerial photographs and field data does provides a useful tool for detecting and monitoring of vegetal cover in the study area.

7. References

Barret, E.C. and Curtis, L.F. ( 1982). Introduction to Environ Remote Sensing. Chapman and Hall, London, 352 p.
Brady, N.C. ( 1974) . ady, N.C. ( 1974) . The Nature and Properties of Soil. Machlillan, New York, 639 p.
Canaas, A.D. and Barnett, M.E. ( 1985). The generation and interpretation of false color composite principal component images. International Journal of Remote Sensing 6:867-81.
Eastman, J.R. ( 1990). IDRISI: A grid-Based Geographic Analysis System. Clark Univ., Worcester, 363 p.
Lee, R. ( 1980. Forest Hydrology. Columbia Uni. Press, N. York.
Low, K.S. ( 1972). Interception loss in the humid forested areas ( with special reference to sungal Lui Catchment , West Malaysia. Malayan Nature Journal 25: 104-11.
Manokaran, N. ( 1979). Streamflow, Throughfall and Rainfall Interception in a Lowland Tropical Rainforest in Peninsular Malaysia. Malayan Forester 42d (3), pp 174 – 201.
Melaka, State Govt. ( 1991). Laporan Jawatan Kuasa Siastan Berhubunq Kehilangan Air Di Empang Durian Tunggal, Melaka State Govt. 46 pp.
Teoh, T.S. ( 1973). Some effects of Hevea Plantations on rainfall redistribution, in E. Soepadmo & K.G. Singh (eds. ), Proc. Sympo. Bio1. & Nat. Devl., Malayan Nature Soc., K. Lumpur ,pp. 73-83.
Troiler, L.J.& Philipson, ( 1986). Visual analysis of Landsat Thematic Mapper Image for hydrologic land use and cover. Photogrammetric Engineering and Remote Sensing, 52:9 1531-38.