INGENIEUR
of the oil palm land scheme for the year 1988 and 2020 . Each colour represents a different change trajectory . Overall , most changes gravitate towards vegetation with NDVI values of more than 0.6 which is considered natural since the oil palm detections were carried out from the year the oil palm was planted until recently .
( v ) Terrain & Digital Elevation Model ( DEM ) Basically , the digital elevation model ( DEM ) was able to show the elevation of the study area . The three dimensional view of the lots shows that the land scheme is located in a hilly area which is not considered an optimal site . This is can be related to the problematic soil growing conditions in hilly tropical areas where the soils weather and leach easily ( Pirker et al ., 2016 ) .
This analysis was then paired with the Topographic Wetness Index ( TWI ) and it correlated with the saturated condition in the northern part of the land scheme . The results from the TWI map agreed with the conditions stated by the local people where they stated that the northern part of the land scheme is most likely to get flooded during the rainy season due to its low elevation and the overflow of the river located on the northwest of the land scheme .
( vi ) River Detection Evaluating river features such as length , width and temporal variance may offer valuable insight into surface water availability , transportation , delivery and dynamism . Rivers that are open water sources have extremely poor visible / nearinfrared spectral reflection and therefore appear as dark , continuous , and curvilinear lines . These spectral properties show the river ’ s special features in remotely sensed optical visible / nearinfrared imaging . The river in the area of interest , however , is too narrow in width with an average of two to three metres during dry weather . The river is also quite muddy and has a greenishbrown clay or sand colour which is about the same as the ground colour in the land scheme . This is also another challenge in detecting the river using the usual pixel clustering technique . So the sub-pixel method was used to detect this specific river or any small water bodies . This technique was able to extract the endmember abundance in a pixel .
( a ) Sub-pixel Analysis The water endmembers cannot be extracted directly from the water bodies in the river reserve of the cadastral layer because of their small inundation extent . To overcome this limitation , the water endmembers were extracted from Sungai Kerian , the closest large waterbody to our study area . To account for the best and closest spectral signatures to capture and detect waterbodies , we extracted the smallest / lightest water endmember . Results ranged from negative to one , where negative values reflect a cell ’ s non-presence of water and positive values show a percentage of a water-covered cell . Values > 1.0 suggested that the cell was covered fully by water bodies . Since Sungai Jambu is narrow in width in dry months and the SPOT-5 image and Landsat TM image has a resolution of 30 metres and 10 metres respectively , this makes it difficult to gain large values to show the presence of water bodies inside the river reserve . This is because the Landsat pixel covers an area of 30 m x 30 m , or 900 square metres while the size of river in this pixel will be , say 3 m x 30 m , that is 90 square metres . Hence the river attribute is minor and non-dominant in the pixel as it occupies an area of only 10 % of the pixel size . However , a comparison of values gained from the pixels in the river reserve and the vegetation lots does provide a measure of contrast to show there might be a river inside the cadastral river reserve . The pixel values of the river located near the study area which were extracted from Landsat and SPOT-5 images shown Figure 8 and Figure 9 respectively were consistent throughout the river reserve . This is another strong reason to show the presence of water bodies in the river reserve of the cadastral map .
( b ) Tasseled Cap Transformation This analysis was generated to detect the contrast between the top part of the river and the bottom part of the river . The map shown in Figure 10 ( northern part of the river ) and Figure 11 ( southern part of the river ) correlates with the wetness index generated during the terrain analysis which shows that the upper part of the lots / land scheme will have a higher moisture index and can get flooded easily . This also showed that the upper part of the
58 VOL 84 OCTOBER-DECEMBER 2020