Basal Stem Rot of oil palms is a well-known problem in oil palm plantations in Malaysia. As this common infection can have a negative impact on palm oil production, the Forest Department of Sarawak is now using airborne hyperspectral imaging to assess the scope of the damage.
The Forest Department is operating a small aircraft equipped with a compact, turnkey RT3052 GPS/INS system and a Specim AISA Eagle hyperspectral sensor. The RT3052 system onboard the plane provides highly accurate position, roll, pitch and true heading measurements. Together with the hyperspectral sensor, the Forestry Department of Sarawak can now detect and map the distribution of infected oil palm trees. The combined system is so accurate that individual trees can be identified. With a map showing the health condition of individual palm trees, the Forest Department of Sarawak can then take appropriate action to treat the oil palm trees in specific locations which have been identified by the hyperspectral survey.
Using geo-referenced hyperspectral technology brings many benefits. The amount of fungicide needed is drastically reduced, saving money and reducing the damage on the environment. The time taken to treat the infected trees is reduced. Early treatment of smaller areas prevents the fungus from spreading, further reducing costs. Overall, a smaller, more efficient team can cover a much wider area and be more effective at it.
As a ﬁrst step in data processing, Specim’s software Caligeo was used to turn the raw hyperspectral data to radiance and to georectify the image to a map. Figure 1 shows part of a georectiﬁed ﬂight line over the plantation, visualized in true colors. Secondly, the Forest Department applied their palm tree detection algorithm to remove the background and create a palm tree map (Figure 2). The purpose of this mapping application is early phase detection of trees infected by Basal Stem Rot fungus. The Forest Department of Sarawak has developed an optimized detection algorithm, which utilizes sensitive spectral information particularly in the red edge region. Figure 3 shows the tree map after the fungus detection algorithm is applied. Infected trees are visualized in range of yellow color, and healthy trees in brown. Finally, Figure 4 just shows the infected areas in red, which will make it easy for ground workers to locate the areas with infected oil palm trees.