A low-cost UAS-based Visible-Near Infrared imaging system helps detect latent stages of deadly fungal infection in Malaysian plantations
Malaysia produces up to 18 million tonnes of palm oil annually, with plantations covering more than 4 million hectares of land. Basal stem rot (BSR) or Ganoderma fungal infection is a catastrophic disease in oil palm plantations. Th is disease could reduce up to 80% oil palm productivity every year. The infection disrupts distribution of water and nutrients in the trees, resulting in appearance of specific foliar symptoms such as yellowing and necrosis leaves, unopened spears, small canopy and skirt-like shape of crown. The only effective way to prevent the spread of this disease is by removing the infected trees. Nowadays, Ganoderma detection in oil palm plantations is well defined by visually detecting lesions and fungus fruiting bodies (mushrooms) on the infected trunks. However, visual monitoring in the field is time consuming and expensive at early stages. The need for rapid, accurate and non-destructive method for detection of BSR is becoming crucial.
UAVs to the rescue
Quantitative remote sensing application requires a combination of high spatial and temporal resolution satellite sensors, which are very costly and most available combinations do not fit the ideal requirement. Alternatives based on manned airborne platforms could provide high spatial resolution and short revisit time, but their use is also limited by their high operational costs. Remote sensing sensors placed on Unmanned Aerial Systems (UAS), or Unmanned Aerial Vehicles (UAV) as they are more popularly known as, could provide low-cost approach to meet the critical requirements of spatial and temporal resolutions. A low-cost UAV-based Visible-Near Infrared (VIS-NIR) imaging system was developed to detect Ganoderma at latent stages of the disease, before it can be visibly detected.
Rotary six wing UAV (Hexacopter) and Crop- Cam fixed wing UAV were chosen to fly over the Universiti Putra Malaysia and Sime Darby oil palm plantations located in Banting, Selangor. The UAVs were mounted with Canon PowerShot SD 780 IS digital camera (visible sensor) and Tetracam ADC (Near Infrared sensor).
The flight altitude was selected depending on the study area to cover a camera field of view (FOV), and the desired spatial resolution. The average flight height was 100 m yielding 10-cm ground resolution imagery.
The raw images were stored on individual Visible image showing the affected oil palm at severity level of G2 and G3 malaysia.indd 60 5/7/2013 7:54:28 PM Geospatial World | May 2013 61 compact flash cards installed in the camera. Image triggering was activated from the ground control station when the UAV reached the desired study site. Photogrammetric techniques were required to register the framebased imagery to map coordinates. Each digital frame comes with position, altitude and timing information allowing for the generation of large mosaics. The main limitations encountered for these platforms were the endurance (15-30 min) and the low flight speed (30-60 km/h), limiting the productivity to 70 – 250 hectares per flight.
On top of that, foliar samples from healthy (G0), slightly damaged (G1), moderately damaged (G2) and heavily damaged (G3) oil palm trees (15-year-old) were collected for field and laboratory spectral measurements using hyperspectral handheld portable ASD field spectroradiometer (FieldSpec HandHeld).
Results from the samples’ spectral reflectance showed that the severities of level G2 and G3 were easily distinguished compared to level G1. Th is is also proven by the image captured by UAV-based VIS-NIR imaging. Special vegetation indices were calculated using the three bands of the multispectral camera and generally agreed, yielding a promising RMSE. As for the detection of level G1, thermal sensor will be used in the next campaign.
Conclusions and future work
UAV system enables data collection in small and inaccessible area, especially for crop monitoring. From the foregoing, it is proven that several sensors can be successfully flown aboard small-scale UAV platform. Application of UAV-based VIS-NIR imaging technique along with development of robust statistical models of discrimination has been proven to provide more efficient and timely management of BSR in oil palm plantation. Future work will include developing thermal infrared imaging capability and hyperspectral remote sensing missions.
(The author would like to thank Siti Hajar and Normahani of Sime Darby Research Sdn Bhd for the in-situ measurements. Field spectroradiometer measurements were assisted by Kay of TSKay Technology Sdn Bhd. Wong (Jurupro Sdn Bhd), Zailani, Hermi and Azmi are acknowledged for the technical support in operating the UAVs for the airborne campaigns.)