28 IREC Farmers' Newsletter No. 198 — Spring 2017 There are a number of advantages of using drones over other technologies. Drones are cheaper, target specific fields, enable higher resolution images to be taken, and take images in a more timely fashion. When mapping a field, drones fly between 30 and 150 metres above the field with varying flight speeds. The lower the flying height, the slower the drone should go (5–25 metres per second ). Drones fly in overlapping transects across the field to get a complete picture. Images taken by the sensors are then “stitched” together to form a high resolution and comprehensive map of the field. It is important that the sensor camera should face straight down. Most drones use a gimbal to adjust for pitch. During flights, operators need to aim for consistent light conditions, i.e. either clear and sunny or overcast. Changing light conditions, as with patchy and scattered clouds, make it difficult to obtain a clear image. Getting the most from drone-gathered data Following Dr Sulik was a presentation from Dr John Hornbuckle, Associate Professor with Deakin University, based at Griffith. Dr Hornbuckle covered the foundations of imaging and the significance of the light/colour spectrum on the images derived from drone-mounted sensors. Like yield maps, it’s all very well to collect the data and make images, but what they can tell us? The type of sensor (camera) used on a drone will be determined by what information the grower requires. All plants reflect different parts (wavelengths) of the light/colour spectrum and by different amounts. The amount of colour reflected is based upon plant health, as well as water and nutrition status. There are also sensors that measure temperature, which can be an indicator of plant health and stress. Dr Hornbuckle discussed four different types of sensors: l true colour (RGB – red green blue) l multispectral l longwave infrared (IR) or thermal l hyperspectral. True colour (RGB) sensors simply produce a colour photograph. Entry level drones usually come with this sensor. In agriculture RGB can be useful for identifying crop variability and pinpointing where samples should be taken. For more accurate plant health analysis, more expensive sensors are required. Multispectral sensors capture specific colour wavebands which are not visible to the human eye. These are professional sensors for agricultural remote sensing. While Normalized Difference Vegetation Index (NDVI) is the primary output of multispectral sensors, output may also be Normalized Difference Red Edge (NDRE), Visible Atmospherically Resistant Index (VARI) and Simplified Canopy Chlorophyll Content Index (SCCCI). The NDRE images indicate variations in nitrogen so these are most commonly used in nitrogen management of crops, such as variable rate top dressing of rice and winter cereals and identifying nitrogen deficiencies in cotton. The SCCCI images show chlorophyll content of leaves. Images appear "greenest" when leaves are approaching physiological maturity and so chlorophyll is at its maximum. These images can be used to identify disease and certain fertiliser deficiencies. Longwave IR or thermal sensors measure the temperature and indicate water or heat stress of plants. They are thus useful in assisting with irrigation management. While these sensors have good qualitative capabilities they do have limitations which include difficultly in stitching images together and difficulty in comparing data/maps taken at different times due to temperature variabilities. Hyperspectral sensors capture a large number of wavebands simultaneously and are used mainly for research purposes and for the classification of plants. Dr Hornbuckle stressed that the real benefits of remotely sensed data collected by drones comes from putting data into a format that makes it easy for irrigators to make decisions around fertiliser and irrigation management, so they can modify practices and improve management. Practical examples of drone technology Dr Sulik presented several case studies demonstrating the practical use of drones in farm management. Citrus tristeza virus, which causes chlorosis and dieback, is the most economically damaging disease to citrus in the world. The virus is transmitted by the brown citrus aphid and has killed over 80 million trees in South Africa, Argentina, Brazil and the USA. Tree decline and total dieback can take years after the first symptoms of chlorotic leaves first appear. Drones fitted with NDVI and NDRE sensors have been used to identify unhealthy trees in orchards. NDVI was used as a general indicator of canopy density, while the NDRE sensor, which is sensitive to chlorophyll content in leaves, and may indicate variability in leaf area. In another case study, Dr Sulik showed how multispectral composite has been used to identify wheat streak mosaic virus infection in wheat. Wheat streak mosaic virus is carried by a mite that lives in volunteer wheat and other grass weeds that act as a green bridge between successive wheat seasons. Early detection is the key to identifying and managing the virus. Infected wheat is not detected by RGB or NDVI sensors. Multispectral sensors have also been used to identify problems in a plum orchard. Chlorophyll maps were produced using data collected from the sensors. Bicarbonate build-up from use of groundwater caused the trees to stress, which could be identified by variations in tree health as indicated by the chlorophyll maps. Early detection not only saved money but also saw trees return to full health shortly after switching back to a surface water source for irrigation.