Are vegetation index maps derived from sUAS-mounted multi-spectral sensors an accurate predictor of yield in potatoes?

Potatoes are an expensive crop to grow, and many inputs such as fertilizers and pesticides are required to ensure that the product is marketable. With advancements in GPS technology over the last three decades, one technology that has come to the forefront in farming is Precision Agriculture (PA). PA assumes that not all parts of a field are uniform, and that by tailoring management decisions to certain areas of a field, farmers can improve production and possibly reduce inputs. Adopting PA in the potato industry makes sense, as there is much to be gained in terms of increasing production as well as mitigating the environmental impacts of industrial farming. Two tools which fall under the umbrella of PA are the yield monitor and Small unmanned aerial systems (sUAS). sUAS have the ability to collect high resolution remotely sensed data for agriculture. When multi-spectral sensors are mounted to sUAS, algorithms (vegetation indices) can be applied to the data to assess spatial characteristics related to field health. Yield monitors are tools which measure the quantity or quality of production throughout the field. They are synced with GPS systems and can assist a farmer to identify which parts of a field produce higher or lower yields than others. The resultant yield data can be viewed as a report card of a field and used in informing management decisions. In this study the question was posed: are vegetation index maps derived from sUAS mounted multi-spectral sensors an accurate predictor of yield in potatoes? This study used sUAS to survey a 30 acre potato field in Indian River, PE, Canada four times throughout the growing season (Once before planting for elevation mapping – May 7th 2016; 39 days after planting – July 13th 2016; 67 days after planting – August 10th 2016; and 98 days after planting – September 10th 2016) . These dates equate to growth stages II, IV and late IV respectively (vegetative growth and tuber bulking) and were chosen at separate growth stages to determine which stage correlated greatest with yield. Growth stage I is considered pre emergence, while growth stage V is considered maturation where photosynthesis decreases and vines die off – these stages were not relevant for capturing imagery. It was expected that the areas of the field that appeared healthiest early in the growing season would produce the greatest yield at harvest. The collected sUAS data were correlated with yield harvest data to examine relationships between in-season field health maps and actual yield in lbs/acre. Correlations between sUAS data collected in July, and yield data collected at harvest, indicate that the farmer can get an idea of which areas of a field will produce the highest yield early in the growing season, and use this data to make informed management decisions on their farm. sUAS technology is rapidly evolving and adoption of these analytical tools on the farm will be more common as they become more affordable and user friendly. As the large range of sUAS collected data becomes more manageable, farmers and agronomists will be able to apply this technology on their crops and thereby, improve the ways they farm in order to maximize yields and minimize inputs.