USING REMOTELY PILOTED AIRCRAFT (RPA) IMAGERY TO MAP THE PROFITABILITY OF COTTON CROPS

ABSTRACT

The goal of this paper was to apply a method for the delineation of vegetation homogeneous zones (HZs) to map the profitability of cotton ( Gossypium sp.) using NDVI (Normalized Difference Vegetation Index) data from RPA imagery. Two irrigated cotton crops, at the FAU (103 ha) and FRP farms (106 ha), located in the western region of Bahia State, Brazil, were studied. The NDVI images were classified into three HZs: “high”, “medium” and “low” plant vigor; using the k-means clustering method. In each HZ the yield was measured, and the profitability estimated. In FAU we found that the lower the HZ’s NDVI values, the lower its profitability, because there was profit in the “high vigor” HZ, of US$ 829.40 ha^-1, and loss in the “low vigor” HZ, of about US$ ‒1.256.09 ha^-1 (‒251.44% compared with the “high vigor”). In the FRP plantation, the lower the NDVI values, the lower the loss of the plantation, as there were losses in all HZs: from “high” to “low vigor” HZ, the percentage difference in profitability was ‒343.43%. Thus, the use of a low-cost modified near-infrared Canon S100 on an RPA enabled the mapping of crop profitability, aiding the search for the factors behind yield variability.

precision agriculture; management zones; clustering analysis; vegetation index