Assessing the potential of images from unmanned aerial vehicles (UAV) to support herbicide patch spraying in maize

The capability of images acquired from unmanned aerial vehicles (UAVs), with multispectral cameras, to detect weed patches, should be tested in operational situations of site-specific weed management. In this regard, different post-emergence herbicide application strategies were evaluated on a total of four silage maize fields in Central Italy. The treatments compared were uniform blanket application, patch spraying according to the application map and an untreated control (the latter treatment only in the second year). Images were acquired a few weeks after maize emergence and were processed into application (i.e. prescription) maps. The accuracy of prescription maps was evaluated on the basis of ground-truth data. Maize and weed biomass data collected at end of the growing season were used to assess differences among the herbicide application strategies. Results showed no differences between uniform and patch spraying treatments for silage maize biomass in the two fields of the first year. In the second year, maize biomass differences were observed between the untreated control and the other two treatments. In terms of weed biomass there were no differences among treatments, for three out of four fields. The use of UAV image data captured early post-emergence in maize lead to a decrease in the use of herbicide without negative consequences in terms of crop yield and, at the same time, increased the silage biomass production as compared to non-treated area. The saving of herbicide calculated in terms of untreated area ranged between 14 and 39.2 % for patch spraying as compared to a uniform blanket application, and saved between 16 and 45 a,notsign ha(-1).