Probabilistic modeling of weed distributions for optimal treatment

Wild oat (Avena fatua L.) is an annual grass weed that is expensive to control in grain growing areas of western Canada and northern United States. Wild oat herbicides represent a significant cash cost of growing a cereal or oilseed crop. Crop yield loss due to wild oat interference is costly.; Optimizing herbicide rates based on the spatial distribution of weeds is possible because of global positioning technology. Exhaustive sampling would provide the exact locations of all wild oat for targeted herbicide application; however, the cost would be prohibitive. Instead, some reasonably spaced samples should be collected to identify weed locations and design an optimum herbicide treatment program. The optimum program will give the most profit, that includes revenue due to increased yield, herbicide cost, and sampling expenses. This research proposes a methodology to implement optimum herbicide rates with consideration of (1) wild oat in western Canadian crops, (2) locally varying herbicide treatments, (3) uncertainty in the predicted maps, and (4) economics in decision making.; Procedures for determining locally varying herbicide rates include kriging and simulation. The revenue from kriging and simulation plus two other weed prescription techniques is compared for different sampling designs. Of the sampling designs assessed, the simulated Square₇ with 98 sample locations in a square pattern for the Stony Plain field is the most profitable while at the Viking field, a simulated Grid ₁₀ design with 100 sampling locations in a rectangular pattern generates the most profit. These designs have the smallest number of sampling locations and the lowest sampling expenses. Applying herbicide with the locally varying rates results in {dollar}3440 more revenue compared to a no herbicide option. Locally varying rates based on simulation or kriging average {dollar}570 field−1 more revenue than the conventional approach of a label rate.; Locally varying herbicide rates provide economic and environmental advantages for consideration. The profit of locally varying rates is based on weed sampling. Costs of weed sampling may decline with secondary date, historical records and satellite imagery. Locally varying rates reduce environmental loading of herbicides by up to 40% compared to the conventional approach. Locally varying herbicide rates need to be integrated into weed control management programs.