A Semi-Automated Approach for Mapping Brassica Oilseeds.

In-season acreage estimates and mapping of canola production is of interest to stakeholders along the oilseed processing supply chain. This study compares the use of freely available medium resolution satellite imagery for inventorying the extent of canola production within the inland PNW. Two Landsat satellites are currently operational and offer a combined 8-day revisit capability. We compare a decision tree approach with a sophisticated machine learning approach, support vector machines, for isolating and identifying canola field during flowering. A semi-automated algorithm will provide geographic information content that is freely available in a web-mapping application that allows for spatial visualization and quantitative analysis. Future research will entail generalizing our approach to the northern Great Plains and other geographies.