SCAN: A New Decision-Support System for Sidedressed N Rate Recommendation in Corn.

Following meta-analyses of about 350 side-dressed N rate trials conducted in North America and numerical optimization, the SCAN (Soil, Crop, Atmosphere for N) decision-support system (DSS) has been tested for performance on 52 commercial farms in Quebec over the seasons 2013 to 2015 inclusive. SCAN use resulted in average benefits ranging from 25 to 49 CAD/ha depending on year as compared to farmers rate. Those benefits resulted from either applying less N fertiliser (as low as -72 kg N/ha compared to farmers rate) with no effect on yield for some farms, or by applying more N fertiliser (as high as +75 kg N/ha compared to farmers rate) for other farms that resulted in yield increases. The 2016 calibration of the SCAN algorithm achieved a RMSE of 9 kg N/ha for the economically optimal N rate (EONR) determination within a Quebec-Ontario database containing close to 300 site-years of N response trials. The RMSE with actual EONR was 38 kg N/ha for a database of 60 independent N response trials conducted in 2014 and 2015 in Quebec. This performance positions SCAN in the top few DSS for optimal N rate recommendations. The main originality of SCAN resides in its use of a fuzzy inference engine handling rules involving soil surface texture, rainfall quantity and distribution in time, previous crop, soil organic matter, vegetation status (if available) and economic ratio to achieve EONR recommendation. The system features real-time access to user-defined (or interpolated) cumulative rainfall and forecasts for the next 15 days by Environment Canada servers. SCAN is available as a web-mapper and the platform is currently being tested by a selected group of collaborators for a commercial release in spring 2017.