Topographical and Hydrological Determinants of Crop Yield in Farm Fields Across Alberta, Canada.

Crop yield is often improved by using improved cultivars or increasing the amount of crop inputs (in-crop nitrogen, plant growth regulators and foliar fungicides). In dryland agriculture in western Canada, these crop inputs may not always result in large yield differences over the landscape. On the other hand, topographical and hydrological variation determines site-specific availability of soil moisture within a farm field, thereby influencing the efficacy of these crop inputs on crop growth and yield. This presentation examines the relationship between elevation-derived topographical and hydrological variables and yield monitor data of wheat, barley and canola taken from fields of ten farms in Alberta, Canada. This relationship is determined using geostatistical methods capable of accounting for spatial correlation inherent in all variables. Topographical variables (slope and aspect) and hydrological variables (flow direction, basins, and flow accumulation) are derived with ESRI’s ArcGIS 10.3.  We use a machine-learning model (Random Forest) to assess the relative importance of agronomic (urea-ammonium nitrate, UAN) and non-agronomic (topographical and hydrological) factors as determinants of crop yield. The results show that UAN is the least important determinant. Thus, topographical and hydrological variables may be more important than the agronomic factors such as UAN in determining the crop yield on a field scale.