Soil erosion in arable fields is intensified on irregular surfaces. Although machine and crop-row patterns following terrain contours reduce runoff and increase water infiltration, these contours are almost never parallel while machine operations always are. In this work, a method is presented to generate patterns of machine paths on sloping land and assess their susceptibility to water erosion. The approach comprises three main process-steps: (1) assembling a comprehensive set of reference tracks and introducing hybrid contour lines; (2) adjust these curved tracks into steerable parallel tracks for agricultural machines; and (3) assess water flow accumulation and susceptibility to soil loss of the corresponding pattern. The methods were implemented in open source software and applied on three case studies concerning sugarcane production in the São Paulo region in Brazil. Our results suggest that soil loss could be reduced fivefold by inserting one single change in the cropping pattern while estimated reductions up to 75% could be obtained by the model when compared to a human-suggested coverage pattern.