This research deals with the control of working width and fertilization quality when using centrifugal fertilizer spreader. From such spinning disc spreader, the stationary spatial distribution of fertilizers is described, in the cylindrical coordinate system, by radius and angle (average and standard deviation). Because of the variability of fertilizer's physical characteristics, the landing position and drop angle of particles typically exhibit a strong variation. Indeed, the latter has been demonstrated by both experimental investigations, such as using the Cemib device at Irstea, and numerical modeling. In this work, we address the construction and identification of a prediction model for controlling the spatial distribution of fertilizers from spinning disc spreader. The proposed model is based on (i) the operating parameters of machine and (ii) the physical characteristics of fertilizer particles. To this aim, we first perform the characterization of particle size and shape by using image analysis. The morphology of particle is characterized by various shape parameters, scaled to values between 0 and 1, describing the changes with respect to a reference geometry (circle or ellipse for instance). The image-based method provides a quantitative, representative and reproducible description of the fertilizer morphology, hence allowing for the study of particle aerodynamic properties in the ballistic flight and the construction of a suitable prediction model for the drag coefficient of fertilizers. This model is subsequently identified through a inverse problem involving experimental data of Cemib. The model also helps calibrating which fertilizer's physical characteristics best describe the spreading pattern of centrifugal spreader. The impact of such prediction on transverse distributions, working widths and fertilization quality (CV curves) is finally characterized by using sensitivity analysis.