In this study we hypothesized that soil water content and topography influence the durum wheat final yield and grain protein content.  The study was carried out on a 10 ha field with rolling landscape in Southern Italy, in a two years wheat (Triticum Durum, Desf.) monoculture (2001/02 – 2002/03). The crop was managed using a conventional agriculture approach commonly used in the study area. The CERES-wheat model was executed for 56 years on 35 locations previously selected, within the field. A Principal Component Analysis (PCA) was carried out to identify the main factors responsible of the yield and grain quality variability. PCA results showed that for both years, most of the field variability was explained by the soil water content variability, which is highly affected by the position in the landscape.

 The model performed well when compared to the yield measured in the two year-study.  The model was able to identify spatially variable zone that behaved consistently throughout the years. The stable zones identified can be potentially managed differently (different cultivar, plant population or nitrogen rates). The measured and simulated yield ranged from 1 t ha^-1 to 4.5 t ha^-1.