PIEteR : a field specific bio-economic production model for decision support in sugar beet growing

To support decisions in sugar beet growing, a model, PIEteR, was developed. It simulates growth and production of the crop in a field specific way, making a tailor-made approach in decision taking possible.PIEteR is based on causal regression analysis of Dutch data of mostly experimental sugar beet fields. Its prototype, which only simulated root and sugar yields, was selected through a test on the performance of four models and extended with a number of parameters: sugar content, (K + Na) and α-amino-N contents, extractability index, tare content, operating receipts (a measure for gross returns), and amounts of leaves and nitrogen in leaves and crowns after harvest. Growth and production rates are corrected by a water balance module.The effects of plant density, nitrogen availability and harvest date were modelled and included in PIEteR, thus improving its applicability and the accuracy of the predictions. The profitability of resowing after a poor crop establishment was studied and critical plant densities were given for several combinations of sowing and resowing dates. The profitability of a delay in harvest depends to a large extent on the question whether the sugar yield has exceeded the sugar quota level or not. A method to allocate equipment costs to crops, making tactical decisions on sugar beet area possible, was described and included in PlEteR.Validation of PIEteR on a set of commercial and experimental sugar beet fields showed average prediction errors for root and sugar yields and financial returns per ha of 12 %, 13 % and 13%, respectively, and the variances accounted for were 52%, 51% and 50%, respectively. A major part of the prediction errors was caused by the prediction error of the sugar content and by the use of average regional instead of local yield and quality levels.Improvements on PIEteR can contribute to successful use in practical applications, such as: 1) decision support at farm and field level; 2) industrial campaign planning; 3) yield gap analysis; 4) analysis of new cropping techniques, new cultivars, etc. Further research on the prediction of local levels of output parameters seems to be the most important option for improvement of PIEteR, followed by addition of modules for weeds, diseases and pests, cultivars and preceding crops.