Modifying the CROPGRO Safflower Model to Simulate Growth, Seed and Floret Yield under Field Conditions in Southwestern Germany

Abstract

The Decision Support System for Agrotechnology Transfer (DSSAT) currently provides a safflower model based on CROPGRO. The model was calibrated with the field data of one cultivar grown in New Mexico in 2013 and 2014. As it is rather new and has not yet been tested with other field data, it is important to evaluate the model in different environments. This study evaluated the CROPGRO safflower model for two different cultivars grown under field conditions in southwestern Germany. In addition, a new approach was added, enabling it to predict the yield of florets, which is of special interest, as these are used as a food colorant in Europe. The default model was evaluated with data from 2017 and 2018, obtained in a field trial in southwestern Germany with two cultivars, with row spacing of 12 and 33 cm and sowing densities of 40 and 75 plants m−2. As the default model was not well adapted to European conditions, model modifications were implemented in the species, ecotype, and cultivar files. With these modifications, observed variables such as leaf appearance over time were well predicted (RMSE: 4.76; d-index: 0.88), and simulations of the specific leaf area and leaf area index were greatly improved (RMSE: 24.14 and 0.82; d-index: 0.78 and 0.73). Simulations of the original New Mexico data set were also improved. The newly-added approach to predict floret yield was successfully integrated into the model. Over two years and two cultivars, floret yield was simulated with a RMSE of 97.24 and a d-index of 0.79. Overall, the extended model proved to be useful for simulating growth, floret yield, and yield of safflower in southwestern Germany.