Abstract: In order to overcome the shortcomings of existing methods in crop growth modeling, gene expression programming (GEP) was introduced and adopted in modeling and prediction of tomato leaf CO₂ exchange rate response to major environmental factors. A new model was established by GEP in this paper, then the performance of the proposed model was compared with two classical modeling methods-regression and neural network. The experimental results on three sets of data showed that, the GEP based model get the highest predictive accuracy and the best predictive time effect, at the same time, the complexity of the GEP based model was numerically similar to neural network. The study indicated that GEP is a good tool in crop modeling, and will be an important supplement for the existing methods.

Key words: crop growth model, gene expression programming, carbon dioxide exchange rate, environmental factors