Evaluation of three grass growth models to predict grass growth in Ireland

SUMMARY Grass growth in temperate regions is highly seasonal and difficult to predict. A model that can predict grass growth from week to week would offer a valuable management and budgeting tool for grassland farmers. Many grass growth models have been developed, varying from simple empirical to complex mechanistic models. Three published grass growth models developed for perennial ryegrass swards in temperate climates were selected for evaluation: Johnson & Thornley (1983) (J&T model), Jouven et al. (2006) (J model) and Brereton et al. (1996) (B model). The models were evaluated using meteorological data and grass growth data from Teagasc Moorepark as a framework for further refinement for Irish conditions. The accuracy of prediction by the models was assessed using root mean square error (RMSE) and mean square prediction error (MSPE). The J&T model over-predicted grass growth in all 5 years examined and predicted a high primary grass growth peak, while the J and B models predicted grass growth closer to that measured. Overall, the J model had the smallest RMSE in 3 of the 5 years and the B model in 2 of the 5 years. In spring (February–April), the B model had the lowest RMSE and MSPE. In mid-season (April–August), the B model had the closest prediction to measured data (lowest RMSE), while in autumn (August–October) the J model had the closest prediction. The models with the greatest potential for grass growth prediction in Ireland, albeit with some modifications, are the J model and the B model.