This thesis investigates different aspects of weather, grass growth and statistical modelling. First, the accuracy of weather forecasts is assessed and improved using bias correction methods at twenty-five Irish locations for weather variables that influence grass growth. For the first time, soil temperature observations measured at six depths are verified and bias corrected. Next, the weather forecasts are included in an Irish grass growth model to investigate how the model accuracy is affected. The model predictions at an Irish farm are compared for weather observations and forecasts in multiple years, as well as to on-farm grass growth observations. These studies show that forecasts can be used in place of observations, and model predictions generally describe weekly grass growth accurately. Finally, grass growth modelling methods for experiments involving multiple species are developed for the analysis of a weed invasion study involving a large number of species. These developments include fitting novel random effects over multiple years to describe pairwise interactions between species parsimoniously and incorporating spatial planting pattern treatment into modelling methods.