Use of stochastically generated weather records with rangeland simulation models.

We compared long-term historical and stochastically generated weather records in terms of their statistical attributes and effects on herbage yield and runoff forecasts calculated from model simulations. The historical and synthetic air temperature and solar radiation records were in good agreement in terms of monthly means and extremes. The synthetic precipitation record failed to simulate extreme precipitation events which significantly reduced forecasted runoff values. Yield forecasts were similar using either historical or synthetic weather records.