The SGS Pasture Model was used to estimate daily total soil water content (mm) for the period 1998–2001 (using experimental data) and from July 1905 to June 2005 (using a 100-year daily climate file). For both model simulations, the correlation coefficient for actual and predicted data was >0.86. Four sites were examined, three in the winter-rainfall zone (Albany, Western Australia; Hamilton, Victoria; and Wagga Wagga, New South Wales) and one in the predominantly summer-rainfall zone (Barraba, New South Wales) of temperate Australia. Trends were detected in long-term rainfall data for lower mean annual rainfall at the Albany site since the mid 1950s and at the Hamilton site since the mid 1970s.

Analyses of days in which herbage accumulation was likely to be limited only by water allowed a minimum soil water content to be defined; below this value, soil water was in deficit. A drought period was defined as occurring after >50 consecutive days of predicted soil water deficit. For the four sites, six categories of drought duration were described, ranging from >50–120 days (seasonal drought) to >364 days (exceptional drought).

Numbers of predicted droughts were 90, 90, 65, and 60, respectively, at Albany, Hamilton, Wagga Wagga, and Barraba sites. At the three winter-rainfall sites, most predicted droughts occurred in a regular pattern and were an extension of a seasonal drought, but at the Barraba site they were of a longer duration and irregular in occurrence.