The evaporative stress index (ESI) is defined as the standardized ratio between evapotranspiration (ET) and potential evapotranspiration (PET). Our previous analysis shows that the ESI derived from version 6 of the Bureau of Meteorology's land surface model (AWRA-L v6) is useful for monitoring fast developing or flash droughts over Australia. Here, we assess the ability of the Bureau's ACCESS-S1 prediction system to predict the ESI on lead times from 2-4 weeks, the typical time scale of flash drought development. Calibrated ACCESS-S1 hindcasts are used to compute the ESI and verified against observed estimates of the ESI computed from AWRA-L. The hindcasts are comprised of 23 years of 11 ensemble members and the calibration consists of a quantile-quantile matching technique from the native 60km output of ACCESS-S1 to the AWRA-L resolution of 5-km. The calibrated variables used are daily values of evapotranspiration, minimum and maximum temperature, surface wind speed, and water vapour partial pressure. PET is calculated using the Penman-Monteith equation, following the United Nations Food and Agriculture Organisation in their Irrigation and Drainage paper 56 (FAO56). Given the high correlation skill for the calibrated variables used here, especially in the first two weeks lead-time, it is anticipated that the ESI will also have high correlation skill when assessed against the AWRA-L derived ESI. This would offer avenues for forecasting ESI out to at least 2 weeks and earlier warning of flash drought occurrence probability.