There has been a long need for routine forecasts across Australia of key hydrological variables – in particular for soil moisture. Operational forecasts of soil moisture would benefit business operations and planning decision-making in a range of sectors that includes agriculture, water management, water and energy utilities, and various levels of government from local to federal.
For several years the Bureau of Meteorology has been producing Australia-wide daily time-scale simulations of soil moisture using the Australian Water Resource Assessment model for landscapes (AWRA-L). These simulations are updated daily using the latest observations of AWAP (tmax, tmin, precipitation) as well as solar radiation, going back to 1911.
The Australian Community Climate and Earth System Simulator for seasonal scale forecasts – known as ACCESS-S – has recently been made operational at the Bureau of Meteorology (since early 2018). ACCESS-S is ideally placed to provide forecasts of the atmospheric forcing necessary to produce hydrological forecasts from models such as AWRA-L.
Here we present an evaluation of monthly-scale ensemble forecasts of soil moisture for Australia, out to 6-months, produced with ACCESS-S forcing. We examine the skill of the forecast at a national scale as well as focus on the Murray-Darling Basin. Metrics used to evaluate the skill include correlation, residual errors, anomaly persistence and ensemble reliability. This initial evaluation shows the skill of hydrological forecasts clearly has seasonal, spatial, and forecast lead-time dependencies.