Hydrological forecasting systems are used to support real time decision making in regulated river basins such as the Murray-Darling Basin. These are however standard in that they target climate variables (mostly rainfall) and river flow. Increasingly in operational decisions, a range of new indicators associated with river health needs to be considered in order to to comply with environmental regulations. Presently, these indicators are seldom considered when developing forecasting systems. Integrating these indicators into forecasting workflows nonetheless pose significant challenges:

• Lack of observed data complicates the use of traditional forecast verification techniques (e.g. CRPS).
• Indicators are generally based on thresholds (e.g. flooding a wetland) that are known to be challenging in a forecasting environment.
• Stastistical properties of these indicators are less studied, hence the development of post-processor and data assimilation methods will require mathematical developments.

These challenges and ensuing opportunities will be addressed in this presentation.