Balancing uncertainty when assessing climate change risk in large river basins: the case of the Murray Darling basin, Australia

Climate change threatens water resources from local to global scales. However, there are significant challenges in assessing climate risk for large river basins, especially those with multiple jurisdictions and competing management objectives. Traditional methods follow a top-down approach, where the impacts of climate projections by climate models are simulated using hydrological and water resource models. While these methods can provide a detailed snapshot of how rivers are impacted under a small number of projected future climates, their computational burden, and challenges in linking water resource models owned by different jurisdictions mean it is difficult to robustly explore the implications of aleatory (from hydroclimate variability) and epistemic (from hydroclimate change) uncertainty. Unlike top-down approaches, bottom-up approaches can be used to better understand vulnerability under a range of possible future climate. Bottom-up approaches begin with a sensitivity analysis of important management objectives to multiple hydroclimate stressors. Unfortunately, bottom-up approaches are constrained when using complex system models in large river basins, as their methodologies typically require many times more simulations than top-down approaches.

The Murray Darling basin (MDB) is Australia’s most significant river basin. Irrigation in the basin supports over $30 billion (AUD) in agriculture and livelihoods for the 2.4 million residents. The MDB has significant environmental values, with RAMSAR wetlands, many endemic and threatened species, and it is the traditional land of over 50 first nations groups. We assessed the impacts of climate change on basin-wide inflows and key indicator sites using both top-down and bottom-up approaches. We stochastically generated multiple sequences of future hydroclimate conditions, which helps separate the influence of climate variability from climate change. We deliberately traded-off detail in our assessment by deriving simple functional relationships between sub-basin inflows and 21 key indicator sites using existing scenarios from the complex jurisdictional water resource models. This allowed us to assess far more replicates of stochastic data, more climate scenarios, and conduct a more rigorous stress test within the bottom-up framework than would normally be permitted using complex models.

The top-down approach provides a scenario-based assessment of likely conditions for water resources in the MDB, and spatially coherent projections of future inflows and river management metrics. The bottom-up approach provides more insight into spatial differences in sensitivity across the river catchments that make up the MDB, and can be used to both augment and help interpret outcomes from the top-down approach. The bottom-up approach also yields important thresholds in hydroclimate conditions which compromise basin-wide objectives (assessed through flow at the Murray River mouth which prevents the important lower lake system from becoming too saline). We consider top-down and bottom-up approaches to be complementary in assessing and adapting river systems to the impacts of climate change.

The simple methods used here are complementary with other more detailed impact models. The ease of undertaking simulations and computational efficiency means simple methods can filter down the range of possible conditions or stressors that contribute to uncertainty, allowing a more targeted set of simulations to be undertaken using detailed, but costly, water resource models.