Improving our capacity to manage hypoxic blackwater events in lowland rivers: The Blackwater Risk Assessment Tool

Blackwater events occur when returning floodwater contains elevated levels of dissolved organic carbon. Normally the export of carbon from floodplains to a river channel is a beneficial process, providing sustenance to lowland river ecosystems. However if the rate of oxygen consumption during decomposition of the organic carbon is faster than it can be replenished from the atmosphere, this can cause hypoxia, with catastrophic short-term consequences for aquatic fauna. Because river regulation has disconnected many floodplains from their source rivers, floodplain inundation is increasingly being actively managed specifically for the benefit of the environment. However, the risk of hypoxic blackwater generation must be considered during such managed floods. This paper describes a new tool for predicting dissolved oxygen and dissolved organic carbon in flood return water, using minimal data inputs. The Blackwater Risk Assessment Tool (BRAT) is based on the same conceptual framework underpinning an earlier, geographically specific, blackwater model. However BRAT differs in a number of significant ways from the earlier model. The BRAT is based on a generic, rather than site-specific, inundation model; and users have the option to use actual hydrographs if these are available. Furthermore, most of the algorithms used to model blackwater generation processes have been refined, reflecting new knowledge generated since the original model was constructed. This includes changes to floodplain litter load estimation methods, algorithms for the temperature dependence of dissolved organic carbon leaching and biotic uptake, and re-aeration rate calculation. Finally, unlike the earlier model, all the algorithms and constants in BRAT are readily accessible so that the model can be customized to suit floodplain ecosystems other than the ones for which it was developed. The model was validated using flood events in Koondrook–Perricoota and Gunbower forests, Australia. Model outputs were in good agreement with observed dissolved oxygen during these flood events with root mean square variation as low as 1.2mgO2L−1.