Regional scale projections of future change in wetland hydrologic regimes in Australia

Wetlands provide key habitat for many species but are threatened by climate change. However, quantitatively projecting climate change impacts on wetland hydrologic regimes is difficult due to the often remote nature of wetlands, leading to a scarcity of data on wetland inundation. In these contexts, remote sensing offers a large scale tool for periodic observations of wetland water extent. In addition, wetlands can display a diverse range of inundation regimes, driven by the different hydrological processes that contribute to inflows. For example, perennial wetlands might have ongoing groundwater contributions or high rainfall that keep water levels elevated throughout the year. But more variable systems such as floodplain wetlands might rely on overbank river flows, such that they might only be inundated every few years. This large range in hydrologic variability can make modelling wetland inundation a difficult task.

We used a timeseries dataset of wetland inundation extents, extracted from Landsat-derived water observations over 1988-2022, to classify the hydrologic regime of 34,890 wetlands in the state of Victoria, Australia. Wetlands were classified as permanent, seasonal, intermittent or episodic systems, which represent increasing variability in inundation. We then calibrated a series of conceptual hydrologic and Long Short-Term Memory (LSTM) neural network models to simulate wetland inundation, driven by climate inputs. Conceptual hydrologic models were able to reasonably simulate wetland inundation for permanent, seasonal and intermittent systems, but struggled in representing the more variable episodic wetlands. For episodic wetlands, LSTM models performed better than conceptual hydrologic models, but many that calibrated well over the historic period showed unrealistic sensitivity to changes in climate inputs.

We then applied a range of climate projections to wetlands models, to understand potential future shifts in wetland hydrologic regimes, based on a subset of 8,334 of the best performing models. Climate change projections substantially reduced the proportion of permanent and seasonal wetlands and increase the proportion of episodic wetlands in Victoria. Our results suggest the biggest risk is to permanent wetlands, where even under a moderate emissions scenario nearly two thirds of permanent wetlands could transition to seasonal or intermittent systems by 2065. “Rare” wetlands (with an average inundation frequency of less than once every 10 years) are predicted to increase eight-fold under a high emissions scenario by 2065. These results demonstrate the significant vulnerability of wetland hydrology to climate change, with potential major implications for wetland habitat for freshwater ecosystems.