Predicting the Thermal Regime Change of a Regulated Snowmelt River Using a Generalised Additive Model and Analogue Reference Streams

Large reservoirs can alter the natural thermal regime and result in the delivery of water with unseasonal water temperatures downstream. Establishing a detailed understanding how a river's thermal regime has changed from its natural state is a critical step required before mitigation options can be investigated. Our study focussed on one of Australia's most iconic rivers, the Snowy River which is regulated by multiple headwater regulating structures. We used mean daily pre-dam water temperature (1962-1967) data from the Snowy River to develop generalized additive models (GAMs) to predict the natural (daily mean, max and min) water temperatures in the Snowy River. Daily modelled natural water temperatures were compared with daily observed water temperatures in the regulated Snowy River and two nearby analogue reference streams. Using the GAMs, we identified a reduced thermal range (max and min) and water temperatures >= 2 degrees C warmer than natural between the cooler months of April and July for at least 50 % of the study period. Temperatures rarely (< 15 % of study period) exceeded 4 degrees C above natural. We show that both GAMs and appropriately selected reference streams are useful for assessing thermal regime change. Future work should focus on clearly identifying if small deviations from natural water temperatures during these periods results in significant ecological consequences in snowmelt streams.