A multi-model assessment of global freshwater temperature under climate change

Water temperature is a key abiotic factor for determining the health, functioning and services provided by aquatic ecosystems. While analysis of existing observational data indicates that freshwaters are warming across the globe, the availability of long-term water temperature monitoring data remains limited in several regions of the world (e.g. Africa, South America, parts of Asia).

Models offer unique possibilities to explore the spatial and temporal dynamics of surface water temperature beyond what is possible through monitoring efforts alone. Water temperature models have been developed and applied for past and future conditions across various spatial scales, from individual lakes and streams to global applications. With a few notable exceptions, comparisons of water temperature simulations across different models are scarce. Aligning with ongoing activities within the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP), here we compare simulations of surface freshwater temperature from 11 surface water quality models (7 river and 4 lake models) that consistently used bias-corrected climate forcing from either CMIP5 (ISIMIP2) or CMIP6 (ISIMIP3).

Our multi-model ensemble suggests that surface water temperatures have risen substantially over the last 40 years, with global average annual water temperatures already 0.5 – 0.8 ºC warmer than at the turn of the century, and that warming will extend and intensify with future climate change throughout the 21^st century. For example, the multi-model ensemble suggests that global average annual water temperatures will rise by approximately +1 ºC under RCP2.6, +2 ºC under RCP4.5, +2.5 ºC under RCP6.0, +3 ºC under RCP7.0 and +4 ºC under RCP8.5 by the end of the century, compared to a historical reference period (1981-2000). Despite the consistent projections of warming, inter-model differences can be substantial. Furthermore, water temperature simulations are demonstrated to be highly sensitive to the meteorological forcing from different global climate models. To further unpack these aspects, in addition to evaluate model performance and better elucidate spatio-temporal patterns in water temperature projections, in this presentation we will display additional analysis on modelled output from three river water temperature models (CWatM-WQ, DynQual and WaterGap2) run using the state-of-the-art ISIMIP3 climatological forcing. To illustrate a potential societal impact of these projected water temperature rises, we quantified the associated reduction in usable capacity of existing thermoelectric powerplants globally.