Author: Francois  Clayer, Ph.D. Researcher (Norwegian Institute for Water Research (NIVA))

Designation: 

Description:
Seasonal climate forecasts provide opportunities to predict water quality (WQ) in response to future harmful climate events. When dealing with water bodies, some predictability can originate from legacy effects. To ensure successful implementation of seasonal forecasts in water management, a rigorous assessment of the source of the predictability (seasonal climate forecasts vs legacy) needs to be performed. Within the WATExR project, we set up workflows for WQ seasonal predictions for three reservoirs and a lake in Australia and Europe. Hydrological models (GR4J/GR6J/SimplyQ) were coupled to hydrodynamic lake models (GLM/GOTM). Historical skills were assessed by comparing seasonal hindcasts to pseudo-observations generated by our models forced by two meteorological datasets: the seasonal forecasting SEAS5 data from ECMWF and the global reanalysis ERA5 data, respectively. The predictive skills of the SEAS5 climate data was also quantified. The performance of our WQ hindcasts was generally low. Nevertheless, for some specific variables, seasons and terciles, WQ hindcasts and SEAS5 data had significant predictive skills. However, skills between WQ hindcasts and SEAS5 data did not always agree which raised questions on the source of the predictability. We thus performed a set of experiments where forcing data over the warm-up, transition and target-season periods were alternatively replaced by random data. These experiments enabled us to attribute most of the forecast skills to either legacy or seasonal climate forecasts, depending on the variable.

Category: Scientific Program Abstract > Special Sessions > SS46 Forecasting is the future: advancing methods and applications of ecological forecasting in aquatic ecosystems