S2S4E – a decision support tool to visualize climate variability weeks and months ahead

14:00 Wednesday 29 May

OC138

Room S2

 

Isadora Christel (Spain) 1; Albert Soret (Spain) 1; Fernando Cucchietti (Spain) 1; Luz Calvo (Spain) 1; Marta Terrado (Spain) 1; Dragana Bojovic (Spain) 1; Llorenç Lledó (Spain) 1; Andrea Manrique (Spain) 1

1 - Barcelona Supercomputing Center

From June 2019, the EU-funded project S2S4E will run an online operational Decision Support Tool (DST) for energy decision makers in solar, wind and hydropower production and energy demand. This DST is an interactive climate service interface where energy users can explore probabilistic predictions of essential climate variables and energy indicators for weeks and months ahead.

Renewable energy growth is central to the transition to less carbon-intensive economies. However, both renewable energy supply and energy demand are strongly influenced by climate variability, which is a major barrier to renewable energy integration in electricity networks. Current energy practices use past conditions – climatology- to inform on expected future conditions, but this doesn’t take into account climate change. Top-notch advances on climate predictions at sub-seasonal and seasonal time scales have the potential to inform decision-making processes in the energy sector with climate conditions in the coming weeks and months. However, a challenge for the creation of a successful DST is the complexity of probabilistic forecasts, and how to communicate forecasts uncertainty to decision makers.

To solve probability and uncertainty communication, the DST map interface included three main features: color, size and a subtractive filter. The prediction in each grid point is shown as a glyph (circle) in a map. The color of the glyph conveys the most probable of three categories (above, below or on normal) in the future trend of a selected variable. We selected a colorblind-safe scale to facilitate fast detection of spatial trends. Alongside the most likely category, users need to visualize its probability, as only categories predicted with high probability might trigger a decision. Hence, users can set a probability threshold that encodes a larger glyph’s size enabling fast visual detection. We included a subtractive filter so the user can remove the predictions that are worst than using climatology or fall below their minimum required quality level. All this enables the user to explore and detect their area of interest. When the user selects a particular glyph, an on-demand display panel provides additional information including, categories probability, forecast quality, forecasts trend over time and current forecast vs. historical data.

We arrived at our solution after a multi-cycle user-centered design process, validated with user interviews and eye-tracking experiments. Results show that users identify clearly and quickly regions with increasing/decreasing trends. We observed high task accuracy and understanding, with users mentioning that location of items and their interactivity was intuitive.