University of Reading virtual workshop on "Next generation challenges in energy‑climate modelling"

A brief personal account of a virtual workshop hosted by the energy meteorology group at the University of Reading, United Kingdom.

I guess the key idea was to explore how past and future weather estimates from meteorologists can be better incorporated into energy system analysis. The workshop emphasized historical weather data and projected climate‑changed weather data for long‑haul system analysis, rather than servicing the operational needs of the energy sector as it stands today (although that remains an important topic in energy meteorology).

In one sense, this is a story about “resolution”. The climate science general circulation models (GCM) can now provide high‑resolution information via reanalysis (of historical satellite data) and dynamic simulation (yielding future estimates). While simultaneously, energy system modelers have steadily increased the geographic, topological (mostly through improved grid representation), and temporal resolutions of their models. So the time has arrived whereby meteorologists can usefully deliver truck loads of weather data to energy system modeling teams.

Hence dialog on the transfer of information between the two communities is required, including semantics, scope, numerical formats, and metadata — these being the “challenges” in the workshop title. At one level, the workshop was simply about getting together and swapping notes.

Some themes that caught my attention follow.

Data semantics

There were several calls for a “dictionary” or “glossary” to better facilitate this cross‑disciplinary work. But there are indeed two nascent projects within the openmod community on data semantics — the Open Energy Ontology (OEO) and the openENTRANCE nomenclature workstream. Some links:

A greater involvement by the meteorology community in these projects would be invaluable.

Data vehicles

Although “data trucks” — a somewhat tongue‑in‑check term coined by Sofia Simoes (NOVA University of Lisbon) to describe the amount of data available from GCM modelers — were discussed, not much attention was in fact devoted to describing, cataloging, curating, hosting, and serving this data. In this regard, the proposed common European energy data space (see here) might have a considerable role to play.

Technical issues

Much discussion on the recalibration and downscaling of GCM outputs — including bias correction (post‑run alignment with observations in place of model tuning) and downscaling, both statistical (stand‑alone) and dynamical (employing numerical weather prediction techniques) — to meet the needs of energy system modelers.

Also discussion on cold, still, cloudy periods (Dunkelflaute in German) now being increasingly considered “extreme meteorological events”.

Energy analysts do not need predicted weather, just representative weather. The term “synthetic” was used in this context — whether that qualifier is useful or even descriptive remains to be seen.

The need for a better understanding how uncertainty propagates through the analysis chain was covered too. With a related topic being the role of energy model intercomparison studies (a technique long used by the GCM community). Ways of dealing with endogenous uncertainy were also discussed, including robust optimization.

Finally, it is possible that windfarm infrastructure can degrade the natural resource by harvesting momentum, beyond wake effects between neighboring farms (which themselves may extend 50 km or more in stable conditions).


Several familiar faces from the openmod community were present, some even having made it on to the organizing committee. And although data access and data licensing were explicitly off‑the‑table (a perfectly reasonable constraint), the idea of open modeling generally gained quite some airplay.

Noting here too that the Pan European Climate Database, used to inform official future power adequacy analysis in Europe, is entirely private and opaque.

Finally, I was a little puzzled to learn of “hourly capacity factors” being used in some analysis. The concept of capacity factor becomes difficult to define for multi‑product processes (like variable ratio cogeneration) and for processes whose real‑time capacities vary markedly from nameplate (open cycle gas turbines, for example). Better, in my view, to use the terms “conversion ratio” and “instantaneous capacity” and perhaps adopt exergy accounting for the purposes of plant and technology comparisons. Even photovoltaic performance depends on thermal influences, degrading significantly in still air and high ambient temperatures.

Thanks to the organizers for the welcoming, timely, productive, and useful discussions.


I participate on the steering committee for the Open Energy Ontology. I am not otherwise involved in the projects discussed in this posting.


@robbie.morrison, great summary! I also attended this insightful workshop and noted down some “needs” as a wrap-up primarily for myself, but I am willing to share!

  • need to look at simultaneous extreme events (low wind, low temp, no sun)
  • need clearer definitions of what characterises stressful events
  • need recommendations how many years of data (or which representative
    year) are needed for robust results (obvsly depending on application, look at spectral properties of time series?)
  • need detailed/comprehensive overview how (historic) reanalysis datasets differ (strengths and weaknesses) and impact the power system
  • need for detailed hydro inflow modelling
  • need to improve cloud cover in GCMs
  • need to investigate whether one can extract features from time-series which are decisive for
    investment (machine learning)
  • need for systematic feasibility checks of optimisation results with more physically accurate models and heuristics to fix infeasibilities
  • need to understand effect of changes in spatial correlation of wind in energy models
  • need to understand effect of changes in frequency of extreme events in energy models
  • need for common vocabulary; “cultivate people who are bilingual, who can move between fields of energy and climate” (Brayshaw)
  • need for energy-climate summer school for learning and networking

Some additional information:
We we will be working on including meteorology terms in the Open Energy Ontology within our LOD-GEOSS project. The climate modelling community already has a glossary, see The terms use there are not always in line on what we tend to use in energy systems modelling (e.g. global radiation vs. surface downsward flux).

For data there is already huge archives of climate simulations, which we already started to use in energy systems modelling, e.g. The use of the data is challenging, on hand through the large amount of data sets and the data data need some more pretreatment before you can use it in an energy systems modell (e.g. intelligent interpolation of solar radiation data which takes into account the diuarnal cycle and solar geometry troughout the day).