It is getting near to Christmas, which means the decorations are coming out, the nights are drawing in, and even Scrooge has conceded that it’s probably time to put the heating on. In winter the UK has a significantly higher electricity demand than it does in summer, due to our increased need for heating and lighting. In past decades meeting this increased demand was relatively easy, as our generation was supplied from a more “traditional” fleet of coal, gas, and oil-fired power stations. These are relatively easy to control when they are on or off and can turn on very rapidly to meet a sudden peak in demand if the EastEnders Christmas special comes on. In order to meet the government’s proposed carbon reduction targets, the type of generators that are used to meet demand are rapidly changing. A transition is underway to more renewable sources such as wind, solar and hydro power. The operation of the renewable generation depends on the present weather conditions, as opposed to the TV scheduling (although perhaps online streaming services are helping the national grid by reducing TV viewership). It is therefore important to understand how the weather sensitivity of the current power system is likely to change when more renewable generation is introduced.
Recent research from the energy-meteorology research group has investigated various aspects of this problem using weather-dependent models of demand, wind power generation and solar power generation (available here if you’d like to have a play with the data). A problem with using measured demand and renewable generation data for this type of analysis is that there are large trends in the data due to the rapid expansion of renewable capacity, or due to changes in economic factors, human behaviour and the energy efficiency of every-day products. We therefore build reconstructions of a fixed power system configuration (i.e. taking the amount of installed renewable generation and level of demand in 2017) and then use a reanalysis dataset to reconstruct what the present day power system would have looked like with the last 40 years of weather. With these long datasets we can look at the meteorological drivers of power systems [1,2] and investigate thoroughly the changing weather conditions associated with extreme power system events as we put more renewables on the system [3,4,5]. Some examples of this are discussed below.
Figure 1: Anomaly composites of the mean of the ten most extreme peak loads from the 1980–2015 winter-mean. These are given for for 2 m temperature (first column) and 10 m wind speed (2nd column) for a UK power system with no renewables (NO-WIND) and one with 45GW of installed wind power. Mean-sea-level-pressure contours for the events are overlaid in black with a 4 hPa interval. The thick contour represents 1016 hPa. The ’H’ represents the location of the centre of the region of high pressure. Adapted from 
Figure 1 shows composites of the ten highest demand events experienced in the UK from 1980-2016 (using the synthetic time series discussed previously). We see that if there is no renewable generation installed on the system then peak demand events are associated with very cold near-surface temperatures and moderately low winds. The key ingredient here is the location of a high pressure system, with a strong pressure gradient over the UK bringing cold continental air from central Europe. If we were to look forward to planned future power systems including 45GW of wind power generation (around twice the amount currently installed) then the weather conditions causing largest system stress are likely to change. Figure 1 shows peak load events (i.e. the demand minus all available wind power generation) are still associated with high pressure, but there is a shift to high pressure events centered over the UK now being most important. These events lead to both a reduction in the magnitude of the temperature anomaly and an increase in the wind speed anomaly over the UK.
It is relatively easy to forecast the broad characteristics of energy demand, as we all tend to go to work and take holiday at certain times of the day/year. It is however challenging to forecast the type of events shown in Figure 1 at timescales greater than a few days. If information was available at timescales of 1 week to 1 month ahead (sub-seasonal timescales) then this would allow for planning decisions to be made about where power is going to be sourced from to meet high demand events . Studies have shown that at these longer lead times, there is relatively more skill at forecasting large-scale weather conditions, rather than local anomalies at the surface. Because of this, some current work as part of the S2S4E project has investigated the potential for two types of large scale patterns to be used to provide information at sub-seasonal timescales for energy system resource planning.
The first pattern-based method is weather regimes. These are patterns defined based on large scale meteorological conditions (daily anomalies of 500hPa geopotential height). Each day during the extended winter season can be assigned into one of the patterns: the positive and negative phase of the North Atlantic Oscillation (NAO+, NAO-), Scandinavian Blocking (ScBL) and the Atlantic Ridge (AR). Rather than clustering on a large-scale field, the second method, newly developed for the project is called Targeted Circulation Types. This assigns each day into one of four patterns based on national energy data (using 28 countries across Europe). These patterns are: Blocked (BL), Zonal (ZL), a European High (EuHi) and the European Trough (EuTr).
Figure 2: Probability of each country’s demand being in the upper climatological tercile of demand in each weather regime (top) and Targeted Circulation Type (bottom). The TCT patterns used are constructed from normalised demand. See  for more details.
Each of these 8 weather patterns can be associated with a set of surface meteorological impacts (e.g., a set of demand, wind power or solar power anomalies). Figure 2 shows the probability of demand being in the upper tercile (top 30%) while each of the patterns has occurred (again using our reanalysis-based power system data to get a longer record for analysis). This shows that particular patterns are more likely to be associated to high demand events, especially those associated with European Blocking. Some patterns do not have a particularly strong link to European demand (e.g. the Atlantic Ridge). As well as this, the link to the impacted system is not as strong using weather regimes as using the Targeted Circulation Types. Current work is investigating if there is a tradeoff in the levels of predictability associated with the two different types of patterns. For further details see  and the S2S4E website.
 Bloomfield, H. C., Brayshaw, D. J. and Charlton-Perez, A. (2019) Characterising the winter meteorological drivers of the European electricity system using Targeted Circulation Types. Meteorological Applications. (2019) https://doi.org/10.1002/met.1858
 van der Wiel, K., Bloomfield, H., Lee, R. W., Stoop, L., Blackport, R., Screen, J. and Selten, F. M. (2019) The influence of weather regimes on European renewable energy production and demand. Environmental Research Letters, 14. 094010. https://doi.org/10.1088/1748-9326/ab38d3
 Bloomfield, H., Brayshaw, D. J., Shaffrey, L., Coker, P. J. and Thornton, H. E. (2018) The changing sensitivity of power systems to meteorological drivers: a case study of Great Britain. Environmental Research Letters, 13 (5). 054028. https://doi.org/10.1088/1748-9326/aabff9
 Bloomfield, H. C., Brayshaw, D. J., Shaffrey, L. C., Coker, P. J. and Thornton, H. E. (2016) Quantifying the increasing sensitivity of power systems to climate variability. Environmental Research Letters, 11 (12). 124025. https://doi.org/10.1088/1748-9326/aabff9
 Drew, D. R., Coker, P. J., Bloomfield, H. C., Brayshaw, D. J., Barlow, J. F. and Richards, A. (2019) Sunny windy Sundays. Renewable Energy, 138. pp. 870-875. https://doi.org/10.1016/j.renene.2019.02.029
 White, C. J., Carlsen, H., Robertson, A. W., Klein, R. J., Lazo, J. K., Kumar, A., … & Bharwani, S. (2017). Potential applications of subseasonal‐to‐seasonal (S2S) predictions. Meteorological applications, 24(3), 315-325. https://doi.org/10.1002/met.1654