UK Carbon Intensity Dashboard

Analysis of UK National Grid carbon intensity data

The UK electricity grid’s carbon intensity — measured in grams of CO₂ per kilowatt-hour (gCO₂/kWh) — varies continuously depending on which energy sources are generating power at any given moment. Renewables like wind and solar push the value down; gas and coal push it up. The UK National Grid publishes both a real-time measurement and a rolling forecast for every 30-minute period.

This dashboard analyses that data across three dimensions: how clean the grid is, how predictable it is, and when — during the day and week — intensity peaks or dips.

Grid Composition Over Time

Each band shows what fraction of 30-minute periods in that week fell into each carbon intensity category — the bands sum to 100%. A growing green base (very low + low) signals a cleaner grid, typically driven by high renewable output. Winter weeks tend to have a wider orange/red band due to higher heating demand and less solar availability.

Weekly Green Energy Trend

“Green hours” counts the share of 30-minute slots where intensity fell into the very low or low categories (roughly below 100 gCO₂/kWh). The 4-week rolling average smooths out week-to-week noise caused by weather variability, making the underlying structural trend easier to read.

Daily Intensity Range

Each day’s carbon intensity does not stay constant — it fluctuates across the 48 half-hour periods. This chart shows that spread: the dark line is the daily average, the inner band covers the interquartile range (p25–p75, where 50% of readings fell), and the outer band extends to the daily minimum and maximum. A wide band indicates a volatile day where intensity swung significantly, often caused by variable wind output or large demand shifts.

Forecast vs Actual

Each point is one day: its x-position is the average actual intensity and its y-position is the average forecast. Points on the dashed diagonal represent perfect predictions. Points above the line mean the forecast overestimated (predicted a dirtier grid than reality); points below mean it underestimated. A systematic tilt in one direction across a whole intensity category reveals structural bias in the forecasting model for that regime.

Forecast Accuracy Calendar

MAE (Mean Absolute Error) measures the average absolute difference between forecast and actual values across all 30-minute periods in a day, in gCO₂/kWh. An MAE of 10 means the forecast was off by ~10 gCO₂/kWh on average that day. Darker cells — higher MAE — tend to cluster around weather events, unexpected demand changes, or periods of rapid renewable ramp-up or ramp-down. Grey cells indicate days with no data in the dataset.

Forecast Accuracy Distribution by Intensity Level

Each box shows the distribution of daily MAE values for days where a given intensity category was dominant. High-intensity days rely more on gas and coal dispatch, which is less predictable than wind and solar — hence the larger errors. The height of the box indicates consistency: a tall box means forecast accuracy itself was variable on those days, not just poor on average.

Weekly Forecast Accuracy

MAE and MAPE (Mean Absolute Percentage Error) at weekly resolution show whether forecasting quality is stable or drifting over time. MAPE normalizes error as a percentage of the actual value, making it comparable across weeks with different average intensity levels: a week with high MAE but low MAPE likely had a high-intensity baseline; a week with low MAE but high MAPE suggests the grid was clean but still hard to predict.

Carbon Intensity by Period

The color shows average actual carbon intensity (gCO₂/kWh) for each combination of time-of-day and day type. Periods are defined as: morning (06:00–12:00), afternoon (12:00–19:00), and night (19:00–06:00). The annotation inside each cell shows MAPE — the average forecast percentage error for that segment. Weekday/weekend separation matters because demand patterns differ substantially: office heating, industrial load, and commuting all shift the grid’s generation mix.

Period Volatility

Volatility is the average standard deviation of carbon intensity readings within each period-day-type segment — a high value means intensity fluctuated significantly during that time window, typical when renewable output is unstable. % high intensity counts the share of readings classified as ‘high’ or ‘very high’, a proxy for how often that period forces the grid to dispatch expensive, high-emission peaking plants.