Event Recap | 17 July 2025 | London South Bank University
Hosted by: CIBSE Building Simulation & Daylight Group
Speakers: John Mardaljevic & Matt Eames
Why This Matters
Weather data is the backbone of building performance simulation. From energy demand to daylighting and overheating assessments, the accuracy of our models hinges on the quality of the weather files we use. But what if the standard approach i.e. using Typical Meteorological Years (TMYs) is no longer fit for purpose?
This event brought together experts to explore whether Actual Meteorological Years (AMYs) or Recent Meteorological Decades (RMDs), particularly using CAMS satellite data, offer a more robust alternative.
The Foundation: What Are TMYs?
Typical Meteorological Years (TMYs) are synthetic weather files created by stitching together the most “average” months from a historical dataset. They are widely used in:
- Energy performance simulations
- Daylight modelling
- Overheating risk assessments
- Compliance and certification (e.g., EPCs)
But TMYs are not real years - they are statistical constructs. Each month may have occurred, but never in the same sequence which raises debates regarding their ability to represent real-world variability and extremes.
Enter CAMS: A Satellite-Based Revolution
The Copernicus Atmosphere Monitoring Service (CAMS) provides high-resolution, satellite-derived solar radiation data. The speakers validated CAMS data against 610 days of ground-based illuminance measurements across the UK and found:
- Excellent agreement in global horizontal illuminance
- Sub-hourly variability captured surprisingly well
- Mean bias error under 10% for high-irradiance days
This suggests CAMS data is not only accurate but also more representative of real-world conditions than traditional TMYs.
Key Findings from the Study
1. TMYs vs CAMS: A Mismatch
- TMYs often fall outside the distribution of actual data from the past decade.
- Particularly poor agreement in beam normal irradiance (BNI).
- Even among TMYs, there is significant variation, raising concerns about consistency.
2. Temporal Resolution Matters
- CAMS offers 15-minute resolution, compared to the standard 60-minute TMYs.
- This finer granularity captures short-term solar variability, which is crucial for:
- Daylight modelling
- HVAC sizing
- Glare and comfort analysis
3. Validation with Ground Truth
- CAMS-derived illuminance was compared to real-world measurements using Perez luminous efficacy models.
- Results showed remarkable agreement, even on highly variable days.
Visualising the Difference
The study used frequency distribution plots and dissimilarity matrices to compare:
- 10 years of CAMS data (15- and 60-minute resolution)
- 4 TMYs from different sources (including IWEC and TMYx)
The results were clear: TMYs are not representative of the recent decade’s solar conditions. In contrast, CAMS data showed tight clustering, indicating consistency and reliability.
Implications for Practice
What We Should Rethink:
- Daylight standards based on outdated TMYs
- Compliance modelling that assumes TMYs are “truth”
- Design decisions based on a single synthetic year
What We Should Consider:
- Using Recent Meteorological Decades (RMDs) for simulations
- Incorporating CAMS data into standard workflows
- Moving toward 15-minute simulation time steps
Final Thoughts from the Panel
“If none of the weather files existed and we had to start from scratch, would we create what we have now?”
This question encapsulates the spirit of the event. The consensus was clear: we need to evolve. With CAMS offering unprecedented accuracy and accessibility, the building simulation community has an opportunity to redefine best practices.
View the Recording (link will be added soon)
Further Reading
The presentation is based on research from the article “Daylight solar radiation AMY data derived from satellite remote sensing: Validation against ground measurements and comparison with TMYs”, J. Mardaljevic, E. Brembilla, and M. Eames, Building Services Engineering Research and Technology, (in press), 2025.