We shared this post with CIBSE for their feedback and we have published their full response on this separate page.
CIBSE have recently released new weather data based on the UKCP18 climate projections. The previous weather dataset was published in 2016, based on the UKCP09 projections. The new set includes files for 28 UK zones (previously 16 locations), and the future files cover 2030, 2050 and 2080 time horizons.
For full disclosure, we are grateful to have been provided with the new weather files by CIBSE at a heavily discounted rate in return for the work we are doing in authoring and testing the update to TM59.
We also acknowledge that weather is notoriously difficult to predict and that producing weather files that are robust for a range of purposes is not an easy task.
Inkling have been reviewing the new data in the context of testing the revision to TM59, which is due out in the next few months. Our focus is therefore on the use of DSY (Design Summer Year) weather files for modelling overheating risk.
With that context in mind, this blog covers concerns we have about the new datasets in terms of how well they represent city centre locations and the urban heat island effect (UHI).
What is the urban heat island effect?
From the Royal Meteorological Society:
The urban heat island (UHI) effect is a phenomenon describing the elevated temperatures felt in towns and cities compared to rural surroundings and particularly felt at night-time as the heat retained by artificial surfaces is slowly released, keeping temperatures higher than in the countryside, combined with other impacts such as the reduced cooling effect of vegetation in urban areas, and the compounding effect of anthropogenic heat. UHI up to 8degC have been felt in UK cities, whilst cities the size of London has (sic) experienced temperatures in the order of 10degC.
As a response to growing awareness of the UHI effect on building design, CIBSE published TM49 in 2014, which established the case for three different weather files to cover London and the South East: urban (London Weather Centre), suburban (Heathrow) and rural areas (Gatwick) .
These three London files were included within the 2016 weather dataset mentioned above.
TM49 states that:
“Urban centres, especially in the south and south east of the UK, will experience more intense and frequent summer hot events, exacerbated by the urban heat island (UHI) effect.”
A BSERT paper (Using the new CIBSE design summer years to assess overheating in London: Effect of the urban heat island on design) published in 2015 reinforces this approach, highlighting the importance of including the UHI effect in the weather file to correctly assess the effectiveness of night cooling (the paper is focused on office buildings). For dwellings this would translate to the cooling effect of natural ventilation at night, which is critical to TM59.
Changes to London and south east England weather files
The 2025 CIBSE weather release has seen a shift from the three weather locations mentioned previously to just two zones for London and the southeast; Z1, which covers London postcodes; and Z2, which covers a large portion of the home counties. CIBSE have produced a handy tool which will specify which weather zone applies to any UK postcode.
Figure 1. Snapshot of the CIBSE 2025 weather dataset zones for England and Wales
The Z1 file is based on data measured at RAF Northolt (see Figure 2), which is considered by TM49 to be a ‘peri-urban’ location, rather than an urban location.
The Z2 file is based on weather data recorded at Manston on the east coast of Kent (see Figure 3). Z2 covers a large area from Milton Keynes to Canterbury and the Isle of Wight. It also includes many of the built-up London suburbs including Slough, Hornchurch and Dartford.
Figure 2. Z1 weather station location in Northolt
Figure 3. Z2 weather station location in Manston
Why have these location changes been made?
In the BSERT paper (published July 2025) Comparative analysis of old and new weather files: City-based versus climate zone-based approaches, written by the team that developed the new weather files, the authors state that city location weather stations (as utilised within the 2016 weather set) are not representative of the wider areas that each file represents. They also comment that there is a lack of clarity over which file should be selected. A particular concern raised is that city based weather locations make it harder to pass overheating assessments in rural areas. The new files prioritise making the weather data more representative of wider geographical areas rather than focusing on their accuracy within cities.
From the paper:
“Overall, the zone-based files are cooler than the city-based files. This is largely because more cooler climates are now included and warmer sites such as Heathrow, which are not representative of the wider area are not selected. In the example of the southeastern most area as represented by zone 2, a specific climate file has been established which is much cooler than the London based file that would have been used previously. This is likely to mean that less thermal discomfort will be experienced and thus less mitigation measures will be required to pass current building regulations, and in the case where mechanical cooling is employed it will result in smaller plant sizes being specified. Subsequently, resulting in lower embodied carbon or unnecessary mitigations measures being implemented in cooler regions.”
This approach makes sense for projects on more rural sites. However, the majority of buildings are located in towns and cities. The DSY weather files are specifically aimed at supporting the mitigation of overheating risk, so we need a clear picture of how that risk manifests in the all the locations we build in.
How does this affect the data?
As UHI is most apparent at night, we have looked at the summer nighttime temperatures in the data used for London locations. The following figures are collated from 50th percentile, high emissions DSY1 files, as would be used for TM59 overheating analysis.
For central London, the average 5am air temp (between May to September) has dropped from 13.2°C (2016 2020s LWC) to 10.6°C (2025 2030s Z1). This 2.6°C drop seems to align with the sort of average difference you might expect between outside and inside an urban heat island.
We’ve also looked at the number of summer night time hours above 22°C which have gone from 64 (2016 2020s LWC) to 20 (2025 2030s Z1) or 38 (2025 2050s Z1).
The chart in figure 4 below is based on the summer (May to September) data for each file (2020s/2030s data). It shows that the peak temperature is higher in Z1 than in any of the 2016 London files, but the Z2 peak is lower than previously.
The minimum Z1 summer temperature is significantly lower in the 2025 data (2.1°C) than in the 2016 data for LWC or Heathrow (7.4°C/5.3°), while the Z2 minimum is slightly higher than under the 2016 Gatwick file. The average dry bulb temperatures over the summer months drop slightly – probably largely due to the cooler nights.
The maximum and average wind speeds (km/hr) have risen in the new data for both Z1 and Z2.
Figure 4. Summer air temperature comparison for 2016 and 2025 CIBSE weather data 2020s and 2030s timelines respectively
Do these changes matter?
There does appear to be a risk that the new zonal weather files are underestimating summer temperatures, particularly nighttime temperatures, in cities. For London and the South East region, the files are also a bit more windy than the previous data. Both of these factors would make overheating assessments easier to pass.
Can we be confident that these weather files are representative of the whole zones that they apply to?
The following weatherspark.com chart shows measured monthly average high and low temperatures for the City of London and RAF Northolt. Despite being less than 30km apart there is a clear difference of approximately 2°C between the nighttime lows.
Figure 5. Monthly measured weather comparison for RAF Northold and City of London
The chart below covers Z2 locations and again there are significant differences. Milton Keynes is much cooler at night than the more coastal locations, and the Isle of Wight has lower daytime peaks.
Figure 6. Monthly measured weather comparison for Milton Keynes, Canterbury and the Isle of Wight
The average Z2 wind speeds according to weatherspark.com are also very different (see below) with much higher wind speeds in Canterbury and the Isle of Wight. We potentially risk underestimating overheating risk in Milton Keynes if we use data with higher wind speeds than are usual for this location.
Figure 7. Monthly measured wind data comparison for Milton Keynes, Canterbury and the Isle of Wight
These charts of measured data suggest there is significant variation in measured weather data within the new geographic weather zones which might mean that higher risk locations are not as well reflected. This could lead to insufficient mitigations being included in the designs for some sites and increased incidence of overheating, or excessive cooling consumption (where present), in operation.
Impact on modelling results
A comparison of modelling results between the 2016 weather files and the 2025 files shows that the new Z1 results indicate lower overheating risk than under the 2016 files (comparing 2020s DSY1 high emissions 50th %ile data and the 2030’s equivalent from the 2025 data).
The results below show the pass/fail results for a terraced house and a single aspect flat modelled in TAS software through four orientations and the 2020s/2030s, 2050s and 2080s data from the 2016 and 2025 weather sets for London. Both results sets show more passes under the new weather data.
Figure 8. Results comparison for terraced house
Figure 9.Result comparison for single aspect flat
The largest hourly reduction in results between the old and new weather data is for bedrooms, presumably due to the cooler nights.
Figure 10. Hourly results for the terraced house (west facing) comparing 2016 weather data (left) with 2025 weather data (right)
Please note that our testing is ongoing and these results may not be fully indicative. We would welcome feedback from anyone who has tested existing projects against the new weather files on how their findings compare with ours.
Summary
We have some concerns about the new datasets in terms of how well they represent city centre locations in the context of urban heat island effect (UHI).
Focusing on the Southeast, the Z1 and Z2 zones do not appear to be representative of the more built-up areas in which most buildings will be located, instead being weighted to be more representative of surrounding rural areas. The effect of this is equivalent to the removal of the UHI from the weather data we model city projects against.
Looking at the central London CIBSE data, the average 5am air temp (between May to September) has dropped by 2.6°C between the weather file versions, from 13.2°C (2016 2020s LWC) to 10.6°C (2025 2030s Z1). Additionally, for London and the South East, the new CIBSE files are windier than the previous data, which would also make modelled overheating assessments easier to pass.
Recorded weather data has been shown to support the existence of nighttime temperature differences between urban and peri-urban locations. For example, showing a 2°C difference between the nighttime low between the City of London and RAF Northolt.
Comparison of modelling results between the 2016 weather files and the new 2025 data shows that the new Z1 files result in significantly lower nighttime overheating risk predictions than the 2016 files.
We would expect similar effects to occur in Manchester, which despite having significant and increasing UHI, is also not represented by a city centre weather zone.
The DSY weather files are specifically aimed at supporting the design of buildings that mitigate overheating risk, so we need a clear picture of how that risk manifests in all the locations we build in. We are concerned that these new files could lead us to underestimate overheating risk in urban areas at the design stage and consequently increase the possibility of delivering buildings that do not perform in operation. Another perspective is that the new files may be fairer to projects in more rural sites.
Have you looked at the new CIBSE weather data? What have you observed?
We would be interested in comparing notes and will provide further updates here when we have them.
Our thanks to Jake Hacker at Arup and Geoff Levermore from Manchester University for reviewing this text.
