the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Nov 2025
Assessment of current and future heat in a large hospital complex based on continuous indoor measurements and climate simulations
Abstract. People with acute illnesses and pre-existing conditions are particularly vulnerable to heat, making hospitals an area of high concern during heatwaves. Further, extreme heat impacts critical medical infrastructure due to increased patient admissions and impacts on workforce. This study assesses indoor heat occurrence and intensity in the University Medical Centre Freiburg, Germany based on measurements and data-driven climate simulations. Measurements were taken from May to September 2023 using a distributed sensor network in 60 rooms in 11 buildings. Measured air temperatures and physiologically equivalent temperatures are evaluated in terms of location, frequency, and intensity, as well as in relation to outdoor conditions, allowing for identification of vulnerable hospital structures and functions. Slight heat stress was most frequent and observed in all rooms, with 49 rooms showing additional occurrence of moderate and 17 rooms strong heat stress during summer 2023. Three heatwaves were identified as periods with high levels of heat stress and limited night-time cooling. Spatial hotspots were found in rooms without windows or air conditioning, located on higher floors, and predominantly in buildings constructed in 1950–1990. Measurements were combined with climate model data to project room-specific future indoor heat occurrence in all 60 rooms. All levels of heat stress are modelled to become more frequent and intense in rooms without air conditioning. Moderate heat stress or higher will increase on average by an additional 24 hours in 2020–2049 relative to 1990–2019. These findings call for immediate and widespread heat adaptation measures to ensure continued provision of critical medical infrastructure.
- Preprint
(2425 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-3871', Anonymous Referee #1, 02 Apr 2026
-
RC2: 'Comment on egusphere-2025-3871', Anonymous Referee #2, 08 Sep 2026
Given recent reports of intensifying heat waves affecting hospitals as most vulnerable sites, the measurement campaigns reported by this study produced thermal exposure data inside and outside a medical facility, which are timely and unique and form the most valuable part of this research. Unfortunately, the biggest strength simultaneously represents the weak point of this study because the link provided for accessing the research data (https://dx.doi.org/10.5281/zenodo.15356528) does not work. Thus, it is a pity that the data are not available for reviewing, especially as the presented analyses do not make use of this data in full capacity.
The authors mention and discuss differences in the occupant population (patients, visitors, staff, students), but do not address these differences by performing population-specific analyses. E.g., PET calculations are performed with a clothing insulation of 0.9 clo and 80 W work metabolism (to be added to basal metabolism, e.g., assumed to be 80 W by DIN 33403-3). While the clothing insulation might represent a resting reclining patient under a blanket on top of a mattress or could alternatively mean a business suit for men, the activity level with total metabolic rate of 160 W would be representative for a person with a low to moderate activity, e.g., a student or visitor who will likely wear less insulating clothing than assumed above. Overall, the assumptions underlying the PET calculations appear to be appropriate to business men working in a hospital, not among the major target populations listed above.
In addition, the assessment metrics are either too simple (indoor air temperature, neglecting the influence of solar irradiation) or not suitable for indoor conditions, as PET is an outdoor biometeorological index (VDI 3787-2). Numerous established metrics for indoor heat stress assessment are available, e.g. operative temperature (weighted mean of air temperature and mean radiant temperature with equal weighting appropriate in low wind indoor conditions, ASHRAE 55), PMV (ISO 7730), WBGT (ISO 7243) or PHS (ISO 7933), and could be easily computed for different populations from the recorded variables, if I understand the paper correctly (I could not verify this because the link to the data does not work) using established open source software (e.g., https://doi.org/10.1016/j.softx.2020.100578, https://doi.org/10.32614/RJ-2016-050).
Summarized, in my opinion an essential requirement before publication of the paper will be a fully functional link to the complete measurement data supplemented with detailed metadata and descriptions. The authors could then decide whether they will ameliorate their analyses by pursuing alternative pathways with higher practical relevance for specific occupant groups or prefer mentioning this as study limitations and provide other researchers with the option for performing such deeper analyses by making their valuable data openly available.
Detailed comments:
- Line 22 (abstract): “Moderate heat stress or higher will increase on average by an additional 24 hours in 2020-2049 relative to 1990-2019.” I suspect that this does not mean one day more heat stress in thirty years (2020-2049). Specify the temporal relation: is it 24 h per month, year, …?
- Lines 55-57: The relationship between ASTA and BAUA (Federal Institute of Occupational Safety and Health) and the ministry is just the opposite as described in the text: ASTA is a committee of the ministry and the committee’s office is housed by BAUA (who in turn is a research institution affiliated to the ministry, “Ressortforschungseinrichtung”), please adjust the text accordingly.
- Lines 58-61: The text omits that the indoor air temperature thresholds from ASR A3.5 will only apply for outdoor air temperatures above 26 °C (cf. ASR A3.5, section 4.4). Not considering this additional criterion will render the results presented for the indicator indoor air temperature Ti non-conforming to ASR A3.5. The authors should either correct their analyses concerning Ti or mention this limitation in the discussion. Another prerequisite for the application of Ti mentioned in that section of ASR A3.5 are suitable measures preventing the room from solar irradiation. The authors will have to check which rooms in their study will fulfill this criterion. Background of these conditions introduced to ASR A3.5 is that Ti is offered as simplified indicator for employers and responsible staff for assessing the room temperature (note that ASR A3.5 is titled “Raumtemperatur” not “Lufttemperatur”). ASHRAE 55 and ISO 7730 propose the indicator “operative temperature” for assessing indoor heat stress, calculated as mean of air temperature and mean radiant temperature weighted by convective and radiative heat transfer, respectively, thus considering the influence of air temperature, air movement and thermal radiation. As mentioned above, in case of low wind speed the operative temperature can be estimated by averaging air and mean radiant temperature. The indicator Ti was introduced in ASR A3.5 to simplify the assessment at workplaces in practical settings, not to simplify research studies, which should at least attempt to follow the state of the art in heat stress assessment. This implies the use of thermal indices as listed before, where operative temperature is the simplest one. The authors should either augment their analyses or discuss this limitation. In any case, as the research data (presumably) include all relevant measurements, the database will have to be made available publicly for later re-use, potentially applying more advanced metrics.
- Line 107: Report the Ti limits from the ‘overheating guidelines’ and compare them to Ti limits in this study.
- Lines 110-111: The motivation for applying PET as established outdoor thermal index to indoor conditions is not convincing. Note that PMV relies on a model of human heat balance based on Fanger’s work. Also note that according to the described measurements, it should be possible to calculate more advanced heat stress assessment metrics like WBGT and PHS, too.
- Line 116: Though research question 3 addresses vulnerability, the aimed study population remains unclear: should it be patients, visitors, students, staff (office workers, physicians, nurses) etc.? The occupants’ health status, their activities and clothing will determine their vulnerability, however, this is barely considered in this manuscript (cf. below). Define the study population(s) and their characteristics relevant to heat stress.
- Line 137: Replace “that” by “than” and verify whether 1.40 K refers to 1961-1990 while 2.26 K refers to 1991-2020. Shouldn’t it be switched?
- Line 155: Report the accuracy for the measured climatic parameters.
- Line 157-158: Please report the range, mean and SD of measured air velocities for motivating the default setting to 0.1 m/s.
- Line 159: Please motivate the choice of clothing insulation (0.9 clo, conforming to a business suit or to lying on a mattress under a blanket) and activity level (‘work metabolism’ of 80 W conforming to a person with low activity in a sitting posture according to DIN 33403-3, which suggests to add another 80 W as basic metabolism to transfer that number to the more conventional energy expenditure, often assumed equivalent to metabolic heat production). At what type of ‘vulnerability’ does this choice aim?
- Line 172-173: Explain the term “window exposure”.
- Lines 234ff: The results are presented rather in a descriptive style, which lacks a systematic analytical comparison between room types, e.g., AC vs non-AC, solar exposure vs no solar exposure, comparison of occupant types or daytime vs nighttime heat stress, relevant for patients’ recovery during sleep etc. Consider to present summarizing analyses contrasting different conditions derived from the research questions.
- Line 240: Explain why the highest PET was recorded in the morning at 9am.
- Line 248: Explain why ZMK had high PET values in all rooms.
- Caption Figure 2: Include an explanation where to find the abbreviations used as the x-axis labels and the meaning of ‘*’
- Line 266-267: Quantify and provide an explanation for the ‘clear differences’ in heat stress.
- Line 278-279: Again, try to explain “clearly visible” differences in PET within and between departments.
- Line 288-289: Would you consider a time series approach for analyzing the time lag between outdoor and indoor heat stress over six days?
- Figures 5 to 7 are hardly readable. Consider to present Ti and PET in different panels, e.g. Ti left, PET right.
- Line 316: How was the average lag of three hours determined? (cf. comment 18)
- Figure 7: Note that means will smooth the dynamics over the day, which could be more accurately determined by fitting a suitable cyclic function to the data and presenting the “mean curve” instead the “curve of means”, which will be different for non-linear trends.
- Line 337-338: Instead of reporting daily maximum and daily minimum values, extreme values reported separately for daytime and nighttime would be more informative on heat stress during daily activities and during sleep recovery (or nighttime work).
- Line 340: Quantify “most rooms”, e.g., by percentage.
- Line 372: In my understanding, the reported numbers (237h or 7%) are based on 24h-assessment, which might not be relevant if the room is occupied only during daytime. Adjust your analyses for ‘typical’ occupant time. (This could affect most figures in the results, e.g., Fig.10 or line 407 ‘80% of the time’). Interestingly, this is discussed in lines 415f. referring to external studies.
- Line 413-414: In my opinion, the conclusion that ‘heat stress indoors is less severe than outdoors’ derived from air temperature values alone will not be tenable as it neglects potential wind cooling, especially during the night. Verify and re-phrase if necessary.
- Line 425: Quantify the statement that ‘higher floors generally receive more solar irradiance’, e.g., by providing statistics on mean radiant temperature.
- Line 438: The statement about high peaks in the daytime, but also limited cooling at night comes without justification from the data. Please quantify.
- Line 442ff: Section 4.2 discusses vulnerable user groups and the importance of occupant time suggesting that corresponding systematic analyses should be integrated in the results section. Please add such analyses.
- Lines 473-477: When discussing extreme PET values in an office, please provide an explanatory example for ‘anthropogenic origin’, what does this actually mean here in an office room?
- Line 514: Another example of insufficient quantification as ‘some of the lowest levels of heat stress’. How many and what level?
- Line 524: Acknowledging the comprehensive measurement campaign of an entire hospital complex, the major limitation in this study is the non-exhaustive approach to data analysis relying on (too) simple or non-appropriate indices and not presenting the contrasts of interest. Therefore, the deposition of the highly relevant measurement data including metadata and documentation at a public repository is crucial for sustainable use of this research.
- Line 577: The link provided to the ZENODO repository does not work. Please provide a functional link to the research data.
Citation: https://doi.org/10.5194/egusphere-2025-3871-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,670 | 1,193 | 154 | 3,017 | 156 | 143 |
- HTML: 1,670
- PDF: 1,193
- XML: 154
- Total: 3,017
- BibTeX: 156
- EndNote: 143
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study addresses an important and timely topic: the prediction of indoor heat stress in healthcare buildings using a data-driven approach. A major strength of the paper is the large-scale indoor monitoring campaign, covering 60 rooms across 11 buildings. The manuscript is generally well written, and the methodology appears sound and clearly explained. That said, the study seems quite similar to previously published work from the same research group, with the main distinction being the building type examined here. The authors may wish to clarify more explicitly the novel contribution of this paper beyond the application to hospital buildings.
Specific comments: