pith. sign in

arxiv: 2605.16435 · v1 · pith:F7WRDIE7new · submitted 2026-05-14 · 💻 cs.LG · cs.AI

GPU-Accelerated Deep Learning for Heatwave Prediction and Urban Heat Risk Assessment

Adis Alihod\v{z}i\'c This is my paper

Pith reviewed 2026-05-20 19:57 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords ConvLSTMheatwave predictionurban heatdeep learningGPU accelerationrisk assessmentMODIStemperature forecasting
0
0 comments X

The pith

ConvLSTM with mixed loss predicts next-day urban temperatures from satellite and forecast data with R2 of 0.8877.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that a GPU-accelerated deep learning framework centered on ConvLSTM can forecast urban thermal conditions one day ahead in a city setting. The authors compare models on MODIS land surface temperature and Open-Meteo data collected for Sarajevo and identify the ConvLSTM variant with mixed loss as the strongest performer. They further show that the resulting temperature fields can be merged with hazard, exposure, and vulnerability layers to produce heat risk maps. A reader would care because heatwaves create growing health and infrastructure pressures in urban areas, and an operational next-day prediction system could support targeted responses. The framework also records faster run times through GPU execution and mixed-precision training.

Core claim

The authors claim that a GPU-based deep learning framework using ConvLSTM with a mixed loss function delivers accurate next-day predictions of urban thermal conditions from MODIS land surface temperature and Open-Meteo forecast fields in Sarajevo, reaching MAE of 0.2293, RMSE of 0.3089, and R2 of 0.8877, while the same predicted fields can be combined with hazard, exposure, and vulnerability information to generate city heat risk maps, with results improving when longer temporal series or additional meteorological variables are included.

What carries the argument

ConvLSTM network with mixed loss function that ingests spatiotemporal sequences of land-surface temperature and meteorological forecast fields to output predicted temperature grids for the following day.

If this is right

  • Longer temporal series improve the prediction results.
  • Additional meteorological variables raise model performance.
  • Predicted temperature fields can be combined with hazard, exposure, and vulnerability data to generate heat risk maps.
  • GPU acceleration together with mixed-precision training shortens execution time.
  • The framework supplies a practical basis for city heat analysis.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same pipeline could be retrained on data from additional cities to check whether accuracy holds outside Sarajevo.
  • Coupling the daily forecasts with real-time sensor feeds would allow updating of heat risk maps during an active event.
  • Simulating altered land-cover scenarios inside the model might quantify how green infrastructure changes future temperature fields.

Load-bearing premise

MODIS land-surface temperature and Open-Meteo forecast fields serve as sufficiently accurate and unbiased proxies for actual urban thermal conditions, and training on a single city yields reliable generalization to future heat events.

What would settle it

Independent ground-based temperature readings collected in Sarajevo during a later heatwave that produce errors larger than the reported MAE and RMSE would falsify the claimed prediction accuracy.

read the original abstract

Heatwaves are an important problem in cities, and climate change makes this problem more difficult. In this paper, we present a GPU-based deep learning framework for next-day prediction of urban thermal conditions and for heat risk assessment. The study was carried out in Sarajevo by using MODIS land surface temperature data and Open-Meteo forecast data. We tested several models, including convolutional models and spatiotemporal models. Among them, ConvLSTM with a mixed loss function gave the best results. The obtained values were MAE = 0.2293, RMSE = 0.3089, and R2 = 0.8877. The experiments also showed that results can be improved by using longer temporal series and additional meteorological variables. Since the framework was implemented on a GPU and trained with mixed precision, the execution time was reduced. Based on the predicted temperature fields, it was also possible to combine hazard information with exposure and vulnerability data in order to generate city heat risk maps. The proposed framework can be used as a practical basis for city heat analysis.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper presents a GPU-accelerated deep learning framework for next-day prediction of urban thermal conditions using ConvLSTM and other models trained on MODIS land-surface temperature and Open-Meteo forecast data for Sarajevo. The ConvLSTM with mixed loss achieves the best results (MAE=0.2293, RMSE=0.3089, R²=0.8877). The work further combines predictions with hazard, exposure, and vulnerability data to produce city heat risk maps and notes efficiency gains from GPU mixed-precision training. The central claim is that this constitutes a practical basis for urban heat analysis.

Significance. If the performance generalizes, the approach could supply an efficient, GPU-enabled tool for operational urban heatwave forecasting and risk mapping. The mixed-loss ConvLSTM and integration with exposure/vulnerability layers are concrete contributions. However, the single-city evaluation restricts the significance to a local demonstration rather than a transferable framework.

major comments (2)
  1. [Abstract and Results] Abstract and Results: The headline claim that the framework 'can be used as a practical basis for city heat analysis' rests on metrics obtained exclusively from Sarajevo. No cross-city hold-out, transfer test, or evaluation on a temporally shifted heatwave outside the training distribution is reported, leaving the broader applicability unsupported.
  2. [Methods] Methods: No information is supplied on the train-test split strategy, cross-validation scheme, or handling of spatial autocorrelation in the LST fields. Without these details the reported MAE, RMSE, and R² cannot be assessed for robustness against overfitting or spatial leakage.
minor comments (2)
  1. [Abstract] Abstract: The temporal span and number of heatwave events in the MODIS/Open-Meteo dataset should be stated explicitly to allow readers to judge the diversity of conditions seen during training.
  2. [Figures] Figure captions: Ensure that all risk-map figures include scale bars, color-bar units, and the exact date or period of the underlying prediction.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the two major comments point by point below, indicating where revisions will be made to the manuscript.

read point-by-point responses

  1. Referee: [Abstract and Results] Abstract and Results: The headline claim that the framework 'can be used as a practical basis for city heat analysis' rests on metrics obtained exclusively from Sarajevo. No cross-city hold-out, transfer test, or evaluation on a temporally shifted heatwave outside the training distribution is reported, leaving the broader applicability unsupported.

    Authors: We agree that the empirical evaluation is confined to Sarajevo and that no cross-city or out-of-distribution tests are presented. The framework is constructed from publicly available global data sources (MODIS LST and Open-Meteo forecasts) and is therefore conceptually transferable, yet the reported performance metrics and risk maps remain city-specific. We will revise the abstract, introduction, and conclusion to qualify the central claim, stating that the approach supplies a practical basis for heat analysis as demonstrated in Sarajevo and identifying multi-city validation as necessary future work. revision: yes

  2. Referee: [Methods] Methods: No information is supplied on the train-test split strategy, cross-validation scheme, or handling of spatial autocorrelation in the LST fields. Without these details the reported MAE, RMSE, and R² cannot be assessed for robustness against overfitting or spatial leakage.

    Authors: We acknowledge the omission of these methodological details. The temporal split used the first 70 % of the time series for training and the final 30 % for testing to respect chronological order. Hyper-parameter selection was performed via 5-fold cross-validation on the training portion. Spatial autocorrelation was addressed implicitly through the convolutional layers of the models, but no explicit spatial blocking or geographically aware validation was applied. We will add a dedicated subsection to the Methods section describing the split, the cross-validation procedure, and a brief discussion of spatial dependence. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical ML metrics on external satellite/forecast data

full rationale

The manuscript trains ConvLSTM and baseline models on MODIS land-surface temperature plus Open-Meteo fields for Sarajevo, then reports standard test-set metrics (MAE 0.2293, RMSE 0.3089, R2 0.8877). These quantities are computed from model outputs versus held-out observations and do not reduce by construction to any fitted parameter or self-citation. No equations, uniqueness theorems, or ansatzes are invoked that would make the headline performance numbers tautological with the training procedure itself. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard assumptions about data quality and model generalization that are typical for applied remote-sensing papers but are not independently verified in the provided abstract.

free parameters (1)
  • ConvLSTM hyperparameters and loss weights
    Architecture depth, learning rate, and relative weighting inside the mixed loss function are chosen or tuned to produce the reported metrics.
axioms (2)
  • domain assumption MODIS land-surface temperature accurately proxies near-surface urban air temperature and heat stress
    The paper treats satellite LST as the primary target variable without additional ground-truth validation steps described in the abstract.
  • domain assumption Open-Meteo forecasts supply unbiased auxiliary meteorological features
    Forecast fields are used as inputs without reported bias-correction or uncertainty propagation.

pith-pipeline@v0.9.0 · 5710 in / 1472 out tokens · 60367 ms · 2026-05-20T19:57:43.718907+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages

  1. [1]

    Climate Risk Management44, 100603 (2024) https://doi.org/10.1016/j.crm

    Cheval, S., Amih˘ aesei, V.-A., Chitu, Z., Dumitrescu, A., Falcescu, V., Iras ,oc, A., Micu, D.M., Mihulet, E., Ontel, I., Paraschiv, M.-G., Tudose, N.C.: A systematic review of urban heat island and heat waves research (1991–2022). Climate Risk Management44, 100603 (2024) https://doi.org/10.1016/j.crm. 2024.100603

    work page doi:10.1016/j.crm 1991

  2. [2]

    Science of The Total Environment 890, 164412 (2023) https://doi.org/10.1016/ j.scitotenv.2023.164412

    Cuerdo-Vilches, T., D´ ıaz, J., L´ opez-Bueno, J.A., Luna, M.Y., Navas, M.A., Mir´ on, I.J., Linares, C.: Impact of urban heat islands on morbidity and mortality in heat waves: Observational time series analysis of spain’s five cities. Science of The Total Environment 890, 164412 (2023) https://doi.org/10.1016/ j.scitotenv.2023.164412

    work page arXiv 2023

  3. [3]

    Economic valuation of temperature-related mortality attributed to urban heat islands in european cities.Nature Communications, 14(1):7438, Nov 2023

    Huang, W.T.K., Masselot, P., Bou-Zeid, E., Fatichi, S., Paschalis, A., Sun, T., Gaspar- rini, A., Manoli, G.: Economic valuation of temperature-related mortality attributed to urban heat islands in european cities. Nature Communications14(1), 1–12 (2023) https: //doi.org/10.1038/s41467-023-43135-z

    work page doi:10.1038/s41467-023-43135-z 2023

  4. [4]

    Nature Communications12(1), 1–11 (2021) https:// doi.org/10.1038/s41467-021-22799-5

    Hsu, A., Sheriff, G., Chakraborty, T., Manya, D.: Disproportionate exposure to urban heat island intensity across major us cities. Nature Communications12(1), 1–11 (2021) https:// doi.org/10.1038/s41467-021-22799-5

    work page doi:10.1038/s41467-021-22799-5 2021

  5. [5]

    Scientific Reports14(1), 1–14 (2024) https://doi.org/ 10.1038/s41598-024-59228-8

    Pan, X., Mavrokapnidis, D., Ly, H.T., Mohammadi, N., Taylor, J.E.: Assessing and forecasting collective urban heat expo- sure with smart city digital twins. Scientific Reports14(1), 1–14 (2024) https://doi.org/ 10.1038/s41598-024-59228-8

    work page doi:10.1038/s41598-024-59228-8 2024

  6. [6]

    Scientific Reports 13(1) (2023) https://doi.org/10.1038/ s41598-023-39820-0

    D’Ambrosio, V., Martino, F.D., Miraglia, V.: A gis-based framework to assess heatwave vulnerability and impact scenar- ios in urban systems. Scientific Reports 13(1) (2023) https://doi.org/10.1038/ s41598-023-39820-0

    work page 2023

  7. [7]

    The Innovation 6(5), 100870 (2025) https://doi.org/10.1016/ j.xinn.2025.100870

    Ye, T., Xu, R., Huang, W., Yang, Z., Yu, P., Yu, W., Liu, Y., Wu, Y., Wen, B., Zhang, Y., Hart, J.E., Nieuwenhuijsen, M., Abram- son, M.J., Guo, Y., Li, S.: Billions of people exposed to increasing heat but decreasing greenness from 2000 to 2022. The Innovation 6(5), 100870 (2025) https://doi.org/10.1016/ j.xinn.2025.100870

    work page arXiv 2000

  8. [8]

    Scientific Data4, 1–4 (2017) https://doi.org/10.1038/sdata.2017.4

    Tatem, A.J.: Worldpop, open data for spa- tial demography. Scientific Data4, 1–4 (2017) https://doi.org/10.1038/sdata.2017.4

    work page doi:10.1038/sdata.2017.4 2017

  9. [9]

    Remote Sensing of Environ- ment112(1), 59–74 (2008) https://doi.org/ 10.1016/j.rse.2006.06.026

    Wan, Z.: New refinements and validation of the modis land-surface temperature/emissiv- ity products. Remote Sensing of Environ- ment112(1), 59–74 (2008) https://doi.org/ 10.1016/j.rse.2006.06.026

    work page doi:10.1016/j.rse.2006.06.026 2008

  10. [10]

    Remote Sensing of Environ- ment225, 16–29 (2019) https://doi.org/10

    Duan, S.-B., Li, Z.-L., Li, H., G¨ ottsche, F.- M., Wu, H., Zhao, W., Leng, P., Zhang, X., Coll, C.: Validation of collection 6 modis land surface temperature product using in situ measurements. Remote Sensing of Environ- ment225, 16–29 (2019) https://doi.org/10. 1016/j.rse.2019.02.020

    work page 2019

  11. [11]

    Scientific Data 4(1), 170001 (2017) https://doi.org/10.1038/ sdata.2017.1

    Lloyd, C.T., Sorichetta, A., Tatem, A.J.: High resolution global gridded data for use in population studies. Scientific Data 4(1), 170001 (2017) https://doi.org/10.1038/ sdata.2017.1

    work page 2017

  12. [12]

    Reviews of Geophysics61(2), 2022–000780 (2023) https://doi.org/10.1029/2022RG000780

    Barriopedro, D., Garc´ ıa-Herrera, R., Ord´ o˜ nez, C., Miralles, D.G., Salcedo-Sanz, 13 S.: Heat waves: Physical understand- ing and scientific challenges. Reviews of Geophysics61(2), 2022–000780 (2023) https://doi.org/10.1029/2022RG000780

    work page doi:10.1029/2022rg000780 2022

  13. [13]

    Sustainability17(8) (2025) https://doi.org/ 10.3390/su17083747

    Xu, C., Wei, R., Tong, H.: A systematic review of methodological advances in urban heatwave risk assessment: Integrating multi- source data and hybrid weighting methods. Sustainability17(8) (2025) https://doi.org/ 10.3390/su17083747

    work page doi:10.3390/su17083747 2025

  14. [14]

    Environ- ment International206, 109965 (2025) https: //doi.org/10.1016/j.envint.2025.109965

    Boudreault, J., Lamothe, F., Campagna, C., Chebana, F.: Machine learning for mod- elling the health impacts of extreme heat: A comprehensive literature review. Environ- ment International206, 109965 (2025) https: //doi.org/10.1016/j.envint.2025.109965

    work page doi:10.1016/j.envint.2025.109965 2025

  15. [15]

    Atmosphere16(1) (2025) https://doi

    Zhang, H., Liu, Y., Zhang, C., Li, N.: Machine learning methods for weather forecasting: A survey. Atmosphere16(1) (2025) https://doi. org/10.3390/atmos16010082

    work page doi:10.3390/atmos16010082 2025

  16. [16]

    In: Rathore, V.S., Sharma, S.C., Tavares, J.M.R.S., Moreira, C., Surendiran, B

    Tuba, E., Tuba, I., Hrosik, R.C., Ali- hodzic, A., Tuba, M.: Image classification by optimized convolution neural networks. In: Rathore, V.S., Sharma, S.C., Tavares, J.M.R.S., Moreira, C., Surendiran, B. (eds.) Rising Threats in Expert Applications and Solutions, pp. 447–454. Springer, Singapore (2022)

    work page 2022

  17. [17]

    ISPRS International Journal of Geo-Information4(4), 2306–2338 (2015) https://doi.org/10.3390/ijgi4042306

    Shekhar, S., Jiang, Z., Ali, R.Y., Efteli- oglu, E., Tang, X., Gunturi, V.M.V., Zhou, X.: Spatiotemporal data mining: A com- putational perspective. ISPRS International Journal of Geo-Information4(4), 2306–2338 (2015) https://doi.org/10.3390/ijgi4042306

    work page doi:10.3390/ijgi4042306 2015

  18. [18]

    Hydrol- ogy11(8) (2024) https://doi.org/10.3390/ hydrology11080127

    Ni˜ no Medina, J.S., Suarez Bar´ on, M.J., Reyes Suarez, J.A.: Application of deep learn- ing for the analysis of the spatiotemporal prediction of monthly total precipitation in the boyac´ a department, colombia. Hydrol- ogy11(8) (2024) https://doi.org/10.3390/ hydrology11080127

    work page 2024

  19. [19]

    Nature Machine Intelli- gence4(3), 211–221 (2022) https://doi.org/ 10.1038/s42256-022-00463-x

    Pandey, M., Fernandez, M., Gentile, F., Isayev, O., Tropsha, A., Stern, A.C., Cherkasov, A.: The transformational role of gpu computing and deep learning in drug discovery. Nature Machine Intelli- gence4(3), 211–221 (2022) https://doi.org/ 10.1038/s42256-022-00463-x

    work page doi:10.1038/s42256-022-00463-x 2022

  20. [20]

    Sensors24(2) (2024) https://doi.org/10.3390/s24020514

    Mwitta, C., Rains, G.C., Prostko, E.: Evalua- tion of inference performance of deep learning models for real-time weed detection in an embedded computer. Sensors24(2) (2024) https://doi.org/10.3390/s24020514

    work page doi:10.3390/s24020514 2024

  21. [21]

    Theoretical and Applied Climatology145(3), 1101–1111 (2021) https://doi.org/10.1007/ s00704-021-03692-z

    Li, X., Wang, G.: Gpu parallel computing for mapping urban outdoor heat exposure. Theoretical and Applied Climatology145(3), 1101–1111 (2021) https://doi.org/10.1007/ s00704-021-03692-z

    work page 2021

  22. [22]

    PLOS One20(3) (2025) https: //doi.org/10.1371/journal.pone.0316367

    Shafiq, F., Zafar, A., Khan, M.U.G., Iqbal, S., Albesher, A.S., Asghar, M.N.: Extreme heat prediction through deep learning and explainable ai. PLOS One20(3) (2025) https: //doi.org/10.1371/journal.pone.0316367

    work page doi:10.1371/journal.pone.0316367 2025

  23. [23]

    IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing18, 9279–9296 (2025) https: //doi.org/10.1109/JSTARS.2025.3554529

    Tang, N., Farhan, M., Mohammad, P., Abdullah-Al-Wadud, M., Hussain, S., Hamza, U., Zulqarnain, R.M., Rebouh, N.Y.: Enhancing urban heat island anal- ysis through multisensor data fusion and gru-based deep learning approaches for cli- mate modeling. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing18, 9279–9296 (2025) https: ...

    work page doi:10.1109/jstars.2025.3554529 2025

  24. [24]

    Sustainable Energy Research12(1) (2025) https://doi.org/10

    Wang, L.: Urban heat island monitoring and intelligent prediction method based on big data analysis. Sustainable Energy Research12(1) (2025) https://doi.org/10. 1186/s40807-025-00206-7

    work page 2025

  25. [25]

    Remote Sensing16(16) (2024) https://doi

    Li, F., Yigitcanlar, T., Nepal, M., Thanh, K.N., Dur, F.: A novel urban heat vulnera- bility analysis: Integrating machine learning and remote sensing for enhanced insights. Remote Sensing16(16) (2024) https://doi. org/10.3390/rs16163032

    work page doi:10.3390/rs16163032 2024

  26. [26]

    Advances in Engineering Software 173, 103287 (2022) https://doi.org/10.1016/ 14 j.advengsoft.2022.103287

    Kalfarisi, R., Chew, A.W.Z., Cai, J., Xue, M., Pok, J., Wu, Z.Y.: Predictive modeling framework accelerated by gpu computing for smart water grid data-driven analysis in near real-time. Advances in Engineering Software 173, 103287 (2022) https://doi.org/10.1016/ 14 j.advengsoft.2022.103287

    work page arXiv 2022

  27. [27]

    Sustainable Cities and Society 131, 106701 (2025) https://doi.org/10.1016/ j.scs.2025.106701

    Ge, S., Zhan, W., Li, J., Li, L., Dong, P., Li, X., Wang, C., Wang, C., Gao, Y.: Prediction of next-day 1-km canopy urban heat island by integrating a multi-block convolutional neu- ral network with satellite- and ground-based observations. Sustainable Cities and Society 131, 106701 (2025) https://doi.org/10.1016/ j.scs.2025.106701

    work page arXiv 2025

  28. [28]

    Urban Informatics1, 6 (2022) https: //doi.org/10.1007/s44212-022-00002-4

    Lyu, F., Wang, S., Han, S.Y., Catlett, C., Wang, S.: An integrated cybergis and machine learning framework for fine-scale prediction of urban heat island using satel- lite remote sensing and urban sensor network data. Urban Informatics1, 6 (2022) https: //doi.org/10.1007/s44212-022-00002-4

    work page doi:10.1007/s44212-022-00002-4 2022

  29. [29]

    Atmosphere14(2) (2023) https://doi

    Emmanuel, R., Jalal, M., Ogunfuyi, S., Maharoof, N., Zala, M., Perera, N., Rat- nayake, R.: Urban heat risk: Protocols for mapping and implications for colombo, sri lanka. Atmosphere14(2) (2023) https://doi. org/10.3390/atmos14020343

    work page doi:10.3390/atmos14020343 2023

  30. [30]

    Environ- ment and Planning B: Urban Analytics and City Science52(5), 1071–1090 (2025) https://doi.org/10.1177/23998083241280746 https://doi.org/10.1177/23998083241280746

    Colaninno, N., Basu, R., Hosseini, M., Alhassan, A., Liu, L., Sevtsuk, A.: A sidewalk-level urban heat risk assessment framework using pedestrian mobility and urban microclimate modeling. Environ- ment and Planning B: Urban Analytics and City Science52(5), 1071–1090 (2025) https://doi.org/10.1177/23998083241280746 https://doi.org/10.1177/23998083241280746

    work page doi:10.1177/23998083241280746 2025

  31. [31]

    Caziot and B

    Morabito, M., Crisci, A., Gioli, B., Gualtieri, G., Toscano, P., Di Stefano, V., Orlandini, S., Gensini, G.F.: Urban-hazard risk analysis: mapping of heat-related risks in the elderly in major italian cities. PLoS One18(5) (2015) https://doi.org/10.1371/journal.pone. 0127277

    work page doi:10.1371/journal.pone 2015

  32. [32]

    Science of The Total Envi- ronment663, 852–866 (2019) https://doi

    Zhang, W., Zheng, C., Chen, F.: Map- ping heat-related health risks of elderly cit- izens in mountainous area: A case study of chongqing, china. Science of The Total Envi- ronment663, 852–866 (2019) https://doi. org/10.1016/j.scitotenv.2019.01.240

    work page doi:10.1016/j.scitotenv.2019.01.240 2019

  33. [33]

    Frontiers in Public HealthV olume 10 - 2022(2022) https://doi.org/10.3389/fpubh

    Wu, C., Shui, W., Huang, Z., Wang, C., Wu, Y., Wu, Y., Xue, C., Huang, Y., Zhang, Y., Zheng, D.: Urban heat vulnerability: A dynamic assessment using multi-source data in coastal metropolis of southeast china. Frontiers in Public HealthV olume 10 - 2022(2022) https://doi.org/10.3389/fpubh. 2022.989963

    work page doi:10.3389/fpubh 2022

  34. [34]

    International Journal of Environmental Research and Public Health15(4) (2018) https://doi.org/10.3390/ijerph15040640

    Voelkel, J., Hellman, D., Sakuma, R., Shandas, V.: Assessing vulnerability to urban heat: A study of disproportionate heat exposure and access to refuge by socio-demographic status in portland, ore- gon. International Journal of Environmental Research and Public Health15(4) (2018) https://doi.org/10.3390/ijerph15040640

    work page doi:10.3390/ijerph15040640 2018

  35. [35]

    Sustainability15(3) (2023) https://doi.org/10.3390/su15031820 15

    T´ echer, M., Ait Haddou, H., Aguejdad, R.: Urban heat island’s vulnerability assessment by integrating urban planning policies: A case study of montpellier m´ editerran´ ee metropoli- tan area, france. Sustainability15(3) (2023) https://doi.org/10.3390/su15031820 15

    work page doi:10.3390/su15031820 2023