Pith. sign in

REVIEW 5 major objections 6 minor 59 references

Predicting Air Temperature from Volumetric Urban Morphology with Machine Learning

T0 review · 5 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a machine-learning model using only per-pixel building volume—computed by a fast CityGML-to-voxel conversion—can predict the spatial distribution of 2 m air temperature across cities, and that evaluating predictions…

desk verdict A useful engineering shortcut for CityGML voxelization wrapped in a predictive claim that is not yet supported because the evaluation lacks baselines and an elevation control. read the letter →

arxiv 2501.09469 v1 pith:57MZOZPA submitted 2025-01-16 cs.LG cs.AI

Berk K{i}v{i}lc{i}m , Patrick Erik Bradley This is my paper · ORCID
classification cs.LGcs.AI
keywords CityGMLVoxelAirtemperatureUrbanplanningmorphologyMachinelearningheatislandGaussianblur
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that a single spatial input—the total volume of buildings in each grid cell, derived from open 3D city models—is enough for a machine-learning model to predict where night-time summer air temperatures are higher and lower across a city. To make this practical, it introduces a CityGML-to-voxel conversion that assigns one height per building and runs in a single loop per building, which the authors argue is fast enough for city-scale, metre-resolution data. It also shows that blurring the building-volume raster with a Gaussian kernel raises the correlation between volume and temperature, and that evaluating predictions with SSIM and LPIPS alongside MSE reveals which models capture spatial patterns rather than merely minimizing error. If these claims hold, urban planners could edit building volumes in a plan and see the predicted thermal consequences without running expensive simulations.

What carries the argument

The load-bearing object is the voxelized building-volume raster: 2D building footprints rasterized at 1 m from CityGML, combined with a single height per building to construct one voxel column per building, then aggregated into total volume per 1 km air-temperature cell. This raster format makes Gaussian blur applicable and turns the prediction task into an image-to-image mapping, allowing SSIM and LPIPS to evaluate spatial similarity. The conversion's speed comes from iterating once per building rather than checking per-voxel intersections.

What would settle it

Apply the trained volume-only model to two held-out cities with similar per-pixel building volumes but very different elevations or regional climates; if the model gives nearly identical temperature fields while the measured fields differ by several degrees Celsius, the premise that volume alone drives night-time temperature fails. A simpler check: compare a flat, low-density city with a mountainous city of matched building volume; the model would have to reproduce the measured offset to be right.

Watch

Extended reading notes

Core claim

The central claim is that the spatial distribution of 2 m air temperature—here, the July 01:00 average—can be predicted from the volumetric morphology of buildings alone, with building volume per air-temperature pixel as the only input feature. The authors demonstrate the claim with voxelized building data from ten cities in a German state, training Random Forest and XGBoost on seven cities and testing on three, and report that the Random Forest model reproduces the overall urban-heat-island pattern of held-out cities even though its MSE is higher than XGBoost's. The paper further claims that the proposed CityGML-to-voxel conversion, which sacrifices per-building detail by assigning a single height value, is computationally fast at high resolution on city scale, and that Gaussian blurring of the volume raster meaningfully increases correlation with temperature.

Load-bearing premise

The load-bearing premise is that at 01:00 in July, the air temperature over a city cell is mostly determined by how much building volume sits in and around that cell, so a model with no information about elevation, land cover, regional climate, or weather can still predict held-out city temperatures.

Editorial extensions

If this is right

  • Urban planners can alter building volumes in a proposed layout and read off predicted changes in nocturnal air temperature, making heat effects visible during plan design.
  • Because voxel resolution is adjustable, the same pipeline can consume finer meteorological rasters as they become available without re-architecting the model.
  • Low MSE does not guarantee a correct spatial pattern: models with deeper trees achieved MSE near 0.20–0.45 °C² but produced overfitted, spatially wrong maps, so image-similarity metrics should accompany standard error metrics.
  • Random Forest with depth-limited trees generalizes urban-heat-island patterns to unseen cities, whereas XGBoost yields systematically urban-biased errors despite lower MSE.
  • Gaussian blurring of the volume input raises the correlation with air temperature in all ten cities, from ranges of 0.48–0.85 to 0.65–0.93, in denser cities more strongly.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If volume-only prediction is as robust as claimed, the same raster pipeline could be tested on additional environmental fields such as humidity, wind, or pollutant concentration by associating voxel volumes with those rasters, though the paper does not test this.
  • The single-height-per-building simplification removes roof geometry and height variation within footprints; a plausible extension is to encode height distributions per cell rather than one height, which might improve predictions for complex buildings with multiple height components.
  • The correlation gains from Gaussian blur suggest that neighbourhood scale matters; a testable extension would vary kernel radius systematically to find the spatial scale at which building mass best explains night-time temperature.
  • Applying the model to cities outside central Europe with different climates, building materials, or topography would test whether volume alone remains predictive or whether additional covariates become necessary.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper proposes a CityGML-to-voxel conversion method that assigns a single height per building footprint, aggregates building volumes to a 1-km pixel grid, applies Gaussian blurring to incorporate spatial neighbourhood effects, and trains Random Forest and XGBoost models to predict July 01:00 2-m air temperature for ten Thuringian cities. The authors report MSE, SSIM, and LPIPS scores and argue that Random Forest, despite higher MSE, reproduces the spatial pattern of urban heat islands better than XGBoost. The broader claim is that a model using only per-pixel building volume can predict the spatial distribution of air temperature and support urban planning decisions.

Significance. If the central claim were established, the work would be useful: the voxelization pipeline is an interesting engineering contribution that avoids expensive geometric intersections, and casting the prediction problem in raster form enables spatial evaluation metrics that are rarely used in this literature. The paper is also commendable for using open data and releasing code and for reporting correlation changes before and after Gaussian blurring. However, the predictive claim is not currently supported because the evaluation lacks a baseline, uses a very small and spatially autocorrelated sample, applies different test splits to the two models, and selects the reported model post hoc from visual inspection. The elevation-confounding concern is concrete and currently unaddressed, so the paper needs substantial additional analysis before the central claim can be accepted.

major comments (5)
  1. [Section 3, first paragraph and Table 2] The reported MSE values (0.84-0.92 C^2 for the selected models) are not interpretable without a baseline. A trivial model that predicts each city's mean temperature, or a model using elevation as the only predictor, would provide the necessary reference point. Because Table 1 shows raw correlations between building volume and temperature of only 0.48-0.85, it remains possible that the volume-based model adds little predictive power beyond a simple spatial mean or a topographic proxy. Please add at least a per-city mean baseline and an elevation-only baseline, and report the resulting MSE/SSIM/LPIPS for the same test cities.
  2. [Section 3, Figures 5-8] The Random Forest and XGBoost models are evaluated on different held-out city splits: RF is tested on Erfurt, Suhl, and Sonneberg, while XGBoost is tested on Sondershausen, Schmalkalden, and Erfurt. This makes the direct model comparison in Table 2 invalid, since the test sets differ in difficulty (e.g., Suhl and Schmalkalden are the two cities with the largest reported prediction-range discrepancies). All models should be compared on the same held-out split, with per-city metrics reported for every city.
  3. [Section 3, first paragraph] The reported model was selected after multiple training runs using a trial-and-error approach, with the authors explicitly stating that the presented model is the one that 'qualitatively provided the most accurate results' and showing that other hyperparameter settings gave lower MSE but were rejected as overfitting. This post hoc selection on the test cities means the reported generalization numbers are optimistic and not a valid estimate of performance on unseen data. Please describe a fixed model-selection protocol (e.g., a validation split or nested cross-validation) and report results for all configurations tried, or clearly label the reported numbers as retrospective and not predictive of future performance.
  4. [Sections 2.3 and 4] The central claim that building volume alone predicts air temperature is vulnerable to confounding by elevation. All ten cities are in Thuringia, a state with substantial relief, and the largest prediction errors occur precisely in the two cities in the Thuringian Forest (Suhl and Schmalkalden). Since settlements and high building volumes tend to be located in valleys, building volume may be serving as a proxy for elevation, and the July nocturnal temperature field is strongly controlled by adiabatic lapse rate. Please add elevation (or a digital terrain model) as a covariate or at least as a control in a baseline model, and test transfer to cities outside the same regional topographic regime before claiming that modifying building volume will alter temperatures.
  5. [Section 2.3] The effective sample size is very small: each city contributes a 10-by-8 grid, i.e., 80 pixels, and air temperature at 1-km resolution is strongly spatially autocorrelated. Treating these pixels as independent samples inflates the apparent evidence and makes the reported correlations and MSE values hard to interpret. Please report results at the city level as well as the pixel level, and use a spatial cross-validation scheme or a clustered bootstrap that respects the spatial dependence structure.
minor comments (6)
  1. [Section 2, data availability statement] The data availability statement contains a typo: 'can be fund' should be 'can be found'.
  2. [Section 2] There is a typo in Section 2: 'visuzalitazion' should be 'visualization'.
  3. [Section 2.2.3] The 3x3 patch size for matching building corners is introduced without a sensitivity analysis; the discussion notes that patch size should depend on voxel resolution, so a short experiment varying patch size would strengthen the method description.
  4. [Table 1] Correlation values are reported without confidence intervals or significance tests, and the sample size is only 80 pixels per city; please provide uncertainty estimates.
  5. [Section 3] The statement that Random Forest 'accurately predict[s] overall spatial patterns' is too strong given the large prediction-range errors reported for Suhl and Schmalkalden; please qualify this claim with the actual error magnitudes.
  6. [Section 3, LPIPS description] The LPIPS values are extremely small (on the order of 1e-5 to 1e-4); please clarify whether these values are on the expected scale after resizing to 64x64 and state the normalization applied to the temperature images before computing LPIPS.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the model is fit to observed air temperature and evaluated on held-out cities; self-citations are background only.

full rationale

The paper's derivation chain is linear: CityGML data are voxelized, per-pixel building volumes are aggregated to the 1 km temperature grid, a Gaussian blur is applied to the volume feature, and Random Forest/XGBoost models are trained to predict air temperature from that feature. The target variable is observed 2 m air temperature from the German Weather Service; the model is trained on seven cities and tested on three held-out cities (Erfurt, Suhl, Sonneberg for Random Forest; Sondershausen, Schmalkalden, Erfurt for XGBoost). There is no equation or definition by which the predicted temperature is constructed from building volume; the mapping is learned from data and then checked against separate held-out data. The Gaussian blurring step modifies the input feature and is reported to increase the correlation between that feature and temperature, but this correlation is a descriptive statistic computed before training, not a fitted parameter renamed as a prediction. Hyperparameters were selected by trial and error, which is ordinary model fitting rather than circular reasoning. The paper's self-citations ([16], [17], [18]) are cited as background regarding watertight volumetric model extraction and are explicitly contrasted with the proposed method: 'those methodologies are based on geometric intersection procedures. This means that those algorithms would be insufficient for a direct voxelization of large 3D city datasets.' They are not used as a load-bearing uniqueness theorem or as the justification for the predictive claim. The Discussion does acknowledge limitations, such as loss of building detail and LPIPS vulnerability, but these are not circularity. The potential confounding of building volume with elevation is a scientific validity concern, not a circularity concern, and does not make the prediction equivalent to its input by construction. Therefore no circular step can be exhibited from the paper's own definitions or equations, and the circularity score is 0.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the single-feature sufficiency of building volume, hand-chosen preprocessing and hyperparameters, and small convenience samples of cities. No new entities are introduced.

free parameters (7)
  • Gaussian blur sigma = 0.85
    Hand-chosen kernel spread for blurring building volume input; directly affects input features and reported correlation increases (Section 2.3).
  • Gaussian blur radius = 1
    Hand-chosen kernel radius, paired with sigma (Section 2.3).
  • RF hyperparameters = n_estimators=100000, max_depth=3, min_samples_split=4, min_samples_leaf=2, max_features='sqrt'
    Set via trial and error; model selection presented was based on qualitative appearance (Section 2.3).
  • XGBoost hyperparameters = n_estimators=300000, max_depth=3, learning_rate=0.000003
    Set via trial and error; the reported XGBoost model was selected on qualitative grounds (Section 2.3).
  • Data augmentation parameters = n_samples=100, noise_level=0.01
    Added to mitigate overfitting; these choices are arbitrary and not justified (Section 2.3).
  • Voxel intersection patch size = 3x3 meters
    Chosen to increase footprint-corner intersections; affects which buildings are represented (Section 4).
  • Temporal window = July, 01:00 AM
    Chosen based on literature that correlation peaks at this time; a selection that conditions the entire result (Section 2.1).
assumptions (5)
  • domain assumption Building volume alone is a sufficient predictor of 2m air temperature at 01:00 in July in the study cities
    The model uses only per-pixel building volume as input, ignoring elevation, land cover, weather, and regional climate (Sections 2.3, 3).
  • domain assumption CityGML data from Thuringia and DWD temperature grids from the same year are temporally consistent
    The paper aligns datasets by matching the year of CityGML creation, but does not validate against station data (Sections 2.1, 2.2).
  • domain assumption The ten selected cities are representative of German urban form for generalization
    Generalization claims rest on three held-out cities; no diversity across climate zones or building typologies is shown (Sections 1, 3).
  • ad hoc to paper Gaussian blurring with sigma=0.85, radius=1 captures the spatial neighborhood effects of urban morphology
    No spatial autocorrelation analysis or physical basis is given for these specific blur parameters (Section 2.3).
  • domain assumption The 1km temperature grid can be meaningfully linked to 1m voxel data aggregated to 1km cells
    The paper aggregates voxels to 1km cells to match DWD resolution, but does not discuss representativeness of a single grid cell (Section 2.3).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Predicting Air Temperature from Volumetric Urban Morphology with Machine Learning." pith.science (2026). https://pith.science/paper/57MZOZPA

@misc{pith2026250109469,
  author       = {Pith},
  title        = {Pith review of: Predicting Air Temperature from Volumetric Urban Morphology with Machine Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/57MZOZPA}},
  note         = {Machine review of arXiv:2501.09469}
}
read the original abstract

In this study, we firstly introduce a method that converts CityGML data into voxels which works efficiently and fast in high resolution for large scale datasets such as cities but by sacrificing some building details to overcome the limitations of previous voxelization methodologies that have been computationally intensive and inefficient at transforming large-scale urban areas into voxel representations for high resolution. Those voxelized 3D city data from multiple cities and corresponding air temperature data are used to develop a machine learning model. Before the model training, Gaussian blurring is implemented on input data to consider spatial relationships, as a result the correlation rate between air temperature and volumetric building morphology is also increased after the Gaussian blurring. After the model training, the prediction results are not just evaluated with Mean Square Error (MSE) but some image similarity metrics such as Structural Similarity Index Measure (SSIM) and Learned Perceptual Image Patch Similarity (LPIPS) that are able to detect and consider spatial relations during the evaluation process. This trained model is capable of predicting the spatial distribution of air temperature by using building volume information of corresponding pixel as input. By doing so, this research aims to assist urban planners in incorporating environmental parameters into their planning strategies, thereby facilitating more sustainable and inhabitable urban environments.

Figures

Figures reproduced from arXiv: 2501.09469 by the authors.

Figure 1
Figure 1. Workflow of the voxelisation process 2.2.1 Retrieving 2D Building Footprint Areas Initially, the CityGML data, downloaded via [2], covered an area of 2 km × 2 km. Therefore, these data were merged to create a single comprehensive CityGML file for each city. This resulting CityGML file was then converted into the CityJSON format and imported into QGIS using the “CityJSON Loader” plugin. Subsequently, GDAL’s rasterisa… view at source ↗
Figure 2
Figure 2. The left column represents the CityGML visualization of Gotha [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. A brief workflow demonstration of data pre-processing steps for [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: This figure illustrates the spatial distribution of building volumes [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Comparison of the air temperature prediction results obtained [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 6
Figure 6. Figure 6: Comparison of the predictions obtained from the model trained [PITH_FULL_IMAGE:figures/full_fig_p018_6.png]
Figure 7
Figure 7. Figure 7: Comparison of the air temperature prediction results obtained from [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: Comparison of the predictions obtained from the model trained [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 56 canonical work pages

  1. [1]

    https://opendata.dwd.de/climate_environment/CDC/ grids_germany/hourly/hostrada/air_temperature_mean/

    Dwd climate data center (cdc): Annual mean of station obser- vations of daily air temperature at 2 meter above ground in °c for germany. https://opendata.dwd.de/climate_environment/CDC/ grids_germany/hourly/hostrada/air_temperature_mean/

    work page

  2. [2]

    https: //geoportal.thueringen.de/gdi-th/download-offene-geodaten/ download-3d-gebaeudedaten

    Open source citygml data of thuringia/germany. https: //geoportal.thueringen.de/gdi-th/download-offene-geodaten/ download-3d-gebaeudedaten

    work page

  3. [3]

    paraview: Parallel visualization application

    Sandia national labs, kitware inc, and los alamos national labs. paraview: Parallel visualization application. https://www.paraview.org/

    work page

  4. [4]

    Two decades of urban climate research: a review of turbulence, exchanges of energy and water, and the urban heat island

    A John Arnfield. Two decades of urban climate research: a review of turbulence, exchanges of energy and water, and the urban heat island. International Journal of Climatology: a Journal of the Royal Meteoro- logical Society, 23(1):1–26, 2003

    work page 2003

  5. [5]

    Do water bodies play an important role in the relationship between urban form and land surface temperature? Sustainable cities and society , 39:487–498, 2018

    Zhi Cai, Guifeng Han, and Mingchun Chen. Do water bodies play an important role in the relationship between urban form and land surface temperature? Sustainable cities and society , 39:487–498, 2018

    work page 2018

  6. [6]

    The use of climate knowledge in urban planning

    Ingeg¨ ard Eliasson. The use of climate knowledge in urban planning. Landscape and urban planning , 48(1-2):31–44, 2000

    work page 2000

  7. [7]

    Exploring the relationship between air temperature and urban morphology factors using machine learning under local climate zones

    Chengliang Fan, Binwei Zou, Jianjun Li, Mo Wang, Yundan Liao, and Xiaoqing Zhou. Exploring the relationship between air temperature and urban morphology factors using machine learning under local climate zones. Case Studies in Thermal Engineering , 55:104151, 2024

    work page 2024

  8. [8]

    Ma- chine learning applications in urban building energy performance fore- casting: A systematic review

    Soheil Fathi, Ravi Srinivasan, Andriel Fenner, and Sahand Fathi. Ma- chine learning applications in urban building energy performance fore- casting: A systematic review. Renewable and Sustainable Energy Re- views, 133:110287, 2020

    work page 2020

Show all 59 references
  1. [9]

    GDAL/OGR geospatial data abstraction software library, 2024

    GDAL/OGR contributors. GDAL/OGR geospatial data abstraction software library, 2024. 24

    2024

  2. [10]

    Attacking perceptual similarity metrics

    Abhijay Ghildyal and Feng Liu. Attacking perceptual similarity metrics. arXiv preprint arXiv:2305.08840 , 2023

    2023 arXiv

  3. [11]

    Characterizing the impact of urban morphology heterogeneity on land surface temperature in guangzhou, china

    Guanhua Guo, Xiaoqing Zhou, Zhifeng Wu, Rongbo Xiao, and Yingbiao Chen. Characterizing the impact of urban morphology heterogeneity on land surface temperature in guangzhou, china. Environmental Modelling & Software, 84:427–439, 2016

    2016

  4. [12]

    Toward mitigating urban heat island effects: Investigating the thermal-energy impact of bio-inspired retro-reflective building envelopes in dense urban settings

    Yilong Han, John E Taylor, and Anna Laura Pisello. Toward mitigating urban heat island effects: Investigating the thermal-energy impact of bio-inspired retro-reflective building envelopes in dense urban settings. Energy and Buildings , 102:380–389, 2015

    2015

  5. [13]

    Semantically enriched voxels as a common representation for comparison and evaluation of 3d building models

    Medhini Heeramaglore and Thomas H Kolbe. Semantically enriched voxels as a common representation for comparison and evaluation of 3d building models. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences , 10:89–96, 2022

    2022

  6. [14]

    Modeling the impact of 2d/3d urban indicators on the urban heat island over different seasons: A boosted regression tree approach

    Yunfeng Hu, Zhaoxin Dai, and Jean-Michel Guldmann. Modeling the impact of 2d/3d urban indicators on the urban heat island over different seasons: A boosted regression tree approach. Journal of environmental management, 266:110424, 2020

    2020

  7. [15]

    Building volume effects on ambient temperature in the kuala lumpur city

    Nurul Amirah Isa, Siti Aekbal Salleh, Wan Mohd Naim Wan Mohd, Andy Chan, Maggie Chel Gee Ooi, Nur Hidayah Zakaria, and Md Amirul Islam. Building volume effects on ambient temperature in the kuala lumpur city. In IOP Conference Series: Earth and Environ- mental Science, volume ...

    2020

  8. [16]

    M. Jahn. Distributed & Parallel Data Management to Support Geo- Scientific Simulation Implementations . PhD thesis, Karlsruhe Institute of Technology, 2022

    2022

  9. [17]

    Jahn and P.E

    M.W. Jahn and P.E. Bradley. Computing watertight volumetric mod- els from boundary representations to ensure consistent topological op- erations. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences , VIII-4/W2-2021:21–28, 2021

    2021

  10. [18]

    Jahn and P.E

    M.W. Jahn and P.E. Bradley. A robustness study for the extraction of watertight volumetric models from boundary representation data. IS- PRS International Journal of Geo-Information , 11(4):224, 2022

    2022

  11. [19]

    Web enabled spatio-temporal semantic analysis of traffic noise using citygml

    Amol Konde and Sameer Saran. Web enabled spatio-temporal semantic analysis of traffic noise using citygml. J Geomatics, 11(2):248–59, 2017. 25

    2017

  12. [20]

    High-resolution grids of hourly meteorological variables for germany

    S Kr¨ ahenmann, A Walter, S Brienen, F Imbery, and A Matzarakis. High-resolution grids of hourly meteorological variables for germany. Theoretical and Applied Climatology, 131:899–926, 2018

    2018

  13. [21]

    Imagenet clas- sification with deep convolutional neural networks

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet clas- sification with deep convolutional neural networks. Advances in neural information processing systems, 25, 2012

    2012

  14. [22]

    How do urban buildings impact sum- mer air temperature? the effects of building configurations in space and time

    Yuliang Lan and Qingming Zhan. How do urban buildings impact sum- mer air temperature? the effects of building configurations in space and time. Building and Environment , 125:88–98, 2017

    2017

  15. [23]

    Investigating the relationship between air temperature and the intensity of urban development using on-site measurement, satellite imagery and machine learning

    Tsz-Kin Lau and Tzu-Ping Lin. Investigating the relationship between air temperature and the intensity of urban development using on-site measurement, satellite imagery and machine learning. Sustainable Cities and Society, 100:104982, 2024

    2024

  16. [24]

    Cityjson: A compact and easy-to-use encoding of the citygml data model

    Hugo Ledoux, Ken Arroyo Ohori, Kavisha Kumar, Bal´ azs Dukai, Anna Labetski, and Stelios Vitalis. Cityjson: A compact and easy-to-use encoding of the citygml data model. Open Geospatial Data, Software and Standards, 4(1):1–12, 2019

    2019

  17. [25]

    Quantifying 3d building form effects on urban land surface temperature and modeling seasonal correlation patterns

    Huifang Li, Yanan Li, Tao Wang, Zhihua Wang, Meiling Gao, and Huan- feng Shen. Quantifying 3d building form effects on urban land surface temperature and modeling seasonal correlation patterns. Building and Environment, 204:108132, 2021

    2021

  18. [26]

    How does urban heat island differ across urban functional zones? insights from 2d/3d urban morphology using geospatial big data

    Anqi Lin, Hao Wu, Wenting Luo, Kaixuan Fan, and He Liu. How does urban heat island differ across urban functional zones? insights from 2d/3d urban morphology using geospatial big data. Urban Climate , 53:101787, 2024

    2024

  19. [27]

    How urban morphology relates to the urban heat island effect: A multi-indicator study

    Biao Liu, Xian Guo, and Jie Jiang. How urban morphology relates to the urban heat island effect: A multi-indicator study. Sustainability, 15(14):10787, 2023

    2023

  20. [28]

    Climate-conscious spatial morphology optimization strategy using a method combining local climate zone parameterization concept and ur- ban canopy layer model

    Lin Liu, Jing Liu, Lei Jin, Liru Liu, Yunfei Gao, and Xinpei Pan. Climate-conscious spatial morphology optimization strategy using a method combining local climate zone parameterization concept and ur- ban canopy layer model. Building and Environment , 185:107301, 2020

    2020

  21. [29]

    A machine learning-based method for the large-scale evaluation of the qualities of 26 the urban environment

    Lun Liu, Elisabete A Silva, Chunyang Wu, and Hui Wang. A machine learning-based method for the large-scale evaluation of the qualities of 26 the urban environment. Computers, environment and urban systems , 65:113–125, 2017

    2017

  22. [30]

    Automatic Repair of 3D City Building Model Using a Voxel– based Repair Method

    DT Mulder. Automatic Repair of 3D City Building Model Using a Voxel– based Repair Method. PhD thesis, Master Thesis, Delft Univ. Technol- ogy, Netherlands, 2015

    2015

  23. [31]

    Towards planning and practical understanding of the need for meteorological and climatic information in the design of high-density cities: A case-based study of hong kong

    Edward Ng. Towards planning and practical understanding of the need for meteorological and climatic information in the design of high-density cities: A case-based study of hong kong. International Journal of Cli- matology, 32(4):582–598, 2012

    2012

  24. [32]

    Voxelization algorithms for geospatial applications: Computational methods for voxelating spatial datasets of 3d city models containing 3d surface, curve and point data models

    Pirouz Nourian, Romulo Gon¸ calves, Sisi Zlatanova, Ken Arroyo Ohori, and Anh Vu Vo. Voxelization algorithms for geospatial applications: Computational methods for voxelating spatial datasets of 3d city models containing 3d surface, curve and point data models. MethodsX, 3:69–86, 2016

    2016

  25. [33]

    The heat island of the urban boundary layer: char- acteristics, causes and effects

    Timothy R Oke. The heat island of the urban boundary layer: char- acteristics, causes and effects. Wind climate in cities , pages 81–107, 1995

    1995

  26. [34]

    The application of citygml food water energy ade to es- timate the biomass potential for a land use scenario

    Rushikesh Padsala, Ernst Gebetsroither-Geringer, Keyu Bao, and Volker Coors. The application of citygml food water energy ade to es- timate the biomass potential for a land use scenario. In CITIES 20.50– Creating Habitats for the 3rd Millennium: Smart–Sustainable–Climate Neutr...

    2021

  27. [35]

    Klaus Pusacker, Volker Coors, J¨ org-Detlef Eckhardt, and Isabel Rupf. A concept for 3d geological and urban subsurface modeling with a unified voxel model examined by a case study for the city center of stuttgart (baden-w¨ urttemberg), germany.ISPRS Annals of the Photogrammet...

    2024

  28. [36]

    Qgis geographic information system

    QGIS Development Team. Qgis geographic information system

  29. [37]

    Understanding urban temperature differences through 2d/3d urban morphology

    Tom Raaymakers. Understanding urban temperature differences through 2d/3d urban morphology. 2024. 27

    2024

  30. [38]

    Developing a rapid method for 3-dimensional urban morphology extraction using open-source data

    Chao Ren, Meng Cai, Xinwei Li, Yuan Shi, and Linda See. Developing a rapid method for 3-dimensional urban morphology extraction using open-source data. Sustainable Cities and Society , 53:101962, 2020

    2020

  31. [39]

    Capturing and characterizing human activities using building locations in america

    Zheng Ren, Bin Jiang, and Stefan Seipel. Capturing and characterizing human activities using building locations in america. ISPRS Interna- tional Journal of Geo-Information , 8(5):200, 2019

    2019

  32. [40]

    3d vector- ization and rasterization of citygml standard in wind simulation

    Nurfairunnajiha Ridzuan, Uznir Ujang, and Suhaibah Azri. 3d vector- ization and rasterization of citygml standard in wind simulation. Earth Science Informatics, 16(3):2635–2647, 2023

    2023

  33. [41]

    Local climate zones for urban tem- perature studies

    Ian D Stewart and Tim R Oke. Local climate zones for urban tem- perature studies. Bulletin of the American Meteorological Society , 93(12):1879–1900, 2012

    1900

  34. [42]

    Machine learning for urban heat island (uhi) analysis: Predicting land surface temperature (lst) in urban environments

    Ghazaleh Tanoori, Ali Soltani, and Atoosa Modiri. Machine learning for urban heat island (uhi) analysis: Predicting land surface temperature (lst) in urban environments. Urban Climate, 55:101962, 2024

    2024

  35. [43]

    Predicting urban heat island in european cities: A comparative study of gru, dnn, and ann models using urban morphological variables

    Alireza Attarhay Tehrani, Omid Veisi, Yasin Delavar, Sasan Bahrami, Saeideh Sobhaninia, Asma Mehan, et al. Predicting urban heat island in european cities: A comparative study of gru, dnn, and ann models using urban morphological variables. Urban Climate, 56:102061, 2024

    2024

  36. [44]

    Reviewing the application of machine learning methods to model urban form indicators in planning decision support systems: Potential, issues and challenges

    Stephane Cedric Koumetio Tekouabou, El Bachir Diop, Rida Azmi, Remi Jaligot, and Jerome Chenal. Reviewing the application of machine learning methods to model urban form indicators in planning decision support systems: Potential, issues and challenges. Journal of King Saud Uni...

    2022

  37. [45]

    Hyperlocal mapping of urban air temperature using remote sensing and crowd- sourced weather data

    Zander S Venter, Oscar Brousse, Igor Esau, and Fred Meier. Hyperlocal mapping of urban air temperature using remote sensing and crowd- sourced weather data. Remote Sensing of Environment , 242:111791, 2020

    2020

  38. [46]

    Cityjson in qgis: Development of an open-source plugin.Transactions in GIS, 24(5):1147– 1164, 2020

    Stelios Vitalis, Ken Arroyo Ohori, and Jantien Stoter. Cityjson in qgis: Development of an open-source plugin.Transactions in GIS, 24(5):1147– 1164, 2020

    2020

  39. [47]

    Thermal remote sensing of urban climates

    James A Voogt and Tim R Oke. Thermal remote sensing of urban climates. Remote sensing of environment , 86(3):370–384, 2003. 28

    2003

  40. [48]

    Effects of urban surface geometry on remotely-sensed surface temperature

    James A Voogt and TR Oke. Effects of urban surface geometry on remotely-sensed surface temperature. International Journal of Remote Sensing, 19(5):895–920, 1998

    1998

  41. [49]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing , 13(4):600–612, 2004

    2004

  42. [50]

    Se- mantic 3d city models serving as information hub for 3d field based simulations

    BRUNO Willenborg, Maximilian Sindram, and THOMAS H Kolbe. Se- mantic 3d city models serving as information hub for 3d field based simulations. L¨ osungen f¨ ur eine Welt im Wandel, pages 54–65, 2016

    2016

  43. [51]

    Development of a 3-d urbanization index using digital terrain models for surface ur- ban heat island effects

    Chih-Da Wu, Shih-Chun Candice Lung, and Jihn-Fa Jan. Development of a 3-d urbanization index using digital terrain models for surface ur- ban heat island effects. ISPRS Journal of Photogrammetry and Remote Sensing, 81:1–11, 2013

    2013

  44. [52]

    Urban sustainability: an inevitable goal of landscape re- search, 2010

    Jianguo Wu. Urban sustainability: an inevitable goal of landscape re- search, 2010

    2010

  45. [53]

    Application of building geometry indexes to assess the correla- tion between buildings and air temperature

    Zhiwei Yang, Yingbiao Chen, Zihao Zheng, Qingyao Huang, and Zhifeng Wu. Application of building geometry indexes to assess the correla- tion between buildings and air temperature. Building and Environment, 167:106477, 2020

    2020

  46. [54]

    Challenges of influenza a viruses in humans and animals and current animal vaccines as an effective control measure

    Sung J Yoo, Taeyong Kwon, and Young S Lyoo. Challenges of influenza a viruses in humans and animals and current animal vaccines as an effective control measure. Clinical and experimental vaccine research , 7(1):1–15, 2018

    2018

  47. [55]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 586–595, 2018

    2018

  48. [56]

    The higher, the cooler? effects of building height on land surface temperatures in residential areas of beijing

    Zhong Zheng, Weiqi Zhou, Jingli Yan, Yuguo Qian, Jia Wang, and Weifeng Li. The higher, the cooler? effects of building height on land surface temperatures in residential areas of beijing. Physics and Chem- istry of the Earth, Parts A/B/C , 110:149–156, 2019

    2019

  49. [57]

    Revealing centrality in the spa- tial structure of cities from human activity patterns

    Chen Zhong, Markus Schl¨ apfer, Stefan M¨ uller Arisona, Michael Batty, Carlo Ratti, and Gerhard Schmitt. Revealing centrality in the spa- tial structure of cities from human activity patterns. Urban Studies , 54(2):437–455, 2017. 29

    2017

  50. [58]

    Does spatial con- figuration matter? understanding the effects of land cover pattern on land surface temperature in urban landscapes

    Weiqi Zhou, Ganlin Huang, and Mary L Cadenasso. Does spatial con- figuration matter? understanding the effects of land cover pattern on land surface temperature in urban landscapes. Landscape and urban planning, 102(1):54–63, 2011

    2011

  51. [59]

    Towards 3d raster gis: On developing a raster engine for spatial dbms

    Sisi Zlatanova, Pirouz Nourian, Romulo Gon¸ calves, and Anh Vu Vo. Towards 3d raster gis: On developing a raster engine for spatial dbms. In ISPRS WG IV/2 Workshop: Global Geospatial Information and High Resolution Global Land Cover/Land Use Mapping , 2016. 30

    2016

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.