A preview of this full-text is provided by Springer Nature.
Content available from Remote Sensing in Earth Systems Sciences
This content is subject to copyright. Terms and conditions apply.
Vol.:(0123456789)
Remote Sensing in Earth Systems Sciences (2024) 7:681–698
https://doi.org/10.1007/s41976-024-00156-6
RESEARCH
Integrative Remote Sensing Approaches Using Generative Adversarial
Networks forUrban Heat Island Analysis andMitigation
GanapathySundar1· PitchaimuthuPatchaiammal1· BalajiVijayanVenkateshwarulu2· ThangavelPradeeshKumar3·
KesavamoorthyRajamannar4· RajeshKumarTripathi5
Received: 3 September 2024 / Revised: 1 October 2024 / Accepted: 19 October 2024 / Published online: 28 October 2024
© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024
Abstract
The phenomenon of urban heat islands (UHI) presents a critical challenge for urban sustainability, exacerbating local
temperatures, increasing energy demands, and impairing public health. Traditional methods for addressing UHI are often
resource-intensive and slow. This study introduces a novel approach, utilizing a Hybrid CycleGAN-SVM (HCGS) model
that leverages the synergy of Generative Adversarial Networks (GANs) and Support Vector Machines (SVMs) to efficiently
analyze and mitigate UHI effects through high-resolution satellite imagery and temperature data. The model incorporates
Enhanced Vision Transformers (EViTs) for superior feature extraction, adept at capturing intricate spatial and spectral urban
patterns. The CycleGAN component of the model generates high-quality synthetic imagery, enhancing the dataset and
addressing class imbalances, thereby bolstering the SVM classifier’s ability to precisely pinpoint heat-prone urban areas.
Implemented in Google Colab, the HCGS model demonstrated exceptional performance, achieving a classification accuracy
of 0.98. This indicates its potential as an effective tool for urban heat mitigation, offering actionable insights for urban plan-
ning and policy-making. By integrating advanced machine learning techniques with remote sensing data, the HCGS model
paves the way for innovative climate adaptation strategies, fostering more sustainable and resilient urban environments.
Keywords Urban heat island· UHI effect· Hybrid CycleGAN-SVM· HCGS model· Generative Adversarial Networks·
GANs· Support Vector Machines· SVM· Enhanced Vision Transformers· EViTs· Satellite imagery
1 Introduction
The urban heat island (UHI) is an increase in temperature
over the urban area compared to rural areas, originated
and amplified by human activities and artificial salts. This
is primarily driven by the high use of impervious surfaces
that store and release heat, as well lack of vegetation that
typically provides cooling via evapotranspiration [1, 2]. The
UHI effect has troubling consequences: more energy must be
consumed, emissions of air pollutants and greenhouse gases
become higher in urban areas than surrounding regions,
and human health and comfort are increasingly compro-
mised while water resources will get limited. Reducing the
urban heat island effect is critical for making cities more
* Ganapathy Sundar
sundarganapathy66@gmail.com
Pitchaimuthu Patchaiammal
sarandsk1@gmail.com
Balaji Vijayan Venkateshwarulu
psgbala.vijayan@gmail.com
Thangavel Pradeesh Kumar
pradeeshkumar.t@vit.ac.in
Kesavamoorthy Rajamannar
Kesavamoorthycse@gmail.com
Rajesh Kumar Tripathi
rajesh.tripathi@gla.ac.in
1 Department ofComputer Applications, Sindhi College,
Chennai, India
2 Department ofInformation Science andEngineering,
HKBK College ofEngineering, VTU, Belagum, Nagawara,
Bengaluru, India
3 VIT School ofAgricultural Innovations andAdvanced
Learning (VAIAL), VIT, Vellore, India
4 Department ofComputer Science andEngineering, CMR
Institute ofTechnology, Bengaluru, India
5 Department ofComputer Engineering andApplications,
GLA University, Uttar Pradesh, Mathura, India
Content courtesy of Springer Nature, terms of use apply. Rights reserved.