Suyeon Choi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Suyeon Choi has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Yeonjoo Kim
Department of Civil & Environmental Engineering, Yonsei University, Seoul, South Korea
3 shared publications Diana Rechid
2 shared publications- Eric Samakinwa
Institute of Geography, University of Bern, Bern, Switzerland
2 shared publications - Guy Brasseur
Environmental Modeling Group, Max Planck Institute for Meteorology, Hamburg, Germany
2 shared publications - JiHyun Kim
Department of Geography, Kyung Hee University, Seoul, Republic of Korea
2 shared publications - Dan Li
Department of Earth and Environment Boston University Boston MA USA
1 shared publication
Latest Publications
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
FOCAL Urban Pilot: Efficient exploration of climate data locally for data-driven decision-support in urban climate adaptation planning
Climate change impacts are increasingly manifested at local scales, where mitigation and adaptation strategies are implemented. Despite the growing wealth of available climate data and services, their effective usage in local climate impact assessment and decision-making processes for mitigation and adaptation planning remains limited due to scale mismatches, computational constraints, complexity, and usability barriers for non-domain experts. Addressing these challenges requires both advanced computational methods and improved access to climate data and analysis tools. The EU Horizon project, FOCAL, bridges the gap between data, services, and their users by implementing an open compute platform that combines intelligent workflow management with high-performance computing (HPC) resources to allow for an efficient exploration of climate data on a local scale. In addition, innovative artificial intelligence (AI) tools are developed and made available to enhance climate data analysis in terms of speed, robustness, pattern detection, and localization; thereby expanding the toolkit of climate data analysis and impact assessment methods. A main objective of FOCAL is to support science-based, actionable decision-making processes in forestry and urban planning through its provided tools. In a co-design process involving developers and potential platform users from two forest pilot regions with contrasting ecological and management contexts (Forest Pilots) as well as a pilot city (Urban Pilot), web applications for intuitive user-platform-interaction and workflows, grounded in state-of-the-art climate science, to address concrete user questions in forestry and urban planning have been specified. As a result, decision makers can efficiently use climate data for the development of climate adaptation strategies.This contribution focuses on the Urban Pilot, implemented for the pilot city Constance (Baden-Württemberg, southern Germany), located at the western end of Lake Constance. Three core workflows have been developed:1) Regional climate change workflow: provision of robust regional climate change information for the past and the future under different global warming levels for urban areas, based on regional climate model and localized climate data, serving multi-sectoral local climate impact assessments;2) Urban hot and cool spot workflow: detection and high-spatial-resolution visual exploration of hot and cool spots in urban environments, supporting exposure assessment by integrating additional data (e.g., population or infrastructure data), risk assessment, and the planning of urban heat resilience measures and cooling spaces;3) Urban blue spot workflow: identification of blue spots (rainfall accumulation hazards) and provision of blue spot data in urban landscapes using processed precipitation data and extreme precipitation scenarios, supporting applications in hydrological modeling, flood risk management, and climate adaptation.By leveraging HPC-based data processing and AI-assisted analysis, these workflows translate complex climate data into actionable, locally relevant information. While demonstrated for the pilot city Constance, the methods and workflows are transferable to other urban areas, contributing to scalable and reproducible climate services.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessFOCAL Urban Pilot: Efficient exploration of climate data locally for data-driven decision-support in urban climate adaptation planning
Climate change impacts are increasingly manifested at local scales, where mitigation and adaptation strategies are implemented. Despite the growing wealth of available climate data and services, their effective usage in local climate impact assessment and decision-making processes for mitigation and adaptation planning remains limited due to scale mismatches, computational constraints, complexity, and usability barriers for non-domain experts. Addressing these challenges requires both advanced computational methods and improved access to climate data and analysis tools. The EU Horizon project, FOCAL, bridges the gap between data, services, and their users by implementing an open compute platform that combines intelligent workflow management with high-performance computing (HPC) resources to allow for an efficient exploration of climate data on a local scale. In addition, innovative artificial intelligence (AI) tools are developed and made available to enhance climate data analysis in terms of speed, robustness, pattern detection, and localization; thereby expanding the toolkit of climate data analysis and impact assessment methods. A main objective of FOCAL is to support science-based, actionable decision-making processes in forestry and urban planning through its provided tools. In a co-design process involving developers and potential platform users from two forest pilot regions with contrasting ecological and management contexts (Forest Pilots) as well as a pilot city (Urban Pilot), web applications for intuitive user-platform-interaction and workflows, grounded in state-of-the-art climate science, to address concrete user questions in forestry and urban planning have been specified. As a result, decision makers can efficiently use climate data for the development of climate adaptation strategies.This contribution focuses on the Urban Pilot, implemented for the pilot city Constance (Baden-Württemberg, southern Germany), located at the western end of Lake Constance. Three core workflows have been developed:1) Regional climate change workflow: provision of robust regional climate change information for the past and the future under different global warming levels for urban areas, based on regional climate model and localized climate data, serving multi-sectoral local climate impact assessments;2) Urban hot and cool spot workflow: detection and high-spatial-resolution visual exploration of hot and cool spots in urban environments, supporting exposure assessment by integrating additional data (e.g., population or infrastructure data), risk assessment, and the planning of urban heat resilience measures and cooling spaces;3) Urban blue spot workflow: identification of blue spots (rainfall accumulation hazards) and provision of blue spot data in urban landscapes using processed precipitation data and extreme precipitation scenarios, supporting applications in hydrological modeling, flood risk management, and climate adaptation.By leveraging HPC-based data processing and AI-assisted analysis, these workflows translate complex climate data into actionable, locally relevant information. While demonstrated for the pilot city Constance, the methods and workflows are transferable to other urban areas, contributing to scalable and reproducible climate services.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Urban Heat Effects and Targeted Adaptation under Extreme Events in the Lakeside City: A Case Study of Konstanz, Germany
Urban climate conditions in lakeside cities are shaped by the interaction between the urban morphology and atmospheric processes modified by the lake. It is essential to understand how and where the urban effect intensifies in order to develop effective adaptation strategies under increasingly frequent and intense extreme climate events. In this study, the city of Konstanz, located on Lake Constance, is used as a case study to examine urban-induced climate responses during extreme conditions.To assess urban climate impacts, high-resolution (1 km) simulations were conducted using the Weather Research and Forecasting (WRF) model with Building Effect Parametrization and Building Energy Model (BEP/BEM) integration. We simulated multiple extreme weather events, including heatwave and heavy rainfall events. Urban effects were quantified by contrasting current urban land use patterns with a hypothetical non-urban surface representation, allowing the evaluation of urban climate signals under lake-influenced conditions.Results indicate that the urban thermal effect is particularly pronounced under nighttime conditions. Compared to non-urban conditions, urban areas exhibit enhanced nighttime warming, with surface skin temperatures rising by approximately 1–2.5°C during heatwaves. These patterns suggest that urban heat storage and release significantly contribute to the nighttime thermal conditions. Furthermore, this nighttime warming varies with lake proximity and land-use characteristics, indicating that these factors influence the spatial distribution of urban heat during extreme events.Based on the findings, spatially targeted urban adaptation strategies are explored through the application of mitigation measures in areas experiencing persistent thermal stress. This study suggests that targeted approaches can effectively reduce local heat stress while limiting the extent of mitigation strategy application, emphasizing the potential for more strategic and efficient urban climate adaptation. This perspective provides useful context for climate-responsive urban planning approaches that give priority to impact-prone areas under increasing extreme events.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessUrban Heat Effects and Targeted Adaptation under Extreme Events in the Lakeside City: A Case Study of Konstanz, Germany
Urban climate conditions in lakeside cities are shaped by the interaction between the urban morphology and atmospheric processes modified by the lake. It is essential to understand how and where the urban effect intensifies in order to develop effective adaptation strategies under increasingly frequent and intense extreme climate events. In this study, the city of Konstanz, located on Lake Constance, is used as a case study to examine urban-induced climate responses during extreme conditions.To assess urban climate impacts, high-resolution (1 km) simulations were conducted using the Weather Research and Forecasting (WRF) model with Building Effect Parametrization and Building Energy Model (BEP/BEM) integration. We simulated multiple extreme weather events, including heatwave and heavy rainfall events. Urban effects were quantified by contrasting current urban land use patterns with a hypothetical non-urban surface representation, allowing the evaluation of urban climate signals under lake-influenced conditions.Results indicate that the urban thermal effect is particularly pronounced under nighttime conditions. Compared to non-urban conditions, urban areas exhibit enhanced nighttime warming, with surface skin temperatures rising by approximately 1–2.5°C during heatwaves. These patterns suggest that urban heat storage and release significantly contribute to the nighttime thermal conditions. Furthermore, this nighttime warming varies with lake proximity and land-use characteristics, indicating that these factors influence the spatial distribution of urban heat during extreme events.Based on the findings, spatially targeted urban adaptation strategies are explored through the application of mitigation measures in areas experiencing persistent thermal stress. This study suggests that targeted approaches can effectively reduce local heat stress while limiting the extent of mitigation strategy application, emphasizing the potential for more strategic and efficient urban climate adaptation. This perspective provides useful context for climate-responsive urban planning approaches that give priority to impact-prone areas under increasing extreme events.
- Open Access
- 13 June 2025
- Published by American Geophysical Union (AGU) in Earth's Future
- Vol. 13 (6)
Deep Learning‐Guided Urban Climate Risk Mitigation Through Optimal Spatial Allocation of Green and Cool Roofs
With cities facing increasing challenges due to climate change, we developed a deep learning‐based surrogate modeling framework to optimize urban roofing strategies for climate risk mitigation. Applied to Seoul, South Korea, the framework utilized the Weather Research and Forecasting model coupled with an Urban Canopy Model (WRF‐UCM) to generate objective indices for heat stress, flash floods, and wind circulation projected to the end of this century under four roof schemes: business‐as‐usual, 25% and 100% cool roofs (CR25 and CR100), and 100% green roofs (GR100). These indices were used to test four deep learning algorithms: UNet, UNet++, UNet3+, and Multi‐ResUNet. Multi‐ResUNet demonstrated superior performance, thus it was employed to develop the surrogate model, which was applied to 262,144 multi‐type roofing scenarios. Two optimal roofing scenarios were identified using the Pareto method, balancing the three climate objectives and economic costs: the first with CR100 covering 95.9% of urban areas, reducing heat stress by over 50% in 34.3% of regions and wind circulation by 10% in 27.7% of regions, and the second with CR100 covering 60.2% of urban areas, achieving a similar heat stress reduction in 21.6% of regions but a stronger reduction in wind circulation. Both scenarios had minimal impact on flash flood mitigation. This study highlights the importance of spatial configuration in maximizing the benefits of urban roofing strategies due to the heterogeneous effects across urban areas. Furthermore, the considerably lower computational time increases the practical utility of the proposed surrogate modeling framework for use in a diverse range of urban contexts, advancing global efforts to mitigate urban climate risks.
- 13 June 2025
- Published by American Geophysical Union (AGU) in Earth's Future
- Vol. 13 (6)
Open AccessDeep Learning‐Guided Urban Climate Risk Mitigation Through Optimal Spatial Allocation of Green and Cool Roofs
With cities facing increasing challenges due to climate change, we developed a deep learning‐based surrogate modeling framework to optimize urban roofing strategies for climate risk mitigation. Applied to Seoul, South Korea, the framework utilized the Weather Research and Forecasting model coupled with an Urban Canopy Model (WRF‐UCM) to generate objective indices for heat stress, flash floods, and wind circulation projected to the end of this century under four roof schemes: business‐as‐usual, 25% and 100% cool roofs (CR25 and CR100), and 100% green roofs (GR100). These indices were used to test four deep learning algorithms: UNet, UNet++, UNet3+, and Multi‐ResUNet. Multi‐ResUNet demonstrated superior performance, thus it was employed to develop the surrogate model, which was applied to 262,144 multi‐type roofing scenarios. Two optimal roofing scenarios were identified using the Pareto method, balancing the three climate objectives and economic costs: the first with CR100 covering 95.9% of urban areas, reducing heat stress by over 50% in 34.3% of regions and wind circulation by 10% in 27.7% of regions, and the second with CR100 covering 60.2% of urban areas, achieving a similar heat stress reduction in 21.6% of regions but a stronger reduction in wind circulation. Both scenarios had minimal impact on flash flood mitigation. This study highlights the importance of spatial configuration in maximizing the benefits of urban roofing strategies due to the heterogeneous effects across urban areas. Furthermore, the considerably lower computational time increases the practical utility of the proposed surrogate modeling framework for use in a diverse range of urban contexts, advancing global efforts to mitigate urban climate risks.
- 23 April 2025
- Published by Elsevier in Journal of Environmental Management
- Vol. 383, 125398
Deep learning-based spatial optimization of green and cool roof implementation for urban heat mitigation
- 23 April 2025
- Published by Elsevier in Journal of Environmental Management
- Vol. 383, 125398
Deep learning-based spatial optimization of green and cool roof implementation for urban heat mitigation
- Open Access
- 1 August 2022
- Published by Copernicus GmbH in Geoscientific Model Development
- Vol. 15 (15) , 5967-5985
Rad-cGAN v1.0: Radar-based precipitation nowcasting model with conditional generative adversarial networks for multiple dam domains
Numerical weather prediction models and probabilistic extrapolation methods using radar images have been widely used for precipitation nowcasting. Recently, machine-learning-based precipitation nowcasting models have also been actively developed for relatively short-term precipitation predictions. This study was aimed at developing a radar-based precipitation nowcasting model using an advanced machine-learning technique, conditional generative adversarial network (cGAN), which shows high performance in image generation tasks. The cGAN-based precipitation nowcasting model, named Rad-cGAN, developed in this study was trained with the radar reflectivity data of the Soyang-gang Dam basin in South Korea with a spatial domain of 128 × 128 pixels, spatial resolution of 1 km, and temporal resolution of 10 min. The model performance was evaluated using previously developed machine-learning-based precipitation nowcasting models, namely convolutional long short-term memory (ConvLSTM) and U-Net. In addition, Eulerian persistence model and pySTEPS, a radar-based deterministic nowcasting system, are used as baseline models. We demonstrated that Rad-cGAN outperformed reference models at 10 min lead time prediction for the Soyang-gang Dam basin based on verification metrics: Pearson correlation coefficient (R), root mean square error (RMSE), Nash–Sutcliffe efficiency (NSE), critical success index (CSI), and fraction skill scores (FSS) at an intensity threshold of 0.1, 1.0, and 5.0 mm h−1. However, unlike low rainfall intensity, the CSI at high rainfall intensity in Rad-cGAN deteriorated rapidly beyond the lead time of 10 min; however, ConvLSTM and baseline models maintained better performances. This observation was consistent with the FSS calculated at high rainfall intensity. These results were qualitatively evaluated using typhoon Soulik as an example, and through this, ConvLSTM maintained relatively higher precipitation than the other models. However, for the prediction of precipitation area, Rad-cGAN showed the best results, and the advantage of the cGAN method to reduce the blurring effect was confirmed through radially averaged power spectral density (PSD). We also demonstrated the successful implementation of the transfer learning technique to efficiently train the model with the data from other dam basins in South Korea, such as the Andong Dam and Chungju Dam basins. We used the pre-trained model, which was completely trained in the Soyang-gang Dam basin. Furthermore, we analyzed the amount of data to effectively develop the model for the new domain through the transfer learning strategies applying the pre-trained model using data for additional dam basins. This study confirmed that Rad-cGAN can be successfully applied to precipitation nowcasting with longer lead times and using the transfer learning approach showed good performance in dam basins other than the originally trained basin.
- 1 August 2022
- Published by Copernicus GmbH in Geoscientific Model Development
- Vol. 15 (15) , 5967-5985
Open AccessRad-cGAN v1.0: Radar-based precipitation nowcasting model with conditional generative adversarial networks for multiple dam domains
Numerical weather prediction models and probabilistic extrapolation methods using radar images have been widely used for precipitation nowcasting. Recently, machine-learning-based precipitation nowcasting models have also been actively developed for relatively short-term precipitation predictions. This study was aimed at developing a radar-based precipitation nowcasting model using an advanced machine-learning technique, conditional generative adversarial network (cGAN), which shows high performance in image generation tasks. The cGAN-based precipitation nowcasting model, named Rad-cGAN, developed in this study was trained with the radar reflectivity data of the Soyang-gang Dam basin in South Korea with a spatial domain of 128 × 128 pixels, spatial resolution of 1 km, and temporal resolution of 10 min. The model performance was evaluated using previously developed machine-learning-based precipitation nowcasting models, namely convolutional long short-term memory (ConvLSTM) and U-Net. In addition, Eulerian persistence model and pySTEPS, a radar-based deterministic nowcasting system, are used as baseline models. We demonstrated that Rad-cGAN outperformed reference models at 10 min lead time prediction for the Soyang-gang Dam basin based on verification metrics: Pearson correlation coefficient (R), root mean square error (RMSE), Nash–Sutcliffe efficiency (NSE), critical success index (CSI), and fraction skill scores (FSS) at an intensity threshold of 0.1, 1.0, and 5.0 mm h−1. However, unlike low rainfall intensity, the CSI at high rainfall intensity in Rad-cGAN deteriorated rapidly beyond the lead time of 10 min; however, ConvLSTM and baseline models maintained better performances. This observation was consistent with the FSS calculated at high rainfall intensity. These results were qualitatively evaluated using typhoon Soulik as an example, and through this, ConvLSTM maintained relatively higher precipitation than the other models. However, for the prediction of precipitation area, Rad-cGAN showed the best results, and the advantage of the cGAN method to reduce the blurring effect was confirmed through radially averaged power spectral density (PSD). We also demonstrated the successful implementation of the transfer learning technique to efficiently train the model with the data from other dam basins in South Korea, such as the Andong Dam and Chungju Dam basins. We used the pre-trained model, which was completely trained in the Soyang-gang Dam basin. Furthermore, we analyzed the amount of data to effectively develop the model for the new domain through the transfer learning strategies applying the pre-trained model using data for additional dam basins. This study confirmed that Rad-cGAN can be successfully applied to precipitation nowcasting with longer lead times and using the transfer learning approach showed good performance in dam basins other than the originally trained basin.