Semi-automated workflow for multi-scale analysis of heatwaves : A case study of Lyon, France
Popal, Abuzar (2021)
Popal, Abuzar
2021
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2021102718960
https://urn.fi/URN:NBN:fi:amk-2021102718960
Tiivistelmä
Globally, the frequency, severity, and length of heatwaves have intensified and are expected to increase. While many methods for estimating heatwave risk and mapping methodologies have been developed, a comprehensive, harmonized multiscale assessment of the most impacted areas remains absent. Currently, heatwave risk assessment is either too generic and does not adequately represent the heterogeneous urban landscape, or it requires a large amount of data, is computationally intensive, and takes a long time at the microscale level. This research developed a novel semi-automated approach to analyze heatwaves at two different scales: city and neighborhood. Thirteen of the most frequently used machine learning algorithms were used in this research to determine which algorithms delivered the best results for heatwave hotspot identification. The XGBoost classifier achieved the highest accuracy of 94% and was chosen as the basis for forecasting heatwave hotspots. The normalized difference building index (NDBI), the enhanced vegetation index (EVI), the percent of the industrial area (Industrial A P), the albedo, the percent of low skilled workers (Workers P), and the digital elevation model (DEM) are the factors that contributed the most to the projection of heatwave hotspots. NDBI was the most significant factor in the model, accounting for 30% of the total. The temperature was positively correlated with NDBI, Industrial A P, and Worker P, and EVI, albedo, and DEM were inversely correlated with temperature. The workflow combined city-scale analysis into neighborhood-scale analysis by examining the most severely affected areas in more detail, and greening scenarios were applied to simulate the appropriate heatwave mitigation threshold. Greening 50% of the three most impacted areas was sufficient to reduce the risk from Extreme to High, resulting in a 0.4°C to 0.5°C reduction. It is crucial for decision-makers to quickly explore hotspots at different scales within a heatwave-affected region to efficiently allocate emergency operations in a timely manner and plan future mitigation strategies to reduce the effect of a heatwave in the most impacted areas.