In this report, the Urban Heat Island (UHI) effect was studied by introducing the Local Climate Zone (LCZ) concept, which is an effective tool for the differentiation of local urban climates. With the first objective of developing a universal and accurate method for LCZ maps derivation, Random Forest supervised classification was used on remote sensing images covering the Metropolitan City of Milan. Then, satellite images of summers 2006 to 2009 and 2020 to 2022 were leveraged to study the variation of land surface temperature (LST) in the light of the urban morphology and land cover. Complementary, the air temperature (AT) variation between LCZs was investigated as well using official and crowd-sourced in-situ measures from 2018 to 2021. As a result, an overall accuracy of 94% has been achieved for the LCZ map using Landsat-8 images. However, the optimal final result was obtained by adding a building height dataset, reducing the universality of the method. With respect to temperature variations, the results confirmed that: (1) The artificial LCZs had higher LSTs than natural ones with a mean difference statistically and practically significant of 6.7°C. (2) Two main factors participated in the reduction of LST: the height of building, introducing shadows on the ground, and the vegetation, increasing the portion of pervious materials. (3) In turn, vegetation was retrieved as factor of AT reduction, but the building height turned out to have the opposite effect as high buildings retain the heat and prevent air from circulating. This was illustrated by a statistically significant mean difference of 1.0°C between the Open low-rise and the Compact mid-rise classes. (4) The LCZs with higher building density were associated with slightly lower ATs, especially in winter and autumn. Finally, the analysis of records over 2018 to 2021 revealed a strong linear correlation of 0.9 between the AT and the LST. This study showed great potential of both satellite thermal data and crowdsourced information for climate research. In addition, the findings might be of great interest for urban climate specialists and urban planners in the pursuit of heat mitigation strategies.
L'effetto dell'isola di calore urbana (UHI) è stato studiato introducendo il concetto di zona climatica locale (LCZ), che è uno strumento efficace per la differenziazione dei climi urbani locali. Con il primo obiettivo di sviluppare un metodo universale e accurato per la derivazione delle mappe LCZ, è stata utilizzata la classificazione supervisionata Random Forest su immagini satellitari della Città Metropolitana di Milano. Successivamente, sono state utilizzate le immagini satellitari delle estati dal 2006 al 2009 e dal 2020 al 2022 per studiare la variazione della temperatura superficiale del suolo (LST) alla luce della morfologia urbana e dell'uso del suolo. Inoltre, è stata studiata la variazione della temperatura dell'aria (AT) tra le LCZ utilizzando in-situ misure ufficiali e crowd-sourced dal 2018 al 2021. Di conseguenza, è stata raggiunta una overall accuracy di 94% per la mappa delle LCZ utilizzando le immagini Landsat-8. Tuttavia, il risultato ottimale è stato ottenuto con l'utilizzo di immagini Landsat-8. Tuttavia, il risultato finale ottimale è stato ottenuto aggiungendo un set di dati sull'altezza degli edifici, riducendo l'universalità del metodo. Per quanto riguarda le variazioni di temperatura, i risultati hanno confermato che: (1) Le LCZ artificiali avevano LST più elevate rispetto a quelle naturali, con una differenza media statisticamente e praticamente significativa di 6.7°C. (2) Due fattori principali hanno contribuito alla riduzione dellla LST: l'altezza degli edifici, che introduce ombre sul suolo, e la vegetazione, che aumenta la porzione di materiali permeabili. (3) La vegetazione è stata individuata come fattore di riduzione della AT, ma l'altezza dell'edificio si è rivelata avere l'effetto opposto, poiché gli edifici alti trattengono il calore e impediscono la circolazione dell'aria. Ciò è stato dimostrato da una differenza media statisticamente significativa di 1.0°C tra le classi Open low-rise e Compact mid-rise. (4) Le LCZ con maggiore densità di edifici sono state associate a valori di AT leggermente inferiori, soprattutto in inverno e in autunno. Infine, l'analisi dei dati relativi al periodo 2018-2021 ha rivelato una forte correlazione lineare di 0.9 tra l'AT e la LST. Questo studio ha dimostrato il grande potenziale dei dati termici satellitari e delle informazioni raccolte in crowdsourcing per la ricerca sul clima. Inoltre, i risultati potrebbero essere di grande interesse per gli specialisti del clima urbano e per i pianificatori urbani nella ricerca di strategie di mitigazione del calore.
Analysis of temperature variation with respect to the LCZs using in situ measures and satellite imagery : study case of the metropolitan city of Milan
Puche, Mathilde Danielle Melanie
2021/2022
Abstract
In this report, the Urban Heat Island (UHI) effect was studied by introducing the Local Climate Zone (LCZ) concept, which is an effective tool for the differentiation of local urban climates. With the first objective of developing a universal and accurate method for LCZ maps derivation, Random Forest supervised classification was used on remote sensing images covering the Metropolitan City of Milan. Then, satellite images of summers 2006 to 2009 and 2020 to 2022 were leveraged to study the variation of land surface temperature (LST) in the light of the urban morphology and land cover. Complementary, the air temperature (AT) variation between LCZs was investigated as well using official and crowd-sourced in-situ measures from 2018 to 2021. As a result, an overall accuracy of 94% has been achieved for the LCZ map using Landsat-8 images. However, the optimal final result was obtained by adding a building height dataset, reducing the universality of the method. With respect to temperature variations, the results confirmed that: (1) The artificial LCZs had higher LSTs than natural ones with a mean difference statistically and practically significant of 6.7°C. (2) Two main factors participated in the reduction of LST: the height of building, introducing shadows on the ground, and the vegetation, increasing the portion of pervious materials. (3) In turn, vegetation was retrieved as factor of AT reduction, but the building height turned out to have the opposite effect as high buildings retain the heat and prevent air from circulating. This was illustrated by a statistically significant mean difference of 1.0°C between the Open low-rise and the Compact mid-rise classes. (4) The LCZs with higher building density were associated with slightly lower ATs, especially in winter and autumn. Finally, the analysis of records over 2018 to 2021 revealed a strong linear correlation of 0.9 between the AT and the LST. This study showed great potential of both satellite thermal data and crowdsourced information for climate research. In addition, the findings might be of great interest for urban climate specialists and urban planners in the pursuit of heat mitigation strategies.| File | Dimensione | Formato | |
|---|---|---|---|
|
2022_10_Puche.pdf
accessibile in internet per tutti
Descrizione: Text of the thesis
Dimensione
22.96 MB
Formato
Adobe PDF
|
22.96 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/194835