Evaluation of Land Surface Temperature based on Local Climate Zones Scheme inGreater Copenhagen Area
Author
Mocanu, Bogdan
Term
4. term
Publication year
2020
Submitted on
2020-08-03
Pages
82
Abstract
Dette speciale undersøger variationer i landoverfladetemperatur (LST) i Storkøbenhavn i relation til bystruktur ved hjælp af Local Climate Zones (LCZ). LCZ-kort blev udarbejdet for 2015–2019 ud fra Landsat 8-billeder i Google Earth Engine med en Random Forest-klassificering (samlet nøjagtighed omkring 0,75), med særlig opmærksomhed på det hastigt urbaniserede Nordhavn. For hvert år blev klassedynamik og LST beregnet på grundlag af Landsat 8-samlingen, herunder arbejdet med emissivitet, og resultaterne peger på en stigende LST-tendens i perioden med værdier i intervallet 20,57–29,33. Analysen indikerer en forbindelse mellem højere LST og nye bebyggede områder, men understreger usikkerheder, bl.a. klassifikationsfejl, behov for kalibrering og validering af LST samt effekter af arealanvendelsesændringer. Der foreslås forbedringer som brug af digitale højdemodeller til udtræk af bebyggede klasser, udvidelse af studieområdet og validering mod temperaturmålinger. Studiet demonstrerer en reproducerbar, skybaseret arbejdsgang til at koble LCZ-kortlægning med LST for at belyse urbane varmeforhold i København.
This thesis evaluates how land surface temperature (LST) varies across the Greater Copenhagen Area in relation to urban form using the Local Climate Zones (LCZ) framework. LCZ maps were produced for 2015–2019 from Landsat 8 imagery in Google Earth Engine with a Random Forest classifier (overall accuracy around 0.75), with particular attention to the rapidly developing Nordhavn district. For each year, LCZ class dynamics and LST were derived from the Landsat 8 image collection, including emissivity work, and the results indicate an increasing LST trend during 2015–2019 with values in the range 20.57–29.33. The analysis suggests a connection between higher LST and newly built-up areas, while highlighting uncertainties such as classification errors, the need for LST calibration and validation, and land-use change effects. Recommended improvements include using digital surface models for built-up class extraction, extending the study area, and validating satellite-derived temperatures against records. The study demonstrates a reproducible, cloud-based workflow linking LCZ mapping with LST to explore urban heat patterns in Copenhagen.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents