Back to search

National Technical University of Athens (NTUA)

Applying computational methods for processing thermal satellite images of urban areas

Abstract

dc:description

The land surface temperature (LST) is the skin temperature of the boundary layer between the Earth’s surface and atmosphere and a key variable in the physics of many terrestrial biophysical processes. Today the only practical way to acquire LST over extensive regions on the surface of the Earth is from space using thermal remote sensing. However, due to technical and physical constrains, satellite instruments cannot provide LST datasets with a spatial and temporal resolution that matches the characteristic scale of the LST diurnal cycle. A promising solution to this problem is to increase the spatial resolution of temporally-dense geostationary LST using statistical downscaling methods that rely on LST disaggregation kernels available at the desired fine spatial resolution. This thesis contributes to this body of research and is concerned with the identification of LST disaggregation kernels that perform robustly over various times-of-day, land covers and landscapes. To that end, this work demonstrates that multitemporal LST annual climatology data in the form of the annual cycle parameters (ACP)—namely the mean annual LST, the yearly amplitude of LST, and the LST phase shift from the spring equinox—offer these attributes. Such LST predictors provide information about the LST spatial distribution at different times-of-day and their use can improve the downscaled LST (DLST) spatial patterns, the DLST accuracy and the DLST diurnal range. This thesis also investigates how LST downscaling, when applied to continuous temporally-dense LST data, affects the spatiotemporal interrelationships of the DLST. To achieve this, a quarter-hourly, 4 km, three-month long LST time series from SEVIRI (Spinning Enhanced Visible and Infrared Imager) is statistically downscaled to 1 km and (i) the accuracy, reliability, and consistency of the downscaling method; (ii) the shape, size, and location of the DLST spatial patterns; and (iii) the capability of the DLST time series to emulate the diurnal and seasonal characteristics of the original LST data are assessed. The results suggest that the downscaling process can operate consistently and generate DLST data that reproduce the diurnal cycle without any artefacts. However the results also reveal that LST downscaling can produce DLST time series with increased autocorrelation. Finally, to facilitate the analysis of temporally-dense geostationary LST datasets, this thesis proposes a method for summarising a LST time series to a set of gap-free climatological cycle parameters that describe both the diurnal and annual LST cycles. The proposed method is based on the ACP sine-fitting method and is coupled with a spatiotemporal fusion scheme so as to increase the coarse spatial resolution of the geostationary ACP. The results suggest that the derived gap-free parameters are more informative than individual LST images and can provide an observation-based spatially consistent background for studying and characterising the thermal behaviour of the surface and also a dataset to support climate classification at a finer spatial resolution.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2018

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Sismanidis, Panagiotis
  • Σισμανίδης, Παναγιώτης

Subjects

dc:subject × 34

Rights

dc:rights
Statement dc:rights
  • BY_NC
Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/43916

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Sismanidis, Panagiotis; Σισμανίδης, Παναγιώτης. Applying computational methods for processing thermal satellite images of urban areas. National Technical University of Athens (NTUA), 2018. http://hdl.handle.net/10442/hedi/43916