Croatian Science Foundation · IP-2025-02 · 2025–2028
UTOMIKA
Urban Heat Islands of the Mediterranean and Continental Croatia
Cities trap heat. UTOMIKA measures how much, where, and why — across Zagreb and Split, over 37 years of satellite imagery — and turns the result into a map of which neighbourhoods are most at risk.
EN | HR
3 years
November 2025 – October 2028
1991–2027
period under analysis
2
cities: Zagreb and Split
1,400+
satellite images processed
Host
University of Zagreb, Faculty of Geodesy
About the project
An urban heat island is the temperature gap between a city and the countryside around it. It can reach ten degrees Celsius, and it is not spread evenly: heat concentrates where vegetation was replaced by asphalt, concrete and metal, where buildings stand close and tall enough to trap warm air, and where surfaces are dark enough to absorb sunlight all day and release it all night.
The effect is measurable but poorly mapped. Weather stations give precise readings at a handful of fixed points, rarely the ones that matter most; satellite thermal imagery covers everything but is limited by how often a satellite passes overhead and how much cloud is in the way. UTOMIKA combines the two, over Zagreb on the continental side and Split on the Mediterranean coast, so the two climates can be compared directly.
Working across 1991 to 2027, the project computes vegetation, surface temperature and environmental criticality values from more than 1,400 satellite images, checks them against its own field measurements, and feeds the whole set into machine learning models. The outcome is a graded map: which parts of each city are already heat islands, and which are most likely to become one.
Objectives
01
How the data connects
Establish how the different input datasets relate to one another, so together they can support decisions about detecting, monitoring and preventing heat islands. This is the project’s main scientific outcome, reached once everything else is in place.
02
A catalogue of data sources
Build a catalogue of open data servers useful for detecting and analysing heat islands, plus a list of commercial sources that can improve on them. Completed in the first three months, because everything else depends on it.
03
Detect and analyse the heat islands
Map heat islands across Zagreb and Split from satellite imagery, May to September, for 1991 to 2027, with field measurements in 2026 and 2027 to check the results against reality.
04
Digital atlas
Publish a digital atlas of averaged vegetation index, land surface temperature and environmental criticality values for both cities across the full period. Produced in the final six months.
05
Risk grading
Divide Zagreb and Split into zones graded by how likely each is to develop a heat island, based on the relationships established in objective 01.
Key activities
Seven strands of work run through the three years, each serving one or more of the objectives above.
Objective 02
Data source catalogue
The project surveys national, European and international platforms publishing meteorological and geospatial data relevant to urban heat — satellite imagery, climate records, land use — and assembles them into a single catalogue. Alongside it sits a shorter list of commercial sources, mainly high-resolution imagery, for use where open data is not enough.
Objectives 01 · 03
Climatological baseline
A climatological study covers Zagreb and Split for 1991 to 2024, drawing on the Grič, Maksimir, Pleso and Marjan weather stations. It analyses mean, minimum and maximum air temperature, precipitation, soil temperature at 2 and 5 cm and wind, establishing the baseline against which the satellite results are read.
Objective 03
Satellite processing
Thermal imagery from Landsat 5, 8 and 9 provides daytime coverage and MODIS the night-time picture, while Sentinel-2 supplies the visible bands used for land classification and vegetation index. From these the project computes surface temperature and environmental criticality values for May to September across the whole period.
Objective 03
Field validation
Between June and September in 2026 and 2027 the team measures at five fixed locations in each city: air temperature, contact soil temperature and soil temperature at 5 cm depth, plus surface temperature by infrared thermometer and readings from mobile weather stations. Daytime measurements are timed to Landsat overpasses and night-time ones to MODIS, so ground and satellite data describe the same moment.
Objectives 01 · 05
Urban geometry
Alongside the thermal data, the project analyses the shape of the built environment: sky view factor as a measure of how openly a place can radiate heat away, the ratio of building height to street width as it governs airflow, and surface albedo to identify materials that drive overheating. These turn out to be among the strongest predictors of where heat accumulates.
Objectives 01 · 05
Machine learning and risk grading
Convolutional neural networks identify heat island patterns in the imagery, while recurrent and long short-term memory models read the time series for long-term trends and seasonal variation. The combined output segments Zagreb and Split into zones graded by risk of heat island formation.
Objective 04
Atlas and dissemination
The averaged vegetation, temperature and criticality results are compiled into a digital atlas covering 1991 to 2027, published online and reachable through a web application. The team publishes in peer-reviewed journals and presents at international conferences throughout the project, including the ISPRS Congress and the EARSeL symposia.
Partners
University of Zagreb, Faculty of Geodesy
Host institution. Leads satellite image processing, the data source catalogue and the digital atlas.
Andrija Krtalić — principal investigator
Ana Kuveždić Divjak — data catalogue and atlas
University of Split, FESB
Machine learning and multi-criteria analysis, and the risk grading of both cities.
Ljiljana Šerić — machine learning and risk grading
University of Split, FGAG
Field measurement campaigns and heat island analysis for the Split study area.
Meteorological data is provided by the Croatian Meteorological and Hydrological Service (DHMZ), supplemented by the volunteer weather station network of the Pljusak association.
About the organisations
University of Zagreb, Faculty of Geodesy
The faculty teaches and researches geodesy and geoinformatics — surveying, photogrammetry, remote sensing, cartography and geographic information systems. It hosts UTOMIKA and carries out the satellite image processing, the catalogue of data sources and the cartographic work behind the digital atlas.
University of Split, FESB
The Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture works on artificial intelligence and distributed systems, with a research line applying them to environmental management. On UTOMIKA it is responsible for the machine learning models and for grading the two cities by heat island risk.
University of Split, FGAG
The Faculty of Civil Engineering, Architecture and Geodesy brings expertise in construction materials, the built environment and geodetic survey. On UTOMIKA it runs the field measurement campaigns and the analysis of the Split study area, the project’s Mediterranean case.
Funded by the Croatian Science Foundation
This research has been supported by the Croatian Science Foundation under project number IP-2025-02-REPLACE.
UTOMIKA — Urban Heat Islands of the Mediterranean and Continental Croatia. Hosted by the University of Zagreb, Faculty of Geodesy. Principal investigator: Andrija Krtalić. November 2025 – October 2028.