Daily Air Temperature Mapping in Urban Areas from Satellite Land Surface Temperature and ERA5 Data

Urban areas are increasingly affected by environmental and public health challenges driven by rising temperatures. Due to ongoing climate change and the increased presence of greenhouse gases in the atmosphere, the frequency and duration of heatwaves are expanding. These phenomena have serious implications for human health, particularly among vulnerable populations such as the elderly, individuals with pre-existing cardiovascular or respiratory conditions, and disadvantaged socio-economic communities.

Air temperature at 2 meters above the surface is a critical variable for assessing climate change impacts and thermal stress, especially in densely populated urban environments. However, ground-based observations of air temperature are often sparse, mostly concentrated in developed regions, and frequently suffer from temporal gaps. This spatial and temporal inconsistency limits our ability to monitor urban thermal conditions effectively. On the other hand, satellite data provide continuous and global measurements of land surface temperature (LST), but do not directly measure air temperature. Since LST and air temperature are not equivalent, translating satellite-based LST into reliable air temperature estimates remains challenging.

In this work, we developed a statistical approach that leverages MODIS satellite observations and ERA5-Land model data across the 70 largest and most populous cities worldwide, geographically distributed with a maximum of three cities per country to prevent national over-representation and ensure global balance. The dataset spans from 2012 to 2023 and is categorized by three latitudinal zones equatorial (0 to ±15°), tropical-temperate (±15° to ±45°), and temperate-subpolar (±45° to ±75°) and by month, distinguishing between day and night observations.

For each geographical and temporal class, we fit the parameters of the equation:

T_air = a * LST_day + b * LST_night + c   [1]

This resulted in parameter triplets (a, b, c) specific to each month and latitude band. These parameters were then applied to MODIS and VIIRS data, for the year 2024, to assess the inter-sensor scalability. The resulting air temperature estimates are obtained at the native spatial resolution of the input datasets (1 kilometer). The method operates on a daily basis, leveraging both daytime and nighttime satellite acquisitions to ensure consistent and temporally detailed air temperature estimates, an essential feature for capturing urban thermal dynamics and short-term variability, such as the urban heat island effect.

The model was validated using data from the year 2024 for the same cities, with the ERA5-Land dataset (https://cds.climate.copernicus.eu/datasets/reanalysis-era5-land?tab=overview) serving as a reference. Pearson correlation coefficients ranged from 84% to 93% for daytime temperatures and from 77% to 93% for nighttime temperatures.

The approach is also adaptable to ongoing and future satellite missions with improved spatial resolution (e.g. ECOSTRESS). Looking toward future developments, the integration of Artificial Intelligence could further enhance this methodology by incorporating additional weather variables, improving the representation of complex ambient conditions. This work represents a promising advancement in the field of high-resolution, daily thermal comfort assessments across urban areas, offering a scalable and flexible tool for heat-related stress monitoring.

[1] Hooker, J. et al., A. A global dataset of air temperature derived from satellite remote sensing and weather stations. (2018). https://doi.org/10.1038/sdata.2018.246