Pérez-Planells et al. (2026) Annual and diurnal temperature cycle modelling of a merged multi-annual ECOSTRESS and Landsat land surface temperature dataset

Identification

• Journal: International Journal of Applied Earth Observation and Geoinformation
• Year: 2026
• Date: 2026-04-06
• Authors: Lluís Pérez-Planells, Frank-M. Göttsche, Jan Cermak
• DOI: 10.1016/j.jag.2026.105276

Research Groups

• Mediterranean Center for Environmental Studies (CEAM), Meteorology and Climatology Area, Paterna, Spain
• IMKASF, Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany

Short Summary

This study models annual and diurnal land surface temperature (LST) dynamics using a five-parameter Annual-Diurnal Temperature Cycle (ADTC) model fitted to a novel six-year (2018–2023) merged high-resolution (70 m) ECOSTRESS and Landsat LST dataset over four European cities. The high-resolution ADTC parameters demonstrate good agreement with MODIS-derived parameters for most variables and show clear potential for detailed urban thermal studies and land cover change monitoring.

Objective

• To investigate the ADTC model’s performance with a merged multi-annual dataset containing high-resolution data from arbitrary overpasses.
• To evaluate the high-resolution ADTC parameters and study potential links to other surface features.

Study Configuration

• Spatial Scale: Four European cities (Karlsruhe, Paris, Madrid, Valencia). High-resolution LST data (70 m for ECOSTRESS, 30 m for Landsat, upscaled to 1 km for comparison with MODIS).
• Temporal Scale: Six years (2018–2023) of multi-annual LST data.

Methodology and Data

• Models used: Annual-Diurnal Temperature Cycle (ADTC) model, combining the ATCsza model for annual dynamics and the GOT09 model for diurnal dynamics. The ADTC model uses five controlling parameters: annual minimum temperature (TA,0), annual amplitude of the annual minimum temperature (TA,a), annual maximum daily amplitude (TAD,a), annual mean thermal noon (tAD,m), and lag with respect to summer solstice (dx).
• Data sources:
  • Satellite: ECOSTRESS LST and Emissivity (LST&E) product (Level 2, 70 m spatial resolution). Landsat 8/9 LST product (USGS, 30 m spatial resolution). MODIS LST products (MOD11A1 and MYD11A1, Level 3, 1 km spatial resolution) for comparison.
  • Auxiliary: Corine Land Cover (CLC) classification map (version CLC2018, 250 m spatial resolution).

Main Results

• The mean Root Mean Square Error (RMSE) of the modelled LST over the study areas was 3.8 K.
• Overall RMSE of the ADTC fits for the high-resolution data was 3.8 ± 0.7 K, with lowest RMSEs over water bodies (~2 K) and highest over agricultural areas (4.1 ± 0.5 K).
• Comparison of upscaled high-resolution ADCPs with MODIS ADCPs showed good agreement for TA,0 (RRMSE 1.6 K), TA,a (RRMSE 2.0 K), tAD,m (RRMSE 0.2 hours), and RMSE of the fits (RRMSE 0.4 K).
• Larger RRMSE values were observed for TAD,a (6.5 K) and dx (6.8 days), attributed to insufficient high-resolution observations around thermal noon.
• ADCPs exhibited distinct patterns related to land cover, allowing for characterization of water bodies, forests, and agricultural areas.
• The ADTC model parameters showed potential for discriminating between different urban elements (e.g., TAD,a showed differences up to 8.8 K between industrial and vegetated areas in Paris).
• Topographic effects on LST dynamics were successfully captured by ADCPs, with higher TA,0 and TAD,a on south-facing slopes and shifted tAD,m on west/east-facing slopes.

Contributions

• First-time application of the ADTC model to a six-year merged multi-annual high-resolution (70 m) ECOSTRESS and Landsat LST dataset, leveraging ECOSTRESS's variable overpass times to sample the diurnal cycle.
• Comprehensive evaluation of high-resolution ADTC parameters against MODIS-derived parameters, quantifying agreement and identifying limitations related to data sampling around thermal noon.
• Demonstrated the capability of ADTC parameters to characterize land covers and urban elements based on their thermal properties, as well as to capture topographic effects on LST dynamics.
• Highlighted the potential of ADTC parameters for detailed urban thermal studies, land cover change monitoring, and as predictors for land cover classification.

Funding

• The North American Space Agency (NASA) and the U.S. Geological Survey provided the ECOSTRESS and Landsat LST products.

Citation

@article{PérezPlanells2026Annual,
  author = {Pérez-Planells, Lluís and Göttsche, Frank-M. and Cermak, Jan},
  title = {Annual and diurnal temperature cycle modelling of a merged multi-annual ECOSTRESS and Landsat land surface temperature dataset},
  journal = {International Journal of Applied Earth Observation and Geoinformation},
  year = {2026},
  doi = {10.1016/j.jag.2026.105276},
  url = {https://doi.org/10.1016/j.jag.2026.105276}
}