Rautiainen et al. (2025) An operational SMOS soil freeze–thaw product

Identification

• Journal: Earth system science data
• Year: 2025
• Date: 2025-10-15
• Authors: Kimmo Rautiainen, Manu Holmberg, Juval Cohen, Arnaud Mialon, Mike Schwank, Juha Lemmetyinen, Antonio de la Fuente, Yann H. Kerr
• DOI: 10.5194/essd-17-5337-2025

Research Groups

• Finnish Meteorological Institute (FMI), Helsinki, Finland
• CESBIO, Université Toulouse 3, CNES/CNRS/IRD/INRAe/UPS, Toulouse, France
• GAMMA Remote Sensing Research and Consulting AG, Gümligen, Switzerland
• Swiss Federal Institute WSL, Birmensdorf, Switzerland
• ESA – ESRIN, Frascati, Italy

Short Summary

This paper introduces an updated operational SMOS Level-3 Soil Freeze–Thaw (FT) product, detailing its L-band passive microwave-based classification algorithm with enhanced noise reduction. Validation against in situ measurements and reanalysis data demonstrates improved accuracy in detecting the day of first freezing, providing crucial data for carbon cycle studies in high-latitude environments.

Objective

• To introduce and detail the updated SMOS Level-3 (L3) Soil Freeze–Thaw (FT) product and its threshold-based classification algorithm.
• To validate the product's ability to detect soil freeze–thaw transitions, particularly the day of first freezing, with improved accuracy against in situ and reanalysis data.

Study Configuration

• Spatial Scale: Global, focusing on Northern Hemisphere land surfaces, with data gridded to an Equal-Area Scalable Earth 2 (EASE-2) grid at a 25 km × 25 km resolution.
• Temporal Scale: Daily monitoring of soil freeze–thaw state. SMOS data used for freezing periods between 2010 and 2024, with reference values derived from 1 January 2014 to 4 September 2023. SMOS provides global coverage twice in 3 days, with daily overflights in northernmost land areas.

Methodology and Data

• Models used:
  • Threshold-based classification algorithm for soil freeze–thaw detection using normalized polarization ratio (NPR).
  • Kalman filtering for temporal noise reduction in NPR time series.
  • Soil Freeze–Thaw Index (SFTI) for in situ data comparison, derived from soil volumetric liquid water content and soil temperature.
  • Regression model (Gregow et al., 2011) for estimating soil frost depth (SFD) based on freezing degree days and snow depth for representativeness checks.
• Data sources:
  • Satellite:
    • Soil Moisture and Ocean Salinity (SMOS) satellite L-band (1.4 GHz) passive microwave brightness temperature measurements.
    • CATDS (Centre Aval de Traitement des Données SMOS) Level-3 brightness temperatures (L3TB), version 331.
  • Observation/Ancillary:
    • ECMWF (European Centre for Medium-Range Weather Forecasts) 2 m air temperature data (Atmospheric Model High-Resolution 10 d Forecast for operational processing, ERA5-Land for reprocessing).
    • USNIC (United States National Ice Center) IMS (Interactive Multisensor Snow and Ice Mapping System) Daily Northern Hemisphere Snow and Ice Analysis (4 km resolution, reprojected to 25 km).
    • ISMN (International Soil Moisture Network) in situ soil moisture and soil temperature data (from SNOTEL, SCAN, USCRN, RISMA, BNZ LTER, FMI networks) for validation.
    • ESA CCI Land Cover time series v2.0.7 (1992–2015) for land cover distribution.
  • Reanalysis:
    • ERA5-Land reanalysis data (2 m air temperature, soil temperature in layer 1 (0–7 cm depth), snow depth) for validation and auxiliary data.

Main Results

• The updated SMOS FT product (version 3.01) provides daily monitoring of the freeze–thaw state of Northern Hemisphere land surfaces at a 25 km spatial resolution.
• Validation against in situ data shows improved accuracy in detecting the day of first freezing (DoFF). For descending orbits, the bias is -6.3 days, Pearson correlation (R) is 0.71, and standard deviation of difference (SDD) is 18.6 days. For ascending orbits, the bias is -5.0 days, R is 0.71, and SDD is 19.2 days. Ascending orbits (morning conditions) detect freezing slightly earlier than descending orbits (evening conditions).
• Comparison with ERA5-Land soil temperature data reveals broadly similar DoFF patterns, but the SMOS FT product generally estimates later freezing (median differences of +10.7 days for ascending and +12.3 days for descending) compared to ERA5-Land.
• Discrepancies between SMOS FT and ERA5-Land are more pronounced in regions with dense forest cover, strong radio frequency interference (RFI), and complex terrain.
• Limitations include reduced effectiveness in dry soils, thin soil layers, and during spring melt periods due to wet snow, which can obscure actual soil thawing. RFI, particularly over Eurasia, significantly hampers data continuity and quality, especially after spring 2022.

Contributions

• Introduction of an updated, operational SMOS Level-3 Soil Freeze–Thaw (FT) product (version 3.01) with enhanced noise removal through temporal Kalman filtering.
• Comprehensive validation of the product's accuracy in identifying the day of first freezing (DoFF) using a diverse set of in situ soil moisture and temperature networks and ERA5-Land reanalysis.
• Demonstration of the product's utility for carbon cycle studies, particularly for improving methane flux estimates in high-latitude environments (e.g., integration into the CarbonTracker Europe inverse modelling system).
• Detailed description of the threshold-based classification algorithm, including data selection, quality filtering, and the integration of auxiliary datasets for robust soil state determination.

Funding

• European Space Agency (ESA – ESRIN) projects:
  • SMOS L3 Freeze/Thaw L3 Data Service (ESRIN contract no. 4000124500/18/I-EF SMOS F/T Service)
  • ESA SMOS ESL2020+ (ESRIN contract no. 4000130567/20/I-BG)
• Research Council of Finland:
  • Academy Research Fellowship project EMonSoil (grant no. 364034)
  • Flagship programmes Advanced Mathematics for Sensing Imaging and Modelling (FAME; grant no. 359196)

Citation

@article{Rautiainen2025operational,
  author = {Rautiainen, Kimmo and Holmberg, Manu and Cohen, Juval and Mialon, Arnaud and Schwank, Mike and Lemmetyinen, Juha and Fuente, Antonio de la and Kerr, Yann H.},
  title = {An operational SMOS soil freeze–thaw product},
  journal = {Earth system science data},
  year = {2025},
  doi = {10.5194/essd-17-5337-2025},
  url = {https://doi.org/10.5194/essd-17-5337-2025}
}