Hu et al. (2025) Global retrieval of harmonized microwave land surface emissivity leveraging multi-sensor measurements from GMI, AMSR2 and MWRIs

Identification

Journal: Remote Sensing of Environment
Year: 2025
Date: 2025-12-08
Authors: Jiheng Hu, Rui Li, Peng Zhang, Yu Wang, Shengli Wu, Husi Letu, Fuzhong Weng
DOI: 10.1016/j.rse.2025.115169

Research Groups

School of Earth and Space Sciences, CMA-USTC Laboratory of Fengyun Remote Sensing, University of Science and Technology of China
Meteorological Observation Center, China Meteorological Administration
National Satellite Meteorological Center, China Meteorological Administration
State Key Laboratory of Remote Sensing Science, Aerospace Information Research Institute, Chinese Academy of Sciences
Earth System Modeling and Prediction Centre, China Meteorological Administration
Department of Climate and Space Sciences and Engineering, University of Michigan

Short Summary

This study develops an innovative framework to retrieve a global harmonized microwave land surface emissivity (MLSE) database by integrating measurements from five passive microwave sensors (GMI, AMSR2, and three MWRIs) and six geostationary visible/infrared imagers. The framework, employing a simultaneous conical overpass (SCO) recalibration technique, achieves exceptionally strong consistency among the harmonized MLSE subsets, with Pearson R ≈0.95, RMSD <0.011, and mean bias within ±0.005.

Objective

To generate a global harmonized microwave land surface emissivity (MLSE) database for the GPM era (2014–2023) by mitigating discrepancies from individual sensors through an innovative retrieval framework and sophisticated cross-sensor calibrations.

Study Configuration

Spatial Scale: Global coverage (within approximately ±68° latitude for GMI, other sensors also global). Output data is projected to a 0.1° × 0.1° climate model grid. Passive microwave sensor ground resolutions range from 3 km × 5 km to 85 km × 51 km. Geostationary cloud mask resolutions are 2 km to 5 km at nadir.
Temporal Scale: January 2014 – December 2023. Hourly reanalysis data, daily and monthly satellite products. Diurnal variations of emissivity are also investigated.

Methodology and Data

Models used:
- Radiative Transfer for TIROS Operational Vertical Sounder (RTTOV, version 13.2) for estimating atmospheric radiative contributions.
- Two-stream (2S) microwave land surface RT module, incorporated into the Community Radiative Transfer Model (CRTM), for simulating surface emissivity over various landscapes.
- Sine function model for fitting diurnal emissivity curves.
Data sources:
- Passive Microwave Sensors (Top-of-Atmosphere Brightness Temperatures):
  - GPM-CO/GMI (GPM L1C V07 recalibrated data)
  - GCOM-W1/AMSR2 (GPM L1C V07 recalibrated data)
  - Fengyun-3B/MWRI (recalibrated data)
  - Fengyun-3C/MWRI (recalibrated data)
  - Fengyun-3D/MWRI (recalibrated data)
- Geostationary Visible/Infrared Imagers (Clear-Sky Masks):
  - NOAA GOES-16/ABI (Level 2 Clear Sky Mask data)
  - JMA Himawari-8/AHI (Level 2 cloud property product)
  - JMA Himawari-9/AHI (Level 2 cloud property product)
  - EUMETSAT MSG-1/SEVIRI (Cloud mask data)
  - EUMETSAT MSG-2/SEVIRI (Cloud mask data)
  - EUMETSAT MSG-3/SEVIRI (Cloud mask data)
- Reanalysis Data (Geophysical Parameters):
  - ECMWF ERA5 hourly reanalysis (atmospheric profiles of specific humidity, air temperature, geopotential height).
  - ECMWF ERA5-Land hourly reanalysis (surface skin temperature, surface pressure, snow cover).
- Evaluation Data:
  - In-situ measurements: Truck-mounted radiometer (X-band brightness temperatures) from the Soil Moisture Experiment in the Luan River (grass and crop fields).
  - Reference emissivity datasets: NASA/GSFC GMI emissivity (MunchakMLSE), CUNY/CREST AMSR-E emissivity (CRESTMLSE), AER AMSR-E emissivity (AERMLSE), TELSEM emissivity atlas (TELSEMMLSE).
- Auxiliary Data:
  - MODIS land cover data set MCD12C1.
  - MODIS MYD13A3 (16-day composited normalized difference vegetation index - NDVI).
  - NASA GPM IMERG level 3 rainfall product (monthly rainfall).
  - MODIS MOD10A2 Snow Cover data set.
  - FAO 5-min 16-category soil texture map.
  - Global Land Surface Satellite (GLASS) product (leaf area index - LAI, fractional vegetation cover - FVC).

Main Results

The harmonized MLSE datasets from the five passive microwave sensors exhibit exceptionally strong self-consistency, with Pearson R ≈0.95, RMSD <0.011, and mean bias within ±0.005 across all channels. AMSR2MLSE showed the highest correlation with GMIMLSE (R > 0.99, RMSD < 0.005).
Integration of rescaled MLSEs from multiple sensors significantly improves the robustness of reconstructed diurnal variations, particularly for horizontal polarization, with an improvement of 0.002 to 0.005.
Validation against in-situ radiometer measurements at 10.65 GHz shows errors generally within ±0.01 for vertical polarization and a systematic underestimation of approximately −0.02 for horizontal polarization (mean bias = −0.019, RMSE = 0.021), with unbiased RMSE within 0.01.
Global evaluation against four reference emissivity datasets demonstrates strong consistencies: R values typically range from 0.8 to 0.95, with RMSD generally below 0.015 (~1.5 %) for vertical polarization and below 0.02 (~2 %) for horizontal polarization on a monthly scale.
Water fraction is identified as the most dominant factor impacting intercalibration performance, with residual differences rising from approximately −0.0005 in dry areas to 0.01 at 20–30 % water fraction.

Contributions

Established the first global harmonized microwave land surface emissivity (MLSE) database for the GPM era (2014–2023) by integrating observations from five passive microwave sensors (GMI, AMSR2, and three MWRIs).
Developed an innovative framework for MLSE retrieval that includes sophisticated cross-sensor calibrations using the simultaneous conical overpass (SCO) technique and leverages multi-geostationary visible/infrared observations for comprehensive clear-sky masking.
Quantitatively demonstrated the significant improvement in consistency among MLSEs from different sensors, achieving high accuracy that meets the demands of physics-based precipitation retrieval algorithms.
Provided the first ground-based validation of satellite-retrieved MLSE, showing reliable accuracy against in-situ radiometer measurements in grass and crop fields.
Revealed the potential of the harmonized MLSE dataset for tracking large-scale sub-daily surface ecohydrological dynamics and providing crucial observational constraints for land surface models and precipitation algorithms.

Funding

National Natural Science Foundation of China NSFC (Grant No. 42330602, 42275139, 41830104)
National Key Research and Development Program of China (Grant No. 2024YFF0809401)
Anhui Provincial Natural Science Foundation (Grant No. 2208085UQ02)
Innovation Center for Fengyun Meteorological Satellite Special Project (Grant No. FY-APP-ZX-2022.0211)
CMA-USTC Joint Laboratory of Fengyun Remote Sensing
The Youth Innovation Team of the “Fengyun Satellite Remote Sensing Product Verification” of the China Meteorological Administration (CMA2023QN12)

Citation

@article{Hu2025Global,
  author = {Hu, Jiheng and Li, Rui and Zhang, Peng and Wang, Yu and Wu, Shengli and Letu, Husi and Weng, Fuzhong},
  title = {Global retrieval of harmonized microwave land surface emissivity leveraging multi-sensor measurements from GMI, AMSR2 and MWRIs},
  journal = {Remote Sensing of Environment},
  year = {2025},
  doi = {10.1016/j.rse.2025.115169},
  url = {https://doi.org/10.1016/j.rse.2025.115169}
}

Original Source: https://doi.org/10.1016/j.rse.2025.115169