Miao et al. (2026) SM2RAIN–dual: a global rainfall fusion product derived from multi-source satellite soil moisture observations

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-03-07
• Authors: Linguang Miao, Zushuai Wei, Lingkui Meng, Lei Li, Wenxiao Zhang, Xi Wang, Hao Wang, Zhe Wang, W. J. Zhu, Luca Brocca
• DOI: 10.1016/j.jhydrol.2026.135261

Research Groups

• School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430079, China
• School of Artificial Intelligence, Jianghan University, Wuhan 430056, China
• Research Institute for Geo-Hydrological Protection, National Research Council, Perugia, Italy

Short Summary

This study addresses regional disparities in SM2RAIN-derived rainfall estimates by developing a rainfall data fusion scheme using multi-source satellite soil moisture products. The research generated a global, more reliable rainfall product called SM2RAIN–Dual, which combines SMAP and ASCAT data.

Objective

• To address region-dependent uncertainties and limited global applicability of single-source SM2RAIN rainfall estimates by developing a multi-source satellite soil moisture rainfall data fusion scheme to generate a global rainfall product.

Study Configuration

• Spatial Scale: Global, with in situ observations from Italy, the United States, Australia, and India. The final product has a spatial resolution of 10 km.
• Temporal Scale: From 2015 to 2022, with daily intervals.

Methodology and Data

• Models used: SM2RAIN algorithm, Triple-collocation (TC) analysis.
• Data sources:
  • Satellite soil moisture products: Soil Moisture Active and Passive (SMAP), Advanced SCATterometer (ASCAT), Advanced Microwave Scanning Radiometer 2 (AMSR2), Soil Moisture and Ocean Salinity (SMOS), European Space Agency Climate Change Initiative (ESA CCI).
  • In situ observations: 1009 uniformly distributed points across Italy, the United States, Australia, and India.
  • Reference rainfall products: MSWEP, GPCC, GPM-LR.

Main Results

• For individual satellite products, SMAP-based rainfall estimates showed the highest Pearson’s correlation coefficient (R) in the United States, while ASCAT-based estimates achieved the best R performance in Italy and superior Nash–Sutcliffe efficiency (NS) in India.
• Fusion at the rainfall level outperformed fusion at the soil moisture level. The combination of SMAP and ASCAT yielded the best rainfall-level results (median R = 0.62; median NS = 0.38), surpassing ESA CCI-derived rainfall (median R = 0.55; median NS = 0.34).
• A fused remote sensing rainfall product, SM2RAIN–Dual, was generated from 2015 to 2022 based on SMAP and ASCAT, featuring a spatiotemporal resolution of 10 km and daily intervals. Triple-collocation analysis with GPCC and GPM-LR showed that SM2RAIN–Dual achieved lower error variance in 88% of global regions and higher correlation in 34% of regions compared to the other two products.

Contributions

• Introduces a novel rainfall-level fusion strategy that overcomes the regional limitations of single-source satellite rainfall products.
• Provides a more reliable global precipitation input (SM2RAIN–Dual) for hydrological modeling, particularly in areas with sparse ground-based observations.

Funding

• Not specified in the provided text.

Citation

@article{Miao2026SM2RAINdual,
  author = {Miao, Linguang and Wei, Zushuai and Meng, Lingkui and Li, Lei and Zhang, Wenxiao and Wang, Xi and Wang, Hao and Wang, Zhe and Zhu, W. J. and Brocca, Luca},
  title = {SM2RAIN–dual: a global rainfall fusion product derived from multi-source satellite soil moisture observations},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.135261},
  url = {https://doi.org/10.1016/j.jhydrol.2026.135261}
}