A multi-frequency framework for soil moisture retrieval from time series radar data

Liujun Zhu, Jeffrey P. Walker, Leung Tsang, Huanting Huang, Nan Ye, Christoph Rüdiger

Research output: Contribution to journalArticleResearchpeer-review

3 Citations (Scopus)


The increased availability of spaceborne radar data projected over the next decade provides a great opportunity for operational soil moisture mapping with high spatial (<50 m) and temporal (<3 days) resolution, by combining the data from multiple SAR missions. Accordingly, a multi-frequency soil moisture retrieval framework has been proposed, being applicable for SAR missions operating at the commonly used remote sensing frequencies of L-, C- and X-band. A combination of numerical, physical and semi-empirical scattering models was selected to build a series of forward modeling look up tables (LUTs) covering the typical radar configurations and nature surface conditions. An unsupervised change detection method was integrated to identify areas with abrupt roughness and vegetation changes, so that time-series data collected from different SARs can be combined with the assumption of time-invariant roughness and vegetation. The multi-frequency backscattering coefficient (σ0) with negligible scattering from soil surface (equivalent to calibration uncertainty) was then removed before soil moisture retrieval. Finally, soil moisture retrieval was carried out independently for each landcover type using an optimization method and forward LUTs. Evaluation based on the Soil Moisture Active Passive Experiment-5 dataset consisting of L-band airborne data, C-band RADARSAT-2 data and X-band COSMO-SkyMed data showed an acceptable overall root mean square error (RMSE) of 0.058 cm3/cm3 at the paddock scale (~0.1 – 0.5 km). The comparison with single and dual frequency retrieval suggests that multi-frequency retrieval is not necessary to have the highest accuracy. However, it is still valuable to joint use multi-frequency data consider the limited deterioration in accuracy and the significantly enhanced temporal resolution.

Original languageEnglish
Article number111433
Number of pages15
JournalRemote Sensing of Environment
Publication statusPublished - 15 Dec 2019


  • Multi-frequency
  • Multi-temporal
  • Soil moisture
  • Synthetic aperture radar

Cite this