Well, plenty of literature on this topic. First, it depends on the spatial and temporal scale in which you are interested to. Then, where you are? Finally, temporal coverage and timeliness.
Generally, SMAP, ASCAT, and SMOS currently provide the best satellite soil moisture products
The satellite itself isn't as important as the bands available. For moisture content the best means of capture are through NIR (750 nm-950 nm) and SWIR (1400nm-3000nm) wavelengths. So any multispectral or hyperspectral satellite with these wavelengths will work, but ultimately the exact choice comes down to the specific goals of your project.
Well, plenty of literature on this topic. First, it depends on the spatial and temporal scale in which you are interested to. Then, where you are? Finally, temporal coverage and timeliness.
Generally, SMAP, ASCAT, and SMOS currently provide the best satellite soil moisture products
The SMAP Level-2 and Level-3 standard and enhanced radiometer soil moisture products from the NASA National Snow and Ice Data Center Distributed Active Archive Center (NSIDC DAAC) are being updated in summer 2019. The Level-2 updates will introduce an improved version of the existing Dual Channel Algorithm, DCA (option 3). The new version, known as M-DCA, achieves better retrieval performance by modeling the mixing of vertically and horizontally polarized brightness temperature channels and using new estimates of single-scattering albedo and roughness coefficients. M-DCA (option 3) will supersede optional algorithms MPRA (option 4) and E-DCA (option 5), so the following data fields will be removed from the SMAP Level-2 standard and enhanced radiometer soil moisture products:
soil_moisture_option4
vegetation_opacity_option4
retrieval_qual_flag_option4
soil_moisture_option5
vegetation_opacity_option5
retrieval_qual_flag_option5
In addition, a clay fraction field and a bulk density field will be added to both the Level-2 and Level-3 products. The baseline retrieval algorithm remains SCA-V, which is unchanged from the previous release.
Access to each data set and related documentation can be found at:
SMAP L2 Radiometer Half-Orbit 36 km EASE-Grid Soil Moisture, Version 5
DOI: https://doi.org/10.5067/SODMLCE6LGLL
SMAP Enhanced L2 Radiometer Half-Orbit 9 km EASE-Grid Soil Moisture, Version 2
DOI: https://doi.org/10.5067/K4A1SNL5DLON
SMAP L3 Radiometer Global Daily 36 km EASE-Grid Soil Moisture, Version 5
DOI: https://doi.org/10.5067/ZX7YX2Y2LHEB
SMAP Enhanced L3 Radiometer Global Daily 9 km EASE-Grid Soil Moisture, Version 2
Well, see our very recent paper on Lake Urmia region: Jalilvand, E., Tajrishy, M., Hashemi, S.A.G., Brocca, L. (2019). Quantification of irrigation water using remote sensing of soil moisture in a semi-arid region. Remote Sensing of Environment, 231, 111226, doi:10.1016/j.rse.2019.111226. https://doi.org/10.1016/j.rse.2019.111226
SMAP Level-3 soil moisture data has the best performance in North of China. And SMAP data has high resolution. But soil moisture from remote sensing satellite could not monitor deep level soil moisture. Is there any data sets with good performance in 10-20cm level soil moisture ?
It depends on many factors such as the extension of your study area, the study period, and the purpose of your research. Beside SMOS and SMAP, you may also consider LDAS products (e.g., GLDAS, FLDAS) available at: https://ldas.gsfc.nasa.gov/
The LDAS data are produced by integrating satellite- and ground-based observational data products. The data are available at different spatial resolution ranging from 0.125º × 0.125º to 1.0º × 1.0º .
It is better to apply soil moisture downscaling algorithm if you need higher resolution. And soil moisture products also can be produced from optical satellite, such as Landsat8 or Sentinel-2, but with lower accuracy.
I would add another perspective to the interesting discussion here. The performance can also depend on the density of in-situ sensors used for validation.
In our study, we validated eight passive, two active, and one merged soil moisture products over CONUS. These include operational SMAP, SMOS, AMSR2, ASCAT products. We considered 1058 ISMN stations, which include sparse networks.
Our results indicate that SMAP performs well, followed by SMOS and ASCAT with comparable accuracy over CONUS. AMSR2 (JAXA algorithm) product accuracy got affected due to the fact that the algorithm doesn't filter the retrievals during winter periods (snow, frozen ground).
The point I would like to convey is that the results may vary if we consider only dense networks for validation.
Go through the enclosed research article, this will provide you, information about which reanalysis datasets are best for hydrological studies over the Indian region.
Article Does ERA‐5 Outperform Other Reanalysis Products for Hydrolog...