Soil moisture is an important hydrosphere parameter that regulates the partitioning of water and energy fluxes on land surfaces, with wide-ranging practical applications in numerical weather prediction, crop yield forecasts, drought monitoring, landslide risk assessment, etc. Since early 80's, passive microwave remote sensing has become a primary means to conduct frequent and large-scale mapping of soil moisture from space. Given its unique "see-through" capability under cloudy and lightly vegetated covers and high sensitivity to soil moisture at appropriate frequencies, microwave emission remains a preferred remote sensing tool for routine global estimates of soil moisture at high accuracy under all weather conditions, day and night. A more comprehensive review of the application of microwave radiometry to soil moisture retrieval can be found in the following book chapter:
Soil Moisture from the Advanced Microwave Scanning Radiometer (AMSR) Instruments in Comprehensive Remote Sensing, Earth Systems and Environmental Sciences, vol. 4, pp.191-223, 2018. Njoku, E., and S. Chan, Elsevier. ISBN: 9780128032206. https://doi.org/10.1016/B978-0-12-409548-9.10356-2
Antenna theory for real-aperture antennas dictates the antenna size and weight on a spaceborne platform for a given "footprint" size projected on the Earth's surface. The size of this footprint is a measure of the native resolution of the antenna. Clearly, the finer the footprint, the finer the resolvable spatial features will be, meaning that the resulting soil moisture estimates will be able to capture variation of soil moisture at a finer spatial scale.
The native resolution is a function of wavelength (λ) divided by antenna aperture diameter (D). For a given native resolution or solid angle of a given angular width, shorter wavelengths require a smaller (a lighter) antenna; longer wavelengths require a larger (and heavier) antenna. For this reason, it is not surprising that the history of passive microwave remote sensing of soil moisture from space started with smaller and lighter instruments that operate at shorter wavelengths (thus higher frequencies), and only relatively recently began to see the deployment of larger and heavier instruments that operate at longer wavelengths (thus lower frequencies).
This MEaSUREs (Making Earth System Data Records for Use in Research Environments) project focuses on the development and production of a global long-term soil moisture data record using passive microwave observations acquired by past and current radiometers acquired at L-band (1.4 GHz) and X-band (10.7 GHz) frequencies. The data record will be derived from L-band and X-band separately, allowing the resulting soil moisture data to more fully capture the distinct physics observed at these two frequency bands.
Radiometers that operate at L-band (1.4 GHz) frequencies include the ESA Soil Moisture and Ocean Salinity (SMOS) mission (2009-present) and the NASA Soil Moisture Active Passive (SMAP) mission (2015-present), and the NASA Aquarius mission (2011-2015). At present, SMOS and SMAP are the only operational L-band radiometry missions.
The value of L-band radiometry to passive remote sensing of soil moisture has been recognized since early 80's due to the high sensitivity of brightness temperature (TB) in response to soil moisture change at this frequency range. For example, TB from bare soils with a smooth surface could exhibit a change of ~90 K between dry soil (~5% water by volume) and wet soil (~35% water by volume) conditions. With a typical radiometric uncertainty of < 1 K in modern radiometers, the resulting large signal-to-noise ratio allows for accurate estimation of soil moisture. L-band radiometry also exhibits other advantages that are beneficial for soil moisture retrieval, such as its larger penetration depth of soil moisture signals over vegetated areas, and lower sensitivity to physical aspects (e.g. surface roughness, soil texture, diurnal variation of skin temperature, etc.) that would otherwise confound the interpretation of soil moisture retrieval.
The zeroth order radiative transfer model (a.k.a. the "tau-omega" (τ-ω) model) is a common inversion model that relates soil moisture to TB observations. Over the last few decades, its usefulness has been demonstrated at various spatial scales based on agreement between in situ ground truth and ground-based, airborne and spaceborne TB observations. Besides its decent modeling accuracy, the model is also relatively straightforward to deploy over large spatial scales due to its modest parameterization requirements. The model provides an end-to-end physics-based description of how the impact of soil moisture on soil dielectric properties affects the TB from soils as well as the TB interaction between soils and vegetation through scattering and absorption. Operationally, a full formulation of the model often requires additional ancillary data to provide TB correction before soil moisture retrieval is attempted. Common ancillary data include (1) land/water mask to correct for TB contamination due to water near coastlines or open-water bodies, (2) vegetation indices such as LAI or NDVI to correct for TB scattering, absorption and emission by vegetation, (3) surface roughness and soil temperature for surface emissivity estimation, and (4) soil texture as inputs to soil dielectric models.
Radiometer-based or passive soil moisture retrieval begins with solving for the estimated soil moisture from the τ-ω inversion model with actual TB observations and prior information from the ancillary data listed above as constraints. The retrieval process is often of iterative numerical nature, in that an initial numerical guess is used as a "seed" to search for an estimated soil moisture that predicts the actual TB observations according to the model either analytically as with single-channel TB observations or in a least-squared sense as with multi-channel TB observations that involve concurrent TB observations at multiple observation angles, frequency channels, and/or polarization planes. Both single-channel and multi-channel soil moisture retrieval algorithms have been extensively studied in field experiments and tested with airborne TB observations in field campaigns and spaceborne TB observations. State-of-the-art L-band soil moisture retrieval algorithms from some of these missions have been validated using in situ ground truth to demonstrate a retrieval accuracy of an unbiased RMSE of less than 0.04 m3/m3 and a correlation of greater than 0.80.
X-band (10.7 GHz) frequencies lack the deeper soil sensing depth that is possible L-band frequencies. Nonetheless, X-band TB observations represent the lion share of all available microwave observations to date since the dawn of passive remote sensing of soil moisture from space. these observations began with the Scanning Multichannel Microwave Radiometer (SMMR) on Nimbus-7 in 1978, followed by the TRMM Microwave Imager (TMI) on TRMM in 1997, the Advanced Scanning Microwave Radiometer on EOS Aqua (AMSR-E) in 2002, the WindSat on Coriolis in 2003, the Advanced Scanning Microwave Radiometer (AMSR2) on GCOM-W in 2012, and the GPM Microwave Imager (GMI) on GPM in 2014. Future radiometers that are equipped with X-band channels include the AMSR3 (successor to AMSR2) and the Copernicus Microwave Imaging Radiometer (CIMR). By their sheer abundance, soil moisture estimates derived from all X-band TB observations provide unparalleled frequent global coverage unattainable with L-band TB observations alone.
X-band geophysical inversion physics mirrors closely its L-band counterpart described above, sharing the same analytical τ-ω inversion model and ancillary data, differing mostly in their respective model parameterization schemes. Prior field campaigns and validation studies on X-band soil moisture estimates indicated that an unbiased RMSE of less than 0.06 m3/m3 is achievable.