Soil Moisture Retrieval Algorithms: The SMOS Case

Abstract : After the successful acquisition by a coarse L-band radiometer on board Skylab in the early seventies, the potential of L-band radiometry was made clear in spite of a strict limitation linked to minimum antenna dimensions required for appropriate spatial resolution. More than 20 years later new antenna concepts emerged to mitigate this physical constraint. The first to emerge, in 1997, and to become a reality, was the Soil Moisture and Ocean Salinity (SMOS) mission (0265 ; 0270). It is European Space Agency’s (ESA’s) second Earth Explorer Opportunity mission (Kerr et al., 2001), launched in November 2009. It is a joint program between ESA, CNES (Centre National d’Etudes Spatiales), and CDTI (Centro para el Desarrollo Tecnologico Industrial). SMOS carries a single payload, an L-band 2D interferometric radiometer in the 1400–1427 MHz protected band. This wavelength penetrates well through the atmosphere, and hence, the instrument probes the Earth surface emissivity from space. Surface emissivity can be related to the moisture content in the first few centimeters of soil, and after some surface roughness and temperature corrections, to the sea surface salinity over ocean. Soil moisture retrieval from SMOS observations with a required accuracy of 0.04 m3/m3 is challenging and involves many steps. The retrieval algorithms are developed and implemented in the ground segment, which processes level 1 and level 2 data. Level 1 consists mainly of directional brightness temperatures, while level 2 consists of geophysical products in swath mode, i.e., for successive imaging snapshots acquired by the sensor during a half orbit from pole to pole. Level 3 consists in composites of brightness temperatures, or geophysical products over time and space, i.e., global maps over given temporal periods from 1 day to 1 month. In this context, a group of institutes prepared the soil moisture and ocean salinity Algorithm Theoretical Basis Documents (ATBD), used to in operational soil moisture and sea salinity retrieval algorithms (Kerr et al., 2010a). The principle of the level 2 soil moisture retrieval algorithm is based on an iterative approach, which aims at minimizing a cost function. The main component of the cost function is given by the sum of the squared weighted differences between measured and modeled brightness temperature (TB) at horizontal and vertical polarizations, for a variety of incidence angles. The algorithm finds the best set of parameters, e.g., soil moisture (SM) and vegetation characteristics, which drive the TB model and minimizes the cost function. From this algorithm, a more sophisticated one was developed to take into account multiorbit retrievals (i.e., level 3). Subsequently, after several years of data acquisition and algorithm improvements, a neural network approach was developed so as to be able to infer soil moisture fields in near-real time. In parallel, several simplified algorithms were tested, the goal being to achieve a seamless transition with other sensors, along with other studies targeted on specific targets such as dense forests, organic rich soils, or frozen and snow-covered grounds. Finally, it may be noted that most of these approaches deliver not only the surface soil moisture but also other quantities of interest such as vegetation optical depth, surface roughness, and surface dielectric constant. The goal of this article is to give an overview of these different approaches and corresponding results and adequate references for those wishing to go further. Sea surface salinity is not covered in this article, while the focus is SMOS data.
Complete list of metadatas

https://hal-insu.archives-ouvertes.fr/insu-01676542
Contributor : Catherine Cardon <>
Submitted on : Friday, January 5, 2018 - 5:28:02 PM
Last modification on : Friday, October 25, 2019 - 6:44:03 PM

Identifiers

  • HAL Id : insu-01676542, version 1

Citation

Yann H. Kerr, Ali Mahmoodi, Arnaud Mialon, A. Al Biltar, Nemesio Rodriguez-Fernandez, et al.. Soil Moisture Retrieval Algorithms: The SMOS Case. Shunlin Liang (eds.). Comprehensive Remote Sensing, 4, Elsevier, pp.156-190, 2018, 978-0-12-803221-3. ⟨insu-01676542⟩

Share

Metrics

Record views

818