Surface rock effects on soil moisture retrieval from L-band passive microwave observations
Introduction
Soil moisture is a key variable in global water, energy, and carbon cycling, which is fundamental to hydrology, meteorology, and agriculture (Sellers et al., 1997). Due to its high variability in time and space, it is difficult to measure or predict the spatial and temporal distribution of soil moisture at regional and global scales (Crow et al., 2012; Ryu and Famiglietti, 2006). However, the first satellite dedicated to measuring global soil moisture was launched on November 2nd, 2009. This Soil Moisture and Ocean Salinity (SMOS) mission, led by the European Space Agency (ESA) in collaboration with the Centre National d'Etudes Spatiales (CNES) in France and the Centro para el Desarrollo Tecnologico Industrial (CDTI) in Spain, measures soil water content in the top ~5 cm soil every 2 to 3 days with a target accuracy of better than 0.04 m3/m3, using a 2-D interferometric radiometer operating at L-band (1.413 GHz; Kerr et al., 2010). Likewise the National Aeronautics and Space Administration (NASA) developed the Soil Moisture Active Passive (SMAP) mission to measure soil moisture using a combination of L-band (1.41 GHz) radiometer and L-band (1.26 GHz) radar to increase the resolution of soil moisture products from 40 km to 10 km (Entekhabi et al., 2010). Since the approach was to first downscale the brightness temperature measurements according to the spatial patterns in the radar data, with the soil moisture then retrieved using the standard passive microwave algorithms, any errors in the radiometer data at the native resolution would be carried through to the higher resolution products. Even though the radar unfortunately malfunctioned shortly after launch, this same issue applies to the alternative downscaling approaches currently being proposed.
The passive microwave remote sensing technique has been adopted for soil moisture measurement as it is unaffected by cloud, has a direct relationship with soil moisture through the soil dielectric constant, and has a reduced sensitivity to land surface roughness and vegetation canopy, compared with optical, infrared, and active microwave techniques (Jackson and Schmugge, 1989; Njoku et al., 2002). However, the spatial resolution of L-band space-borne radiometer measurements is restricted by the size of antenna, meaning that a resolution on the order of ~40 km is achieved based on the current level of antenna technology. For the SMOS mission, 69 elementary radiometers are distributed along three Y-shaped arms of 4.5 m in length, to produce an elliptical shaped footprint of ~43 km in size using synthetic aperture techniques. In contrast, the SMAP mission uses a rotating mesh antenna of 6 m (20 ft) in diameter to directly observe a single brightness temperature of ~40 km spatial resolution.
The volumetric soil moisture of each SMOS and SMAP radiometer pixel is subsequently retrieved from the brightness temperature observations through the use of radiometric transfer models. Although the SMOS soil moisture retrieval algorithm is capable of three different surface types (bare soil, vegetated soil, and forest) within the SMOS footprint (Kerr et al., 2010; Kerr et al., 2012), the accuracy of the SMOS and SMAP soil moisture retrieval will suffer from land surface heterogeneity at such a coarse scale. In addition, the impacts of surface rock, standing water, and urban areas within the sensor's field of view have not been well studied and accounted for in the current soil moisture retrieval models, causing an uncertainty in the soil moisture retrieval accuracy (Delwart et al., 2008). While a few model simulation studies have been performed to explore the rock cover fraction threshold for the SMOS target soil moisture accuracy of 0.04 m3/m3 (Kerr et al., 2010; Loew, 2008), there has been no rigorous assessment of the expected rock impact globally. The rock fraction thresholds of 0.11 and 0.15–0.20 were obtained from Kerr et al. (2010) and Loew (2008) respectively, assuming that rock behaves as very dry bare soil with a fixed dielectric constant and roughness, but these results have not been confirmed through field data.
According to dielectric constant measurements on a range of rock types at a frequency of 400 MHz and 35 GHz, the real part of the dielectric constant of rock ranges from 2.4 to 9.6 (Ulaby et al., 1986). A value of 5.7 - j × 0.074 has been suggested in the SMOS Algorithm Theoretical Basis Document (ATBD) as an appropriate value for the dielectric constant of rock (Kerr et al., 2010). To date only a few experiments (e.g. Cano et al., 2010; Jackson et al., 1992; Monerris et al., 2008) have been conducted to explore the impact of surface rock on L-band brightness temperature observations. These have found that i) rock has a very low porosity meaning that it does not absorb any appreciable amount of water; ii) the low dielectric constant of rock reduces the effective dielectric constant of the wet soil surface mixed with rocks; iii) the presence of rock results in an increase of soil surface roughness; and iv) no improvement of soil moisture retrieval accuracy was found for brightness temperature modelling if rocks were assumed as a smooth surface with a fixed dielectric constant. The joint impact of these four aspects makes microwave emission from the rock covered land surface complex.
The objective of this study was to further investigate the effect of rock on brightness temperature observation and soil moisture retrieval accuracy using both synthetic and observational data from a field experiment. These results were then used to demonstrate the expected impact of rock on L-band brightness temperature observation and soil moisture retrieval accuracy globally, by identifying SMOS and SMAP pixels with rock-induced brightness temperature contribution in excess of the 4 K, which is equal to the SMOS brightness temperature error budget, and/or soil moisture error in excess of the 0.04 m3/m3 target soil moisture accuracy of SMOS and SMAP missions (Entekhabi et al., 2010; Kerr et al., 2010). Consequently, these maps can be used to mask or flag adversely affected pixels in the absence of accounting for rock in the soil moisture retrieval process.
Section snippets
Data sets and study areas
The airborne passive microwave observations, ground sampling data, and monitoring stations measurements collected during the SMOS Arid Zone Experiment in Australia 2009 (Rüdiger et al., 2014) were used in this study to validate the rock fraction effect determined from model simulation. The field experiment was conducted in August 2009 over three 50 km × 50 km study areas in central Australia, aiming to identify vicarious calibration sites for the on-orbit calibration of SMOS using airborne
Soil moisture retrieval model
The volumetric water content in the top 5 cm layer of soil can be retrieved from L-band passive microwave observations using a radio-brightness transfer model such as the L-band Microwave Emission of the Biosphere (L-MEB; Wigneron et al., 2007), Community Microwave Emission Model (CMEM; de Rosnay et al., 2009), or the Land Parameter Retrieval Model (LPRM; Owe et al., 2001). These models were developed based on the ‘τ-ω model’ (Mo et al., 1982) in which τ and ω are the optical depth and the
Methodology
In this paper, rock was assumed to behave as soil but with the fixed dielectric constant of 5.7-j × 0.074 (Jackson et al., 1992; Kerr et al., 2010; Ulaby et al., 1986), and thus the brightness temperature of rock was estimated as soil using the L-MEB. The overall brightness temperature of a rock mixed pixel (bulk brightness temperature TBBp) is defined as the sum of rock component (TBRp) and soil component (TBSp) weighted by the rock cover fraction (fR) such that
Synthetic study of rock fraction impact on soil moisture retrieval accuracy
A land surface representation consisting of soil and rock components with a rock cover fraction of fR was used to simulate the soil moisture retrieval error induced solely by rock. The dual polarized brightness temperature values of the land surface representation were simulated for incidence angles from 0° to 60° for four scenarios including bare and vegetated soil having low and high moisture content mixed with rock (Fig. 2). The parameters used in the L-MEB model are summarized in Table 1,
Rock impact estimation at global scale
In the SMOS Level 2 soil moisture retrieval algorithm, Ecoclimap (Masson et al., 2003) is used as the reference land cover dataset, due to its high resolution and fine land surface classification (Kerr et al., 2010). The Ecoclimap was developed to initialize the soil-vegetation-atmosphere transfer schemes (SVATs) in meteorological and climate models. Two hundred and fifty ecosystems with homogeneous vegetation were derived from a combination of two global land cover datasets (IGBP/DIS and
Conclusions
Both the European Space Agency's (ESA) Soil Moisture and Ocean Salinity (SMOS) and the National Aeronautics and Space Administration's (NASA) Soil Moisture Active Passive (SMAP) satellites employ a L-band (1.41 GHz) radiometer to measure microwave emission from the land surface globally every 2–3 days, which is used to retrieve the top ~5 cm soil moisture through a radiometric transfer model. However, this technique suffers from its coarse 40 km spatial resolution, due to technical limitations
Acknowledgements
This study has been conducted within the framework of the MoistureMap project funded by the Australian Research Council (DP0879212). Access and accommodation at Wirrangula Hill was provided by Trevor Williams of the Nilpinna cattle station and his hospitality is hereby gratefully acknowledged. The authors would also like to acknowledge the invaluable field work assistance of Mahdi Allahmoradi, Justin Costelloe, Susan Hayes, Jon Johansson, Alan Marks, Vjeko Matic, Sandy Peischl, Peter
References (35)
- et al.
The SMOS Mediterranean Ecosystem L-Band characterisation EXperiment (MELBEX-I) over natural shrubs
Remote Sens. Environ.
(2010) - et al.
Vegetation effects on the microwave emission of soils
Remote Sens. Environ.
(1991) Impact of surface heterogeneity on surface soil moisture retrievals from passive microwave data at the regional scale: the Upper Danube case
Remote Sens. Environ.
(2008)- et al.
L-band microwave emission of the biosphere (L-MEB) model: description and calibration against experimental data sets over crop fields
Remote Sens. Environ.
(2007) - et al.
Upscaling sparse ground-based soil moisture observations for the validation of coarse-resolution satellite soil moisture products
Rev. Geophys.
(2012) - et al.
SMOS validation and the COSMOS campaigns
IEEE Trans. Geosci. Remote Sens.
(2008) - et al.
Microwave dielectric behavior of wet soil-part II: dielectric mixing models
Geoscience and Remote Sensing, IEEE Transactions on
(1985) - et al.
The soil moisture active passive (SMAP) mission
Proc. IEEE
(2010) - et al.
Global land cover classification at 1 km spatial resolution using a classification tree approach
Int. J. Remote Sens.
(2000) - et al.
Rock fraction effects on the interpretation of microwave emission from soils
IEEE Trans. Geosci. Remote Sens.
(1992)
Passive microwave remote sensing system for soil moisture: some supporting research
IEEE Trans. Geosci. Remote Sens.
SMOS Level 2 Processor Soil Moisture Algorithm Theoretical Basis Document (ATBD)
The SMOS soil moisture retrieval algorithm
IEEE Trans. Geosci. Remote Sens.
The SMOS mission: new tool for monitoring key elements of the global water cycle
Proc. IEEE
Weather and Climate
The IGBP-DIS global 1 km land cover data set, DISCover: first results
Int. J. Remote Sens.
Development of a global land cover characteristics database and IGBP DISCover from 1 km AVHRR data
Int. J. Remote Sens.
Cited by (2)
The Soil Moisture Active Passive Experiments: Validation of the SMAP Products in Australia
2021, IEEE Transactions on Geoscience and Remote Sensing