International Journal of Applied Earth Observation and Geoinformation
DisPATCh as a tool to evaluate coarse-scale remotely sensed soil moisture using localized in situ measurements: Application to SMOS and AMSR-E data in Southeastern Australia
Graphical abstract
Introduction
Although soil moisture only represents a small part of the liquid freshwater on Earth (about 0.15% (Dingman, 1994)), soil moisture observations over large areas and long time series are increasingly required in a range of environmental applications including meteorology, climatology, water resources management and hydrology. It controls interactions between the land surface and the atmosphere, thereby influencing climate and weather (Entekhabi, 1995). It also influences many processes related to plant growth, as well as a range of soil hydrologic processes such as evaporation, infiltration and runoff.
Soil moisture is highly variable both in space and time, mainly as a result of the heterogeneity in soil properties, topography, land cover, rainfall and evapotranspiration. Various approaches have been developed over the past two decades to deduce Surface Soil Moisture (SSM) from remote sensing measurements (Wagner et al., 2007, Kerr et al., 2001, Njoku et al., 2003). Several approaches have been first developed to retrieve SSM from data collected with C-and X-band radiometers, like the Advanced Microwave Scanning Radiometer-Earth observing system (AMSR-E), launched in 2002 (Owe et al., 2001, Njoku et al., 2003). The recent Soil Moisture and Ocean Salinity (SMOS) mission, launched in 2009, operates at L-band (the optimal microwave band to estimate SSM) and is the first space mission dedicated to observe SSM globally (Kerr et al., 2010). This mission is being complemented by the new NASA satellite mission SMAP (Soil Moisture Active Passive) launched on the 31st of January 2015. This mission ensures the continuity of L-band passive microwave data for global SSM monitoring (Entekhabi et al., 2010). The estimated SSM must be validated in order to assess the quality of the acquisition and retrieval processes by the comparison of the product with reference often in situ data sources.
One major difficulty in calibrating/validating SSM retrieval algorithms in the passive microwave domain is the coarse-scale resolution (about 40 km for SMOS and 50 km for AMSR-E) of spaceborne observations, and the small spatial representativeness scale (several m or so) of localized in situ measurements. SSM is highly variable in both time and space across different scales (Famiglietti et al., 2008). In consequence, the severity of the validation challenge would be much eased by reducing the contrast between the spatial support of ground-based observations and that of the satellite-based SSM retrievals. To help solve the huge scaling issue and to circumvent the direct comparison, two distinct approaches can be considered, namely (i) the upscaling of localized in situ measurements at the observation resolution and (ii) the downscaling of satellite data at the representativeness scale of ground measurements.
Many validation strategies of satellite SSM data using in situ measurements have been based on the assumption that local ground observations are representative of a much larger spatial extent (Grayson and Western, 1998, Cosh et al., 2008, Jackson et al., 2010, Jackson et al., 2012). Even if short term intensive field campaigns have been used for calibration and validation, mainly in North America and Australia (Bindlish et al., 2006, Merlin et al., 2008b, Walker et al., 2006, Panciera et al., 2014) these provide reliable estimates for a subset of physical and climate conditions only. Hence, the representativeness issue of in situ data is commonly addressed by aggregating (e.g. averaging) the ground measurements of relatively dense networks (Crow et al., 2012, de Rosnay et al., 2009, Cosh et al., 2008). In the heterogeneous case where the spatial uniformity assumption of SSM does not hold (due to static influence of soil, vegetation and topography), various upscaling approaches have been developed (Cosh et al., 2004, Grayson and Western, 1998, Mohanty et al., 2000). In general, upscaling approaches suggest that current ground instrumentation is adequate for satellite mission validation needs (Crow et al., 2012). However, the performance of these approaches is site-dependent. The upscaling issue is indeed directly correlated with the presence of extensive horizontal variability in SSM fields (Crow et al., 2012). Alternatively, statistical tools like triple collocation have been also applied using land surface modeling, footprint-scale soil moisture products and single ground based station with existing low density ground networks (Miralles et al., 2010).
Another approach that can be used to circumvent the direct comparison between satellite and in situ SSM is the downscaling of satellite data. This technique consists in disaggregating remote sensing data to produce SSM at a spatial resolution closer to the representativeness scale of ground measurements.
The main problem of this approach is the potentially large uncertainty in the disaggregation output. Since data aggregation is a way to decrease random errors in SSM estimates, one may state that the higher the downscaling resolution, the more uncertain is the downscaled data. Nevertheless, data disaggregation is always a trade-off between output accuracy and spatial representation so that in heterogeneous conditions, the systematic differences (between low-resolution and in situ SSM) that is associated with sub-pixel variability may exceed the random errors in high-resolution disaggregated data. In this case, the downscaled SSM would be more accurate at the validation scale than the original coarse-scale observation, and would hence provide valuable information for validation purposes.
One promising approach for obtaining accurate estimates of high-resolution SSM is the disaggregation of microwave-derived SSM using MODIS (MODerate resolution Imaging Spectroradiometer) like thermal infrared and visible/near-infrared data. The relationship between SSM, land surface temperature and vegetation cover has been commonly represented as an empirical polynomial relationship based on the “universal triangle” of Carlson et al. (1994). Since then, many efforts have been made to improve the triangle method. For instance, Piles et al. (2011) developed a new polynomial-fitting method, based on the work of Chauhan et al. (2003), by merging SMOS and MODIS data to provide SSM data at 10 km and 1 km resolution. Merlin et al. (2008a) replaced the polynomial function by a semi-physical model of soil evaporative efficiency. Kim and Hogue (2012) developed a new evaporation-based disaggregation method, which is based on the formulation of surface evaporative fraction derived by Jiang et al. (2003) and a linear scaling relationship between surface evaporative fraction and SSM. DisPATCh (Disaggregation based on Physical and Theoretical scale Change (Merlin et al., 2012)) is an improved version of Merlin et al. (2008a). The new algorithm now includes the effect of vegetation water stress (Moran et al., 1994), the “universal trapezoid” has replaced the “universal triangles”, and a simple correction for elevation effects has been implemented (Merlin et al., 2013). Nevertheless, despite the significant gain in maturity of those algorithms, to the knowledge of the authors, none of them has still been used for validation of coarse-scale satellite SSM products.
In this context, this study aims to investigate the potential of DisPATCh for improving the validation strategies of coarse-scale microwave-derived SSM data using localized in situ measurements. The main idea is to assess the uncertainty in 1 km resolution DisPATCh data and to compare it with the systematic differences-associated with the subpixel heterogeneity- between the low-resolution and in situ SSM data. The methodology is tested using SMOS and AMSR-E level 3 soil moisture products over the Murrumbidgee River catchment located in southeastern Australia, and the in situ data collected by 38 stations during a one year period in 2010 and 2011.
Section snippets
Site and soil moisture data description
This analysis is based on the comparison between a disaggregation data set obtained from the SMOS (Centre Aval de Traitement des donnes SMOS, CATDS) and AMSR-E (Vrije Universiteit Amsterdam, VUA) level-3 products, and the in situ soil moisture measurements collected across the Murrumbidgee River catchment in southeastern Australia (Smith et al., 2012).
DisPATCh method
DisPATCh aims to provide 1 km resolution SSM data from coarse-scale microwave-derived SSM and the 1 km resolution Soil Evaporative Efficiency (SEE, defined as a ratio of actual to potential soil evaporation) derived from thermal infrared and visible/near-infrared MODIS data. Briefly, the soil evaporation from the 0–5 cm soil layer and the vegetation transpiration from the root zone soil layer are partitioned by separating MODIS LST (Land Surface Temperature) into its soil and vegetation
Validation strategy
In this paper, the SSM data sets are validated for different temporal (daily, seasonally and yearly) and spatial (point, zone and catchment) scales. Three evaluations are performed in this study: temporal comparison against one single in situ station. The station named Y3 was selected for the temporal comparison because (1) it is one of 12 stations used in a former study (Draper et al., 2009) and (2) it is also located in the Yanco area, well known for soil moisture validation studies. Moreover
Results and discussion
In this section, DisPATCh is run over the entire study period (from 06/2010 to 05/2011) for both SMOS and AMSR-E data sets, and the validation strategies described in the previous section are implemented. DisPATCh results are presented to address the three following key questions: first, does disaggregation improve SSM data sets at the downscaling resolution? Second, can disaggregation be useful for validating coarse-scale SSM observation? And third, by how much the disaggregated SMOS and
Conclusion
DisPATCh is an algorithm dedicated to the disaggregation of microwave-derived SSM observation using MODIS like HR LST and NDVI data. Such a downscaling approach can help solve the disparity of spatial scales between satellite observations (e.g. SMOS, ASMR-E) and in situ measurements. This study aims to examine the potential of DisPATCh for evaluating coarse-scale (several tens of km resolution) SMOS and AMSR-E products using localized ground data by explicitly representing the sub-pixel
Acknowledgements
Initial setup and maintenance of the Murrumbidgee monitoring network used in this study was funded by the Australian Research Council (DP0343778, DP0557543) and by the CRC for Catchment Hydrology. This study was supported by the CNES “Terre, Océan, Surfaces Continentales, Atmosphère” program and by the French “Agence Nationale de la Recherche” MIXMOD-E project (ANR-13-JS06-0003-01).
References (46)
- et al.
Towards soil property retrieval from space: an application with disaggregated satellite observations
J. Hydrol.
(2015) - et al.
Soil moisture mapping and AMSR-E validation using the PSR in SMEX02
Remote Sens. Environ.
(2006) - et al.
Watershed scale temporal and spatial stability of soil moisture and its role in validating satellite estimates
Remote Sens. Environ.
(2004) - et al.
Temporal persistence and stability of surface soil moisture in a semi-arid watershed
Remote Sens. Environ.
(2008) - et al.
Multi-scale soil moisture measurements at the Gourma meso-scale site in Mali
J. Hydrol.
(2009) - et al.
An evaluation of AMSR-E derived soil moisture over Australia
Remote Sens. Environ.
(2009) - et al.
Towards areal estimation of soil water content from point measurements: time and space stability of mean response
J. Hydrol.
(1998) - et al.
A sequential model for disaggregating near-surface soil moisture observations using multi-resolution thermal sensors
Remote Sens. Environ.
(2009) - et al.
Self-calibrated evaporation-based disaggregation of SMOS soil moisture: an evaluation study at 3 km and 100 m resolution in Catalunya, Spain
Remote Sens. Environ.
(2013) - et al.
Towards deterministic downscaling of SMOS soil moisture using MODIS derived soil evaporative efficiency
Remote Sens. Environ.
(2008)
The NAFE’06 data set: towards soil moisture retrieval at intermediate resolution
Adv. Water Resour.
Validation of AMSR-E soil moisture using L-band airborne radiometer data from National Airborne Field Experiment 2006
Remote Sens. Environ.
Estimating crop water deficit using the relation between surface-air temperature and spectral vegetation index
Remote Sens. Environ.
A method to make use of thermal infrared temperature and NDVI measurements to infer surface soil water content and fractional vegetation cover
Remote Sens. Rev.
Spaceborne soil moisture estimation at high resolution: a microwave-optical/IR synergistic approach
Int. J. Remote Sens.
Upscaling sparse ground-based soil moisture observations for the validation of coarse-resolution satellite soil moisture products
Rev. Geophys.
Physical Hydrology
Recent advances in land–atmosphere interaction research
Rev. Geophys.
The soil moisture active passive (SMAP) mission
Proc. IEEE
Field observations of soil moisture variability across scales
Water Resour. Res.
Water resources and management overview: Murrumbidgee catchment, NSW Office of Water, Sydney
Validation of soil moisture and ocean salinity (SMOS) soil moisture over watershed networks in the U.S.
IEEE Trans. Geosci. Remote Sens.
Validation of advanced microwave scanning radiometer soil moisture products
IEEE Trans. Geosci. Remote Sens.
Cited by (68)
Improved downscaling of microwave-based surface soil moisture over a typical subtropical monsoon region
2023, Journal of HydrologyA web-based system for satellite-based high-resolution global soil moisture maps
2023, Computers and GeosciencesSoil moisture downscaling using multiple modes of the DISPATCH algorithm in a semi-humid/humid region
2021, International Journal of Applied Earth Observation and GeoinformationEstimating catchment scale soil moisture at a high spatial resolution: Integrating remote sensing and machine learning
2021, Science of the Total EnvironmentDetecting and mapping irrigated areas in a Mediterranean environment by using remote sensing soil moisture and a land surface model
2021, Journal of HydrologyCitation Excerpt :In fact, a value of correlation equal to 0.59 was found for the 40 km resolution product, while correlation values were 0.67 for the 3 km version and 0.73 and 0.86 for two different configurations of the 100 m resolution product. A similar study was carried out by Malbéteau et al. (2015), which evaluated the performances of coarse and DISPATCH downscaled (1 km) versions of SMOS and AMSR-E soil moisture products against in situ observations over the South East of Australia. The study highlighted enhancements in the correlation with ground measurements over a semi-arid region obtained during summer for the disaggregated products with respect to their coarser resolution versions.