Soil Moisture Retrieval and New Techniques

Measuring soil moisture using remote sensing techniques has been a key application for many years; measurement techniques have been developed using both active and passive sensors with considerable success. However, these techniques encounter difficulties in arid and semi-arid terrain.  Satellite altimetry presents an attractive option for retrieval of soil surface moisture in these areas. From initial work in the ESA SMALT project (1) surface soil moisture has now been successfully retrieved from Cryosat2 data in LRM mode, using a completely re-developed modeling technique, as part of the EU LOTUS project [2, 3].   A series of studies have confirmed that this technique is also applicable to Cryosat2 SAR mode data (3). The key to retrieving accurate soil surface moisture estimates is the development of very detailed Dry Earth Models (DREAMS) using multi-mission recalculated backscatter, cross-calibrated over desert calibration sites whose surface characteristic variation is very well understood and informed by field campaign survey data. This implementation requires that the surface be dry for part of the year to give a baseline for the cross-calibration. Therefore, DREAMS have been developed over desert and semi-arid terrain.

New DREAMS have now been developed for Sentinel3, using the augmented methodology developed for Cryosat2. These DREAMS are fully built for the Simpson, Gibson, and Victoria deserts in Australia plus the vast Arabian desert, and the Tenere desert in the Sahara. For all these models the new techniques utilising waveform analysis and filtering, and, crucially, fusing with detailed ground truth, have been fully implemented. Results for all models are presented.

The methodology and models have been assessed with both Cryosat2 and Jason2 data to retrieve soil moisture time series.  Results are presented for all deserts, with validation of these time series with in-situ rain gauge data and independent remote sensed soil moisture data, derived using both active and passive sensors [4].

A first look at Sentinel3 SRAL data is included; the first step is the detailed backscatter and waveform shape analysis and cross-calibration over the desert calibration areas.

The outcomes of a study into spatial sampling and resolution are presented, and the results compared with other remote sensing soil moisture datasets.  These results show very clearly that the detailed along-track estimates that can be achieved from satellite radar altimetry have a unique contribution to make to soil moisture retrieval in arid and semi-arid terrain.  

1  http://tethys.eaprs.cse.dmu.ac.uk/SMALT/

2 Berry, P.A.M., Dowson, M., Carter, J.O., 2013. Surface soil moisture from multi-mission satellite radar altimetry. SP-722 Proc. ESA Living Planet Symposium 2013.

3 P.A.M. Berry, R. Balmbra, 2015.  Surface Soil Moisture From SRAL Satellite Radar Altimetry. Proc. Third Space for Hydrology Workshop, in press

4 Wagner, W., W. Dorigo, R. de Jeu, D. Fernandez, J. Benveniste, E. Haas, M. Ertl (2012) Fusion of active and passive microwave observations to create an Essential Climate Variable data record on soil moisture, ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences (ISPRS Annals), Volume I-7, XXII ISPRS Congress, Melbourne, Australia, 25 August-1 September 2012, 315-321

Airborne L-band brightness temperature and soil moisture acquisitions were carried out with Passive Active L-band System (PALS) in Texas and Arizona in 2015. These measurements coincided with NASA’s SMAP (Soil Moisture Active Passive) and ESA’s SMOS (Soil Moisture Ocean Salinity) measurements. The area flown by PALS was 36 km by 36 km in Texas and 36 km by 108 km in Arizona which covered at least one footprint of each mission at about 1km spatial resolution. The Arizona measurements were part of the SMAP Validation Experiment 2015 (SMAPVEX15) campaign. PALS covered the Texas site four times in May and June, and the SMAPVEX15 domain seven times in August. The brightness temperature and soil moisture trends between PALS, SMAP and SMOS will be compared to each other and to in situ soil moisture observations carried out at these locations. 

Soil moisture is a measure for the quantity of water contained in the soil. As processes like plant growth depend on the available water, soil moisture is a highly relevant variable for agricultural purposes. Besides, soil moisture is a key player in the energy and water cycle and thus important for weather and climate. In the last decades, several satellite missions (e.g., SMOS (Soil Moisture and Ocean Salinity) and SMAP (Soil Moisture Active Passive)) have been launched to derive especially large scale products of the soil moisture distribution on the Earth. However, validation data is often scarce and always contains uncertainties. Therefore there is always a need for good in situ soil moisture measurements. Since autumn 2013, we are running a new experimental GNSS setup with three permanently installed low-cost GPS (Global Positioning System) receivers at the meteorological study site of the German Weather Service (DWD) in Munich. Two GPS sensors are installed within the soil matrix at a depth of 10 cm and one as a reference above the soil surface. All sensors record the globally and freely broadcasted GPS L1-band data for all 32 GPS satellites continuously. Soil moisture was successfully derived based on GPS signal strength losses and soil models based on the complex permittivity. With this approach we are able to observe the integrated soil moisture of the soil volume above the antennas installed in the ground, whereas conventional in situ methods typically measure a rather small soil volume. The results show high accordance with meteorological and validation data recorded continuously with conventional FDR probes as well as manual gravimetric and TDR probes which were taken throughout the entire measuring period. We will report on a complete period of 1.5 years of measured data covering different soil moisture situations affected by light and intense precipitation events as well as drought periods. Due to its non-invasive, low cost and low power setup, networks of these in-situ GNSS (Global Navigation Satellite System) sensors could be used for potential applications also in remote areas, e.g., to cover a better geographical distribution of in situ soil moisture stations, aiming to serve as satellite validation data and to support the sectors agriculture, water supply and climate change.

The Copernicus Global Land Service (CGLS) provides a series of bio-geophysical products on the status and evolution of land surface at global scale (http://land.copernicus.eu/global/). Most of the CGLS data products have a spatial resolution of 1 km or better, the exceptions being the land surface temperature product with a spatial sampling of 0.05° and the Soil Water Index (SWI) product with 0.1°. The SWI quantifies the moisture condition at various depths in the soil and is derived from ASCAT surface soil moisture time series as provided by EUMETSAT’s Satellite Application Facility on Support to Operational Hydrology and Water Management (H-SAF). To make the spatial resolution of the SWI data commensurate with the rest of the CGLS data products, 1 km surface soil moisture data retrieved from Sentinel-1 Synthetic Aperture Radar (SAR) data can be used as additional input into the SWI algorithm. A wide range of methods for the retrieval of soil moisture data from SAR measurements exist, from simple empirical approaches to the application of machine learning algorithms trained using theoretical models and/or a range of observational data sets. Our approach to retrieve surface soil moisture data from Sentinel-1 at 1 km scale is to invert a backscatter model that is trained using historic SAR observations. This approach has been developed and validated on a continental scale using C-band backscatter data acquired by the Advanced Synthetic Aperture Radar (ASAR) on board of ENVISAT (Pathe et al., 2009; Doubkova et al., 2012; Dostalova, 2014). The algorithm was slightly adapted to the specific measurement configuration of Sentinel-1 by Hornacek et al. (2012), and has in the meantime been implemented on the supercomputing facilities operated by the EODC Earth Observation Data Centre for Water Resources Monitoring (Wagner, 2015) in close collaboration with its Cooperation Partners. Within the framework of CGLS, the production of 1 km Sentinel-1 surface moisture data, in the form of an SWI product, is currently being tested and validated over a number of in situ soil moisture networks in Europe. If validation results are satisfactory then the service may be rolled out first to Europe, and once enough Sentinel-1 data are available for all continental land surface areas to a global scale.

Dostálová, A., Doubková, M., Sabel, D., Bauer-Marschallinger, B. and Wagner, W., 2014: Seven Years of Advanced Synthetic Aperture Radar (ASAR) Global Monitoring (GM) of Surface Soil Moisture over Africa. Remote Sensing 6 (8), 7683-7707.

Doubková, M., Van Dijk, A.I.J.M., Sabel, D., Wagner, W. and Blöschl, G., 2012: Evaluation of the predicted error of the soil moisture retrieval from C-band SAR by comparison against modelled soil moisture estimates over Australia. Remote Sensing of Environment 120 (0), 188-196.

Hornacek, M., Wagner, W., Sabel, D., Truong, H.L., Snoeij, P., Hahmann, T., Diedrich, E. and Doubkova, M., 2012: Potential for High Resolution Systematic Global Surface Soil Moisture Retrieval via Change Detection Using Sentinel-1. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 5 (4), 1303-1311.

Pathe, C., Wagner, W., Sabel, D., Doubkova, M. and Basara, J.B., 2009: Using ENVISAT ASAR Global Mode data for surface soil moisture retrieval over Oklahoma, USA. IEEE Transaction on Geoscience and Remote Sensing 47 (2), 468-480.

Wagner, W., 2015. Big Data infrastructures for processing Sentinel data. In: D. Fritsch (Editor), Photogrammetric Week 2015. Wichmann/VDE, Stuttgart, pp. 93-104.

Earth observation of near-surface soil moisture (SSM) content is of tremendous scientific and application interest for its bearing on such things as the understanding of the interactions between water, energy and biochemical fluxes, the forecasting of meteorological and hydrological events, crop yield predictions, and so on. Presently, products at coarse resolution, such as ESA SMOS [Kerr et al., 2010], EUMETSAT ASCAT [Wagner et al., 2013] and NASA SMAP [Entekhabi et al., 2010], are (or will soon be) operationally delivered. However the challenge to retrieve SSM fields at high resolution (e.g., 0.1-1.0 km) still remains. Coping with this challenge is a daunting task because it means accounting for the impact of the surface heterogeneity in terms of topography, soil properties, vegetation cover, soil roughness etc. Progress is needed to significantly enhance our capability and thus meet the needs for monitoring Earth’s terrestrial water [Wood, et al., 2011].

The ESA Sentinel-1 (S-1) mission, with its improved observational capabilities in terms of spatial/temporal resolution and radiometric accuracy, can play a pivotal role in addressing this challenge leading to high spatial resolution SSM retrieval over agricultural and sparsely vegetated areas. The concept has been demonstrated within an ESA feasibility study, where a soil moisture retrieval algorithm (referred to as Soil MOisture retrieval from multi-temporal SAR data, SMOSAR) has been developed, calibrated and assessed over S-1 simulated data. The rationale of SMOSAR is that temporal changes of surface parameters affecting the radar backscatter apart from SSM (e.g., soil roughness, canopy structure, vegetation water content, etc.) usually take place at longer temporal scales than SSM changes (excluding cultivation practices). Therefore, SAR time series with a sufficiently short repeat cycle are expected to track changes in SSM only, since other parameters affecting the radar backscatter can be considered as constant. In SMOSAR this approximation is made and implies that the retrieval approach is extremely simplified and fast with respect to approaches based on the cumbersome inversion of scattering models depending on a large number of surface parameters [Balenzano et al., 2011]. SMOSAR can be applied to bare and vegetated surfaces dominated by surface scattering as, for instance, cereal crops. This means that before retrieval a masking process, obscuring those areas dominated by volume scattering, is required.

The objective of this study is to assess the SMOSAR performance using time series of S-1A data acquired over three calibration/validation sites, located in Europe (i.e., Segezia, Italy), USA (Little Washita, Oklahoma) and Australia (Yanco, New South Wales). Each site is equipped with a hydrologic network continuously measuring SSM values, which will be used to validate the S-1A retrieved values at high resolution. In addition, a comparison at lower spatial resolution between S-1A, SMOS, ASCAT and SMAP SSM values will be carried out in order to better understand the SMOSAR performance at the various spatial scales. Initial results obtained by analysing S-1A time series acquired over the Segezia site in Italy [Mattia et al., 2015] indicate a SSM retrieval error between 0.05 and 0.07 cm³/cm³.

References

Balenzano, A., et al., “Dense temporal series of C- and L-band SAR data for soil moisture retrieval over agricultural crops”, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 4, no.2, pp. 439-450, 2011.

Entekhabi, D., et al. “The Soil Moisture Active Passive (SMAP) Mission.” Proceedings of the IEEE, vol.58, no.5, pp. 704-716, 2010.

Kerr, J., et al., “The SMOS soil moisture retrieval algorithm”, IEEE Transactions on Geoscience and Remote Sensing, vol. 50, no. 5, pp., 1384- 1403, 2012.

Mattia, F., et al., “Sentinel-1 for wheat mapping and soil moisture retrieval”, Proceedings of the IEEE 2015 International Geoscience and Remote Sensing Symposium, IGARSS 2015, Milano, Italia, 26-31 July, 2015, IEEE Publications (USA, 2015), pp. 2832-2835.

Wagner, W., et al., “The ASCAT soil moisture product: A review of its specifications, validation results, and emerging applications”, Meteorologische Zeitschrift, vol. 22, no.1, pp. 5-33, 2013.

Wood, E.F., et al., “Hyperresolution global land surface modeling: Meeting a grand challenge for monitoring Earth’s terrestrial water”, Water Resources research, vol. 47, w05301, 2011.