The Soil Moisture and Ocean Salinity (SMOS) satellite has provided, for the first time, systematic passive L-band (1.4 GHz) measurements from space with a spatial resolution of ~ 40 km. SMOS was launched in 2009 by ESA and CNES and was followed by two other L-band missions by NASA : Aquarius on SAC-D (2011-2015) and SMAP (Soil Moisture Active Passive, launched in 2015).

L-band measurements provide accurate and absolute surface soil moisture measurements even under dense vegetation canopies. It was demonstrated that SMOS data can be used as reference to derive soil moisture (SM) from other non-dedicated sensors working at higher frequencies, enabling to build long time data records. L-band surface SM measurements have also been used to estimate root zone soil moisture, to derive drought indices, to enable food security monitoring and to improve satellite precipitation estimates. SMOS brightness temperatures are operationally monitored at ECMWF and a specific neural network SM product will be assimilated operationally from 2019. The L-band vegetation optical depth (L-VOD) retrieved simultaneously to the soil moisture thanks to the multi-incidence angle capabilities of SMOS has been used to estimate above ground biomass and to study trends in the global carbon stocks. Over the ocean, SMOS L-band measurements have provided the longest time series of remote sensing sea surface salinity and settled the basis of the ESA Sea Surface Salinity Climate Change Initiative (CCI). These data are complementary to existing in situ measurements but, thanks to the higher spatial resolution, they enabled to study mesoscale variability, coastal oceans and marginal seas, and boundary currents. L-band passive radiometer observations also allow to measure strong winds over the ocean. Regarding the cryosphere, SMOS data have shown that passive L-band radiometry can be used to measure thin sea ice thickness, to measure the temperature of deep layers of ice in Antarctica and to monitor melting events in Antarctica's coastal regions.

However, higher spatial resolution is needed for many applications related to water resources management (interstate water distribution, irrigation) and food security (yield forecasts, phytosanitary risks, droughts), for instance. Downscaling strategies have been developed and implemented using auxiliary higher resolution data such as visible/infrared or synthetic aperture radar observations to estimate SM at ~1 km scale. Nevertheless, all downscaling approaches have intrinsic limitations in their applicability domain (cloud cover, surface roughness, fine-scale calibration, etc.) so that the higher the spatial resolution of the initial SM dataset, the more accurate the final downscaled dataset will be. In addition, operational hydrological models and land data assimilation schemes are expected to run globally at sub-kilometric resolutions in the next decade. In order address this requirement, higher spatial resolution SM from L-Band microwave is needed. Over the ocean as well as in coastal areas, higher resolution will bring the possibility to study in detail mesoscale processes and salinity (and density) variations closer to the coast. Over ice, higher spatial resolution will allow to monitor melting events in the coastal regions of Antarctica, for instance.

In this contribution we will present the SMOS-HR (High Resolution) project, which is currently in Phase 0 at the French Centre National d’Etudes Spatiales (CNES). The goal is to ensure continuity of L-band measurements while increasing the spatial resolution to ~10 km. This would require a typical antenna size of ~18 meters. Taking into account the difficulty of deploying a real aperture of this size in space and the successful alternative approach used by SMOS, the new SMOS-HR satellite will perform aperture synthesis using an array of 160 small antennas distributed in a 12 meters-side square. SMOS-HR will also benefit from developments done for the ULID (Unconnected L-Band Interferometer Demostrator) nano-satellites.

NASA’s Soil Moisture Active Passive (SMAP) mission was launched on January 31st, 2015. SMAP utilized an L-band radar and radiometer sharing a rotating 6-meter mesh reflector antenna until July 7th, 2015, when the SMAP radar encountered an anomaly and became inoperable. During the SMAP post-radar phase, many ways have been explored to recover the high-resolution soil moisture capability of the SMAP mission. One of the feasible approaches is to substitute the SMAP radar with other available SAR data. Sentinel 1A/1B SAR data has been found most suitable for combining with the SMAP radiometer data because of its almost similar orbit configuration that allows overlapping of their swaths with minimal time difference, which is key to the SMAP active-passive algorithm. The Sentinel SDV mode acquisition also provides the co-pol and cross-pol observations required for the SMAP Active Passive algorithm. Some differences do exist between the SMAP SAR data and Sentinel SAR data. They are mainly: 1) Sentinel has C-band SAR and SMAP is L-band; 2) Sentinel has multi incidence angle within its swath, and SMAP has single incidence angle; and 3) Sentinel swath width is ~300 km as compared to SMAP 1000 km swath width. On any given day, the narrow swath width of the Sentinel observations will significantly reduce the spatial coverage of the SMAP active-passive approach as compared to the SMAP swath coverage. The temporal resolution (revisit interval) is also degraded from 3 days to 12 days. One advantage of using Sentinel 1A/1B data in the SMAP Active Passive algorithm is the potential of obtaining the disaggregated brightness temperature and thus soil moisture at a much finer spatial resolution of 3 km. The Beta version of the SMAP-Sentinel Active Passive high-resolution product is now available to the public (released on Nov 1st, 2017) through NSIDC (NASA DAAC). The duration of this product is from May 2015 to current date. The presentation will focus on the basics of the Active Passive algorithm, the characteristics and evolution of the algorithm parameters over different landcover and seasons, and major details of the SMAP-Sentinel Active Passive product. We will also discuss potential of high resolution soil moisture data application to agriculture.

Surface soil moisture (SM) is one of the main factors controlling soil evaporation under non-energy limited conditions. High resolution (HR) SM estimates are needed for various applications ranging from hydrological to agricultural. However, current remote sensing SM estimates have a very low resolution, of tens of kilometers, which proves to be insufficient for the majority of these applications, and hence, high resolution products are needed. Amongst existing algorithms for downscaling SM from remote sensing data, DISPATCH (DISaggregation based on a Physical And Theoretical scale CHange) has been shown to accurately improve the resolution of SMOS (Soil Moisture and Ocean Salinity) soil moisture data using the soil evaporative efficiency (SEE). SEE is defined as the ratio of actual to potential soil evaporation and can be derived from auxiliary land surface temperature (LST) and normalized difference vegetation index (NDVI) data. These data can be provided by various sensors, such as Sentinel-3 SLSTR (Sea and Land Surface Temperature Radiometer), or MODIS (MODerate resolution Imaging Spectroradiometer). More specifically, DISPATCH relies on two separate SEE models: a temperature-based SEE model driven by LST data and a SM-based SEE model driven by SMOS data. This study is aimed at improving the physics and calibration of both SEE models, and hence the robustness of HR SM products, by integrating within DISPATCH a soil energy balance model forced by available meteorological data. This lead to the development of a new algorithm: DISPATCH-E. The approach is tested over two different sites: a mixed dry and irrigated area in Catalonia, Spain for 2011-2012, and over a dry area in Morocco, for 2016-2017. The data used to force the energy balance model was provided by the SAFRAN (Système d'analyse fournissant des renseignements atmosphériques à la neige) model for the Catalan site and by ECMWF (European Centre for Medium-Range Weather Forecasts) for the Moroccan site. Both approaches were validated against in situ measurements. As a first step, HR SM products from both DISPATCH and DISPATCH-E temperature-based SEE models are temporally compared to SMOS SM for each SMOS pixel. The high resolution SM product derived from the temperature-based SEE model generally has a better slope of the linear regression but worse root mean square difference (RMSD) and correlation coefficient when meteorological data are used as input. As a second step, both DISPATCH and DISPATCH-E algorithms are run using the SM-based SEE model fitted to the temperature-based SEE model. The mean correlation and slope of the linear regression between disaggregated and in situ SM is increased when integrating meteorological data within DISPATCH, respectively, all the while keeping stable values of the RMSD. Moreover, the results obtained over the Moroccan site provided insights into developing a new synergistic method between the two approaches, which improves the robustness in homogenous conditions: fully dry or fully humid conditions.

Global data of soil moisture (SM) are currently available at about 40 km resolution from L-band microwave radiometers. Recently, some efforts have been made to disaggregate 40 km coarse resolution SM data to 1 km resolution. But to make them accessible for the benefit of field-scale agriculture and other hydrological purposes, SM data at significantly higher (typically 100 m) resolution are still required. To fill the gap between low resolution (LR) and high resolution (HR) data, a disaggregation strategy of Soil Moisture and Ocean Salinity (SMOS) SM is developed by using disaggregation approach based on physical and theoretical scale change (DISPATCH) algorithm. DISPATCH algorithm provides a spatial link between SM at LR and the Soil Evaporative Efficiency (SEE, defined as the ratio of actual to potential soil evaporation) which is estimated at HR from the land surface temperature (LST) and fractional vegetation cover (Fv) derived from HR optical/thermal data within a LR pixel. SEE is expressed as a linear function of the soil temperature and the minimum and maximum soil temperatures calculated from the LST-Fv feature space. Based on this algorithm, the 40 km resolution SMOS SM is, as a first step, disaggregated to 1 km resolution using optical/thermal data from Moderate Resolution Imaging Spectroradiometer (MODIS) and, as a second step, disaggregated 1 km resolution SM is further disaggregated to 100 m resolution using Landsat data. Figure 1 shows an illustration image of SMOS SM, disaggregated 1 km SM and disaggregated 100 m resolution SM over a 30 x 30 km2 validation area in central Morocco. The main objective of this paper is i) to demonstrate the usefulness of an intermediate spatial resolution (ISR) between 40 km and 1 km before the application of DISPATCH at 100 m resolution using Landsat data and ii) evaluate the 100 m resolution SM data using in situ measurements. The sequential disaggregation (from 40 km to ISR and from ISR to 100 m) is assessed over 22 irrigated field crops where in-situ SM measurements were collected from January to May 2016 using thetaprobes and gravimetric samples. A sensitivity analysis indicates that ISR varied from 1 km to 40 km significantly impacts the accuracy in 100 m resolution SM. An optimal ISR is found at around 10 km. Results indicate that 100 m resolution disaggregated SM has a better unbiased RMSD with in-situ SM (in the range 0.06 to 0.09 m3/m3) when ISR is set to 10 km as compared to other ISR's. This is due to a compromise between the accuracy in temperature endmembers and LR SM, which increases with ISR, and the spatial representativeness at 100 m resolution of ISR SM, which decreases with ISR. Moreover, we proposed a Monte-Carlo method to significantly to reduce the random uncertainty which occurs during the disaggregation of SM at 100 m resolution. Monte-Carlo method is based on a multiple application of DISPATCH to 25 ISR (set to 10 km) resolution grids separated by 2 km in the east-west and north-south directions. The RMSD between disaggregated and in-situ SM is further decreased by 0.01 m3/m3.

Index terms: Soil moisture, Disaggregation, DISPATCH, SMOS, Landsat

Two new 1km soil moisture (SM) products based on Sentinel-1 SAR observations expand the portfolio of geophysical variables of Copernicus Global Land Service (CGLS). The CGLS provides freely accessible a large number of harmonised and co-formatted data products on energy, water and vegetation variables in Near-Real-Time (NRT).

Complementing the already operational 0.1° Metop-ASCAT Soil Water Index Version3 (SWIV3), the Surface Soil Moisture (SSM) from Sentinel-1 SAR data (named SSM1km) and a SWI product based on fused SSM data from ASCAT and Sentinel-1 observations (named SWI1km) enable continental and global monitoring of SM at a 1km sampling. They feature SM dynamics with much higher detail, allowing the analysis of local hydrological patterns, such as impacts from convectional rains and irrigation activities, and thus meet the requirements of many users.

The 1km Sentinel-1 SSM gives in percent the relative soil moisture saturation of the upmost 5 cm, globally every 3-6 days, every 1.5-4 days over Europe. The employed algorithm processes the SAR observations at the initial 20m scale, at which a per-pixel signal analysis drives the algorithm parameter retrieval. Subsequent resampling of pre-processed SAR data to 1km mutes artefacts stemming from the complex radar interactions at the 20m scale, and increases the sensitivity to soil moisture dynamics.

The 1km SCATSAR-SWI gives in percent the relative water content in eight soil layers of the upmost metre, with a daily full coverage over operated areas. ASCAT and Sentinel-1 SSM are combined per pixel and data fusion parameters based on the full archives of Sentinel-1 and ASCAT (2014-2017) account for differences in scaling, accuracy and local dynamics. Through combination of both, the high temporal resolution of ASCAT-SSM and the high spatial resolution of Sentinel-1- SSM, it is the first pre-operational dataset that allows resolving SM dynamics with a daily frequency and a 1km sampling.

Here, we demonstrate the results of the evaluation of the two products, illuminating also their potential for estimating rainfall and irrigation impacts. Both SSM1km and SWI1km were compared over Europe against in-situ measurements, reference data from ASCAT SSM, 1km SM model data, and rainfall observations. For the SWI1km, the experiments yielded comprehensively high agreement with all reference datasets. While the Sentinel-1 signal appeared to attenuate, ASCAT’s signal dynamics were fully transferred to the SWI1km and benefit from the Sentinel-1 parametrisation. The SWI1km showed excellent capability to reproduce rainfall observations. The experiments for the SSM1km yielded a consistent set of model parameters and product masks, unperturbed by Sentinel-1 coverage discontinuities. The SSM1km showed high agreement over plains and agricultural areas and low agreement over strong topography and forests, owing to wrongly attributed vegetation dynamics. While positive biases during the growing season were detected, excellent capability to capture small-scale soil moisture changes as such from convectional rainfall or irrigation was evident, which are often missed by coarse-scale products.

This presentation aims to revisit the state of art of the applications derived from users of low frequency PMW missions over land such as SMOS, SMAP and AMSR. A wealth of applications and science topics have emerged from those missions, many being of operational value (Kerr et al. 2016, Muñoz-Sabater et al. 2016, Mecklenburg et al. 2016). We will be reviewing those applications that have been shown to be key to monitor the water and carbon cycles.

Over land soil moisture measurements have enabled to get access to root zone soil moisture, yield forecasts, fire and flood risks, drought monitoring, improvement of rainfall estimates, etc. Various disaggregation approaches are providing operational 1 km spatial resolutions and are very useful for pest control or irrigation management.
The assimilation of L Band data (brightness temperatures and soil moisture maps) from SMOS mission into eco-hydrological models has showed a significant increase in the skill of the surface and root zone soil moisture estimates. Those results showed the potential of SMOS for the monitoring of agricultural drought. Several other studies addressed the estimation of root zone soil moisture and drought indexes from L Band data (Al Bitar et al. 2013, Kerr et al. 2016, Reichle et al. 2016). These products are already used in drought early alert systems by the USDA.

The vegetation optical depth retrieved by SMOS in L-band (L-VOD) shows a significant correlation with tree height (Rahmune et al. 2015) and above ground biomass for all biomes (Vittucci et al, 2016; Rodriguez-Fernandez et al. 2018) with no signs of saturation for high biomass values. L-VOD is thus suitable for large scale carbon stocks monitoring and it has been used by Brandt et al. (2018) to monitor annual aboveground biomass-carbon changes in sub-Saharan Africa between 2010 and 2016 showing the importance of drylands in the global carbon balance.

Terrestrial carbon and water cycles are tightly coupled through stomatal conductance. This means that observations of atmospheric carbon dioxide can be used to constrain water fluxes (see, e.g. Kaminski et al., 2012) and, vice versa, observed soil moisture provides a constraint on the carbon cycle. In their Carbon Cycle Data Assimilation System (CCDAS) Scholze et al. (2016) demonstrated that the soil moisture product derived from SMOS is an important observational constraint that, in particular over dry regions, is complementary to the constraint from atmospheric CO2 observations and achieves, together with atmospheric CO2, uncertainty reductions in simulated Net Primary Productivity (NEP) or Net Ecosystem Productivity (NEP) of 70 to 95 %.

L-band sea surface salinity data have considerably improved the monitoring of SSS variations related with climate variability such as El Niño-Southern Oscillation, Indian Ocean Dipole, and Madden-Julian Oscillation, and the linkages of the ocean with different components of the water cycle such as evaporation and precipitation and continental runoff. They have enhanced the understanding of various ocean processes such as tropical instability waves, Rossby waves, mesoscale eddies and related salt transport, salinity fronts, hurricane haline wake, river plume variability, cross-shelf exchanges. They have also been used to study ocean biogeochemistry such as the role of ocean circulation and rain freshening on the variability of air-sea CO2 fluxes. Moreover, L-band measurements have also proved to have many other potential applications such as thin sea ice estimate (up to 50-100 cm) as well as freeze thaw dates, and potentially snow density and liquid water content.
Moreover, L-band measurements have also proved to have many other potential applications such as thin sea ice estimate (up to 50-100 cm) as well as freeze thaw dates, and potentially snow density and liquid water content.

Continuity of low frequency microwave observations (L, C) is now being proposed by the Copernicus Microwave Imaging Radiometer (CIMR) high priority candidate mission (Kilic et al 2018) and the SMOS-HR (High Resolution) project, currently in Phase 0 at the French Centre National d’Etudes Spatiales (CNES) to ensure continuity of L-band measurements increasing the spatial resolution to ~10 km (Rodriguez-Fernandez 2018).