Ali Mahmoodi

Soil Moisture Retrieval Using Neural Networks: Application to SMOS

A methodology to retrieve soil moisture (SM) from Soil Moisture and Ocean Salinity (SMOS) data is presented. The method uses a neural network (NN) to find the statistical relationship linking the input data to a reference SM data set. The input data are composed of passive microwaves (L-band SMOS brightness temperatures,

Soil moisture retrieval from SMOS observations using neural networks

T_{b}

RFI in SMOS measurements: Update on detection, localization, mitigation techniques and preliminary quantified impacts on soil moisture products

s) complemented with active microwaves (C-band Advanced Scatterometer (ASCAT) backscattering coefficients), and Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) . The reference SM data used to train the NN are the European Centre For Medium-Range Weather Forecasts model predictions. The best configuration of SMOS data to retrieve SM using an NN is using

Calibrating the effective scattering albedo in the SMOS algorithm: Some first results

T_{b}

Analyzing the impact of using the SRP (Simplified roughness parameterization) method on soil moisture retrieval over different regions of the globe

s measured with both H and V polarizations for incidence angles from 25° to 60°. The inversion of SM can be improved by ~10% by adding MODIS NDVI and ASCAT backscattering data and by an additional ~5% by using local information on the maximum and minimum records of SMOS Tbs (or ASCAT backscattering coefficients) and the associated SM values. The NN-inverted SM is able to capture the temporal and spatial variability of the SM reference data set. The temporal variability is better captured when either adding active microwaves or using a local normalization of SMOS Tbs. The NN SM products have been evaluated against in situ measurements, giving results of comparable or better (for some NN configurations) quality to other SM products. The NN used in this paper allows to retrieve SM globally on a daily basis. These results open interesting perspectives such as a near-real-time processor and data assimilation in weather prediction models.

Enhancements of SMOS level 2 soil moisture products over Canada

A methodology to retrieve soil moisture (SM) from multiinstrument remote sensing data is presented. The method uses a Neural Network (NN) to find the statistical relationship linking the input data to a reference SM dataset. The input data is composed of passive microwaves (L-band SMOS brightness temperatures), active microwaves (C-band ASCAT backscattering coefficients), and visible and infrared observations by MODIS. The reference SM data used to train the NN are ECMWF model predictions or SMOS L3 SM. After determining the best configuration of input data to retrieve SM using a NN, the NN soil moisture product is evaluated with respect to other global SM products and with respect to in situ measurements. The NN is able to capture the spatial and temporal dynamics of SM, and the SM computed with NNs compares well with the other SM datasets.

SMOS and applications: First glance at synergistic and new results

In this communication we present an update on the RFI detection used in the SMOS processing chain and some elements on quantified impact of RFIs on level 2 soil moisture products. The level 2 soil moisture algorithms which included since the beginning a screening mechanism to reject contaminated brightness temperatures is now stricter. New approaches at the level 1 processors are also emerging and will be operational at their next release in 2014. Despite these strengthen procedures, RFIs are still impacting strongly SMOS observations and examples of quantified deterioration are given.

SMOS after six years in operations: First glance at climatic trends and anomalies

This study focuses on the calibration of the effective scattering albedo (ω) of vegetation in the soil moisture (SM) retrieval at L-Band. Currently, in the SMOS Level 2 and 3 algorithms, the value of ω is set to 0 for low vegetation and ~ 0.06 - 0.08 for forests. Different parameterizations of vegetation (in terms of ω values) were tested in this study. The possibility of combining soil roughness and vegetation contributions as a single parameter (“combined” method) leads to an important simplification in the algorithm and was also evaluated here. Following these assumptions, retrieved values of SMOS SM were compared with SM data measured over many in situ sites worldwide from the International Soil Moisture Network. These validation sites were classified using the International Geosphere-Biosphere Programme (IGBP) classification scheme. In situ SM measurements and SM retrievals were compared, and statistical scores were computed. The optimum albedo configuration was then found for each class of the IGBP landcover classification. Preliminary results yield values of albedo between 0.07 to 0.12 under the assumption of homogeneous pixels.

Optimizing the algorithm for retrieving soil moisture from SMOS data

This paper focuses on a new approach to account for soil roughness effects in the retrieval of soil moisture (SM) at L-band in the framework of the SMOS (Soil Moisture and Ocean Salinity) mission: the Simplified Roughness Parameterization (SRP). While the classical retrieval approach considers SM and τNAD (vegetation optical depth) as retrieved parameters, this approach is based on the retrieval of SM and the TR parameter combining τNAD and soil roughness (TR = τNAD + HR/2). Different roughness parameterizations were tested to find the best correlation (R), bias and unbiased RMSE (ubRMSE) when comparing homogeneous retrievals of SM and in situ SM measurements carried out at the VAS (Valencia Anchor Station) vineyard field. The highest R (0.68) and lowest ubRMSE (0.056 m3 m-3) were found using the SRP method. Using the SMOS observations comparisons against several SM networks were also made: AACES, SCAN, watersheds and SMOSMANIA. SM was retrieved over all these stations. The SRP and another similar approach (SRP2) improved the averaged ubRMSE, while the SRP2 method leaded to higher correlation values (R). A global underestimation of SM was noticed, which may be linked to the differences in the sampling depths of the L-band observations (~ 0-3cm for both Elbara-II and SMOS) and of the in situ measurements (~ 0-5 cm).

The SMOS Soil Moisture Retrieval Algorithm

The Soil Moisture Ocean Salinity (SMOS) mission was launched in 2009 and provides derived soil moisture globally using a forward modelling approach that incorporates a number of auxiliary data sets. By default, the SMOS mission uses global land cover and soils data sets to run the soil moisture retrieval models. This study examines the use of national data sets from Agriculture and Agri-Food Canada (AAFC) to determine if improvements in land cover and soils accuracies achieved using these national data sets can provide an improvement in SMOS soil moisture retrieval. Results show that changing the land cover produced the greatest differences, with a reduction in the fraction of the land area identified as forest, but also an increase in the number of failed model retrievals. The use of the AAFC soils resulted in a greater fraction of clay in the surface soil layer, but this did not have a large impact on the overall retrieval accuracy at the study sites. This suggests that the default SMOS parameterization can provide adequate estimation of soil moisture over most sites, but areas where forest, wetland or urban land cover may be over or underestimated should be more closely evaluated.