Amine Merzouki

Canadian Experiment for Soil Moisture in 2010 (CanEx-SM10): Overview and Preliminary Results

The Canadian Experiment for Soil Moisture in 2010 (CanEx-SM10) was carried out in Saskatchewan, Canada, from 31 May to 16 June, 2010. Its main objective was to contribute to Soil Moisture and Ocean Salinity (SMOS) mission validation and the prelaunch assessment of the proposed Soil Moisture Active and Passive (SMAP) mission. During CanEx-SM10, SMOS data as well as other passive and active microwave measurements were collected by both airborne and satellite platforms. Ground-based measurements of soil (moisture, temperature, roughness, bulk density) and vegetation characteristics (leaf area index, biomass, vegetation height) were conducted close in time to the airborne and satellite acquisitions. Moreover, two ground-based in situ networks provided continuous measurements of meteorological conditions and soil moisture and soil temperature profiles. Two sites, each covering 33 km × 71 km (about two SMOS pixels) were selected in agricultural and boreal forested areas in order to provide contrasting soil and vegetation conditions. This paper describes the measurement strategy, provides an overview of the data sets, and presents preliminary results. Over the agricultural area, the airborne L-band brightness temperatures matched up well with the SMOS data (prototype 346). The radio frequency interference observed in both SMOS and the airborne L-band radiometer data exhibited spatial and temporal variability and polarization dependency. The temporal evolution of the SMOS soil moisture product (prototype 307) matched that observed with the ground data, but the absolute soil moisture estimates did not meet the accuracy requirements (0.04 m3/m3) of the SMOS mission. AMSR-E soil moisture estimates from the National Snow and Ice Data Center more closely reflected soil moisture measurements.

Mapping Soil Moisture Using RADARSAT-2 Data and Local Autocorrelation Statistics

The purpose of this study is to evaluate the capability of surface radar backscatter models to estimate soil moisture over agricultural fields from fully polarimetric RADARSAT-2 C-band synthetic aperture radar (SAR) responses. For validation purposes, ground measurements over 44 sampling sites in eastern Ontario, Canada were carried out in the spring of 2008 simultaneously with satellite data acquisitions. Soil moisture retrieval was accomplished using two semi-empirical scattering models (Dubois and Oh) and the SAR image backscatter. Discrepancies between measured radar backscatter coefficients and those predicted by the models were previously reported, requiring correction factors to reduce biases associated with these semi-empirical approaches. Soil moisture was estimated by explicitly solving the two backscatter equations of the Dubois model, and using a look-up table (LUT) approach applied to the Oh model. Results showed that the Oh model in a cross-polarization (HH-HV) and Dubois in a co-polarization (HH-VV) inversion scheme provide the best estimates. These model configurations were implemented to produce multi-date soil moisture maps for the eastern Ontario site. To expand the range of validity of these soil moisture estimates, the maps produced by the Dubois and Oh models were uniquely combined. These estimates of absolute soil moisture were then used to derive spatial patterns of near-surface moisture content using the Getis statistic. The Getis statistic maps provide meaningful spatial information, demonstrating the potential of combining the Getis statistic and RADARSAT-2 data in predicting soil moisture conditions.

Sensitivity of C-band SAR polarimetric variables to unvegetated agricultural fields

The interaction of linear polarized microwaves with agricultural features is well understood. Much less is understood about polarimetric data and the potential use of these data to improve surface parameter retrieval models. This paper explores the soil surface information provided by quad-polarimetric SAR through a review of previous work and an empirical sensitivity study of RADARSAT-2 data at four incidence angles. Soil moisture, surface roughness, and crop residue data are quantitatively sampled in unvegetated fields. RADARSAT-2 variables include: linear backscatter and polarization ratios; copolarized phase difference and magnitude of the copolarized complex correlation; pedestal height; extrema of the scattered intensity, completely polarized, and completely unpolarized components; and parameters of the Cloude–Pottier scattering decomposition. Results demonstrated that sensitivities of field averaged linear backscatter were reproduced from previous reports, lending confidence to the experiment. Field averaged pedestal height and copolarized complex correlation coefficient showed significant relationships to crop residue and surface roughness, suggesting an ability to characterize volume and multiple scattering. Similarly, dynamic range of the degree of polarization showed significant relationships with crop residue cover at higher incidence angles. Target averaged or variance of copolarized phase difference did not produce a consistent relationship with the surface parameters, in contrast to qualitative based suggestions of previous experiments. Extremas of the scattered intensity and completely polarized components indicated comparable relationships to surface features as the like polarized linear intensity channels, suggesting sensitivities of these variables to surface scattering. The extrema of the completely unpolarized component showed comparable relationships to surface features as the pedestal height. Results of this paper contribute to identification of optimal SAR variables for use in agricultural monitoring and evaluate potential contributions of polarimetric data for improving surface parameter retrieval models.

Monitoring Soil Moisture to Support Risk Reduction for the Agriculture Sector Using RADARSAT-2

Monitoring the amount of moisture held in the soil is critical in the management of risk for the agriculture sector. Extremes in soil moisture can lead to devastating consequences. Early assessment of soil moisture reserves, and monitoring of changes in available soil moisture, could assist in risk reduction strategies for the agriculture sector and effective delivery of government programs. Agriculture and Agri-Food Canada has been acquiring RADARSAT-2 data since 2008 to evaluate the accuracy with which this sensor can provide soil moisture to assist with implementing risk reduction strategies for the Canadian agriculture sector. The calibrated Integral Equation Model (IEM) was used to estimate soil moisture for 15 RADARSAT-2 data sets acquired over an eastern and western Canadian test site. Using this approach, field level soil moisture was estimated to a mean average error of 7.95%, although considerable scatter in the results was observed. Removing fields which had significant residue cover improved site specific soil moisture errors, but only for the fall campaign prior to spring tillage and seed bed preparation. Higher errors were also observed for data sets where angles between the RADARSAT-2 look direction and field tillage structures were largest. When soil moisture estimates were evaluated at a regional scale, mean errors fell to 3.14%. The IEM was also able to detect increases and decreases in soil moisture which followed periods of rainfall and drying.

Evaluation of the Dubois, Oh, and IEM radar backscatter models over agricultural fields using C-band RADARSAT-2 SAR image data

Backscatter from synthetic aperture radar (SAR) is correlated with surface characteristics such as soil dielectric properties, surface roughness, and vegetation cover. The purpose of this study is to evaluate the capability of surface radar backscatter models to estimate soil moisture over agricultural fields from fully polarimetric RADARSAT-2 C-band SAR responses. For validation purposes, ground measurements over 44 and 42 sampling sites in eastern Ontario and southern Manitoba, respectively, were carried out in the spring of 2008 simultaneous with satellite data acquisitions. A comparison was made between the backscatter coefficient results derived from three scattering models (the semi-empirical models of Dubois and Oh and the theoretical integral equation model (IEM)) and the SAR image backscatter. Discrepancies between measured radar backscatter coefficients and those predicted by the models have been investigated. Overall, the results show that semi-empirical approaches tend to overestimate the radar response, but correction factors of about 3.5 and 2.0 dB, respectively, were found sufficient to correct the Dubois HH and VV backscatter coefficients. The Oh model backscatter estimations were less variable than those derived from the Dubois model; however, a correction factor of about 5.0 dB was necessary in this case. Data simulated by the IEM showed significant fluctuations. This result was somewhat expected, since previous studies have shown that correlation length measurements are very sensitive to profile length, and a relatively short profile length was used in this study. Correlation agreements were significantly improved when using the IEM semi-empirical calibration technique. These results indicate that further work is needed to assess the requirement for correction factors, such that these models can be operationally implemented to estimate soil moisture in support of agricultural monitoring.

Potential of mapping soil moisture by combining radar backscatter modeling and PolSAR decomposition

The purpose of this study is to evaluate the capability of the Oh backscattering model in combination with the Freeman Durden decomposition to estimate soil moisture over agricultural fields from fully polarimetric RADARSAT-2 C-band SAR responses. Initially, soil moisture multi-polarization retrieval was accomplished by using a look-up table (LUT) approach applied to the Oh model. Two methods were considered: the multi-polarization method and the one-unknown configuration. Of the two methods, results showed that the HH-HV inversion provided the best estimates. In the second phase, the Freeman Durden decomposition was applied to the polarimetric data. The conceptual approach for retrieving soil moisture using the surface scattering component of the total power was implemented in a LUT inversion. The algorithm attempts to minimize the difference between measured single scattering power obtained by applying the Freeman Durden decomposition and simulated total power using Oh model. When compared with the multi-polarization approach, this polarimetry-based method improves the accuracy of soil moisture estimates.

Evaluating TerraSAR-X for the identification of tillage occurrence over an agricultural area in Canada

Tillage practices can affect the long term sustainability of agricultural soils as well as a variety of soil processes that impact the environment. The benefits of reduced tillage and no-till practices over agriculture fields are well documented and include: (1) significant reductions in wind and water erosion mitigating nutrient and pesticide runoff into waterways; (2) increasing and/or maintaining soil organic matter; (3) increasing biological activity and improving soil structure; and (4) increasing soil carbon and its sequestration. Information on tillage activities assists in implementing policies and programs to promote beneficial management practices (BMPs), and in monitoring the success of these initiatives. Agriculture and Agri-Food Canada supports environmentally responsible agriculture and has identified this as one of their priorities. Thus, tillage information requirements have become increasingly important to a number of programs and policies within the department. Rapid, accurate and objective methods are required to map and monitor tillage activities. Earth observing satellites can assist with targeting and monitoring land management activities. For the last decade, research has clearly demonstrated that complementary information provided by both optical and radar satellite sensors are fundamental in developing an agricultural land management monitoring system. Launched in June 2007, the TerraSAR-X is a radar satellite acquiring data at the X-band frequency (9.6 GHz). The application of TerraSAR-X data for conservation tillage mapping has been somewhat limited, and thus this study investigates its use in determining tillage occurrence. An HH-HV TerraSAR-X image was acquired on November 4, 2009 and ground data were also collected characterizing tillage conditions at the time of acquisition. Backscatter responses were analyzed to identify tillage occurrence and to differentiate between untilled, chiseled and moldboard ploughed fields. Preliminary analysis showed that HH polarization can better contribute to tillage discrimination than compared to HV polarization and that the backscatter response can be used to discriminate untilled fields from ones that are moldboard ploughed. However, chiseled fields were often confused with highroughness (rms height~1.30 cm) untilled fields and moldboard ploughed fields. Fully polarimetric X-band radar datasets could potentially contribute more information to mapping tillage conditions.

Monitoring soil moisture to support risk reduction for the agriculture sector using RADARSAT-2

Monitoring the amount of moisture held in the soil is critical in the management of risk for the agriculture sector. Extremes in soil moisture can lead to devastating consequences. Early assessment of soil moisture reserves, and monitoring of changes in available soil moisture, could assist in risk reduction strategies for the agriculture sector and effective delivery of government programs. Agriculture and Agri-Food Canada has been acquiring RADARSAT-2 data since 2008 to evaluate the accuracy with which this sensor can provide soil moisture to assist with implementing risk reduction strategies for the Canadian agriculture sector. The calibrated Integral Equation Model (IEM) was used to estimate soil moisture for 15 RADARSAT-2 data sets acquired over an eastern and western Canadian test site. Using this approach, field level soil moisture was estimated to a mean average error of 7.95%, although considerable scatter in the results was observed. Removing fields which had significant residue cover improved site specific soil moisture errors, but only for the fall campaign prior to spring tillage and seed bed preparation. Higher errors were also observed for data sets where angles between the RADARSAT-2 look direction and field tillage structures were largest. When soil moisture estimates were evaluated at a regional scale, mean errors fell to 3.14%. The IEM was also able to detect increases and decreases in soil moisture which followed periods of rainfall and drying.

Sentinel-1 high resolution soil moisture

The systematic retrieval of near surface soil moisture (SSM) fields at high resolution (e.g., 0.1–1.0 km) is a challenging task that requires the exploitation of new retrieval algorithms and SAR data with advanced observational capabilities (in terms of spatial/temporal resolution, radiometric accuracy, very large swath, long-term continuity and rapid data dissemination). The launch of the Sentinel-1 (S-1) constellation provides these capabilities and calls for the development and validation of pre-operational SSM products at high resolution. The objective of this paper is to present and initially assess a SSM retrieval algorithm developed in view of S-1 data exploitation. The activity is supported by a large scientific community engaged in fostering a more effective interaction between researchers working in the field of high and low resolution SSM retrieval.