A joint approach for estimating vertically continuous soil moisture profiles by combining P-band SAR polarimetry with soil hydrological modeling is proposed. The approach compares the decomposed soil scattering component from remotely sensed P-band SAR observations of NASA’s Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) mission with an ensemble of simulated counterparts based on the hydrological model HYDRUS-1D and the soil scattering model multi-layer small perturbation method (SPM). From the best fit between remote sensing and soil modeling, the most probable soil moisture profile can be retrieved. Estimated soil moisture profiles at individual monitoring stations across the U.S. are compared to in situ measurements, as well as the European ReAnalysis (ERA5) land and AirMOSS L4 products. Pearson’s coefficient of determination between estimated and auxiliary products prove the overall feasibility of the proposed method with respective R2of 0.92, 0.95, and 0.87.

Accurate soil carbon data are important for understanding the permafrost response and potential carbon release to future climate change. However, there is a large discrepancy in current soil organic carbon (SOC) estimates in the Arctic, where sparse measurements are unable to capture SOC complexity over the vast and remote region. Polarimetric Synthetic Aperture Radar (SAR) data are sensitive to roughness and moisture conditions of soil and vegetation, and may provide useful information on surface and profile SOC properties ( Yi et al., 2021 , 2022 ). The NASA Arctic Boreal Vulnerability Experiment (ABoVE) airborne campaign acquired an abundance of full-polarimetric P- and L-band SAR data across Alaska and western Canada ( Miller et al., 2019 ), which provides opportunities to test new remote sensing applications. The main objective of this study is to investigate the potential of low-frequency polarimetric SAR data for regional SOC retrieval in the Arctic through data analysis and modeling. We chose the Alaska North Slope as our study area due to more in-situ data available in this area.

A method for estimating complex soil permittivity (or moisture) and penetration depth based on SAR decomposition is presented. By combining model- and eigen-based decomposition techniques, SAR observations are separated into single scattering components (from soil & vegetation). The proposed method incorporates a multi-layer rough surface scattering model to simulate the soil scattering contribution. From the decomposed soil scattering component, permittivity and thus penetration depth can be estimated from SAR observations. Results are presented for the AirMOSS campaign within the MOISST site, OK, USA. As first results, a median value of 16.44 + 2.02 for the complex permittivity was estimated from the 19 P-band data takes at the SoilSCAPE in situ station in Canton, OK, USA. Overall, the retrieval results for the real part of the complex permittivity are similar to in situ values at 30 cm soil depth, with differences in respective median values of 3.44. The median of penetration depths at the Canton site is 23.91 cm,

The aim of this paper is to systematically combine complementary LiDAR and synthetic aperture radar (SAR) observations to map tree canopy cover in a boreal forest. LiDAR data can provide direct measurements of vegetation structures but are limited by the sparse spatial coverage of observations. SAR systems can perform wall-to-wall high-resolution mapping without weather constraints but the information about vegetation and ground subsurface are mixed in the backscatter data. In this paper, we adopted the Random Forests algorithm to train an upscaling function using tree canopy cover (TCC) and canopy height model (CHM) derived from Goddard's LiDAR, Hyperspectral and Thermal Imager (G-LiHT) point cloud data. The regression model was then applied to the L-band Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) data acquired during the 2017 Arctic-Boreal Vulnerability Experiment (ABoVE) airborne campaign to map the TCC and CHM over the Delta Junction area in interior Alaska.

We propose an inversion model for retrieving the root-zone soil moisture profile using the P-band radar observations and validate it using data collected during the Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) mission. The forward model utilized in this article uses an existing small perturbation method (SPM) to calculate the electromagnetic wave scattered from multilayered rough surface dielectric structures. To solve the inverse problem, which is formulated as a least-squares problem, we utilize a novel global optimization technique that takes the advantage of the speed of a local optimization method while ensuring convergence to the global minimum. We present the numerical results to demonstrate significant improvements in the inversion results and computational speed compared to the previously used AirMOSS baseline inversion algorithm. The presented method is validated with in situ soil moisture measurements. Finally, by utilizing a closed-form analytical solution to Richards' equation, we demonstrate improvements in the accuracy of our soil moisture retrieval results.

The Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) P-band synthetic aperture radar (SAR) was flown more than 1200 h from August 2012 to September 2015, covering regions of 2500 km2spread over nine major biomes in North America. The flights, as a part of the NASA AirMOSS Earth Venture Suborbital 1 (EVS-1) mission, collected radar data used to map root-zone soil moisture (RZSM) at 3-arcsec resolution. We previously reported the baseline retrieval algorithm and demonstrated its performance for a semiarid shrubland (Walnut Gulch, AZ, USA); we represented the RZSM profile as a continuous quadratic function and solved a radar scattering nonlinear optimization problem to obtain the unknown polynomial coefficients. In this article, we expand the retrievals to other AirMOSS sites that, in addition to the semiarid shrubland, include grassland and crops (MOISST, OK, USA), woody savanna (Tonzi Ranch, CA, USA), temperate conifer forest (Metolius, OR, USA), and boreal forest (Saskatchewan, Canada). Due to a wide range of land covers, soil types, and soil moisture regimes, we parameterize the forward model and constrain the inverse algorithm for each site separately. We present the full set of retrievals for these sites, validating the results against in situ observations. Error sources and strategies to minimize their effects are discussed. The concept of sensing depth is introduced. We find that the retrieval errors are smallest for the top 25 cm of soil with a root-mean-square error (RMSE) of less than 0.05 m3/m3. The RMSE remains around 0.06 m3/m3even for depths reaching 45 cm, which is the typical sensing depth for the sites considered. These AirMOSS RZSM products (known as Level-2/3 RZSM, or L2/3-RZSM, products) are the first of their kind in that it is the first time RZSM has been retrieved directly from remote sensing observation.

In this paper, a new set of measurements is presented to investigate the relationship between the hydraulic and electromagnetic properties of the Arctic permafrost active layer soil, which includes high organic matter content. Low-frequency dielectric permittivity and the soil water matric potential have been measured for over 50 soil samples collected from northern Alaska for a full range of soil moisture value. The organic matter content is included as a new input parameter and the hydraulic properties of soil have also been taken into account. The result is a new dielectric mixing model for soils containing organic matter and surpassing the range of validity of other existing models.

In this paper, a harmonized soil parametrization that accounts for both soil organic matter (SOM) content and degree of decomposition are used to parametrize the hydraulic properties of permafrost soils. Using the wilting point calculated from the soil-water retention curve to inform the amount of bound water in a soil medium, we construct a soil dielectric mixing model effective across the full range of SOM levels. The active layer soils are parametrized as profile functions of SOM and soil moisture in the radar retrievals. The NASA P-band Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) data acquired as part of 2017 Arctic-Boreal Vulnerability Experiment (ABoVE) airborne campaign are used to demonstrate the potentials of retrieving SOM and soil moisture profiles from P-band synthetic aperture radar (SAR).

We propose a method to estimate the active layer properties, including soil dielectric profiles and active layer thickness (ALT), in permafrost regions using time-series P-band polarimetric synthetic aperture radar (SAR) observations. The active layer and underlying permafrost are modeled as a three-layer dielectric structure with the layer dielectric constants representing the soil moisture and freeze-thaw state of the layer. To resolve the ambiguity of the retrieved layer thicknesses, an approach of finding the largest possible depths (LPDs) is combined with time-series observations, where the ALT is assumed time-invariant between the maximum thaw and before the upward freezing front rises significantly from the permafrost table. The LPD-assisted time-series retrieval algorithm is applied to the radar data acquired by the Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) P-band SAR in August and October 2014 and 2015 over the Alaska North Slope. The results show that the retrieved ALT values are generally underestimated for the sites where the in situ ALT is larger than the P-band sensing depth, with a retrieval bias ranging from -0.05 to -0.24 m as validated against the in situ ALT collected at Circumpolar Active Layer Monitoring (CALM) sites. For the sites where the in situ ALT is smaller than 0.55 m, the retrieval errors are generally less than 0.1 m. The retrieval results also show that the active layer properties are strongly influenced by the land cover types at the regional scale, and not as much by the North-South temperature gradient across a 180-km-long transect along the Deadhorse flight line.

In this paper, a permafrost active layer retrieval algorithm using time-series P-band SAR observations is presented wherein both layer dielectric constants (representing unfrozen water content) and layer thicknesses (representing water table or thaw depths) of the active layer are retrieved. The time-series observations were acquired after the maximum thaw, so that variables such as active layer thickness (ALT) can be assumed constant as they are nearly time-invariant between the data acquisitions. Having these time-invariant variables can reduce the number of unknowns and improve the retrieval accuracy with limited number of observations. Comparison of the retrieval results with in-situ observations suggests that the retrieved ALT values are generally underestimated.

In this paper, we characterize vertical dielectric profile of permafrost soils to account for different soil horizons (or-ganic/mineral/permafrost) and seasonal variations of soil moisture and freeze/thaw state. The in-situ soil profile measurements from SoilSCAPE sites in Alaska and field observations are analyzed. We proposed three profile functions (i.e., constant, second-order polynomial, and exponential) to be used for modeling permafrost soils in radar retrievals. Constant profile has the lowest errors between in-situ and fitted profiles as well as between measured and simulated radar backscatter values. However, constant profile can only represent sharp transitions across different soil boundaries, whereas the exponential function can describe more realistic transitions at both ends of the active layer while having similar errors as constant profile. Second-order polynomial is only suitable for root-zone soil moisture (RZSM) in non-permafrost regions as permafrost soils have multiple abrupt changes throughout the soil column.

In this paper, a dual-frequency permafrost active layer thickness (ALT) retrieval algorithm is presented wherein P-band AirMOSS and L-band UAVSAR data are used to retrieve the layered soil dielectric constants and layer thicknesses. Using the known radar calibration accuracy, radar sensing depth is defined to help understand the limit of radar backscatter sensitivity to subsurface soil conditions and used as a consistency check when determining ALT. Retrieved maps of soil dielectric constants and layer thicknesses show high spatial correlation with the vegetation type and organic layer thickness. It is shown that the retrieval accuracy for ALT is between 5 cm and 14 cm when the ALT is smaller than the sensing depth.

Precise evaluation of spatial and temporal changes in soil moisture is critical for environmental studies. In this regard, Synthetic Aperture Radar (SAR), providing multi-channel dataset with wide coverage, is a reliable tool for monitoring this parameter. This study proposes Modified Improved Water Cloud Model (MIWCM) as semi-analytical method for predicting soil moisture. C, L and P-band data were obtained in 2003 by Airborne Synthetic Aperture Radar (AIRSAR). Results clearly shows superiority of the proposed method comparing to previous models, with R2in the 95%-98% range and RMSE in the 0.00012-0.0016 range.

NASA's Airborne Microwave Observatory of Sub-canopy and Subsurface (AirMOSS) project, with more than 1200 flight hours of a P-band synthetic aperture radar (SAR), was completed in September 2015. The goal of this mission was to improve the estimates of the North American Net Ecosystem Exchange (NEE) through providing high-resolution observations of root zone soil moisture (RZSM) over nine regions representative of the major North American biomes. AirMOSS flights covered areas of approximately 100 km by 25 km containing FLUXNET tower sites in regions that ranged from boreal forests in Saskatchewan, Canada, to tropical forests in La Selva, Costa Rica. These sites are marked in Fig. 1.

In this paper, an active layer thickness (ALT) retrieval algorithm is presented wherein time-series of P- and L-band radar observations are used simultaneously to retrieve the depth from ground surface to permafrost table. Several model assumptions for two active layer soil conditions (maximum thaw and partially frozen) are made based on observations of in situ soil temperature and soil moisture data. It is expected that P- and L-band radar measurements can provide different aspects of active layer soil structure for retrieving accurate ALT. Monte Carlo numerical simulations are performed to show the potentials of the proposed inversion scheme and its capability of resolving subsurface features in a layered soil structure.

We have concluded that before calculation of the AirMOSS project overall RMSE, not only the behavior and location of in-situ soil moisture probes need to be carefully evaluated but also the present error biases-associated with the in-situ probes, radar calibration, vegetation parameterization, forward model inaccuracy, and the inversion algorithm bias-need to be removed. Several measures have been taken over the course of AirMOSS mission to increase the accuracy of RZSM products. The overall RMSE was reported for each study site. We can show that the overall AirMOSS retrieval error for the top 25 cm in BERMS, Metolius, MOISST, Tonzi Ranch, and Walnut Gulch meets the mission RMSE requirement. This error would be calculated in terms of RMSE between the retrieved and actual moisture values over all qualified validation points, all sites, and all dates. The retrieval performance is expected to improve even more with reprocessing of some of the flights using recalibrated radar data.

We propose a new model for estimating subsurface soil moisture using P-band radar data over barren, shrubland, and vegetated terrains. The unknown soil moisture profile is assumed to have a second-order polynomial form as a function of subsurface depth with three unknown coefficients that we estimate using the simulated annealing algorithm. These retrieved coefficients produce the value of soil moisture at any given depth up to a prescribed depth of validity. We use a discrete scattering model to calculate the radar backscattering coefficients of the terrain. The retrieval method is tested and developed with synthetic radar data and is validated with measured radar data and in situ soil moisture measurements. Both forward and inverse models are briefly explained. The radar data used in this paper have been collected during the Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) mission flights in September and October of 2012 over a 100 km by 25 km area in Arizona, including the Walnut Gulch Experimental Watershed. The study area and the ancillary data layers used to characterize each radar pixel are explained. The inversion results are presented, and it is shown that the RMSE between the retrieved and measured soil moisture profiles ranges from 0.060 to 0.099 m3/m3, with a Root Mean Squared Error (RMSE) of 0.075 m3/m3over all sites and all acquisition dates. We show that the accuracy of retrievals decreases as depth increases. The profiles used in validation are from a fairy dry season in Walnut Gulch and so are the accuracy conclusions.

Most traditional radar retrieval techniques utilize radar forward models that assume or require homogeneous scenes. NASA's upcoming Soil Moisture Active Passive (SMAP) mission launching in October 2014 will carry an L-band radar delivering a pixel size of 3 km × 3 km. Even though this resolution is finer than the SMAP 36 km radiometer pixels, it is still coarse enough that modeling such pixels assuming homogeneous scenes is not realistic. This highlights the need to develop spatial aggregation and disaggregation techniques using radar forward scattering models that assume homogeneity over fine-scale sub-pixels and derive tailor-made models for their contribution to the coarse-scale satellite pixel for successful soil moisture retrieval. For radar based soil moisture retrieval to be a strong candidate for future products, this uncertainty has to be successfully mitigated. To study the heterogeneity at landscape level, high-resolution P-band data from the Earth Venture 1 (EV-1) Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) mission is used as its pixel resolution is fine enough to assume homogeneity over a pixel. In this work, a sampling area over a well-known AirMOSS study site is chosen, an aggregation study based on landcover type is carried out and a coarse resolution radar backscatter function over the sampling area is found to be a weighted linear average of the contributions of the fine-scale pixels.

We present an analytical solution for coherent scattering of electromagnetic waves from a two-layer rough surface structure with uncorrelated random rough interfaces. The Kirchhoff approximation is used to predict the coherent (specular) component of the scattered wave from a layered rough surface that is assumed to have radii of curvature much larger than the wavelength to allow the application of the method. The roughness on both boundaries is assumed small such that the coherent component of the scattered wave is dominant. The derived solution includes all orders of the scattered wave and has a simple algebraic expression that can be readily computed. We validate the solution against a numerical method and present simulation results for various cases.

A generalized radar scattering model to predict backscattering for different frequencies and polarizations is crucial in the endeavor to understand the relationship between radar measurement and properties of both the vegetation and soil, and for allowing the estimation of both vegetation and soil information from radar data. In this work, we develop a combined radar scattering model of vegetation and multilayered soil structure to represent realistic soils with multiple layers of various textures, depths, and moisture contents. A multilayered soil scattering model based on first order small perturbation method (SPM) is integrated [1] to account for direct ground scattering. Furthermore, the coherent interaction between the vegetation and layered ground is included using a recently developed model based on the Kirchhoff approximation [2].

The Airborne Observatory of Subcanopy and Subsurface (AirMOSS) is one of the five Earth Venture-1 missions selected in May 2010, and seeks to improve the estimates of the north American net ecosystem carbon exchange (NEE) through high-resolution observations of root zone soil moisture (RZSM). To obtain estimates of RZSM and assess its heterogeneities, AirMOSS will fly a P-band (430 MHz) synthetic aperture radar (SAR) over 2500 km2areas within nine major biomes of north America. Retrieval of RZSM in the presence of substantial vegetation requires the construction of accurate radar scattering models that account for diverse vegetation conditions as well as subsurface inhomogeneities. In this paper we provide a summary of recent advances in this area, emphasizing several coherent and incoherent models of scattering from multilayered inhomogeneous rough surfaces, as well as strategies for using these forward models in retrieval algorithms.

In this paper, we study the radar retrieval of soil moisture as well as canopy parameters in a range of boreal forests. The retrieval is formulated as an optimization problem where the difference between data and prediction of a forward scattering model is minimized. The forward model is a discrete scatterer radar model, and the optimization algorithm is a global optimization scheme known as simulated annealing. The inversion method is first applied to synthetic data assuming hypothetical allometric relationships to make the retrieval possible by reducing the number of unknown vegetation parameters. The inversion algorithm is then validated using the data acquired with the National Aeronautics and Space Administration (NASA)/Jet Propulsion Laboratory (JPL) Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) in June 2010 in central Canada boreal forests in support of the prelaunch calibration and validation activities of NASA's Soil Moisture Active and Passive (SMAP) mission. The inversion results for synthetic data show that the absolute retrieval error in soil moisture and relative retrieval error in canopy height are small, while the relative output error in trunk density could be large. The inversion results for actual field data show a great accuracy in soil moisture retrieval for Old Jack Pine and Young Jack Pine forests but show large retrieval errors for many of the radar pixels in the Old Black Spruce site. This paper shows that L-band radar is capable of retrieving surface soil moisture in forests with a high biomass where the forest structure allows soil moisture information to be carried by scattering mechanisms.

In this paper, we study the radar retrieval of soil moisture and canopy parameters in a forested area. The for ward model is a discrete scatterer radar model and the inverse model implements a global optimization scheme known as simulated annealing. Inversion is first applied to synthetic data assuming hypothetical allometric relationships to make the retrieval possible by reducing the number of unknowns. The inversion algorithm is then validated using data from the Canadian Experiment for Soil Moisture in 2010 (CanEx-SMlO) acquired with the NASA/JPL UAVSAR in June 2010 in central Canada boreal forests in support of the pre-launch calibration and validation activities of NASA's Soil Moisture Active and Passive (SMAP) mission.

We study the retrieval of surface and deep moisture of bare soil from noisy radar observations using simulated annealing. Due to moisture variations with depth, we model bare soil with a stratified dielectric profile with a rough surface on top. Small perturbation method (SPM) is used as the forward model. We use the full moisture profile for radar data synthesis and study the retrieval accuracy by varying the number of layers that represent the soil profile during inversion. The effect of measurement frequency on the accuracy of deep moisture retrieval is investigated. This work is intended for assessing the effect of subsurface profile on soil moisture retrieval from radar observations of NASA's Soil Moisture Active and Passive (SMAP) mission and future lower frequency airborne or spaceborne systems that may follow SMAP.

We previously derived the bistatic scattering coefficients of a 3-D two-layer dielectric structure with slightly rough boundaries using the small perturbation method (SPM). The use of SPM raises the question about its region of validity, which pertains to the conditions on each layer roughness, slope, and permittivity for which the first-order SPM is accurate within a specified error bound. To this end, the SPM solution needs to be compared with an accurate solution that does not impose roughness restrictions. We use the method of moments to solve an integral equation to analyze electromagnetic scattering from a large ensemble of two-layer structures. Simulations are performed for 1-D rough surfaces represented by zero-mean stationary random processes, separating homogeneous dielectric layers. Observations are reported on the accuracy of the first-order SPM for TM incidence at a fixed incidence angle of 45°.

This work addresses the noise sensitivity of the simulated annealing method in inversion of subsurface parameters of layered rough surfaces to measurement noise. We consider two different noise models and assess the noise response of the inversion algorithm for each of the models. Conclusions are made based on the calculated average and standard deviation of the output error in the retrieved model parameters.

In this paper, the model parameters of a two-layer dielectric structure with random slightly rough boundaries are retrieved from data that consist of the backscattering coefficients for multiple polarizations, angles, and frequencies. We use the small perturbation method to solve the forward problem. The inversion problem is then formulated as a least square problem and is solved using a global optimization method known as simulated annealing, which is shown to be a robust retrieval algorithm for our purpose. The algorithm performance depends on several parameters. We make recommendations on these parameters and propose a technique for exiting local minima when encountered. We test the sensitivity of the inversion scheme to measurement noise and present the noise analysis results.

We have shown that the method of simulated annealing (SA) is capable of inverting subsurface parameters of a three-dimensional, two-layer dielectric structure with slightly rough interfaces. In this work, we address the sensitivity of the Simulated Annealing to measurement noise. We also study the impact of the number of measurement parameters, i.e., number of frequency points and number of observation angles, on sensitivity of the inversion algorithm.

We previously applied an optimization technique known as Simulated Annealing (SA) to the inverse problem associated with a 3D two-layer dielectric structure with slightly rough interfaces and showed that simulated annealing methods are capable of globally minimizing cost functions with many local minima [1]. Nonlinearity of the cost function is a major factor that decreases the performance of the inversion algorithm. With a fixed set of measurement and inversion parameters, as the number of unknown model parameters increases, the cost function nonlinearity becomes more severe, decreasing the efficiency of inversion and making the annealing process perform like an inefficient brute force search. The focus of this work is on strategies to choose the optimal set of measurement parameters for retrieval of the largest possible number of parameters of a layered dielectric structure.

We use the simulated annealing (SA) method to find the global minimum of a cost function associated with scattering from a layered dielectric structure with rough boundaries. In the case of a two layer structure, even a simple implementation of the SA algorithm is capable of retrieving more parameters than is possible with local optimization methods. We discuss different strategies that can improve the performance of this algorithm for our purpose.

An analytical method to calculate the bistatic-scattering coefficients of a three-dimensional layered dielectric structure with slightly rough interfaces is presented. The interfaces are allowed to be statistically distinct, but possibly dependent. The waves in each region are represented as a superposition of an infinite number of up- and down-going spectral components whose amplitudes are found by simultaneously matching the boundary conditions at both interfaces. A small-perturbation formulation is used up to the first order, and the scattered fields are derived. The calculation intrinsically takes into account multiple scattering processes between the boundaries. The formulation is then validated against known solutions to special cases. New results are generated for several cases of two- and three-layer media, which will be directly applicable for modeling of the signals from radar systems and subsequent estimation of a layered medium subsurface properties, such as moisture content and layer depths

An analytical method to calculate the scattering coefficients of a three-layer 3D rough surface is introduced. The two rough interfaces are assumed to be distinct. The waves in each region are represented as a superposition of an infinite number of up- and down-going spectral wave components, whose amplitudes are found by applying the boundary conditions. A small-perturbation formulation is used in the process and the scattering coefficients are derived to the first order. The formulation is validated against known solutions for special cases of flat interfaces as well as the single rough surfaces. Results are then generated for several cases of the three-layer rough interface medium, to be used for modeling of the backscattered signals from a tower-based radar system and subsequent estimation of multilayered soil properties such as moisture content and subsurface layer height