Deep learning estimation of northern hemisphere soil freeze-thaw dynamics using satellite multi-frequency microwave brightness temperature observations.

Satellite microwave sensors are well suited for monitoring landscape freeze-thaw (FT) transitions owing to the strong brightness temperature (TB) or backscatter response to changes in liquid water abundance between predominantly frozen and thawed conditions. The FT retrieval is also a sensitive climate indicator with strong biophysical importance. However, retrieval algorithms can have difficulty distinguishing the FT status of soils from that of overlying features such as snow and vegetation, while variable land conditions can also degrade performance. Here, we applied a deep learning model using a multilayer convolutional neural network driven by AMSR2 and SMAP TB records, and trained on surface (~0-5 cm depth) soil temperature FT observations. Soil FT states were classified for the local morning (6 a.m.) and evening (6 p.m.) conditions corresponding to SMAP descending and ascending orbital overpasses, mapped to a 9 km polar grid spanning a five-year (2016-2020) record and Northern Hemisphere domain. Continuous variable estimates of the probability of frozen or thawed conditions were derived using a model cost function optimized against FT observational training data. Model results derived using combined multi-frequency (1.4, 18.7, 36.5 GHz) TBs produced the highest soil FT accuracy over other models derived using only single sensor or single frequency TB inputs. Moreover, SMAP L-band (1.4 GHz) TBs provided enhanced soil FT information and performance gain over model results derived using only AMSR2 TB inputs. The resulting soil FT classification showed favorable and consistent performance against soil FT observations from ERA5 reanalysis (mean percent accuracy, MPA: 92.7%) and in situ weather stations (MPA: 91.0%). The soil FT accuracy was generally consistent between morning and afternoon predictions and across different land covers and seasons. The model also showed better FT accuracy than ERA5 against regional weather station measurements (91.0% vs. 86.1% MPA). However, model confidence was lower in complex terrain where FT spatial heterogeneity was likely beneath the effective model grain size. Our results provide a high level of precision in mapping soil FT dynamics to improve understanding of complex seasonal transitions and their influence on ecological processes and climate feedbacks, with the potential to inform Earth system model predictions.

Real-Time 3D Microwave Medical Imaging With Enhanced Variational Born Iterative Method.

In this paper, we present a new variational Born iterative method (VBIM) for real-time microwave imaging (MWI) applications. The S-parameter volume integral equation and waveport vector Green's function are implemented to utilize the measured signal of the MWI system. Meanwhile, the real and imaginary separation (RIS) approach is used at each iterative step to simultaneously reconstruct the dielectric permittivity and conductivity of unknown objects. Compared with the Born iterative method and distorted Born iterative method, VBIM requires less computational time to reach the convergence threshold. The graphics processing unit based acceleration technique is implemented for real-time imaging. To demonstrate the efficiency and accuracy of this VBIM-RIS method, synthetic analysis of a complex multi-layer spherical phantom is first conducted. Then, the algorithm is tested with measured data using our new MWI system prototype. Finally, a synthetic brain-tumor phantom model under a thermal therapy procedure is monitored to exemplify the real-time imaging with about 5 seconds per reconstruction frame.

Embedded Temporal Convolutional Networks for Essential Climate Variables Forecasting.

Forecasting the values of essential climate variables like land surface temperature and soil moisture can play a paramount role in understanding and predicting the impact of climate change. This work concerns the development of a deep learning model for analyzing and predicting spatial time series, considering both satellite derived and model-based data assimilation processes. To that end, we propose the Embedded Temporal Convolutional Network (E-TCN) architecture, which integrates three different networks, namely an encoder network, a temporal convolutional network, and a decoder network. The model accepts as input satellite or assimilation model derived values, such as land surface temperature and soil moisture, with monthly periodicity, going back more than fifteen years. We use our model and compare its results with the state-of-the-art model for spatiotemporal data, the ConvLSTM model. To quantify performance, we explore different cases of spatial resolution, spatial region extension, number of training examples and prediction windows, among others. The proposed approach achieves better performance in terms of prediction accuracy, while using a smaller number of parameters compared to the ConvLSTM model. Although we focus on two specific environmental variables, the method can be readily applied to other variables of interest.

A Versatile and Shelf-Stable Dielectric Coupling Medium for Microwave Imaging.

OBJECTIVE: To develop a new class of emulsions using a protein-based emulsifier as the coupling fluid for microwave imaging systems. METHODS: In this paper, we provide a theoretical basis for engineering shelf-stable dielectric fluids, a step-by-step formulation method, and measurements of complex dielectric properties in the frequency range of 0.5-3 GHz, which can be applicable for many of the recent microwave imaging systems. RESULTS: This medium was primarily designed for long-term stability while providing a controllable range of complex dielectric permittivities given different fractions of its constituents. Consequently, this emulsion shows dielectric stability in open air throughout a 7-day experiment and temperature insensitivity over the range of 0 ° C to 60 ° . CONCLUSIONS: This control over dielectric permittivity enables formulations that tune the background-to-target contrast to the linearizable regime of iterative inverse scattering algorithms. Accordingly, the emulsion conductivity can also be controlled and reduced to maintain the required signal-to-noise ratio within the dynamic range of the imaging system. The new formulation overcomes the practical challenges of engineering coupling fluids for microwave imaging systems, e.g., temporal stability, non-toxic, low sensitivity to temperature variation, and easy formulation from readily available and inexpensive materials. SIGNIFICANCE: The achieved properties associated with this new fluid are of particular benefit to microwave imaging systems used in thermal therapy monitoring.

Wearable magnetic induction-based approach toward 3D motion tracking.

Activity recognition using wearable sensors has gained popularity due to its wide range of applications, including healthcare, rehabilitation, sports, and senior monitoring. Tracking the body movement in 3D space facilitates behavior recognition in different scenarios. Wearable systems have limited battery capacity, and many critical challenges have to be addressed to gain a trade-off among power consumption, computational complexity, minimizing the effects of environmental interference, and achieving higher tracking accuracy. This work presents a motion tracking system based on magnetic induction (MI) to tackle the challenges and limitations inherent in designing a wireless monitoring system. We integrated a realistic prototype of an MI sensor with machine learning techniques and investigated one-sensor and two-sensor configuration setups for motion reconstruction. This approach is successfully evaluated using measured and synthesized datasets generated by the analytical model of the MI system. The system has an average distance root-mean-squared error (RMSE) error of 3 cm compared to the ground-truth real-world measured data with Kinect.

Permafrost Dynamics Observatory-Part I: Postprocessing and Calibration Methods of UAVSAR L-Band InSAR Data for Seasonal Subsidence Estimation.

Interferometric synthetic aperture radar (InSAR) has been used to quantify a range of surface and near surface physical properties in permafrost landscapes. Most previous InSAR studies have utilized spaceborne InSAR platforms, but InSAR datasets over permafrost landscapes collected from airborne platforms have been steadily growing in recent years. Most existing algorithms dedicated toward retrieval of permafrost physical properties were originally developed for spaceborne InSAR platforms. In this study, which is the first in a two part series, we introduce a series of calibration techniques developed to apply a novel joint retrieval algorithm for permafrost active layer thickness retrieval to an airborne InSAR dataset acquired in 2017 by NASA's Uninhabited Aerial Vehicle Synthetic Aperture Radar over Alaska and Western Canada. We demonstrate how InSAR measurement uncertainties are mitigated by these calibration methods and quantify remaining measurement uncertainties with a novel method of modeling interferometric phase uncertainty using a Gaussian mixture model. Finally, we discuss the impact of native SAR resolution on InSAR measurements, the limitation of using few interferograms per retrieval, and the implications of our findings for cross-comparison of airborne and spaceborne InSAR datasets acquired over Arctic regions underlain by permafrost.

Author Correction: Human activity recognition using magnetic induction-based motion signals and deep recurrent neural networks.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Human activity recognition using magnetic induction-based motion signals and deep recurrent neural networks.

Recognizing human physical activities using wireless sensor networks has attracted significant research interest due to its broad range of applications, such as healthcare, rehabilitation, athletics, and senior monitoring. There are critical challenges inherent in designing a sensor-based activity recognition system operating in and around a lossy medium such as the human body to gain a trade-off among power consumption, cost, computational complexity, and accuracy. We introduce an innovative wireless system based on magnetic induction for human activity recognition to tackle these challenges and constraints. The magnetic induction system is integrated with machine learning techniques to detect a wide range of human motions. This approach is successfully evaluated using synthesized datasets, laboratory measurements, and deep recurrent neural networks.

Electromagnetic Imaging of Dielectric Objects Using a Multi-Directional Search-Based Simulated Annealing.

In this paper, we introduce a global optimization method that is a novel combination of the simulated annealing method and the multi-directional search algorithm. We demonstrate the use of the algorithm for a microwave-imaging system to obtain the electrical properties of objects. The proposed global optimizer significantly improves the performance and speed of the simulated annealing method by utilizing a nonlinear simplex search, starting from an initial guess, and taking effective steps in obtaining the global solution of the minimization problem. Due to the efficient performance of the proposed global optimization method, we are able to obtain the shape, location, and material properties of the target without considering any a priori information about them. The accuracy and applicability of the proposed imaging method is demonstrated with some numerical results in which two-dimensional images of multiple objects are successfully reconstructed.

Real-Time Three-Dimensional Microwave Monitoring of Interstitial Thermal Therapy.

We report a method for real-time three-dimensional monitoring of thermal therapy through the use of noncontact microwave imaging. This method is predicated on using microwaves to image changes in the dielectric properties of tissue with changing temperature. Instead of the precomputed linear Born approximation that was used in prior work to speed up the frame-to-frame inversions, here we use the nonlinear distorted Born iterative method (DBIM) to solve the electric volume integral equation (VIE) to image the temperature change. This is made possible by using a recently developed graphic processing unit accelerated conformal finite difference time domain method to solve the forward problem and update the electric field in the monitored region in each DBIM iteration. Compared to our previous work, this approach provides a far superior approximation of the electric field within the VIE, and thus yields a more accurate reconstruction of tissue temperature change. The proposed method is validated using a realistic numerical model of interstitial thermal therapy for a deep-seated brain lesion. With the new DBIM, we reduced the average estimation error of the mean temperature within the region of interest from 2.5 to 1.0 for the noise-free case, and from 2.9 to 1.7 for the 2% background noise case.

Spatial and temporal variability of root-zone soil moisture acquired from hydrologic modeling and AirMOSS P-band radar.

The accurate estimation of grid-scale fluxes of water, energy, and carbon requires consideration of sub-grid spatial variability in root-zone soil moisture (RZSM). The NASA Airborne Microwave Observatory of Subcanopy and Subsurface (AirMOSS) mission represents the first systematic attempt to repeatedly map high-resolution RZSM fields using airborne remote sensing across a range of biomes. Here we compare 3-arc-sec (~100-m) spatial resolution AirMOSS RZSM retrievals from P-band radar acquisitions over 9 separate North American study sites with analogous RZSM estimates generated by the Flux-Penn State Hydrology Model (Flux-PIHM). The two products demonstrate comparable levels of accuracy when evaluated against ground-based soil moisture products and a significant level of temporal cross-correlation. However, relative to the AirMOSS RZSM retrievals, Flux-PIHM RZSM estimates generally demonstrate much lower levels of spatial and temporal variability, and the spatial patterns captured by both products are poorly correlated. Nevertheless, based on a discussion of likely error sources affecting both products, it is argued that the spatial analysis of AirMOSS and Flux-PIHM RZSM fields provide meaningful upper and lower bounds on the potential range of RZSM spatial variability encountered across a range of natural biomes.

Characterizing permafrost active layer dynamics and sensitivity to landscape spatial heterogeneity in Alaska.

An important feature of the Arctic is large spatial heterogeneity in active layer conditions, which is generally poorly represented by global models and can lead to large uncertainties in predicting regional ecosystem responses and climate feedbacks. In this study, we developed a spatially integrated modelling and analysis framework combining field observations, local scale (~ 50 m resolution) active layer thickness (ALT) and soil moisture maps derived from airborne low frequency (L+P-band) radar measurements, and global satellite environmental observations to investigate the ALT sensitivity to recent climate trends and landscape heterogeneity in Alaska. Modelled ALT results show good correspondence with in situ measurements in higher permafrost probability (PP ≥ 70%) areas (n = 33, R = 0.60, mean bias = 1.58 cm, RMSE = 20.32 cm), but with larger uncertainty in sporadic and discontinuous permafrost areas. The model results also reveal widespread ALT deepening since 2001, with smaller ALT increases in northern Alaska (mean trend = 0.32 ± 1.18 cm yr-1) and much larger increases (> 3 cm yr-1) across interior and southern Alaska. The positive ALT trend coincides with regional warming and a longer snow-free season (R = 0.60 ± 0.32). A spatially integrated analysis of the radar retrievals and model sensitivity simulations demonstrated that uncertainty in the spatial and vertical distribution of soil organic carbon (SOC) was the largest factor affecting modeled ALT accuracy, while soil moisture played a secondary role. Potential improvements in characterizing SOC heterogeneity, including better spatial sampling of soil conditions and advances in remote sensing of SOC and soil moisture, will enable more accurate predictions of active layer conditions and refinement of the modelling framework across a larger domain.

Combined Radar-Radiometer Surface Soil Moisture and Roughness Estimation.

A robust physics-based combined radar-radiometer, or Active-Passive, surface soil moisture and roughness estimation methodology is presented. Soil moisture and roughness retrieval is performed via optimization, i.e., minimization, of a joint objective function which constrains similar resolution radar and radiometer observations simultaneously. A data-driven and noise-dependent regularization term has also been developed to automatically regularize and balance corresponding radar and radiometer contributions to achieve optimal soil moisture retrievals. It is shown that in order to compensate for measurement and observation noise, as well as forward model inaccuracies, in combined radar-radiometer estimation surface roughness can be considered a free parameter. Extensive Monte-Carlo numerical simulations and assessment using field data have been performed to both evaluate the algorithm's performance and to demonstrate soil moisture estimation. Unbiased root mean squared errors (RMSE) range from 0.18 to 0.03 cm3/cm3 for two different land cover types of corn and soybean. In summary, in the context of soil moisture retrieval, the importance of consistent forward emission and scattering development is discussed and presented.

Data Assimilation to extract Soil Moisture Information from SMAP Observations.

This study compares different methods to extract soil moisture information through the assimilation of Soil Moisture Active Passive (SMAP) observations. Neural Network (NN) and physically-based SMAP soil moisture retrievals were assimilated into the NASA Catchment model over the contiguous United States for April 2015 to March 2017. By construction, the NN retrievals are consistent with the global climatology of the Catchment model soil moisture. Assimilating the NN retrievals without further bias correction improved the surface and root zone correlations against in situ measurements from 14 SMAP core validation sites (CVS) by 0.12 and 0.16, respectively, over the model-only skill and reduced the surface and root zone ubRMSE by 0.005 m3 m-3 and 0.001 m3 m-3, respectively. The assimilation reduced the average absolute surface bias against the CVS measurements by 0.009 m3 m-3, but increased the root zone bias by 0.014 m3 m-3. Assimilating the NN retrievals after a localized bias correction yielded slightly lower surface correlation and ubRMSE improvements, but generally the skill differences were small. The assimilation of the physically-based SMAP Level-2 passive soil moisture retrievals using a global bias correction yielded similar skill improvements, as did the direct assimilation of locally bias-corrected SMAP brightness temperatures within the SMAP Level-4 soil moisture algorithm. The results show that global bias correction methods may be able to extract more independent information from SMAP observations compared to local bias correction methods, but without accurate quality control and observation error characterization they are also more vulnerable to adverse effects from retrieval errors related to uncertainties in the retrieval inputs and algorithm. Furthermore, the results show that using global bias correction approaches without a simultaneous re-calibration of the land model processes can lead to a skill degradation in other land surface variables.

Real-time microwave imaging of differential temperature for thermal therapy monitoring.

A microwave imaging system for real-time 3-D imaging of differential temperature has been developed for the monitoring and feedback of thermal therapy systems. Design parameters are constrained by features of a prototype-focused microwave thermal therapy system for the breast, operating at 915 MHz. Real-time imaging is accomplished with a precomputed linear inverse scattering solution combined with continuous vector network analyzer (VNA) measurements of a 36-antenna, HFSS-modeled, cylindrical cavity. Volumetric images of differential change of dielectric constant due to temperature are formed with a refresh rate as fast as 1 frame/s and 1 (°)C resolution. Procedures for data segmentation and postprocessed S-parameter error-correction are developed. Antenna pair VNA calibration is accelerated by using the cavity as the unknown thru standard. The device is tested on water targets and a simple breast phantom. Differentially heated targets are successfully imaged in cluttered environments. The rate of change of scattering contrast magnitude correlates 1:1 with target temperature.

Self-characterization of commercial ultrasound probes in transmission acoustic inverse scattering: transducer model and volume integral formulation.

A self-contained source characterization method for commercial ultrasound probes in transmission acoustic inverse scattering is derived and experimentally tested. The method is based on modified scattered field volume integral equations that are linked to the source-scattering transducer model. The source-scattering parameters are estimated via pair-wise transducer measurements and the nonlinear inversion of an acoustic propagation model that is derived. This combination creates a formal link between the transducer characterization and the inverse scattering algorithm. The method is tested with two commercial ultrasound probes in a transmission geometry including provisions for estimating the probe locations and aligning a robotic rotator. The transducer characterization results show that the nonlinear inversion fit the measured data well. The transducer calibration and inverse scattering algorithm are tested on simple targets. Initial images show that the recovered contrasts are physically consistent with expected values.

A preclinical system prototype for focused microwave thermal therapy of the breast.

A preclinical prototype of a transcutaneous thermal therapy system has been developed for the targeted treatment of breast cancer cells using focused microwaves as an adjuvant to radiation, chemotherapy, and high-intensity-focused ultrasound. The prototype system employs a 2-D array of tapered microstrip patch antennas operating at 915 MHz to focus continuous-wave microwave energy transcutaneously into the pendent breast suspended in a coupling medium. Prior imaging studies are used to ascertain the material properties of the breast tissue, and these data are incorporated into a multiphysics model. Time-reversal techniques are employed to find a solution (relative amplitudes and phase) for focusing at a given location. Modeling tests of this time-reversal focusing method have been performed, which demonstrate good targeting accuracy within heterogeneous breast tissue. Experimental results using the laboratory prototype to perform focused heating in tissue-mimicking gelatin phantoms have demonstrated 1.5-cm-diameter focal spot sizes and differential heating at the desired focus sufficient to achieve an antitumor effect confined to the target region.

Microwave breast imaging system prototype with integrated numerical characterization.

The increasing number of experimental microwave breast imaging systems and the need to properly model them have motivated our development of an integrated numerical characterization technique. We use Ansoft HFSS and a formalism we developed previously to numerically characterize an S-parameter- based breast imaging system and link it to an inverse scattering algorithm. We show successful reconstructions of simple test objects using synthetic and experimental data. We demonstrate the sensitivity of image reconstructions to knowledge of the background dielectric properties and show the limits of the current model.

Large domain, low-contrast acoustic inverse scattering for ultrasound breast imaging.

We present a full-wave acoustic inverse scattering algorithm designed specifically for ultrasonic breast imaging. At ultrasonic frequencies, the image domain is roughly tens to hundreds of min cubed, where min is the smallest wavelength in the transmit signal spectrum. The expected range of contrasts for the breast imaging problem for density, compressibility, and compressive loss is ± 20% of the background. Because of the low contrast, Born iterations provide the basic structure of the inverse scattering algorithm. However, we use a multi-objective covariance-based least squares cost function in place of the basic least squares cost function to estimate the contrast functions. This cost function provides physically meaningful regularization based on a priori knowledge of the contrasts. Also, due to the size of the imaging domain and because the objects to be imaged are low contrast and inhomogeneous, we use the Neumann series solution as the forward solver. The largest domain imaged in simulation was 50 min x 50 min in 2D.