Sparse and unevenly distributed soil samples across the northern high-latitude region greatly limit the accuracy of soil organic carbon (SOC) mapping. Therefore, substantial discrepancies exist in SOC estimation in this region, which makes it challenging to characterize the SOC spatial variability and its potential responses to climate change and permafrost degradation. To address these challenges, we enhanced a machine learning model for SOC mapping by developing a data resampling approach that accounts for soil samples spatial heterogeneity, using Alaska as a case study. Specifically, in-situ SOC data were resampled with weights proportional to the variance within a 15-km radius, and then fitted using a random forest (RF) regression model. Multiple features, including temporal composites of Sentinel-1 C-band radar backscatter, vegetation indices from Sentinel-2, climate indices including thawing and freezing indices from moderate resolution imaging spectroradiometer (MODIS), and ancillary topography data, were selected as inputs for the RF model after recursive feature elimination to generate top-layer (0-30 cm) SOC content maps in Alaska at a 250-m resolution. The enhanced RF model with data resampling showed improved accuracy compared to the original RF model, with the coefficient of determination (R2) increased from 0.36 to 0.56 and the root mean square error (RMSE) decreased from 16% to 11% for the surface (0-10 cm) SOC content, and slightly improved accuracy for the deeper (10-30 cm) SOC content. Additionally, the enhanced RF model also better captured local-scale variability of SOC than the original RF model and SoilGrids 2.0 dataset, with high-resolution remote sensing indices playing a major role. The improved SOC content estimates were then used to estimate soil bulk density and calculate total SOC stock for Alaska. Our results suggest that Alaskan topsoil (0-30 cm) stores approximately 25.21±17.18 Pg C, with the largest SOC reserves found in shrublands. These findings highlight the importance of accounting for spatial heterogeneity in in-situ samples and leveraging high-resolution remote sensing data for regional soil mapping.

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.

Hoh Xil is an uninhabited extremity secluded on the Tibetan Plateau hinterland. A complete mapping of ground motion variation in Hoh Xil is essential for in-depth understanding of terrain’s responses to climate change on the Tibetan Plateau. However, the inaccessibility and extremely harsh environment impeded extensive field investigations on landform alteration and its formative process. Such difficulty can be resolved by Interferometric Synthetic Aperture Radar (InSAR), which enables a broad detection of subtle permafrost motions at millimeter precision. This study, for the first time, accomplished a Multi-temporal InSAR (MT-InSAR) deformation mapping from 2015 to 2020 in Hoh Xil, with a wide coverage of about 200,000 km2. 1,592 Sentinel-1 images were processed based on the small baseline subset (SBAS) technique. The results show that Hoh Xil was experiencing dynamic permafrost disturbances. Thawing permafrost with both the linear subsidence rate higher than 2 mm/yr and the periodic amplitude over 2 mm was primarily detected in areas of flat or gentle slopes. The InSAR cumulative deformation is highly correlated with permafrost thawing depth. Significant lag times were identified between seasonal oscillation of InSAR deformation and land surface temperature (LST). Thermokarst landforms of retrogressive thaw slumps and thermokarst lakes broadly formed and dynamically evolved as a consequence of permafrost degradation. Particularly, widespread thawing permafrost characterized by the spatial clustering of thermokarst lakes appeared to occur in areas adjacent to large lakes. The discovered dynamic permafrost disturbances in Hoh Xil manifested even the secluded Tibetan Plateau hinterland was facing the threat of climate change.

Knowledge of the spatial and temporal distribution of active layer thickness (ALT) throughout northern Alaska would help to understand the effects of climate change in the region, as well as to quantify how much the permafrost degradation manifestly in progress there is contributing to the accumulation of greenhouse gases in the atmosphere. For this reason, we are developing extensive high-resolution maps of ALT in northern Alaska. We use machine learning along with an extensive set of spatial data layers to upscale ALT from thousands of training pixels taken from high resolution swaths of estimated ALT derived from airborne polarimetric P-band synthetic aperture radar (SAR). The resulting maps of up-scaled ALT have been compared to thousands of validation samples set aside from the PolSAR-derived swaths and to in situ ALT measurements. The maps have achieved root-mean-square errors (RMSEs) of 5–7 cm relative to validation samples, and RMSEs of approximately 10–12 cm relative to in situ ALT measurements.

Accurate vegetation information is essential for analyzing above-ground biomass and understanding subsurface characteristics, such as root biomasss, soilorganicmatter and soil moisture profiles. This paper investigates novel mappings of forest species and canopy height in interior Alaska. We employ Random Forests to train a regression model for canopy height mapping and a classification model for forest species mapping utilizing L-band and P-band Uninhabited Aerial Vehicle Synthetic Aperture Radar(UAVSAR). For canopy height, canopy height model (CHM) data derived from Goddard's LiDAR, Hyperspectral, and Thermal Imager (G-LiHT) are treated as ground truth. For forest species prediction, Tanana Valley State Forest (TVSF) Timber Inventory and Forest Inventory and Analysis (FIA) data are used as reference. The experimental results show the proposed method yields a root-mean-square error of 1.90 m for forest height estimation and overall accuracy of 79.54% for forest species classification. They also demonstrate the feasibility of obtaining precise vegetation information by data-driven methods, which can be further used to enhance forest radar scattering forward models.

We used full-polarimetric L-band and P-band synthetic aperture radar (SAR) data collected from the recent NASA Arctic Boreal Vulnerability Experiment (ABoVE) airborne campaign and Sentinel-1 C-band dual-polarization data to understand the sensitivity of radar backscatter intensity and phase to fire-induced changes in the surface and subsurface soil processes in Arctic tundra underlain by permafrost. The 2007 Anaktuvuk River fire on the Alaska North Slope was used as a case study. At ~10-year postfire, we observed a strong increase (>~3–4 dB) in the low-frequency radar backscatter in severely burned areas during the thaw season, in contrast to limited (1 dB) in burned areas than the adjacent unburned areas. Polarimetric decomposition analysis indicated a general trend toward more random surface scattering, and strong increases in double-bounce scattering and volume scattering power at both P- and L-band in the burned areas. The ice-rich yedoma region shows the largest backscatter increases in burned areas and the highest correlation with burn severity and microtopography changes. The above backscatter changes are attributed to increasing surface roughness and microtopography due to ice-wedge degradation and thermokarst development and increasing subsurface scattering due to an overall drier and deeper active layer in burned areas. Among all frequencies, P-band shows consistently larger contrast in backscatter power and phase between burned and unburned areas, which makes it potentially more useful to study fire–permafrost interactions in the Arctic over decadal time scales.

Detailed information on the spatial and temporal distribution of active layer thickness (ALT) throughout the North Slope of Alaska, were it available, could offer valuable insights into the effects of climate change throughout the region and facilitate the estimation of greenhouse gas emissions resulting from permafrost degradation. We are, therefore, developing extensive high-resolution maps of ALT on the North Slope of Alaska. To do this, we use a machine learning algorithm to extrapolate ALT from high resolution strips of estimated ALT derived from airborne P-band synthetic aperture radar (SAR) acquired over two sets of flights in each of three different years. Our results indicate upscaling root-mean-square error (RMSE) of about 4 cm relative to thousands of randomly-selected SAR-derived ALT validation samples, and RMSE of approximately 10 cm relative to a small number of in-situ ALT measurements.

We used the full polarimetric L-band and P-band SAR data collected from recent NASA Arctic Boreal Vulnerability Experiment (ABoVE) airborne campaign to understand the sensitivity of longwave radar backscatter intensity and phase to the post-fire recovery process of Arctic tundra. The 2007 Anaktuvuk River fire was used as a case study. At 10-years post-fire, we observed a strong increase (>∼4 dB) in both the P- and L-band radar backscatter in the severely burned areas, in contrast to limited backscatter differences (VV, VH) between burned and unburned areas at C-band. The polarimetric target decomposition analysis indicated a general trend towards more random surface scattering, and strong increases of the double-bounce and volumetric scattering power at both P- and L-band in the burned areas. Large differences were also observed in the Pauli phase angle and the dominant-scattering-type Touzi phase angle between burned and adjacent unburned areas. The above changes are likely caused by increasing surface roughness and microtopography due to thermokarst development and ice degradation, and increasing subsurface scattering due to an overall drier and deeper active layer in the burned areas.

Currently, the community lacks capabilities to assess and monitor landscape scale permafrost active layer dynamics over large extents. To address this need, we developed a concept of a remote sensing based Soil Inversion Model for regional Permafrost (SIM-P) monitoring. The current SIM-P framework includes a satellite-based soil process model and a soil dielectric model. We are also working on incorporating a radar scattering model for Arctic tundra into the SIM-P framework. A unified soil parameterization scheme was developed to harmonize key soil thermal, hydraulic and dielectric parameters in the soil process and radar models that can be used in the joint soil-radar inversion framework. The soil parameter retrievals of the SIM-P framework include soil organic content (SOC) and active layer thickness (ALT). Initial tests of SIM-P using in-situ soil permittivity observations showed reasonable accuracy in predicting site-level SOC and soil temperature profiles at an Alaska tundra site and ALT in Arctic Alaska. SIM-P will be further tested using airborne P- and L-band radar data collected during NASA's Arctic Boreal Vulnerability Experiment (ABoVE) to evaluate the sensitivity of longwave radar to active layer properties.

Significant changes have occurred in permafrost and seasonally frozen soils in the Tibetan Plateau (TP) during the last few decades, with potential influence on regional climate, hydrological and ecosystem processes. Land surface temperature (LST) is closely associated with surface energy balance, thus is a critical parameter affecting the frozen soil thermodynamics. Satellite remote sensing provides an efficient way of mapping LST and permafrost changes at a high temporal fidelity and spatial resolution, especially in areas with extremely sparse in-situ observations such as the Tibetan Plateau. Therefore, our main objective is to develop a robust and operational algorithm to retrieval LST under all weather conditions. Multi-frequency AMSR brightness temperatures were incorporated into radiative transfer equations assuming a linear correlation between vertical and horizontal polarization surface emissivities to derive LST and this algorithm was tested using in-situ surface (0cm) temperature measurements at Naqu site in the Central TP. The R and RMSE values for LST retrievals during the growing season (June-August) are 0.869 and 3.64 K for 37GHz and 0.911 and 3.85 K for 89GHz respectively. The LST retrievals using 89GHz show much reduced errors during the transitional season (April-May) than 37GHz. The 89GHz could be potentially very useful for LST retrieval in the TP due to relatively low atmospheric water vapor in this area and high spatial resolution of this channel (~6km).