Understanding the Effects of Optimal Combination of Spectral Bands on Deep Learning Model Predictions: A Case Study Based on Permafrost Tundra Landform Mapping Using High Resolution Multispectral Satellite Imagery.

Deep learning (DL) convolutional neural networks (CNNs) have been rapidly adapted in very high spatial resolution (VHSR) satellite image analysis. DLCNN-based computer visions (CV) applications primarily aim for everyday object detection from standard red, green, blue (RGB) imagery, while earth science remote sensing applications focus on geo object detection and classification from multispectral (MS) imagery. MS imagery includes RGB and narrow spectral channels from near- and/or middle-infrared regions of reflectance spectra.

Use of Very High Spatial Resolution Commercial Satellite Imagery and Deep Learning to Automatically Map Ice-Wedge Polygons across Tundra Vegetation Types.

We developed a high-throughput mapping workflow, which centers on deep learning (DL) convolutional neural network (CNN) algorithms on high-performance distributed computing resources, to automatically characterize ice-wedge polygons (IWPs) from sub-meter resolution commercial satellite imagery. We applied a region-based CNN object instance segmentation algorithm, namely the Mask R-CNN, to automatically detect and classify IWPs in North Slope of Alaska. The central goal of our study was to systematically expound the DLCNN model interoperability across varying tundra types (sedge, tussock sedge, and non-tussock sedge) and image scene complexities to refine the understanding of opportunities and challenges for regional-scale mapping applications.

Detection and Assessment of a Large and Potentially Tsunamigenic Periglacial Landslide in Barry Arm, Alaska.

The retreat of glaciers in response to global warming has the potential to trigger landslides in glaciated regions around the globe. Landslides that enter fjords or lakes can cause tsunamis, which endanger people and infrastructure far from the landslide itself. Here we document the ongoing movement of an unstable slope (total volume of 455 × 10 m) in Barry Arm, a fjord in Prince William Sound, Alaska.

Landscape impacts of 3D-seismic surveys in the Arctic National Wildlife Refuge, Alaska.

Although three-dimensional (3D) seismic surveys have improved the success rate of exploratory drilling for oil and gas, the impacts have received little scientific scrutiny, despite affecting more area than any other oil and gas activity. To aid policy-makers and scientists, we reviewed studies of the landscape impacts of 3D-seismic surveys in the Arctic. We analyzed a proposed 3D-seismic program in northeast Alaska, in the northern Arctic National Wildlife Refuge, which includes a grid 63,000 km of seismic trails and additional camp-move trails.

Tundra water budget and implications of precipitation underestimation.

Difficulties in obtaining accurate precipitation measurements have limited meaningful hydrologic assessment for over a century due to performance challenges of conventional snowfall and rainfall gauges in windy environments. Here, we compare snowfall observations and bias adjusted snowfall to end-of-winter snow accumulation measurements on the ground for 16 years (1999-2014) and assess the implication of precipitation underestimation on the water balance for a low-gradient tundra wetland near Utqiagvik (formerly Barrow), Alaska (2007-2009). In agreement with other studies, and not accounting for sublimation, conventional snowfall gauges captured 23-56% of end-of-winter snow accumulation.

Extrapolating active layer thickness measurements across Arctic polygonal terrain using LiDAR and data sets.

Landscape attributes that vary with microtopography, such as active layer thickness (), are labor intensive and difficult to document effectively through in situ methods at kilometer spatial extents, thus rendering remotely sensed methods desirable. Spatially explicit estimates of can provide critically needed data for parameterization, initialization, and evaluation of Arctic terrestrial models. In this work, we demonstrate a new approach using high-resolution remotely sensed data for estimating centimeter-scale in a 5 km area of ice-wedge polygon terrain in Barrow, Alaska.