Earlier snowmelt may lead to late season declines in plant productivity and carbon sequestration in Arctic tundra ecosystems.

Arctic warming is affecting snow cover and soil hydrology, with consequences for carbon sequestration in tundra ecosystems. The scarcity of observations in the Arctic has limited our understanding of the impact of covarying environmental drivers on the carbon balance of tundra ecosystems. In this study, we address some of these uncertainties through a novel record of 119 site-years of summer data from eddy covariance towers representing dominant tundra vegetation types located on continuous permafrost in the Arctic. Here we found that earlier snowmelt was associated with more tundra net CO2 sequestration and higher gross primary productivity (GPP) only in June and July, but with lower net carbon sequestration and lower GPP in August. Although higher evapotranspiration (ET) can result in soil drying with the progression of the summer, we did not find significantly lower soil moisture with earlier snowmelt, nor evidence that water stress affected GPP in the late growing season. Our results suggest that the expected increased CO2 sequestration arising from Arctic warming and the associated increase in growing season length may not materialize if tundra ecosystems are not able to continue sequestering CO2 later in the season.

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. Current regulations are not adequate to eliminate impacts from these activities. We address issues related to the high-density of 3D trails compared to 2D methods, with larger crews, more camps, and more vehicles. We focus on consequences to the hilly landscapes, including microtopography, snow, vegetation, hydrology, active layers, and permafrost. Based on studies of 2D-seismic trails created in 1984-1985 in the same area by similar types of vehicles, under similar regulations, approximately 122 km2 would likely sustain direct medium- to high-level disturbance from the proposed exploration, with possibly expanded impacts through permafrost degradation and hydrological connectivity. Strong winds are common, and snow cover necessary to minimize impacts from vehicles is windblown and inadequate to protect much of the area. Studies of 2D-seismic impacts have shown that moist vegetation types, which dominate the area, sustain longer-lasting damage than wet or dry types, and that the heavy vehicles used for mobile camps caused the most damage. The permafrost is ice rich, which combined with the hilly topography, makes it especially susceptible to thermokarst and erosion triggered by winter vehicle traffic. The effects of climate warming will exacerbate the impacts of winter travel due to warmer permafrost and a shift of precipitation from snow to rain. The cumulative impacts of 3D-seismic traffic in tundra areas need to be better assessed, together with the effects of climate change and the industrial development that would likely follow. Current data needs include studies of the impacts of 3D-seismic exploration, better climate records for the Arctic National Wildlife Refuge, especially for wind and snow; and high-resolution maps of topography, ground ice, hydrology, and vegetation.

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. We corroborated quantitative error statistics along with detailed visual inspections to gauge the IWP detection accuracies. We found promising model performances (detection accuracies: 89% to 96% and classification accuracies: 94% to 97%) for all candidate image scenes with varying tundra types. The mapping workflow discerned the IWPs by exhibiting low absolute mean relative error (AMRE) values (0.17-0.23). Results further suggest the importance of increasing the variability of training samples when practicing transfer-learning strategy to map IWPs across heterogeneous tundra cover types. Overall, our findings demonstrate the robust performances of IWPs mapping workflow in multiple tundra landscapes.

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. The central objective of this exploratory study is to understand to what degree MS band statistics govern DLCNN model predictions. We scaffold our analysis on a case study that uses Arctic tundra permafrost landform features called ice-wedge polygons (IWPs) as candidate geo objects. We choose Mask RCNN as the DLCNN architecture to detect IWPs from eight-band Worldview-02 VHSR satellite imagery. A systematic experiment was designed to understand the impact on choosing the optimal three-band combination in model prediction. We tasked five cohorts of three-band combinations coupled with statistical measures to gauge the spectral variability of input MS bands. The candidate scenes produced high model detection accuracies for the F1 score, ranging between 0.89 to 0.95, for two different band combinations (coastal blue, blue, green (1,2,3) and green, yellow, red (3,4,5)). The mapping workflow discerned the IWPs by exhibiting low random and systematic error in the order of 0.17-0.19 and 0.20-0.21, respectively, for band combinations (1,2,3). Results suggest that the prediction accuracy of the Mask-RCNN model is significantly influenced by the input MS bands. Overall, our findings accentuate the importance of considering the image statistics of input MS bands and careful selection of optimal bands for DLCNN predictions when DLCNN architectures are restricted to three spectral channels.

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 × 106 m3) in Barry Arm, a fjord in Prince William Sound, Alaska. The slope moved rapidly between 2010 and 2017, yielding a horizontal displacement of 120 m, which is highly correlated with the rapid retreat and thinning of Barry Glacier. Should the entire unstable slope collapse at once, preliminary tsunami modeling suggests a maximum runup of 300 m near the landslide, which may have devastating impacts on local communities. Our findings highlight the need for interdisciplinary studies of recently deglaciated fjords to refine our understanding of the impact of climate change on landslides and tsunamis.

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. Once snowfall and rainfall are bias adjusted, long-term annual precipitation estimates more than double (from 123 to 274 mm), highlighting the risk of studies using conventional or unadjusted precipitation that dramatically under-represent water balance components. Applying conventional precipitation information to the water balance analysis produced consistent storage deficits (79 to 152 mm) that were all larger than the largest actual deficit (75 mm), which was observed in the unusually low rainfall summer of 2007. Year-to-year variability in adjusted rainfall (±33 mm) was larger than evapotranspiration (±13 mm). Measured interannual variability in partitioning of snow into runoff (29% in 2008 to 68% in 2009) in years with similar end-of-winter snow accumulation (180 and 164 mm, respectively) highlights the importance of the previous summer's rainfall (25 and 60 mm, respectively) on spring runoff production. Incorrect representation of precipitation can therefore have major implications for Arctic water budget descriptions that in turn can alter estimates of carbon and energy fluxes.

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

Landscape attributes that vary with microtopography, such as active layer thickness (ALT), 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 ALT 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 ALT in a 5 km2 area of ice-wedge polygon terrain in Barrow, Alaska. We use a simple regression-based, machine learning data-fusion algorithm that uses topographic and spectral metrics derived from multisensor data (LiDAR and WorldView-2) to estimate ALT (2 m spatial resolution) across the study area. Comparison of the ALT estimates with ground-based measurements, indicates the accuracy (r2 = 0.76, RMSE ±4.4 cm) of the approach. While it is generally accepted that broad climatic variability associated with increasing air temperature will govern the regional averages of ALT, consistent with prior studies, our findings using high-resolution LiDAR and WorldView-2 data, show that smaller-scale variability in ALT is controlled by local eco-hydro-geomorphic factors. This work demonstrates a path forward for mapping ALT at high spatial resolution and across sufficiently large regions for improved understanding and predictions of coupled dynamics among permafrost, hydrology, and land-surface processes from readily available remote sensing data.