Search

Significant progress in permafrost carbon science made over the past decades include the identification of vast permafrost carbon stocks, the development of new pan‐Arctic permafrost maps, an increase in terrestrial measurement sites for CO2 and methane fluxes, and important factors affecting carbon cycling, including vegetation changes, periods of soil freezing and thawing, wildfire, and other disturbance events. Process‐based modeling studies now include key elements of permafrost carbon cycling and advances in statistical modeling and inverse modeling enhance understanding of permafrost region C budgets. By combining existing data syntheses and model outputs, the permafrost region is likely a wetland methane source and small terrestrial ecosystem CO2 sink with lower net CO2 uptake toward higher latitudes, excluding wildfire emissions. For 2002–2014, the strongest CO2 sink was located in western Canada (median: −52 g C m−2 y−1) and smallest sinks in Alaska, Canadian tundra, and Siberian tundra (medians: −5 to −9 g C m−2 y−1). Eurasian regions had the largest median wetland methane fluxes (16–18 g CH4 m−2 y−1). Quantifying the regional scale carbon balance remains challenging because of high spatial and temporal variability and relatively low density of observations. More accurate permafrost region carbon fluxes require: (a) the development of better maps characterizing wetlands and dynamics of vegetation and disturbances, including abrupt permafrost thaw; (b) the establishment of new year‐round CO2 and methane flux sites in underrepresented areas; and (c) improved models that better represent important permafrost carbon cycle dynamics, including non‐growing season emissions and disturbance effects. Plain Language Summary: Climate change and the consequent thawing of permafrost threatens to transform the permafrost region from a carbon sink into a carbon source, posing a challenge to global climate goals. Numerous studies over the past decades have identified important factors affecting carbon cycling, including vegetation changes, periods of soil freezing and thawing, wildfire, and other disturbance events. Overall, studies show high wetland methane emissions and a small net carbon dioxide sink strength over the terrestrial permafrost region but results differ among modeling and upscaling approaches. Continued and coordinated efforts among field, modeling, and remote sensing communities are needed to integrate new knowledge from observations to modeling and predictions and finally to policy. Key Points: Rapid warming of northern permafrost region threatens ecosystems, soil carbon stocks, and global climate targetsLong‐term observations show importance of disturbance and cold season periods but are unable to detect spatiotemporal trends in C fluxCombined modeling and syntheses show the permafrost region is a small terrestrial CO2 sink with large spatial variability and net CH4 source [ABSTRACT FROM AUTHOR]

Aims: An Arctic Vegetation Classification (AVC) is needed to address issues related to rapid Arctic-wide changes to climate, land-use, and biodiversity. Location: The 7.1 million km2 Arctic tundra biome. Approach and conclusions: The purpose, scope and conceptual framework for an Arctic Vegetation Archive (AVA) and Classification (AVC) were developed during numerous workshops starting in 1992. The AVA and AVC are modeled after the European vegetation archive (EVA) and classification (EVC). The AVA will use Turboveg for data management. The AVC will use a Braun-Blanquet (Br.-Bl.) classification approach. There are approximately 31,000 Arctic plots that could be included in the AVA. An Alaska AVA (AVA-AK, 24 datasets, 3026 plots) is a prototype for archives in other parts of the Arctic. The plan is to eventually merge data from other regions of the Arctic into a single Turboveg v3 database. We present the pros and cons of using the Br.-Bl. classification approach compared to the EcoVeg (US) and Biogeoclimatic Ecological Classification (Canada) approaches. The main advantages are that the Br.-Bl. approach already has been widely used in all regions of the Arctic, and many described, well-accepted vegetation classes have a pan-Arctic distribution. A crosswalk comparison of Dryas octopetala communities described according to the EcoVeg and the Braun-Blanquet approaches indicates that the non-parallel hierarchies of the two approaches make crosswalks difficult above the plant- community level. A preliminary Arctic prodromus contains a list of typical Arctic habitat types with associated described syntaxa from Europe, Greenland, western North America, and Alaska. Numerical clustering methods are used to provide an overview of the variability of habitat types across the range of datasets and to determine their relationship to previously described Braun-Blanquet syntaxa. We emphasize the need for continued maintenance of the Pan-Arctic Species List, and additional plot data to fully sample the variability across bio- climatic subzones, phytogeographic regions, and habitats in the Arctic. This will require standardized methods of plot-data collection, inclusion of physiogonomic information in the numeric analysis approaches to create formal definitions for vegetation units, and new methods of data sharing between the AVA and national vegetation-plot databases. [ABSTRACT FROM AUTHOR]

The mechanisms driving trends and variability of the normalized difference vegetation index (NDVI) for tundra in Alaska along the Beaufort, east Chukchi, and east Bering Seas for 1982-2013 are evaluated in the context of remote sensing, reanalysis, and meteorological station data as well as regional modeling. Over the entire season the tundra vegetation continues to green; however, biweekly NDVI has declined during the early part of the growing season in all of the Alaskan tundra domains. These springtime declines coincide with increased snow depth in spring documented in northern Alaska. The tundra region generally has warmed over the summer but intraseasonal analysis shows a decline in midsummer land surface temperatures. The midsummer cooling is consistent with recent large-scale circulation changes characterized by lower sea level pressures, which favor increased cloud cover. In northern Alaska, the sea-breeze circulation is strengthened with an increase in atmospheric moisture/cloudiness inland when the land surface is warmed in a regional model, suggesting the potential for increased vegetation to feedback onto the atmospheric circulation that could reduce midsummer temperatures. This study shows that both large- and local-scale climate drivers likely play a role in the observed seasonality of NDVI trends. [ABSTRACT FROM AUTHOR]

Nitrogen availability is considered to be a key limiting factor for plant growth in arctic tundra. Freeze-thaw cycles, which can produce patterned-ground features, may also limit the establishment and growth of arctic plants. In this experiment, we fertilized nonsorted circles, a type of patterned-ground feature, and the surrounding more stable vegetation in moist nonacidic tundra of northern Alaska to determine if nitrogen availability limited plant communities on these disturbance-dominated features. Similar to other studies in moist nonacidic tundra, the fertilized stable vegetation showed an increase in graminoid biomass and a decrease in moss biomass, but no total biomass response. The plant communities on nonsorted circles were less responsive to the addition of nitrogen as compared to the well-vegetated rims. The nonsorted circle vegetation may be limited by additional factors, such as frost heave disturbance, availability of buried seeds, and/or other nutrients. This difference in fertilization response shows that the presence of these features creates spatial heterogeneity in tundra plant community dynamics that should be taken into account when studying tundra responses to environmental change. [ABSTRACT FROM AUTHOR]

The spatial heterogeneity of recent decadal dynamics in vegetation greenness and biomass in response to changes in summer warmth index (SWI) was investigated along spatial gradients on the Arctic Slope of Alaska. Image spatial analysis was used to examine the spatial pattern of greenness dynamics from 1991 to 2000 as indicated by variations of the maximum normalized difference vegetation index (Peak NDVI) and time-integrated NDVI (TI-NDVI) along latitudinal gradients. Spatial gradients for both the means and temporal variances of the NDVI indices for 0.1° latitude intervals crossing three bioclimate subzones were analyzed along two north–south Arctic transects. NDVI indices were generally highly variable over the decade, with great heterogeneity across the transects. The greatest variance in TI-NDVI was found in low shrub vegetation to the south (68.7–68.8°N) and corresponded to high fractional cover of shrub tundra and moist acidic tundra (MAT), while the greatest variance in Peak-NDVI predominately occurred in areas dominated by wet tundra (WT) and moist nonacidic tundra (MNT). Relatively high NDVI temporal variances were also related to specific transitional areas between dominant vegetation types. The regional temporal variances of NDVI from 1991 to 2000 were largely driven by meso-scale climate dynamics. The spatial heterogeneity of the NDVI variance was mostly explained by the fractional land cover composition, different responses of each vegetation type to climate change, and patterned ground features. Aboveground plant biomass exhibited similar spatial heterogeneity as TI-NDVI; however, spatial patterns are slightly different from NDVI because of their nonlinear relationship. [ABSTRACT FROM AUTHOR]

To understand drivers of vegetation type distribution and sensitivity to climate change.Interior Alaska.A logistic regression model was developed that predicts the potential equilibrium distribution of four major vegetation types: tundra, deciduous forest, black spruce forest and white spruce forest based on elevation, aspect, slope, drainage type, fire interval, average growing season temperature and total growing season precipitation. The model was run in three consecutive steps. The hierarchical logistic regression model was used to evaluate how scenarios of changes in temperature, precipitation and fire interval may influence the distribution of the four major vegetation types found in this region.At the first step, tundra was distinguished from forest, which was mostly driven by elevation, precipitation and south to north aspect. At the second step, forest was separated into deciduous and spruce forest, a distinction that was primarily driven by fire interval and elevation. At the third step, the identification of black vs. white spruce was driven mainly by fire interval and elevation. The model was verified for Interior Alaska, the region used to develop the model, where it predicted vegetation distribution among the steps with an accuracy of 60–83%. When the model was independently validated for north-west Canada, it predicted vegetation distribution among the steps with an accuracy of 53–85%. Black spruce remains the dominant vegetation type under all scenarios, potentially expanding most under warming coupled with increasing fire interval. White spruce is clearly limited by moisture once average growing season temperatures exceeded a critical limit (+2 °C). Deciduous forests expand their range the most when any two of the following scenarios are combined: decreasing fire interval, warming and increasing precipitation. Tundra can be replaced by forest under warming but expands under precipitation increase.The model analyses agree with current knowledge of the responses of vegetation types to climate change and provide further insight into drivers of vegetation change. [ABSTRACT FROM AUTHOR]

Intraseasonal patterns of normalized difference vegetation index (NDVI), leaf area index (LAI), and phytomass were compared for four tundra vegetation types at Ivotuk, Alaska, during summer 1999. The vegetation types included moist acidic tundra (MAT), moist nonacidic tundra (MNT), mossy tussock tundra, and shrub tundra. The seasonal curves of NDVI were similar among the vegetation types but with varying magnitudes of the peak values. Peak NDVI in the shrub tundra (0.83) was significantly greater than in MAT (0.76), which was significantly greater than in MNT (0.71) and mossy tussock tundra (0.70). LAI and phytomass exhibited high temporal variability with distinct seasonality only in shrub tundra. Seasonal LAI and NDVI patterns were therefore correlated only in shrub tundra, which was attributed to the high quantity of deciduous shrub foliage present in this community and absent in the other vegetation types. Shrub tundra peak live above-ground phytomass (1256 ± 123 g m-2) was significantly greater than peak live above-ground phytomass for MAT, MNT, and mossy tussock tundra (722 ± 71, 773 ± 53, 703 ± 39 g m-2 respectively, P < 0.05). Relative abundances of deciduous shrubs, mosses, and graminoids were revealed as key' components controlling differences in NDVI, LAI, and phytomass among tundra vegetation types. [ABSTRACT FROM AUTHOR]

Abstract. Two-weekly AVHRR images were used to examine spatial patterns of the normalized difference vegetation index (NDVI) and their relationships with environmental variables for moist acidic tundra (MAT) and moist non-acidic tundra (MNT) along two latitudinal transects in northern Alaska. The NDVI database was derived from a 5-yr time series (1995-1999) of two-weekly AVHRR composites for Alaska. A digital climate map, digital elevation map and vegetation map were processed and overlain with the NDVI grid. Homogeneous vegetation patches for both MAT and MNT were defined as sample sites using infrared aerial photos, MSS images and the vegetation map along the transects. Linear and non-linear regression modeling were performed between NDVI indices and environmental variables, total summer warmth (TSW) and elevation. It was demonstrated that along both western and eastern transects, there were obvious latitudinal trends of peak NDVI (AP-NDVI), average growing season NDVI (GS-NDVI), and early June NDVI (EJ-NDVI). In most cases, MNT had lower NDVI values than MAT throughout the year. There were significant (p<0.01) relations between NDVI (AP-NDVI, GS-NDVI and EJ-NDVI) and total summer warmth (TSW) and elevation in the region. EJ-NDVI showed the strongest correlation with TSW or elevation, making it the most sensitive NDVI indicator along environmental gradients in northern Alaska. NDVI was likely controlled by TSW and elevation, with the former being dominant. [ABSTRACT FROM AUTHOR]

State-of-the-art deep learning technology has been successfully applied to relatively small selected areas of very high spatial resolution (0.15 and 0.25 m) optical aerial imagery acquired by a fixed-wing aircraft to automatically characterize ice-wedge polygons (IWPs) in the Arctic tundra. However, any mapping of IWPs at regional to continental scales requires images acquired on different sensor platforms (particularly satellite) and a refined understanding of the performance stability of the method across sensor platforms through reliable evaluation assessments. In this study, we examined the transferability of a deep learning Mask Region-Based Convolutional Neural Network (R-CNN) model for mapping IWPs in satellite remote sensing imagery (~0.5 m) covering 272 km2 and unmanned aerial vehicle (UAV) (0.02 m) imagery covering 0.32 km2. Multi-spectral images were obtained from the WorldView-2 satellite sensor and pan-sharpened to ~0.5 m, and a 20 mp CMOS sensor camera onboard a UAV, respectively. The training dataset included 25,489 and 6022 manually delineated IWPs from satellite and fixed-wing aircraft aerial imagery near the Arctic Coastal Plain, northern Alaska. Quantitative assessments showed that individual IWPs were correctly detected at up to 72% and 70%, and delineated at up to 73% and 68% F1 score accuracy levels for satellite and UAV images, respectively. Expert-based qualitative assessments showed that IWPs were correctly detected at good (40–60%) and excellent (80–100%) accuracy levels for satellite and UAV images, respectively, and delineated at excellent (80–100%) level for both images. We found that (1) regardless of spatial resolution and spectral bands, the deep learning Mask R-CNN model effectively mapped IWPs in both remote sensing satellite and UAV images; (2) the model achieved a better accuracy in detection with finer image resolution, such as UAV imagery, yet a better accuracy in delineation with coarser image resolution, such as satellite imagery; (3) increasing the number of training data with different resolutions between the training and actual application imagery does not necessarily result in better performance of the Mask R-CNN in IWPs mapping; (4) and overall, the model underestimates the total number of IWPs particularly in terms of disjoint/incomplete IWPs. [ABSTRACT FROM AUTHOR]