PhD defence: Yijing Liu

Yijing Liu defends her thesis,

Upscaling snow cover changes and associated ecological effects based on a snow model – a case study from coastal Greenland
Soil thermal regime, winter soil carbon loss, and land surface phenology  

Supervisors: 
Assistant professor Andreas Westergaard-Nielsen, IGN
Associate Professor Emeritus Birger Ulf Hansen, IGN

Assessment Committee:
Senior Researcher Michele Citterio, GEUS
Associate professor Stef Lhermitte, TU Delft – NL
Associate Professor Ylva Sjöberg (chair), IGN  

Summary:
Arctic terrestrial ecosystems have undergone significant variations due to the amplified effects of climate change in the Arctic region. Seasonal snow cover is one of the predominant features of the Arctic landscape, which persists for up to 7–8 months each year. Changes in snow cover can have profound impacts on various ecological processes from winter to the growing season, including soil temperature changes, ecosystem gas exchange, and seasonal transition dates, particularly under a warmer and wetter Arctic climate. However, the effects of snow cover combined with other environmental factors on these ecological processes remain complex from plot to regional scale. Using ground observations and satellite datasets, a distributed snow model, and statistical models, this thesis analyzed spatial-temporal changes in snow cover, quantified the snow insulation effects, linked snow depth and winter soil respiration as well as ecosystem respiration during the growing season, and explored the responses of seasonal transition dates to multi-climatic conditions at Disko Island, Western Greenland.
Based on an 8-year Arctic ecosystem manipulation experiment (snow addition due to snow fences, open top chamber warming, shrub removal, and control), I found that snow fences significantly altered the year-to-year variations in snow depth and snow phenology from 2012 to 2020. Snow depth was confirmed as the key factor influencing the difference between soil and air temperatures during the snow cover season, with larger snow depths corresponding to higher temperature differences between soil and air. The rate of change in this temperature difference, as the function of snow depth, was slower during the snow accumulation period than during the snow melting period. During the snow-free season, the warming effects of open top chambers and shrub removal were weakened on the snow addition plots, suggesting a delayed impact of snow cover on soil temperature during summer months.
Warmer soil environments under the snowpack keep soil respiration active during winter. Using a first-order exponential model, I quantified the total winter soil respiration carbon (CO2-C) efflux across snow fences, the annual means ranging from 55 to 58 g m-2 year-1 in control plots and 62 to 71 g m-2 year-1 in snow addition plots from 2012 to 2020. The total winter CO2-C emission, as a linear function of accumulated snow depth, significantly increased with larger snow depths (Slope = 0.12, P < 0.01). After upscaling the spatial and temporal variations in snow depth for an approximate 6 km2 dry heath region from 2010 to 2020, I found that wind-induced snow redistribution could account for annual means of CO2-C emissions ranging from 51 to 76 g m-2 year-1 in space, higher CO2-C emissions mainly distributed in low-lying regions and wind-protected areas, where more snow cover was accumulated. Throughout 2010–2020, the 6 km2 dry heath tundra ecosystem was estimated to release 369±38 t of carbon per winter due to accumulated snow depths. During the growing season, although a structural equation model indicated a direct relationship between the seasonal accumulation of snow depth and ecosystem respiration as well as gross ecosystem photosynthesis, the snow addition alone had no significant effects on ecosystem respiration or gross ecosystem photosynthesis measured during the summer season.
Furthermore, I analyzed the direct and indirect effects of multi-environmental factors on seasonal transition dates (start of spring, maximum Normalized Difference Vegetation Index (NDVImax day), and end of fall). On a regional scale (approx. 21 km2 with a 32 m spatial resolution), climatic conditions (air temperature, precipitation, and snow cover) and terrain characteristics (elevation, slope, aspect) exerted different effects on seasonal transition dates for various land surface classes (abrasion, fen, dry heath, wet heath, and tall shrub) in Greenland. The combination of climatic and terrain factors explained 38%–51%, 40%–53%, and 42%–57% changes in the start of spring, NDVImax day, and end of fall, respectively, across 5 land surface classes. Snow ending day, snow depth, and freezing index demonstrated a positive correlation with the start of spring. NDVImax day was primarily influenced by snow ending day, air temperature, and precipitation in June, with snow ending day as the most significant factor; a warmer June led to an earlier NDVImax day. The end of fall exhibited a positive correlation with summer air temperature and precipitation. Moreover, I projected future changes in seasonal transition dates under two different climate scenarios, indicating that a 2–3 °C increase in current air temperature and a 10%–15% increase in current precipitation would advance the start of spring in all land surface classes except for wet heaths. Conversely, the end of fall in abrasion, fen, and dry heaths was expected to be delayed in 2050. The period from the start of spring to the end of fall was projected to extend across all land surface classes except for wet heaths, with the longest duration occurring in tall shrubs, spanning 145–148 days.
This thesis emphasizes the importance of snow cover and the associated ecological effects when considering Arctic climate change, the necessity of understanding ecosystem processes from several perspectives and scales, and the possible change directions of land surface phenology and corresponding annual carbon budgets, pointing out a direction for future land surface model simulations.

A digital version of the PhD thesis can be obtained from the PhD secretary phd@ign.ku.dk