Pan-Arctic linkages between snow accumulation and growing season air temperature, soil moisture and vegetation

Kristina Luus, Yulia R. Gel, John C. Lin, Richard Kelly, Claude Duguay

TL;DR

This study uses a novel non-parametric statistical method (ACE) to uncover complex, non-linear relationships between snow accumulation and growing season conditions across the pan-Arctic. The findings reveal intricate linkages between snow, temperature, soil moisture, and vegetation, providing crucial insights into how Arctic ecosystems might respond to climate change.

What Problem Does This Paper Address?

The Arctic is a region highly sensitive to climate change, with snow cover playing a critical role in regulating its environment. While field studies have hinted at connections between snow accumulation and growing season conditions (like soil moisture and vegetation health), understanding these relationships across the vast and heterogeneous pan-Arctic has been challenging. Existing methods often struggle with the complex, non-linear interactions and the scale-dependent nature of these processes. This paper aims to bridge this gap by providing a comprehensive, regional-scale analysis of these linkages, which is essential for predicting the Arctic's response to a warming climate.

Methods

The core of this study's methodological innovation lies in its application of the Alternating Conditional Expectations (ACE) technique. Unlike traditional linear regression, ACE is a non-parametric method that can uncover and describe complex, non-linear relationships between variables without assuming a predefined functional form. This makes it particularly well-suited for analyzing the intricate interactions within ecological systems like the Arctic. The authors applied ACE to analyze pairwise associations between snow water equivalent (SWE) and growing season variables (air temperature, soil moisture, and vegetation transmissivity) across different vegetation classes and time periods.

Datasets

The study utilizes two key satellite-derived datasets:

  • GlobSnow: Provides snow water equivalent (SWE) data.
  • NTSG (Numerical Terradynamic Simulation Group): Provides growing season estimates of air temperature, soil moisture, and vegetation transmissivity.

Both datasets are derived from AMSR-E (Advanced Microwave Scanning Radiometer for the Earth Observing System) observations. The analysis focuses on the pan-Arctic region north of 60° N, using data from 2003-2008.

Key Insights and Ideas

  • Non-linear is the norm: The study strongly emphasizes that the relationships between snow and growing season variables are predominantly non-linear, a finding that traditional linear models would likely miss or misrepresent.
  • Vegetation as a double-edged sword: Regions with lower vegetation transmissivity (more biomass) tend to accumulate less snow, possibly due to interception and shading. However, vegetation also plays a role in encouraging snow accumulation through wind trapping and providing insulation.
  • Temperature's complex influence: Warmer growing season temperatures are generally associated with diminished snow accumulation. However, warmer mean annual growing season temperatures at high latitudes can lead to more snow accumulation, likely due to increased water vapor availability for precipitation.
  • Soil moisture's role in snow onset: Drier soils preceding snow onset tend to accumulate more snow, possibly because they freeze faster and more thoroughly, creating a more stable base for snow accumulation.
  • ACE as a powerful tool: The study highlights ACE as a valuable technique for unraveling complex ecological relationships, especially when dealing with large, heterogeneous datasets where linear assumptions fail.

Key Results

  • The ACE analysis revealed statistically significant, non-linear associations between SWE and growing season variables across various vegetation classes and time periods.
  • Linear regression and PCA analyses showed very weak or non-existent linear associations, underscoring the importance of the non-linear ACE approach.
  • Specific findings include:
  • Regions with lower vegetation transmissivity (more biomass) accumulated less snow.
  • Warmer growing season temperatures were linked to less snow accumulation.
  • Higher mean annual growing season temperatures at high latitudes were associated with more snow.
  • Drier soils before snow onset led to greater snow accumulation.
  • The study found that ACE-derived linkages and thresholds largely corresponded with in situ findings, despite the coarse resolution of the satellite data.

Limitations and Future Work

  • Data uncertainty: The authors acknowledge uncertainty in passive microwave estimates of the Arctic land surface due to its heterogeneity and high lake fraction.
  • Scale dependence: While ACE helps bridge scales, the authors note that scale dependence and ecological heterogeneity are inherent challenges in pan-Arctic studies.
  • Further field investigations: The paper suggests that further field investigations are needed to better elucidate the drivers of observed associations and clarify their dependence on stated thresholds.
  • Continued monitoring: The authors strongly recommend continued monitoring of Arctic ecosystems at the field scale, complemented by the application of the ACE technique to satellite data, to better understand and predict the impacts of climate change.

Summary updated

Abstract

Arctic field studies have indicated that the air temperature, soil moisture and vegetation at a site influence the quantity of snow accumulated, and that snow accumulation can alter growing season soil moisture and vegetation. Climate change is predicted to bring about warmer air temperatures, greater snow accumulation and northward movements of the shrub and tree lines. Understanding the response of northern environments to changes in snow and growing season land surface characteristics requires: (1) insights into the present-day linkages between snow and growing season land surface characteristics; and (2) the ability to continue to monitor these associations over time across the vast pan-Arctic. The objective of this study was therefore to examine the pan-Arctic (north of 60° N) linkages between two temporally distinct data products created from AMSR-E satellite passive microwave observations: GlobSnow snow water equivalent, and NTSG (growing season air temperature, soil moisture and vegetation transmissivity). Due to the complex and interconnected nature of processes determining snow and growing season land surface characteristics, these associations were analyzed using the modern non-parametric technique of Alternating Conditional Expectations (ACE), as this approach does not impose a predefined analytic form. Findings indicate that regions with lower vegetation transmissivity (more biomass) at the start and end of the growing season tend to accumulate less snow at the start and end of the snow season, possibly due to interception and shading. Warmer air temperatures at the start and end of the growing season were associated with diminished snow accumulation at the start and end of the snow season. High latitude sites with warmer mean annual growing season temperatures tended to accumulate more snow, probably due to the greater availability of water vapor for snow season precipitation at warmer locations. Regions with drier soils preceding snow onset tended to accumulate greater quantities of snow, likely because drier soils freeze faster and more thoroughly than wetter soils. Understanding and continuing to monitor these linkages at the regional scale using the ACE approach can allow insights to be gained into the complex response of Arctic ecosystems to climate-driven shifts in air temperature, vegetation, soil moisture and snow accumulation.
BibTeX
@inproceedings{luus2013pan,
  author = {Kristina Luus and Yulia R. Gel and John C. Lin and Richard Kelly and Claude Duguay},
  title = {{Pan-Arctic linkages between snow accumulation and growing season air temperature, soil moisture and vegetation}},
  booktitle = {Biogeosciences},
  year = {2013},
  doi = {10.5194/bg-10-7575-2013},
}
Figure