SEARCH

SEARCH BY CITATION

Keywords:

  • accumulated growing degree-days;
  • boreal forest;
  • Eurasia;
  • IGBP transects;
  • NDVI;
  • North America;
  • Pathfinder AVHRR Land (PAL);
  • Taiga;
  • Trends;
  • Tundra

Abstract

The International Geosphere–Biosphere Program has delineated five study areas that form a northern high-latitude network for the analyses of vegetation and carbon dynamics. We examined the magnitude and significance of changes in the land surface phenologies of ecoregions within these transects using the NASA Pathfinder Advanced Very High-Resolution Radiometer Land dataset. We applied the seasonal Mann–Kendall (SMK) trend test, a robust and nonparametric approach, to determine the significance of trends in the normalized difference vegetation index (NDVI) over the five transects. The SMK trend test provides an important alternative over the frequently used but unreliable trend analysis based on linear regression.

In addition, we modeled the land surface phenology using quadratic or nonlinear spherical models to relate the NDVI data to accumulated growing degree-days (base 0°C). Nonlinear spherical models parsimoniously describe the green-up dynamics in taiga and tundra ecoregions. Models for each ecoregion within each transect were fitted separately for two time periods (1985–1988 and 1995–1999) and their parameter coefficient estimates were compared. In 10 of 24 ecoregions that comprise 72% of the land area in the transects, the date of the peak NDVI value was significantly earlier (range 2–18 days) in the second study period than in the first study period. This progression was more pronounced in North America than in Siberia (weighted average of 9.3 vs. 6.3 days earlier).

Understanding of what constitutes significant change in land surface phenology amidst background variation is a critical component of global change science. A diversity of datasets, techniques, and study areas has led to a range of conclusions about boreal phenology. We discuss statistical pitfalls in standard analyses and offer a framework to conduct statistically reliable change assessments of land surface phenologies.