Site measurement and model-based results showed similar
ranges of interannual variability in the initiation of the
growing season as radar-based results, while estimates of
annual variability in the termination of the growing season
were generally smaller, averaging only 10 days. The correspondence
between radar estimates and site measurements
of the cessation of the growing season was significant but
generally lower than estimates of the timing of growing
season initiation. Radar and BIOME-BGC estimates of the
cessation of the growing season were not significantly
related. Radar-based results generally delayed the timing
of growing season cessation by more than 2 weeks relative
to site measurements and model results. Additionally, while
radar-based results corresponded to the general timing of
seasonally frozen temperatures in fall, termination of the
growing season as indicated by site observations occurred
from 4 days to 3.5 weeks prior to the arrival of seasonally