Satellite Applications for Detecting Vegetation Phenology - Satellite-based Applications on Climate Change - page 261

processed database at 8-km resampling grid twice monthly from 1982 covering the
globe and at 1-km resolution with biweekly intervals since 1989 covering the
conterminous United States.
The high temporal resolution, moderate spatial resolution, and relatively long-
term continuity make this sensor well suited for examining and monitoring pheno-
logical events for entire ecosystems on regional as well as on global scale. AVHRR-
derived vegetation phenology has widely been used in research areas of vegetation
activity, climate change, land use, and disaster. Vegetation phenology derived from
AVHRR provides unique opportunities for monitoring vegetation activity trends at
large scales. Myneni et al. ( 1997 ) have presented an increase in plant growth
associated with a lengthening of the active growing season from the 1981 to 1991
for the north hemisphere, based on AVHRR Pathinder NDVI data set and the
Global Inventory Monitoring and Modeling Studies (GIMMS) AVHRR NDVI
data set. They have estimated an advance in the active growing season of 8 3

days, a prolongation of the declining phase at 4 2 days, and therefore, a longer

active growing season of 12 4 days over the 1980s. Zhou et al. ( 2001 ) have

investigated the AVHRR-derived northern hemisphere vegetation activity and the
land surface temperature records for the period from 1981 to 1999. Their results
show a persistent increase in growing season NDVI over broad contiguous forests
and woodlands for Europe, a larger increase in growing season NDVI magnitude
and a longer active growing season for Eurasia and North America, and NDVI
decreases due to temperature-induced drought in boreal zones.
In addition to the trend over long periods captured by AVHRR-derived vegeta-
tion phenology, interannual anomalies of phenological dynamics also contain
meaningful information on the response of vegetation to climate change. Asner
et al. ( 2000 ) has revealed that the seasonal NDVI amplitude provided by AVHRR
measurements increased throughout Amazon forest during EI Ni˜o periods when
rainfall was anomalously low. Based on vegetation phenology data set created from
Pathfinder NDVI, European spring phenology has been shown to correlate particu-
larly well with anomalies in winter temperature and winter North Atlantic Oscilla-
tion Index for 20 years from 1982 to 2001 (St¨ckli and Vidale 2004 ).
The phenological changes and attributes have been used as an indicator of land
use changes and land use classification. Reed ( 2006 ) shows that changes in agricul-
tural practices result in a trend toward long duration of season in Saskatchewan,
Canada, based on 8-km AVHRR data set. Wessels et al. ( 2009 ) have used AVHRR-
derived phenology data in a fully supervised decision-tree classification based on
the new biome map of South Africa to identify the phenological attributes that
distinguish between the different biomes.
Furthermore, extensive research has shown the important contribution of vegeta-
tion phenology to the monitoring of environmental disaster. Peckham et al. ( 2008 )
demonstrates that fire has a significant effect on the phenological dates of the Canadian
boreal forest derived from AVHRR-NDVI. They state that the most recently burned
areas have later greenup dates. Brown et al. ( 2008 ) has integrated the 1-km AVHRR-
derived phenological metrics, with climate-based drought index data and other bio-
physical information into a vegetation drought response model (VegDRI) to generate
higher resolution drought monitoring information in near real time.