Perhaps one of the most seminal articles on large-scale land surface phenology
was Myneni et al. (1997), which presented evidence from satellite data that the
photosynthetic activity of terrestrial vegetation in northern latitudes increased from
1981 to 1991 in a manner that is suggestive of an increase in plant growth associated with a lengthening of the active growing season. This was one of first
articles to link large-scale vegetation trends to a warming climate.
As the available time series of remote sensing data increases, so too have the
time spans for LSP studies. As the time series increases, studies are able to better
extract more ‘‘signal’’ from the ‘‘noise’’ and explore correlation between LSP and
climate events and distinguish between natural variability and trends. Ivits et al.
(2012) used LSP derived from the AVHRR from 1982 to 2006 to explore the
correlations between phenology and climate and trends in both. Park et al. (2012)
used two decades of AVHRR data to explore the relationship between El Niño–
Southern Oscillation (ENSO) events and the onset of spring. These are just two
examples among many where the long record from AVHRR is being used to
explore how vegetation responds to a changing climate.
However, there is considerable uncertainty associated with LSP studies that link
climate forcings with vegetation change. At the very heart of the problem is the
uncertainty in methods and techniques used to extract phenology parameters from
a satellite time series. The foundational intercomparison work of White et al.
(2009) demonstrated that start of season estimates vary extensively within and
among methods and that selecting the strongest method is difficult without some
additional ecosystem information. In addition to differences in algorithms, specific
regional phenomena may complicate the analysis. For example, Samanta et al.
(2012) detail how clouds and aerosols complicate LSP studies in the Amazon.
Indeed, we can use the Amazon as an example to expose how uncertainty
associated with LSP studies can lead to controversy and confusion. In 2010, a
Boston University press release
1 stated that a recent study (Samanta et al. 2010)
showed results that were contrary to a previously published report and claims by
the Intergovernmental Panel on Climate Change (IPCC). Because of its potential
relevance to national and international policy, the conclusions of the IPCC can be
highly contentious. The uncertainly involved with discerning trends in the Amazon
(Saleska et al. 2007; Samanta et al. 2010) provided considerable fodder for climate-related blogs,
2,3 On the technical side, it is not surprising that LSP products
may indicate different, somewhat inconsistent results (White et al. 2009; Samanta
et al. 2012). However, if LSP results are to contribute to policy-relevant information, it is important, at the very least, to quantify the uncertainty and, at best,
provide consistent and reliable information. Hopefully advances in sensors,
methods, and validation will help address this uncertainty and lead to greater
utility.
1 http://www.eurekalert.org/pub_releases/2010-03/bumc-nsd031110.php
2 http://scienceblogs.com/deltoid/2010/03/14/its-always-bad-news-for-the-ip/
3 http://www.realclimate.org/index.php/archives/2010/03/saleska-responds-green-is-green/
4 Land Surface Phenology
117
was Myneni et al. (1997), which presented evidence from satellite data that the
photosynthetic activity of terrestrial vegetation in northern latitudes increased from
1981 to 1991 in a manner that is suggestive of an increase in plant growth associated with a lengthening of the active growing season. This was one of first
articles to link large-scale vegetation trends to a warming climate.
As the available time series of remote sensing data increases, so too have the
time spans for LSP studies. As the time series increases, studies are able to better
extract more ‘‘signal’’ from the ‘‘noise’’ and explore correlation between LSP and
climate events and distinguish between natural variability and trends. Ivits et al.
(2012) used LSP derived from the AVHRR from 1982 to 2006 to explore the
correlations between phenology and climate and trends in both. Park et al. (2012)
used two decades of AVHRR data to explore the relationship between El Niño–
Southern Oscillation (ENSO) events and the onset of spring. These are just two
examples among many where the long record from AVHRR is being used to
explore how vegetation responds to a changing climate.
However, there is considerable uncertainty associated with LSP studies that link
climate forcings with vegetation change. At the very heart of the problem is the
uncertainty in methods and techniques used to extract phenology parameters from
a satellite time series. The foundational intercomparison work of White et al.
(2009) demonstrated that start of season estimates vary extensively within and
among methods and that selecting the strongest method is difficult without some
additional ecosystem information. In addition to differences in algorithms, specific
regional phenomena may complicate the analysis. For example, Samanta et al.
(2012) detail how clouds and aerosols complicate LSP studies in the Amazon.
Indeed, we can use the Amazon as an example to expose how uncertainty
associated with LSP studies can lead to controversy and confusion. In 2010, a
Boston University press release
1 stated that a recent study (Samanta et al. 2010)
showed results that were contrary to a previously published report and claims by
the Intergovernmental Panel on Climate Change (IPCC). Because of its potential
relevance to national and international policy, the conclusions of the IPCC can be
highly contentious. The uncertainly involved with discerning trends in the Amazon
(Saleska et al. 2007; Samanta et al. 2010) provided considerable fodder for climate-related blogs,
2,3 On the technical side, it is not surprising that LSP products
may indicate different, somewhat inconsistent results (White et al. 2009; Samanta
et al. 2012). However, if LSP results are to contribute to policy-relevant information, it is important, at the very least, to quantify the uncertainty and, at best,
provide consistent and reliable information. Hopefully advances in sensors,
methods, and validation will help address this uncertainty and lead to greater
utility.
1 http://www.eurekalert.org/pub_releases/2010-03/bumc-nsd031110.php
2 http://scienceblogs.com/deltoid/2010/03/14/its-always-bad-news-for-the-ip/
3 http://www.realclimate.org/index.php/archives/2010/03/saleska-responds-green-is-green/
4 Land Surface Phenology
117