available on the GHRSST website, including the spatial resolution and available
time periods for each dataset (ghrsst.jpl.nasa.gov/GHRSST_product_table.html).
A GHRSST international user community is currently testing and applying
GHRSST data products and services within scientific projects and operational
systems in real-time. The challenge for the GHRSST team is to deliver sustained
production of stable, high-quality SST data products and services, and for the user
community to demonstrate a corresponding requirement for sustained operations.
Long-term gridded datasets are also available that blend satellite and in situ
data to provide greater context for long-term trends in SST. For example, the
NOAA Extended Reconstruction Sea Surface Temperature (ERSST) analysis is
generated using in situ SST data and statistical methods that allow stable reconstruction using sparse data. The monthly ERSST analysis extends from 1854 to
present at 2° spatial resolution (Smith and Reynolds 2004). Although local and
short-term variations have been smoothed, ERSST data are suitable for long-term
global and basin-wide studies. As another example, the NOAA Optimum Interpolation (OI) SST analysis is a weekly and monthly product available from 1981 to
present at 1° spatial resolution (Reynolds et al. 2002) that integrates in situ
measurements, satellite SST, and SST simulated by sea ice cover. The use of reanalysis or reconstruction techniques, however, must be met with caution as differences in data density and acquisition characteristics can lead to different results
within the derived SST output. Nonetheless, these datasets are useful for comparison with other variables and to simulate long-term trends in SST. In essence, in
any application of thermal data, it is important to consider the characteristics,
quality, limitations, caveats and time domain of the SST datasets in order to wisely
select the product(s) most appropriate to the specific application.
12.2.3 Quantifying Trends and Changes
Variations in SST, as indicated through the satellite data record, show that most of
the ocean has warmed. Example of studies undertaken to quantify this and other
temporal trends in SST are presented below.
Using the retrospective AVHRR Pathfinder SST dataset for the period
1985–2006, Strong et al. (2009) evaluated trends in night-time SST of the tropical
oceans (35°S–35°N) to examine implications for tropical ecosystems. Trends were
calculated from monthly mean temperatures and showed that tropical oceans had
generally experienced warming, or only slight cooling, with the exception of a
small region of significant cooling in the north-eastern Pacific Ocean along 20°N
(Fig. 12.1). Observing a shift in Pacific Ocean SST anomalies in the mid-1990s,
Strong et al. (2009) also calculated trends for two eleven-year periods (1985–1995
and 1996–2006). This analysis showed significant regions of the global ocean,
predominantly in the Southern Hemisphere, cooled in the earlier period (western
Pacific, southern Indian and southern Atlantic); however, these same regions
320
S. J. Weeks et al.
time periods for each dataset (ghrsst.jpl.nasa.gov/GHRSST_product_table.html).
A GHRSST international user community is currently testing and applying
GHRSST data products and services within scientific projects and operational
systems in real-time. The challenge for the GHRSST team is to deliver sustained
production of stable, high-quality SST data products and services, and for the user
community to demonstrate a corresponding requirement for sustained operations.
Long-term gridded datasets are also available that blend satellite and in situ
data to provide greater context for long-term trends in SST. For example, the
NOAA Extended Reconstruction Sea Surface Temperature (ERSST) analysis is
generated using in situ SST data and statistical methods that allow stable reconstruction using sparse data. The monthly ERSST analysis extends from 1854 to
present at 2° spatial resolution (Smith and Reynolds 2004). Although local and
short-term variations have been smoothed, ERSST data are suitable for long-term
global and basin-wide studies. As another example, the NOAA Optimum Interpolation (OI) SST analysis is a weekly and monthly product available from 1981 to
present at 1° spatial resolution (Reynolds et al. 2002) that integrates in situ
measurements, satellite SST, and SST simulated by sea ice cover. The use of reanalysis or reconstruction techniques, however, must be met with caution as differences in data density and acquisition characteristics can lead to different results
within the derived SST output. Nonetheless, these datasets are useful for comparison with other variables and to simulate long-term trends in SST. In essence, in
any application of thermal data, it is important to consider the characteristics,
quality, limitations, caveats and time domain of the SST datasets in order to wisely
select the product(s) most appropriate to the specific application.
12.2.3 Quantifying Trends and Changes
Variations in SST, as indicated through the satellite data record, show that most of
the ocean has warmed. Example of studies undertaken to quantify this and other
temporal trends in SST are presented below.
Using the retrospective AVHRR Pathfinder SST dataset for the period
1985–2006, Strong et al. (2009) evaluated trends in night-time SST of the tropical
oceans (35°S–35°N) to examine implications for tropical ecosystems. Trends were
calculated from monthly mean temperatures and showed that tropical oceans had
generally experienced warming, or only slight cooling, with the exception of a
small region of significant cooling in the north-eastern Pacific Ocean along 20°N
(Fig. 12.1). Observing a shift in Pacific Ocean SST anomalies in the mid-1990s,
Strong et al. (2009) also calculated trends for two eleven-year periods (1985–1995
and 1996–2006). This analysis showed significant regions of the global ocean,
predominantly in the Southern Hemisphere, cooled in the earlier period (western
Pacific, southern Indian and southern Atlantic); however, these same regions
320
S. J. Weeks et al.