SST and Long-range Predictions
SST is one of the most important and
famous climate variables that has a pedigree history of measurement and analysis.
The motivation for Climate Prediction Research and Climatology development is the continual improvement reliable
climate prediction uncertainty in support to strategic development through input and advice to climate change assessments.
In this context, SST is required for climate model initialisation, diagnostics and fundamental climate monitoring.
The most important requirement is that the observations are accurate and free of bias. Considering
the best estimates of global warming trends, SST data sets should be exceptionally stable to better than
0.1K/decade if with a mean zero bias.
The Figure shows bias differences between the NOAA Optimal Interpolated SST analysis
Reynolds OIv2.0 SST and the
NOAA/NODC Pathfinder daytime only SST. Clearly seen are different bias differences according to the particular NOAA AVHRR
satellite instrument and other significant atmospheric events, such as the
Mount Pinatubo volcanic eruption in 1991.
The reliable SST data record extends from about 1870 (see the
International Comprehensive Ocean-Atmosphere Data Set (ICOADS)) although considerable effort is required to quality control
biases in the observations caused by changing in the design of ships,
observing practices and the geographical focus of shipping routes. Today, satellite data provide a unique and extensive
source of global coverage SST observations. But in order to meld these together with existing SST climatologies
based on in situ observations alone, considerable effort is required to ensure
that biases due to satellite measurement techniques, instrument drifts, calibration etc. are properly accounted for.
Due to the vast amounts of data that are provided by satellite instruments, even very small errors will have a significant
impact. Cross calibration between follow-on instruments must be planned and executed and adherence
to the ten GCOS climate monitoring principles
is essential. Achieving such a small bias (and verifying that it is true) is a long process of verification and validation using
in situ observations and other analysis The GHRSST-PP Diagnostic Data Sets (DDS)
are developed for this purpose. It can only be done through careful reference to a well documented and calibrated
sub-set SST measurements derived from the in situ network of quality controlled buoys. Through community consensus, this
is the most truthful measure of SST available today. But, there just simply are not enough buoys to measure
all over the ocean.
The GHRSST-PP is making an extremely valuable contribution to climate research through the GHRSST-PP Re-analysis (RAN) program.
The RAN is working to produce a new SST Climate Data Record (CDR) extending across the entire satellite record (from about 1984). The
RAN has already established direct access to all GHRSST-PP operational and delayed mode data which is archived at the
National Oceanographic Data Centre (NODC), USA. Using the
NASA Pathfinder program as an example of how to manage the production
of a SST-CDR within GHRSST-PP, the GHRST-PP RAN will be conducting regular univariate reanalysis runs that consider different
approaches to SST analysis, data preparation and quality control. The first GHRSST-PP RAN products are expected in 2006/7.
For more information, see the GHRSST-PP RAN pages.
SST patterns change realtively slowly and can reasonably be well predicted up to
6 months ahead or longer in some regions of the world. The links between regional SST patterns and the atmosphere can be represented
in computer models of the atmosphere and ocean or statistically related to SST observations
or data-driven analysis. Dynamically
coupled climate models increasingly form the basis of many seasonal
prediction systems. The strongest relationship between SST patterns and seasonal weather trends
are found in tropical regions. Strong signals are associated with the El Niņo phenomenon in the tropical Pacific, roughly
every three or four years, which can disrupt the global pattern of normal
weather including large changes in seasonal rainfall patterns (droughts in some regions
and floods in others). Weaker links between SST and seasonal weather are found in other parts of the globe.
Seasonal forecasts provide information about likely conditions averaged over the next few months based
on long-term averages. The relationship between weather and SST is strongest when long-term weather
averages are used and, because the uncertainty in forecasts generally rises as the forecast range increases,
seasonal forecasts are different in format when compared to the familiar daily forecasts:
- Seasonal forecasts represent average conditions over several months
- Seasonal forecasts are given in terms of probability
The GHRST-PP, as it builds up a global high-resolution SST CDR archive
will provide a unique source of data for seasonal forecasting activities. In the short term, GHRSST-PP global
analysis products can be used to initialise seasonal forecast models and verify seasonal forecasts in hindcast runs.
More on seasonal forecasting and SST can be found at the the following links: