12.07.2015 Views

Book 2.indb - US Climate Change Science Program

Book 2.indb - US Climate Change Science Program

Book 2.indb - US Climate Change Science Program

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Abrupt <strong>Climate</strong> <strong>Change</strong>Figure 4.6. Time series of AMOC variability at 26ºN. (“overturning,” red curve), derived fromthe 2004–05 RAPID array (from Cunningham et al., 2007). Individual contributions to the totalupper ocean flow across the section by the Florida Current (blue), Ekman transport (black), andthe mid-ocean geostrophic flow (magenta) are also shown. A 2-month gap in the Florida currenttransport record during September to November 2004 was caused by hurricane damage to theelectromagnetic cable monitoring station on the Bahamas side of the Straits of Florida.the field of oceanography is rapidly evolvingtoward operational applications of ocean stateestimation analogous to that of atmosphericreanalysis activities. A large number of theseactivities are now underway that are beginningto provide first estimates of the time-evolvingocean “state” over the last 50+ years, duringwhich sufficient observations are available toconstrain the models.There are two basic types of methods, (1) variationaladjoint methods based on control theoryand (2) sequential estimation based on stochasticestimation theory. Both methods involvenumerical ocean circulation models forced byglobal atmospheric fields (typically derivedfrom atmospheric reanalyses) but differ inhow the models are adjusted to fit ocean data.Sequential estimation methods use specifiedatmospheric forcing fields to drive the models,and progressively correct the model fields intime to fit (within error tolerances) the data asthey become available (e.g., Carton et al., 2000).Adjoint methods use an iterative process tominimize differences between the model fieldsand available data over the entire duration of themodel run (up to 50 years), through adjustmentof the atmospheric forcing fields and modelinitial conditions, as well as internal modelparameters (e.g., Wunsch, 1996). Except for thesimplest of the sequential estimation techniques,both approaches are computationally expensive,and capabilities for running global models forrelatively long periods of time and at a desirablelevel of spatial resolution are currently limited.However, in principle these models are able toextract the maximum amount of informationfrom available ocean observations and providean optimum, and dynamically self-consistent,estimate of the time-varying ocean circulation.Many of these models now incorporate a fullsuite of global observations, including satellitealtimetry and sea surface temperature observations,hydrographic stations, autonomousprofiling floats, subsurface temperature profilesderived from bathythermographs, surface133

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!