Combining satellite and in situ data to make improved estimates of upper-ocean thermal variability on eddy to global scales

Large oceanographic datasets, including in situ and remotely sensed measurements, are combined using statistical and dynamical assimilation techniques in order to study upper-ocean variability and heat storage on eddy to global scales. Satellite altimetry, sea-surface temperature and upper-ocean thermal data provide information that is highly complementary, and combinations of these data result in improved space and time resolution as well as improved accuracy relative to the individual data types. The power of integrated datasets is illustrated through several oceanographic examples spanning a wide range of scales.; Using data from a small region in the southwestern Pacific, a statistical technique is developed for combining satellite measurements of sea-surface height and sea-surface temperature with in situ temperature profiles to produce improved estimates of upper-ocean heat content, steric height and temperature variability. The technique relies on satellite data to reduce sampling error caused by the sparsely available profile data. Estimates of interannual variability in heat storage, steric height and temperature are produced and analyzed using nine years of data in the study region.; The technique is applied to global datasets to produce estimates of variability in upper-ocean heat content, temperature and thermosteric sea level for the global oceans. The estimates contain significant interannual variability and imply an oceanic warming rate of 1.02 +/- 0.12 Watts per square meter of ocean (0.35 +/- 0.04 pW) from 1993 to 2003 in the global average. The warming caused sea level to rise at a rate of 1.7 +/- 0.2 mm/yr over the same period. A significant portion of the global trend is shown to be due to regional warming in the Southern Hemisphere that penetrates deep into the water column.; Mesoscale variability is studied using a regional, quasi-geostrophic assimilation model in a small region of the central North Pacific. The assimilation model provides a tool for combining multiple data streams into a dynamically consistent estimate of ocean variability. It shows skill in tracking mesoscale variability and reproducing the subsurface vertical structure responsible for eddy heat transport. Additionally, the relative importance of subsurface and satellite datasets in constraining the model is characterized.