| Variability in 5- to 25-year records of hourly mean in situ sea temperature, ocean currents, and meteorology at diverse shallow-water habitats in the Florida reef tract (FRT) is analyzed. Tidal, diurnal, and annual periodicities generally dominate sea temperature variability, with strong variability apparent in the "weather band" of 3-42 d at one reef-flat site, and at the local inertial period at one offshore site near the shelf break. A statistically significant interannual warming trend is also observed at this one offshore site only. Significant covariability between sea temperature and coincident air temperature, wind speed, sea-surface temperature (SST) gradients, and incident radiation (light) is also found. However, this covariability itself varies with an annual period, and differs between sites with similar depths, apparently due to differences in seafloor slope.;A coastal ocean reef heat budget is estimated from the hourly mean in situ sea temperature, meteorology, satellite SST, and reanalysis data for each site, together with a model of insolation absorption in the water column and heat exchange at the seafloor. A term for smaller-scale heat advection, the so-called horizontal convection (HC) or thermal siphon, previously observed at coral reefs elsewhere in the world, balances the heat budget. At six of the eight sites analyzed, the budget matches the long-term annual climatology of observed in situ sea temperature variability within estimated uncertainty, and matches full seasons at the two other sites. Budget results also match the observed daily sea temperature variability, with R2 > 0.3, root mean squared error < 0.1 K, and bias < 10 mK at most sites. Results are most sensitive to the scaling chosen for the horizontal convection parameterization, to assumed rates of insolation absorption (Kd), and to uncertainties in estimated surface currents and sea temperature gradients. However, estimates for horizontal heat exchange, cross-shore gradients, and insolation absorption rates in the water-column are found to compare well with direct in situ and satellite measurements. The heat budget is also shown to produce reliable results when using only remotely sensed and reanalysis data, providing a mechanism for more reliable monitoring of thermal stress on coral reefs where long in situ records are not available.;Finally, modes of sea-temperature variability, particularly for periods when variability is large or poorly explained by the heat budget, are analyzed in the context of other meteorological and oceanographic data using the techniques of heuristic ecological forecasting, Principal Component Analysis (PCA), and an artificial neural network called a Self-Organizing Map (SOM). Anomalous patterns of meteorological and circulation variability are identified from the in situ and satellite record, that are associated with periods when observed sea temperature variability is not well-explained by the heat budget. A combination of these methods is shown to improve the understanding of past reef ecological impacts related to thermal stress, such as coral bleaching. Applications of the present research for improved understanding of coastal physical oceanography and coral reef ecology in the FRT are briefly discussed. |
