The coupled variability of the Sea Surface Temperature (SST) and atmosphere-ocean surface heat fluxes over the Bay of Biscay (Eastern North Atlantic) has been analyzed. Daily surface heat fluxes from different meteorological reanalyses are combined with a high resolution reconstructed satellite SST data set by means of Lagged Maximum Covariance Analysis (MCA). Lagged MCA is applied at different spatial scales. Its results are interpreted within the framework of Hasselmann's stochastic climate model. The surface heat-flux feedback on SST is confirmed to be generally negative. No clear relation is found between the first MCA expansion series and the leading Sea Level Pressure (SLP) patterns. However, a clear relation is found between the second expansion series and the leading SLP Principal Component (PC) when the atmosphere leads the ocean. Spatial patterns of anomalies of the SST and of the SST tendency are found to be related by a 4 day lag. The same reconstructed satellite SST and reanalysis heat fluxes are combined to estimate the feedbacks related to the surface heat fluxes. The traditional procedure used to compute the surface heat-flux feedbacks from monthly data is adapted for daily data. High resolution maps of the heat-flux feedback are derived for the annual and seasonal cases for the Bay of Biscay. Feedbacks related to turbulent (latent and sensible heat) fluxes are shown to dominate over the radiative ones. Special attention is paid to small-scale features present in both Lagged Covariance patterns and surface heat-flux feedback estimates.