In this work, we present a statistical model to generate relevant reanalysis of geophysical parameters. In particular, we use a stochastic equation to control the temporal and spatial variability of the signal and we take into account the possible error of the observations. We resolve the system iteratively using an ensemble Kalman filter and smoother. We apply the methodology to remote sensing data of Sea Surface Temperature (SST). We use high resolution SST maps provided by an infrared sensor, sensible to the presence of cloud. Comparing the results with the reference SST reanalysis, we demonstrate the capability of our approach to interpolate missing data and keep into account the spatial and temporal consistency of the SST signal.