Due to the connection between atmospheric conditions and the marine climate, previous studies proposed the use of atmospheric data as a predictor to estimate the total water level at the coast. However, none of the previous applications considered the effect that a large and shallow continental shelf may have on the propagation of water level variables, such as the storm surge. The shallow bathymetry facilitates the occurrence of coastal trapped waves, and the storm surge signal in may present a strong component generated in remote regions. This phenomenon must be considered when defining the atmospheric predictor in these areas. This work presents a methodology to define the best atmospheric predictor to describe waves, storm surge and total water level in this particular kind of coasts. The method was applied in a location on the southern coast of Brazil (Southwest Atlantic), where the effect of the shallow bathymetry in the storm surge was previously observed. A statistical relationship between the atmospheric predictor and water level variables at three coastal points was established to assess its ability to estimate atmosphericinduced water levels and their components. High Pearson correlation coefficients (r>0.78 for all variables) and errors comparable to those obtained by traditional numerical methods demonstrate the skill of the predictor to describe the variables related to the water level at the coast. The implications of the atmospheric predictor extend beyond water level estimation including several applications. Some of these applications are shown here, such as the characterization of average and extreme marine climate and the assessment of climate-induced variability.