We report on a pilot demonstration of the usefulness of analog seismograms to improve the database of ocean storms before the 1980s by providing additional data for the quantitative validation of ocean wave modeling, in particular for extreme events. We present a method for automatic digitization of paper seismograms to extract microseismic ground‐motion periods and amplitudes. Each minute of the original paper records is scanned and vectorized. The amplitudes are calibrated based on the original metadata taken from official bulletins. The digitized time series is processed to extract power spectral densities, which are compared with modeled microseisms levels computed using a numerical ocean wave model. As a case study, we focus on one month of data recorded at the Royal Observatory of Belgium (ROB) from January to February 1953, around the “Big Flood” event, a tragic storm surge that flooded the lowlands of England, the Netherlands, and Belgium on 1 February 1953. The reconstructed spectrograms for the three components of ground motion show clear storm signatures that we relate to specific sources in the North Atlantic Ocean. However, our models of the Big Flood event based on these data do not result in the expected amplitudes as modeled compared to the observational data when the storm reached its maximum in the southern North Sea. We suggest that the source of microseisms recorded at ROB is related to the primary microseism generated in the North Sea, at periods of 7–8 s. Other discrepancies identified suggest small modifications of the source locations or energy. Reconstructed horizontal and vertical ground motions are coherent. This is a good news for the purpose of present‐day analyses of constructing twentieth century ocean‐climate models, especially as during much of that time only horizontal seismographs were installed at observatories.