Fifty years later: The Jan 31 - Feb 1 1953 North Sea Gale
revisited via Reanalysis and Reforecasting.
(The storm that nearly killed me.)

Huug Vandendool


From January 31 to February 1, 1953, a period of about 36 hours during a weekend, a very strong windstorm occurred over the North Sea basin. Strong winds are not uncommon in the area, but given the shape of the North Sea, a sustained Northwesterly windstorm can pile up water into the narrower Southern North Sea, and in the extreme, sea level can rise by several meters. On this occasion sustained Northnorthwesterly surface winds in excess of 50 knots were observed for more than 24 hours over a large portion of the North Sea. The very high water caused a major flood in the low lying coastal areas of East central England and, ?turning the corner? in Belgium and The Netherlands. In The Netherlands alone over 1800 people drowned, and tens of thousand of people were displaced. In the UK over 300 people died. Major damage was also reported in Scotland and marine interests of many nations suffered badly with several sunken ships and loss of life. The memory of this storm lives on internationally. In 1999, during the Millenium hype, NOAA scientists voted the 1953 gale into the top 15 list (no particular order) of the most significant meteorological events of the century. And now we approach the 50 year remembrance.

At the time, weather forecasts were made by time tested synoptic techniques not involving any of the numerical models which we have come to rely upon increasingly since the 1960s. Worded forecasts were normally issued only for ?today and tomorrow?, i.e. out to about 24 hours, and distributed by radio, newspaper etc. When warranted the forecast would include a warning for unusual sea level rise valid during the next two (astronomical) high water marks - storm surge forecasts were strictly empirical as well. Sea level rise warnings were sent by telex to various authorities so as to take action. The forecast on this day (which was reasonably accurate) gave the authorities a lead time of at most 18 hours to implement evacuation plans (such as they were), issue warnings, guarding of dikes, emergency sandbagging etc.

With how much more lead time could the 1953 storm have been foreseen, if a modern system for Numerical Weather Prediction had been in place?. While the question is well posed, the situation is somewhat artificial in that we assume modeling and analysis capabilities typical of the mid 1990s, but an observing system as it was in 1953. Of course many other things have changed since 1953. But we leave such other considerations aside. The question, as to the lead time by which this particular storm could have been known to occur, is answered here using the NCEP-NCAR Reanalysis system (Kalnay et al 1996), which covers the entire period from period 1948 to the present. We found the 48 and even 72 & 96 hr forecast to be quite accurate w.r.t. to the strength, direction and duration.

The scope of the disaster would undoubtedly have been much less with that much advance warning. The many improvements in weather forecasting, often too small to notice in any particular year, amount to a revolution when comparing 1953 to the 1990s. Nevertheless, Important details remain difficult even with today?s technology. For instance, the sudden development of this storm as it approached Scotland on the 30th of January is underforecast. The forecast tracks of the storm across the North Sea on the 31st have a northward bias - the observed track from the tip of Scotland to the SE is more dangerous for creating a storm surge in the southern North Sea than the various model forecast tracks toward the ESE or even pure East. We will also compare the contemporaneous hand-analyses to the Reanalysis, and discuss forecasts made by technology that was right around the corner in 1953: a barotropic model.