An analysis of the Galveston Hurricane using the 20CR data set

The Twentieth Century Reanalysis (20CR) is an atmospheric dataset consisting of 56 ensemble members, which covers the entire globe and reaches back to 1871. To assess the suitability of this dataset for studying past extremes, we analysed a prominent extreme event, namely the Galveston Hurricane, which made landfall in September 1900 in Texas, USA. The ensemble mean of 20CR shows a track of the pressure minimum with a small standard deviation among the 56 ensemble members in the area of the Gulf of Mexico. However, there are systematic differences between the assimilated “Best Track” from the International Best Track Archive for Climate Stewardship (IBTrACS) and the ensemble mean track in 20CR. East of the Strait of Florida, the tracks derived from 20CR are located systematically northeast of the assimilated track while in the Gulf of Mexico, the 20CR tracks are systematically shifted to the southwest compared to the IBTrACS position. The hurricane can also be observed in the wind field, which shows a cyclonic rotation and a relatively calm zone in the centre of the hurricane. The 20CR data reproduce the pressure gradient and cyclonic wind field. Regarding the amplitude of the wind speeds, the ensemble mean values from 20CR are significantly lower than the wind speeds known from measurements.

The Samoa Hurricane, 15-16 March 1889

Analysing historical weather extremes such as the tropical cyclone in Samoa in March 1889 could add to our understanding of extreme events. However, up to now the availability of suitable data was limiting the analysis of historical extremes, particularly in remote regions. The new “Twentieth Century Reanalysis” (20CR), which provides six-hourly, three-dimensional data for the entire globe back to 1871, might provide the means to study this and other early events. While its suitability for studying historical extremes has been analysed for events in the northern extratropics (see other papers in this volume), the representation of tropical cyclones, especially in early times, remains unknown. The aim of this paper is to study to the hurricane that struck Samoa on 15-16 March 1889. We analyse the event in 20CR as well as in contemporary observations. We find that the event is not reproduced in the ensemble mean of 20CR, nor is it within the ensemble spread. We argue that this is due to the paucity of data assimilated into 20CR. A preliminary compilation of historical observations from ships for that period, in contrast, provides a relatively consistent picture of the event. This shows that more observations would be available and implies that future versions of surface-based reanalyses might profit from digitizing further observations in the tropical region.