Old Weather and the new climate of the Arctic: crowdsourcing large-scale data rescue for climate research

Large sea ice retreats like that which occurred in 2007, and the extreme meteorology of the past two winters, demonstrate the sensitivity of the Arctic to climate variability and highlight its potential as a harbinger of future change. It is believed that increasing anthropogenic forcing is associated with increasingly large physical and economic impacts, although the early twentieth century warming in the Atlantic – Arctic region also suggests that natural variations can also produce substantial environmental consequences. However, to really understand the underlying climate processes we need longer, higher resolution and spatially comprehensive data sets. In principle these may be built using historical observations. For centuries scientists and sailors have left records of the weather and environmental conditions they encountered, in many cases every hour for years at a time. These observations would allow us to reconstruct long climate time series at the sub-daily resolution required for dynamical reanalysis, but they are inaccessible – buried in thousands of handwritten logbooks and weather journals.

While the manuscript data resource is huge, converting it into a usable data format is difficult. Manual transcription of these records is unavoidable and time consuming; consequently these data have scarcely been examined by climatologists. However, one solution is to use public outreach to attract volunteers to the task. The Old Weather initiative (www.oldweather.org) has proven to be an effective way to obtain rapid and accurate transcriptions of ship logbooks. Within 3 months of its initiation more than 250,000 manuscript logbook pages from World War I era Royal Navy ships were transcribed by some 6000 citizen volunteers. These transcriptions include hourly meteorological variables, sea-surface temperature, as well as information on the movements of the ship and events on board. Quality is assured by the fact that every data point is independently transcribed at least three times.

With the collaboration of partners brought together by the international ACRE initiative (http://www.met-acre.org), we have extended the scope of Old Weather to include ships that sailed the Arctic and high-latitude oceans since the late 18th century. As well as the barometer and thermometer measurements, non-instrumental variables like sea ice reports are also extracted for integration into new or updated data products. Initial evidence gleaned from historical sources demonstrates the potential of crowdsourcing and data archeology. As well as providing direct information about the regional climate, newly digitized observations provide input to future extended surface-input reanalysis products like the 20th Century Reanalysis. Data recovered from high latitudes will greatly enhance reanalysis skill, especially over the sparsely-observed North Pacific and Arctic regions. With these tools not only will we be equipped to distinguish bellwether climate events from rare but ordinary fluctuations, but we will also have better access to their particular meteorological and dynamical underpinnings.