Every winter, some of the automatic instruments at remote stations stop reporting, and most of them stay silent until someone can reach them in spring. Keith Pemberton, an undergraduate at Yale, has spent this fall going through the operating records of nearly 9,000 station winters across Alaska, Greenland, Iceland, the central Arctic Ocean and Antarctica to understand why.
Outages are more common than most people assume. Last winter, roughly 30% of the active automatic weather stations in Alaska went at least a week without sending data.
Cold is a bigger problem than darkness alone. At Alaska seismic stations, the share of days with battery voltage under 12 V rose from 8% above freezing to 74% below minus 30 °C. Weather stations followed the same pattern on a smaller scale, with cold days two to four times more likely to run low.
The most useful finding is that voltage rarely gives warning. Only about a quarter of the week long outages at Alaska weather stations had any low voltage in the two weeks before. The rest came from rime ice, snow burial, failed radio or satellite links and other causes that a battery reading never shows. Some failures build slowly and still go unnoticed. During the MOSAiC expedition, a fuel cell on the sea ice gradually stopped charging one station's batteries. The decline was visible in the data for weeks, but nothing raised an alarm until the station went dark.
Network practices matter too. In Alaska, some networks almost never run low in winter, while others run low on about one in five winter days. The difference likely comes down to how the systems are sized, sited and maintained.
This January, Keith will test a small monitoring box in Fairbanks with a research laboratory. It reads power, temperature, weather and communications together, looks for the patterns that come before a failure, and tells crews what is likely wrong so they can bring the right part on the first trip.
He would like to hear from INTERACT station managers, or to see a rough log of winter outages and repairs. He plans to share the findings back to inform future station design and implementation, and with the open science community. You can reach him at keith.pemberton@yale.edu.
Read more about the research here.
Alaska weather station data courtesy of Synoptic Data.
Photo: Christelle Guesnon