I'm wondering what stations in Alaska you would consider as robust for total snowfall records? Much of my research is in the Brooks Range and it seems that Kotzebue and Bettles have the "best" records for precipitation all year (Kotz better than Bettles). I know that snow records are particularly poor in Alaska, especially in the Arctic. Is there something in the Kotzebue record that makes it not reliable back to the 1930s, even though the GHCN records go back this far?
All snow observations ended in Kotzebue in autumn 2019 (and almost everywhere in rural mainland Alaska). 60-year robust snowfall records are Bettles, Anchorage Airport, Juneau Airport. Denali NP HQ is probably good. Snow measurements ended at Fairbanks Airport about 2020 and are now done at UAF West Ridge, but because of the elevation difference (and so sometimes more snow in the shoulder seasons), that gets an asterisk. There are a few other coop stations with only a little missing you could probably make work, e.g. Alyeska, Auke Bay.
Colin and I are especially interested in the Kotzebue winter precipitation record, which shows a strong upward trajectory, but also an odd feature around 1980 when the baseline appears to shift when the data are aggregated by "winter" (e.g., Nov- Mar). We checked the station meta data and it doesn't appear that there was a change in the instrumentation around that time. It would be great to know if you think the Kotzebue winter precipitation record is robust.
Are the snowfall observations being phased out in favor of winter precipitation, which is a more automated measurement with a longer and more continuous record at most stations? How do the ERA5 snowfall analyses compare with those winter precipitation records? I realize this is a little bit of an apples vs oranges comparison. However, for much of Alaska, almost all of the winter precipitation is snow.
The Weather Bureau/NWS office at Kotzebue was staff (mostly) 24-hours a day from the WW2 era until the 1990s with a standard observation schedule. Without comparative observations it's very difficult to know if the jump in 1979-80 is real or not (ERA does not an abrupt change through the late 1960s and 70s were relatively dry). The post-2018 winter precip record is not robust with multiple missing days, some during big storms (e.g. March 2021 and 2023). Bottom line is that precipitation measurements in windy tundra environments are difficult in the best of circumstances, and unaugmented, un-QC'd automated measurements are not close. Snowfall is even worse, since in such environments even professional, experienced observed can only make educated guesses, with no hope of automation. Indeed, in wind tundra environments it's not clear what it means to say "six inches of snow overnight", when windward sides of buildings have drifts to the roof tops and leeward side scoured down to the ground.
Yes. Measuring snowfall is challenging in windy environments, regardless of whether it's a person or an instrument making the measurements. I'll check the Nome winter precipitation record as soon as the NCEI comes back online. In the meantime, we're really enjoying your newsletter. Thanks for taking the time to share your insights!!
Figured 4 and 5 are in substantial agreement which is both reassuring and perplexing. What might be causing the significant decrease in the Anchorage region?
The big decreases are mainly the mountainous areas on the Kenai Peninsula and on the west side of Cook Inlet. This is likely in part due to higher snow levels and these are also areas that show some decrease in winter precipitation over the past 50 years. The decreased snow at lower elevations mostly the result of higher fraction of cold season precip falling as rain (in the reanalysis).
I'm wondering what stations in Alaska you would consider as robust for total snowfall records? Much of my research is in the Brooks Range and it seems that Kotzebue and Bettles have the "best" records for precipitation all year (Kotz better than Bettles). I know that snow records are particularly poor in Alaska, especially in the Arctic. Is there something in the Kotzebue record that makes it not reliable back to the 1930s, even though the GHCN records go back this far?
All snow observations ended in Kotzebue in autumn 2019 (and almost everywhere in rural mainland Alaska). 60-year robust snowfall records are Bettles, Anchorage Airport, Juneau Airport. Denali NP HQ is probably good. Snow measurements ended at Fairbanks Airport about 2020 and are now done at UAF West Ridge, but because of the elevation difference (and so sometimes more snow in the shoulder seasons), that gets an asterisk. There are a few other coop stations with only a little missing you could probably make work, e.g. Alyeska, Auke Bay.
Colin and I are especially interested in the Kotzebue winter precipitation record, which shows a strong upward trajectory, but also an odd feature around 1980 when the baseline appears to shift when the data are aggregated by "winter" (e.g., Nov- Mar). We checked the station meta data and it doesn't appear that there was a change in the instrumentation around that time. It would be great to know if you think the Kotzebue winter precipitation record is robust.
Are the snowfall observations being phased out in favor of winter precipitation, which is a more automated measurement with a longer and more continuous record at most stations? How do the ERA5 snowfall analyses compare with those winter precipitation records? I realize this is a little bit of an apples vs oranges comparison. However, for much of Alaska, almost all of the winter precipitation is snow.
The Weather Bureau/NWS office at Kotzebue was staff (mostly) 24-hours a day from the WW2 era until the 1990s with a standard observation schedule. Without comparative observations it's very difficult to know if the jump in 1979-80 is real or not (ERA does not an abrupt change through the late 1960s and 70s were relatively dry). The post-2018 winter precip record is not robust with multiple missing days, some during big storms (e.g. March 2021 and 2023). Bottom line is that precipitation measurements in windy tundra environments are difficult in the best of circumstances, and unaugmented, un-QC'd automated measurements are not close. Snowfall is even worse, since in such environments even professional, experienced observed can only make educated guesses, with no hope of automation. Indeed, in wind tundra environments it's not clear what it means to say "six inches of snow overnight", when windward sides of buildings have drifts to the roof tops and leeward side scoured down to the ground.
Yes. Measuring snowfall is challenging in windy environments, regardless of whether it's a person or an instrument making the measurements. I'll check the Nome winter precipitation record as soon as the NCEI comes back online. In the meantime, we're really enjoying your newsletter. Thanks for taking the time to share your insights!!
Nome precipitation has fewer problems than Kotzebue but notably the automated system was OTS during the epic early March 2021 storm.
Thanks! We are really dropping the ball on snow observations, and have been for some time.
Figured 4 and 5 are in substantial agreement which is both reassuring and perplexing. What might be causing the significant decrease in the Anchorage region?
The big decreases are mainly the mountainous areas on the Kenai Peninsula and on the west side of Cook Inlet. This is likely in part due to higher snow levels and these are also areas that show some decrease in winter precipitation over the past 50 years. The decreased snow at lower elevations mostly the result of higher fraction of cold season precip falling as rain (in the reanalysis).