Here are the re-navigated and indexed ice obs 1900-1930 compared to the reanalysis-forced PIOMAS sea ice model run (concentration field). I'm about finished with the rest of the 30s (Northland) and we'll update accordingly. These are rather good results and a very big thank you to everyone who is working so hard on this.
Legend:
Purple line = Hadley 15% ice edge 1979-1985
Pink line = PIOMAS 15% ice edge for specified period
Red dots: ice index = 1 (observed)
Blue dots: ice index = 0 (not observed / water)
Stats, obs compared to model:
Same as model - 5643 obs (74%)
Different from model - 1985 obs (26%)
Most of the misses relate to the 0 index. When there is ice observed in the log the hits are 2875 to 123 misses. There are a number of explanations for 0 index misses -- one being it isn't a positive observation like the 1 index. That is a non-mention isn't the same as a non-presence, perhaps due to low visibility, or the ice is just over the horizon but still in the grid box with the ship. etc.
Cheers,
Kevin
Kevin wrote:Sat Apr 21, 2018 3:19 pm
Here's a little more about what we see in the initial results. As in many cases it is the unexpected, the previously unperceived, that is making my brain melt. First of all, the model is doing a good job, which is remarkable in itself considering it is forced only by the ERA-20C reanalysis. We see the annual cycle of melt and freeze in the Pacific sector looks good, with the Arctic pack more or less situated between the north coast of Alaska and Wrangell Isl. in August, with an open area over by Banks Isl. and the Mackenzie delta (this where the winter-over whalers at Herschel Isl. would go to hunt Bowhead in the spring).
What stands out for me is the situation in June in the northern Bering Sea and southern Chukchi. The pink line is the model ice edge defined by the limit of 15% ice concentration for the 3 decades 1901-1910, 1911-1920, 1921-1930. The purple line is the ice edge from the earliest set of microwave-derived data commonly used today (1979-1985) and the difference is the change. (If you have a look at the map for yesterday, 20 April 2018 https://nsidc.org/arcticseaicenews/ it looks even more stark.) The observations you all have recovered nails this down - it's easy enough to argue that a sea-ice model is getting wrong when there are no obs to validate, but not now.
Implications? The heat budget of the Pacific inflow into the Arctic would likely have been very different, and this would have long term impacts on the multiyear ice cover in Arctic basin consistent with current obs. Our understanding of the state of the Bering Sea ecosystem since the 1970s may be seen in a new context...especially for ice-obligate marine mammals. There will be more.
Kevin wrote:Sat Jun 15, 2019 7:19 pm
Axel's the complicator -- our contribution is whether or not there was ice there in the real world back then. But that's key.
Kevin wrote:Sat Nov 30, 2019 4:50 pm
Now third most-read article in J. Clim. over last 12 mo. (almost 5000 reads) - Arctic sea ice volume variability over 1901-2010: A model-based reconstruction
