Jump to content
  • Member Statistics

    17,605
    Total Members
    7,904
    Most Online
    ArlyDude
    Newest Member
    ArlyDude
    Joined

Weekly Arctic Sea Ice Contest


Recommended Posts

I think we have to accept that the margin of error in ice extent forecasting or measurement is fairly large. It all depends on how they categorize extensive areas of partial ice cover. But that question about the Great Lakes is relevant because Lake Superior must be about to freeze fairly rapidly and it covers a good 50k, then there would be almost that much ready to freeze on combined Lakes Huron and Michigan (actually the world's largest freshwater lake).

I was looking back on maps available at some link involving the University of Chicago where they have maps going back to 1901, looking for data on the Baltic Sea, apparently it froze solid only once in the 20th century after the severe winter of 1947 but had fairly extensive ice a few other times, most notably in 1979 when about half of it was frozen. Otherwise, it seems normal for only the Gulfs of Bothnia, Finland and Riga to freeze in most winters, and a little shore ice elsewhere.

The Gulf of St Lawrence, I believe, freezes in most winters during late January and February and has best ice conditions in March, for the annual seal hunt that causes our country so much p.r. grief -- but this is by no means an annual "sure thing." I think some here would be surprised to see how late in the year ice normally forms off Labrador and Newfoundland. A couple of years ago they were still dealing with harbour ice in June and early July in north shore Newfoundland and that's barely north of 50 deg N. I have not checked this rigorously but I think the maximum extent of sea ice in that region is in April and May.

Link to comment
Share on other sites

  • Replies 377
  • Created
  • Last Reply

This is exactly why I have problems with this contest...what is the resolution of the satellites? Do smaller bodies of water with significant area, like the Great Lakes, count? I've asked these questions before but you never seen to produce an answer.

Feel free to look into it yourself, you have asked me about this many times and I have given you my answers.

However, I am completely satisfied that the data is precise enough for a weekly contest. For one thing the text of their website implies to me that they feel comfortable the data is precise day to day. They have chosen to do a two day rolling average instead of a five day rolling average because they feel their data is precise enough and because they have the satellite data available to do so.

Second of all, the JAXA gains match what one would expect from the satellites. Weeks that look like they gain a lot of ice from the satellite images do in fact gain a lot of ice on Jaxa and vice versa. Our poor scores I believe are a result of not being able to (or not trying to) predict how the weather will affect the ice. However, I believe with careful analysis it is possible to have some skill at this and I think to some extent I have been able to do so.

The product was specifically designed to detect small short term changes in ice concentration at the ice's edge in order to learn more about sea ice processes and for validation of sea ice modeling in GCMs.

http://sharaku.eorc....pdf/alg_des.pdf

As to whether or not the Great Lakes count.. as I said, I don't think they do. It is an index of sea ice. In addition the data comes from AMSR-E which lists the target of its sea ice concentration data as "Both polar regions and the Sea of Okhotsk." This implies to me that AMSR-E does not collect data from the GLs. I will try to find a definitive answer to this.

This product retrieves sea ice concentration over global oceans above 30 degree in north and south latitude. Land areas are masked out by using land-ocean flag information, 25km grid data, provided by algorithm development PI. In addition, sea ice concentration is not retrieved in regions where the sea surface temperatures are above 5 degree of Celsius provided by JAXA. Sea ice concentrations may be retrieved along the coastline because of the accuracy of latitude/longitude position data in L1B data or the effect of antenna side lobe.

http://sharaku.eorc..../sic031224.html

Link to comment
Share on other sites

Feel free to look into it yourself, you have asked me about this many times and I have given you my answers.

However, I am completely satisfied that the data is precise enough for a weekly contest. For one thing the text of their website implies to me that they feel comfortable the data is precise day to day. They have chosen to do a two day rolling average instead of a five day rolling average because they feel their data is precise enough and because they have the satellite data available to do so.

Second of all, the JAXA gains match what one would expect from the satellites. Weeks that look like they gain a lot of ice from the satellite images do in fact gain a lot of ice on Jaxa and vice versa. Our poor scores I believe are a result of not being able to (or not trying to) predict how the weather will affect the ice. However, I believe with careful analysis it is possible to have some skill at this and I think to some extent I have been able to do so.

The product was specifically designed to detect small short term changes in ice concentration at the ice's edge in order to learn more about sea ice processes and for validation of sea ice modeling in GCMs.

http://sharaku.eorc....pdf/alg_des.pdf

As to whether or not the Great Lakes count.. as I said, I don't think they do. It is an index of sea ice. In addition the data comes from AMSR-E which lists the target of its sea ice concentration data as "Both polar regions and the Sea of Okhotsk." This implies to me that AMSR-E does not collect data from the GLs. I will try to find a definitive answer to this.

This product retrieves sea ice concentration over global oceans above 30 degree in north and south latitude. Land areas are masked out by using land-ocean flag information, 25km grid data, provided by algorithm development PI. In addition, sea ice concentration is not retrieved in regions where the sea surface temperatures are above 5 degree of Celsius provided by JAXA. Sea ice concentrations may be retrieved along the coastline because of the accuracy of latitude/longitude position data in L1B data or the effect of antenna side lobe.

http://sharaku.eorc..../sic031224.html

I agree that the general weekly trend on JAXA matches what one would expect looking at satellite imagery, but there are some questions that obviously remain such as, "Does persistent cloud cover affect sampling of certain regions?" and of course, what areas count. Hudson Bay is not part of the polar region in my book, and yet that certainly counts. Also, if they are saying arctic sea ice extent, then why does Hudson Bay count if it's not a sea? Should landlocked bodies of water that are called "seas" such as the Aral count, if they freeze, whereas the Great Lakes don't count because they're called a "lake" even though they resemble an inland sea? What is the difference between a sea and a lake, in the first place, and what difference is Jaxa using?

The poor scores are the result of the fact that we're making guesses...it would take hours to hone a really good estimate...you'd have to study countless models, satellite imagery of cloud cover, SSTs of every area, etc. Most of us have more important things to do than spend 6 hours working towards a guess. That being said, MNTransplant, ElTacoman, you, and I have done the best so far.

Link to comment
Share on other sites

I agree that the general weekly trend on JAXA matches what one would expect looking at satellite imagery, but there are some questions that obviously remain such as, "Does persistent cloud cover affect sampling of certain regions?" and of course, what areas count. Hudson Bay is not part of the polar region in my book, and yet that certainly counts. Also, if they are saying arctic sea ice extent, then why does Hudson Bay count if it's not a sea? Should landlocked bodies of water that are called "seas" such as the Aral count, if they freeze, whereas the Great Lakes don't count because they're called a "lake" even though they resemble an inland sea? What is the difference between a sea and a lake, in the first place, and what difference is Jaxa using?

The poor scores are the result of the fact that we're making guesses...it would take hours to hone a really good estimate...you'd have to study countless models, satellite imagery of cloud cover, SSTs of every area, etc. Most of us have more important things to do than spend 6 hours working towards a guess. That being said, MNTransplant, ElTacoman, you, and I have done the best so far.

I believe it says on the jaxa page or in one of those links I posted that cloud cover doesn't effect the result much at all. It has some small effect. I'd have to read it again.

I'm now not as sure that the GLs are not counted. Some lakes in northern Canada appear on the Jaxa maps that correspond with the ice extent data, and the maps shows them unfrozen in summer and frozen in winter. That implies to me that they might be counted. Either way.. the only lakes that would make much difference are the GLs and even they are not that much area especially considering they don't freeze all at once and some of them often don't freeze.

Agree with the last part.. it would take much more data and time to come up with a good forecast.

Link to comment
Share on other sites

I'm now not as sure that the GLs are not counted. Some lakes in northern Canada appear on the Jaxa maps that correspond with the ice extent data, and the maps shows them unfrozen in summer and frozen in winter. That implies to me that they might be counted. Either way.. the only lakes that would make much difference are the GLs and even they are not that much area especially considering they don't freeze all at once and some of them often don't freeze.

Great Bear Lake and Great Slave Lake are the 7th and 9th largest lakes in the world, located in the NW Territories, with each one covering about 30,000km2 surface area. I assume JAXA counts these lakes as part of their arctic sea ice extent, correct? These would make some difference although their freezing and melting periods are probably somewhat less volatile than more maritime areas of the ice pack.

Link to comment
Share on other sites

+250k

The conditions in the Bering Strait area look quite good this week with north to northeast winds and cold air. Okhotsk is not especially cold, but is cooler than last week with a more northerly wind direction (although it starts the week quite warm). That should allow for some gains there especially considering how little ice there is currently. Barents is highly unfavorable with south winds and above average temps, however I don't think we can lose too much more there considering we are now considerably below average there after last week's torch. I think Hudson Strait (and the area just east of there which may not technically be part of the Strait) will continue to gain as they are right on the edge of that very cold canadian air, and we might get some in Baffin Bay as well although it doesn't get as much of the bitter cold.

Link to comment
Share on other sites

+250k

The conditions in the Bering Strait area look quite good this week with north to northeast winds and cold air. Okhotsk is not especially cold, but is cooler than last week with a more northerly wind direction (although it starts the week quite warm). That should allow for some gains there especially considering how little ice there is currently. Barents is highly unfavorable with south winds and above average temps, however I don't think we can lose too much more there considering we are now considerably below average there after last week's torch. I think Hudson Strait (and the area just east of there which may not technically be part of the Strait) will continue to gain as they are right on the edge of that very cold canadian air, and we might get some in Baffin Bay as well although it doesn't get as much of the bitter cold.

Agree with all this, I just went a little lower because the first 2-3 days look pretty crappy outside of small gains in Hudson Strait/Baffin Bay. Do you think we are going to gain a bunch between Greenland and Svalbard? That area seems to be lacking, but has some cold air for the first 4 days of the week with 850s in the -16C/-20C range.

Should see some bigger gains later in the week as the ice expands south of the Bering Strait, especially on the Russian side, and then in the Sea of Okhotsk.

Link to comment
Share on other sites

Agree with all this, I just went a little lower because the first 2-3 days look pretty crappy outside of small gains in Hudson Strait/Baffin Bay. Do you think we are going to gain a bunch between Greenland and Svalbard? That area seems to be lacking, but has some cold air for the first 4 days of the week with 850s in the -16C/-20C range.

Should see some bigger gains later in the week as the ice expands south of the Bering Strait, especially on the Russian side, and then in the Sea of Okhotsk.

Yep I think that makes sense.. not a large area but looks decent.

Link to comment
Share on other sites

The final was 285k

Week 17 Scores:

skierinvermont     245000
Normal        232000
tacoman        225000
Nzucker        220000
Consensus    198000
Roger Smith      171000
grcjrsc        160000
MN transplant    150000
Mallow        134000
winterwxluvr    121000

Cumulative:

skierinvermont     2173000
Normal        1958000
Consensus    1548000
tacoman        1544000
Nzucker        1492000
MN transplant    1360000
grcjrsc        996000
Roger Smith      826000
winterwxluvr    773000
frivolousz21    733000
Mallow        468000
blazess556    71000
BethesdaBoy    64000
Organizing Low    49000
Ytterbium        -171000

Average Error:

skierinvermont     122176
Normal        134824
tacoman        153500
Consensus    158941
Nzucker        162235
MN transplant    165000
grcjrsc        187750
frivolousz21    188917
winterwxluvr    194786
Roger Smith      201412
Mallow        222471

Also just for fun I decided to tally up the weeks each of us have won

frivolousz21:    0 
skierinvermont:    6 (6,9,13,15,16,17)
Mallow:        1 (11)
MN transplant:    4 (2,8,12,14)
tacoman:       2 (3,7)
grcjrsc:    0
Organizing Low:    0
Nzucker:    2 (1,4)
Ytterbium:    0
Roger Smith:    2 (5,10)
blazess556:    0
winterwxluvr:    0
BethesdaBoy:    0

I had this week won about 4 days in. It was over. Then the sea ice did the exact same thing that the snow has been doing to me. :lol:

Link to comment
Share on other sites

I had this week won about 4 days in. It was over. Then the sea ice did the exact same thing that the snow has been doing to me. :lol:

I feel bad...you had this pretty locked up. The sea ice has been very inconsistent this year and that's what's hurt us most.....incredible to think we can be losing ice during the winter when we're already near record low extent.

Link to comment
Share on other sites

Pretty much going as planned with losses in the Barents and gains in Hudson Strait and Bering strait. Losses in Barents a little more than I expected. Going to come down to if Okhotsk can gain some with the colder air the last 3 days.

Yes, I think we all assumed the last three days would feature the biggest gains. It also depends how quickly the cold air can hit Baffin Bay and areas around Labrador...there seems to be a strong cut-off between brutally cold 850s over Eastern Canada and milder 850s over the Baffin Bay region on the 12z ECM...the really frigid airmass doesn't hit the water until later in the week according to the ECM. Okhotsk is all going to be around -16C to -20C 850s so we're probably good for gains there considering how ice formation is behind schedule over there.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.

×
×
  • Create New...