Support the Arctic Sea Ice Forum and Blog

Author Topic: September Predictions Challenge  (Read 2642 times)

Steven

  • ASIF Citizen
  • Posts: 416
    • View Profile
  • Liked: 35
  • Likes Given: 8
Re: September Predictions Challenge
« Reply #50 on: July 11, 2018, 08:21:42 PM »
June NSIDC cliff is absent or in the wrong direction but the data where it shoots up is compromised by satellite tests, so I'm just treating it as absent rather than an upwards cliff in June cancelled by a downwards one in July.

I don't see a reason to ignore the NSIDC data  (apart from the obvious outlier on 27 June).  In fact, JAXA is in good agreement with NSIDC: Both of them have very high compactness in the second half of June, and a strong decrease of compactness in the first week of July:



May data from Schroeder was 5.3 with confidence given by a 0.5 standard deviation.
June data is 4.7 +- 0.5 but I'm not sure if thats a standard deviation or a confidence range.

It's called forecast error in the Schroeder et al. 2014 paper.  It's obtained by simulating the "predictions" that the model would have generated in previous years, using only data that were available at the moment of the "prediction".  E.g. they simulate a prediction for September 2010 by using the regression equation over 1979-2009 and then feeding the May/June 2010 data into that equation.  They repeat that procedure for every year from 1984 to 2013 and then calculate the standard deviation of the errors of the simulated "predictions", giving them 0.44M km2 in the paper.  I guess the 0.5M km2 is obtained by including the last 4 years of data too (up to 2017) and perhaps rounding to the nearest multiple of 0.1M km2.

Brigantine

  • ASIF Lurker
  • Posts: 71
    • View Profile
  • Liked: 3
  • Likes Given: 17
Re: September Predictions Challenge
« Reply #51 on: July 12, 2018, 12:23:36 AM »
95% CI for prediction = 3.68 to 5.74
Thanks. Definitely sticking with medium confidence then!

uniquorn

  • ASIF Citizen
  • Posts: 243
    • View Profile
  • Liked: 54
  • Likes Given: 8
Re: September Predictions Challenge
« Reply #52 on: July 12, 2018, 11:52:08 PM »
uniquorn
JAXA min 3.5-4.0; medium
NSIDC av 3.75-4.25; medium

Steven

  • ASIF Citizen
  • Posts: 416
    • View Profile
  • Liked: 35
  • Likes Given: 8
Re: September Predictions Challenge
« Reply #53 on: July 14, 2018, 12:30:38 PM »
JAXA minimum extent:      4.25-4.75,  Medium
NSIDC September extent:  4.75-5.25,  Medium

This is based on the calculations discussed here.  Updating those calculations to use the latest July data, the calculated values are slightly lower than when using the June data, but still within the same bin.

Neven

  • Administrator
  • ASIF Royalty
  • *****
  • Posts: 5347
    • View Profile
    • Arctic Sea Ice Blog
  • Liked: 102
  • Likes Given: 86
Re: September Predictions Challenge
« Reply #54 on: July 14, 2018, 01:57:50 PM »
JAXA minimum extent:      4.25-4.75,  Medium
NSIDC September extent:  4.5-5.0,  Medium

My predictions for June didn't have any confidence levels, because I didn't get how it works. Can I still change them to 'low'?

Quote
Neven
June
NSIDC: Between 4.5 and 5.0 million km2, None
JAXA: 4.0 - 4.5 M km^2, None
Compare, compare, compare

Juan C. García

  • ASIF Middle Class
  • Posts: 667
    • View Profile
  • Liked: 44
  • Likes Given: 38
Re: September Predictions Challenge
« Reply #55 on: July 14, 2018, 02:14:55 PM »
JAXA minimum extent:      4.25-4.75,  Medium
NSIDC September extent:  4.5-5.0,  Medium

Same as Neven.
Just a coincidence. I did not copy him.  ;)
________________________________________
Edit:
This month, the level of confidence is medium.
________________________________________
Last month:
June:
JAXA - 3.75-4.25, Medium
NSDIC - 4.0-4.5, Medium


I voted the same way that jdallen.
Confidence? I think that there is a low confidence for everybody, including me. But let's say that I could expect one bin up, but several bins down.
« Last Edit: July 14, 2018, 06:50:12 PM by Juan C. García »
Which is the best answer to Sep-2012 ASI lost (compared to 1979-2000)?
50% [NSIDC Extent] or
73% [PIOMAS Volume]

Volume is harder to measure than extent, but 3-dimensional space is real, 2D's hide ~50% thickness gone.
-> IPCC/NSIDC trends [based on extent] underestimate the real speed of ASI lost.

Richard Rathbone

  • ASIF Citizen
  • Posts: 420
    • View Profile
  • Liked: 6
  • Likes Given: 2
Re: September Predictions Challenge
« Reply #56 on: July 14, 2018, 03:16:26 PM »
JAXA minimum extent:      4.25-4.75,  Medium
NSIDC September extent:  4.5-5.0,  Medium

My predictions for June didn't have any confidence levels, because I didn't get how it works. Can I still change them to 'low'?

Quote
Neven
June
NSIDC: Between 4.5 and 5.0 million km2, None
JAXA: 4.0 - 4.5 M km^2, None

I'm treating None as Very Low, (in all cases, not just yours), which is the option that I put in for those that didn't get it. Its far too late to change.

Almost too late for this month too, a little under 10 hours left from this post.

Neven

  • Administrator
  • ASIF Royalty
  • *****
  • Posts: 5347
    • View Profile
    • Arctic Sea Ice Blog
  • Liked: 102
  • Likes Given: 86
Re: September Predictions Challenge
« Reply #57 on: July 14, 2018, 04:07:52 PM »
Very low is perfect.  :)
Compare, compare, compare