What is PIPS?

Amino,
There is no thickness values in the Cryosphere Today maps you claim proves the volumes from PIOMAS are wrong. You are missing one critical dimension to eyeball, let alone calculate an ice volume.

stevengoddard says:
June 13, 2010 at 10:55 am
After decades of the Navy using PIPS, the only verification I really care about is what happens to the ice this summer. If the ice is thicker as PIPS shows, it will survive. If the ice is thin, it will melt away.
The outcome of this year summer for the Arctic ice is captivating. So is the issue of the correctness of the computer models discussed here.
Only that I would like to read a few words about what variables are biggest factors in the whole issue of possibility of the Arctic sea ice disappearance this year. I think I am not alone in my wishes.
Why everyone is arming himself with concentration/extent/thickness plots putting aside other factors? What about the air temperature within Arctic region? Gaping day by day into Arctic ice extent graphs is boring. 😉 It does not provide intellectual incentive for the readers’ brains. 😉
Perhaps it might have been a good subject for a next post?
Best regards

Przemysław Pawełczyk says:
“Why everyone is arming himself with concentration/extent/thickness plots putting aside other factors? What about the air temperature within Arctic region? Gaping day by day into Arctic ice extent graphs is boring. 😉 It does not provide intellectual incentive for the readers’ brains. ;-)”
____________________
Indeed! For your reading pleasure, you may want to check out the CICE website at:
http://oceans11.lanl.gov/trac/CICE
And have a look at their documentation and user’s guide. Among the little gems there you’ll find a whole list of the factors that CICE uses to calculate sea ice behavior. Among those are:
Freezing/melting potential
Wind velocity
Sea surface temperature
Specific humidity
Sea surface salinity
Air density
Sea surface slope
Air potential temperature
Surface ocean currents
Air temperature
Shortwave radiation (4 bands)
Incoming longwave radiation
Rainfall rate
Snowfall rate
Wind stress
Penetrating shortwave
Sensible heat flux
Fresh water flux
Latent heat flux
Net heat flux to ocean
Outgoing longwave
Salt flux
Evaporated wate
Ice-ocean stress
Surface albedo (4 bands)
Surface temperature
Ice fraction
Humidity (diagnostic)
Absorbed shortwave (diagnostic)
Remember that PIOMAS and PIPS 3.0 are tied into CICE, but alas, our dear old friend PIPS 2.0 is not…

I said in my first comment in this post that some commentors would go on and on about PIOMAS.
http://wattsupwiththat.com/2010/06/12/what-is-pips/#comment-408118
But now I see that it is worse than I thought!

Amino Acids in Meteorites says:
June 13, 2010 at 1:53 pm
I said in my first comment in this post that some commentors would go on and on about PIOMAS.
http://wattsupwiththat.com/2010/06/12/what-is-pips/#comment-408118
But now I see that it is worse than I thought!
_______________
At least PIOMAS has a very specific prediction, down to a map of the extent and concentration of Arctic Sea ice for this September:
http://psc.apl.washington.edu/zhang/IDAO/seasonal_outlook.html
Specific and verifiable prediction is the hallmark of good science, and so, either we’ll know PIOMAS is a good scientific model or not come September. I don’t see any other model (available to the public) making such a prediction…so why shouldn’t we go on and on about PIOMAS? It is specific, verifiable, and has no equal…

Please correct my 2nd “Z” in prior post

R. Gates says:
June 13, 2010 at 1:48 pm
And have a look at their documentation and user’s guide. Among the little gems there you’ll find a whole list of the factors that CICE uses to calculate sea ice behavior. Among those are:
Freezing/melting potential
Wind velocity
Sea surface temperature
Specific humidity
Sea surface salinity
Air density
Sea surface slope
Air potential temperature
Surface ocean currents
Air temperature
Shortwave radiation (4 bands)
Incoming longwave radiation
Rainfall rate
Snowfall rate
Wind stress
Penetrating shortwave
Sensible heat flux
Fresh water flux
Latent heat flux
Net heat flux to ocean
Outgoing longwave
Salt flux
Evaporated wate
Ice-ocean stress
Surface albedo (4 bands)
Surface temperature
Ice fraction
Humidity (diagnostic)
Absorbed shortwave (diagnostic)
Remember that PIOMAS and PIPS 3.0 are tied into CICE, but alas, our dear old friend PIPS 2.0 is not…
You think this complexity and number of factors used is a strength? It is not, it is a weakness. It is linked to the issue of inductiveness or deductiveness in scientific method and particularly the important proposals on this by Carl Popper.
The philosopher of science Carl Popper argued for science to be deductive, based on economic interpretation of measured facts, which can readily be experimentally falsified. Popper rejected “inductive” reasoning in which chains of assumptions are built up. However the age of cheap computing power has caused researchers to fall into the alluring trap of inductive “science”, in which assumptions and hypotheses are built up on eachother like a house of cards.
I like to think of it in terms of the length of the paths that one draws between observation and conclusion. Short and economic (“parsimonious”) = deductive; long and convoluted involving multiple serial assumptions = inductive.
Two teams of scientists, team inductive and team deductive, were given a task: design a speedometer for a car – a device for measuring and displaying the speed that a car is travelling.
So team inductive got to work. This team included a fair number of physicists with computational and modelling skills. It became immediately clear to them that this was a task requiring the procesing of multiple factors all impacting on speed: what was the energy and force driving the car forward, what was the origin of this energy? Chemical and thermodynamic energy from the combustion of fuel needed to be carefully evaluated and modelled. What was the efficiency of this conversion from chemical to kinetic energy – how much was lost in the inefficiency of the motor? Several team members were assigned to modelling these processes. How much energy was lost as friction and heat through the gas exhaust? Simulation of the turbulent fluid flow and associated heat fluxes along the exhaust pipe was clearly called for.
Then of course there were hours of immense fun to be had modelling and evaluating the fluid friction of the air passing over the car. This of course was modified by the dynamics of the air itself – what was the prevailing wind direction? Then of course there was the friction between the tyre and the road. An important input here was the curvature of path of the travelling car and associated sideways force and geometric distortion of the tyre, adding heat to the tyre affecting its friction, and whether or not this induced tyre to road shear and slippage, each in turn calling for further modelling inputs. Of course tyre dynamics were temperature-related so local climate was again a critical factor and another useful variable.
So it became clear to team inductive that to have any hope whatsoever of measuring speed in a credible way, to give an output that would be accepted by internationally recogonised car speed scientists associated with the high profile journals and societies, that a large number of data inputs were needed: chemical measurement probes in the fuel tank to asses the fuel chemical potential energy; probes within the ignition chamber to assess on a millisecond basis pressures and temperatures to illucidate combustion energy. Then multiple sensors were required in the exhaust pipe to provide input for fluid flow modelling of the exhaust gasses. Sensors were also required at many locations on the car’s surface to assess airflow and boundary layer turbulence, as the exact location of the laminar-turbulent transition was a key factor in getting the drag models to work reliably. Sensors were needed within the tyres also. Other factors and associated sensor inputs were also identified and subject to in-depth research and computer simulation.
Thus at the end of the day it was deemed impossible to prove that the “speed” of the car that one measured was correct or not, or that the car was in fact moving at all, or whether it was even in contact with the road, and indeed what it was exactly that one meant by the concept of a “road”. The best one could hope for was an accumulation of evidence on the subject.
Then team deductive got to work. They measured the circumferance of the wheels. And set up a sensor to measure the rate of rotation of the wheels. From this they got a speedometer.

phlogiston says:
June 13, 2010 at 3:15 pm
Phlogiston, elocution is your second strongest point! The first one is the ability to prove why the f[…] cars are so costly nowadays! 🙂
The same you could say also about weather forecast models. Do you really think we should get back to the times where wind direction was detected by salivated finger (pointer one not the middle one)? 😉
Regards

Przemysław Pawełczyk says:
“Gentlemen,
I ask only one question, coming from my curiosity not acridity – how ice volume is validated on constant basis?”
I dare say with the US navy, the sea wolf or LA class submersible sonar platform would be an accurate, available method… But they dont like sharing where they are!
I think the toed radar arrays are quite accurate also(behind low flying aircraft) But thats more to do with civilian efforts. Lasers have issues with snow cover. Being the navy, id put my money on subs.

Amino Acids in Meteorites says:
And no, from looking at the ill comments of some here
correction, should be
from looking at the ill conceived comments of some here

Steve and others supporting PIPS2.0
Here are a couple more links that may be of interest to you on the accuracy of PIPS2.0
First one is a presentation by Pablo Clemente of the National Ice Center
http://www.google.com/url?sa=t&source=web&cd=3&ved=0CCAQFjAC&url=http%3A%2F%2Fwww.star.nesdis.noaa.gov%2Fstar%2Fdocuments%2Fseminardocs%2FClementeColon_2007-02-20.pdf&ei=H8UVTKewDJDMMqvZ1acL&usg=AFQjCNGyTyPpuXCiESzgPIejIeeODTet-Q
This presentation shows comparisons between PIPS3.0, PIPS2.0 and SSM/I sea ice concentrations and shows how poorly the PIPS2.0 model does. It also discusses the improved inputs used in the PIPS3.0 model.
A paper by Michael Van Woert (from NOAA/NESDIS) and co-authored by Walt Meier who works at NSIDC and who previously worked at NIC also discusses accuracy issues of PIPS2.0 (http://journals.ametsoc.org/doi/full/10.1175/1520-0426%282004%29021%3C0944%3AFVOTPI%3E2.0.CO%3B2).
Even Posey discusses in April that more recently the Naval Research Laboratory (NRL) has been using an advanced version of a prediction model relying on HYCOM and CICE as R. Gates has mentioned. I also found a PIPS4.0 in a presentation by Eric Chassignet (http://www.google.com/url?sa=t&source=web&cd=6&ved=0CDEQFjAF&url=http%3A%2F%2Fwww.clivar.org%2Forganization%2Fwgomd%2Fwgomd5%2Fgfdl04%2FChassignet.ppt&ei=H8UVTKewDJDMMqvZ1acL&usg=AFQjCNGb7FTyrFjswy9O3Tzzov7Xnz_gUg).
I think it’s becoming clearer that PIPS2.0 is outdated. But again, I think a real comparison between ICESat data from R. Kwok and the PIPS2.0 data that Steve is using would be worthwhile. Steve, I do believe this is something you should do (i.e. validate your approach), even if it’s just showing two images side-by-side.

From: R. Gates on June 13, 2010 at 2:20 pm (emphasis added)

At least PIOMAS has a very specific prediction, down to a map of the extent and concentration of Arctic Sea ice for this September:
http://psc.apl.washington.edu/zhang/IDAO/seasonal_outlook.html
Specific and verifiable prediction is the hallmark of good science, and so, either we’ll know PIOMAS is a good scientific model or not come September. I don’t see any other model (available to the public) making such a prediction…so why shouldn’t we go on and on about PIOMAS? It is specific, verifiable, and has no equal…

Analysis of Zhang (PIOMAS) Predicative Ability Concerning September Minimum Arctic Sea Ice Extent
Section 1: Data with sources and observations
From SEARCH, Study of Environmental Arctic Change, Sea Ice Outlook. Predictions of the September Arctic sea ice minimum extent from many experts are assembled for these Outlooks, with Zhang (PIOMAS) contributing for 2008 and 2009, which are the completed years posted on the site. Monthly predictions are made, the reports of them are available.
2008 wrap-up (Full Report tab):

How Well Did the Outlook Based on May and July Data Compare to September Observations?
The May Outlook responses had a median value of 4.2 million square kilometers, with a median of 4.9 million square kilometers for the July Outlook (Fig. 5); these compare to a 2008 September observed value of 4.7 million square kilometers. Both Outlook estimates are well below the 1979-2000 mean and below the linear trend line.
(…)
Summaries of Statements Provided by Outlook Participants on the Summer 2008 Sea Ice Season:
Full Contributions can be accessed at the bottom of this report.
(…)
Zhang. Preconditioning and anomalous meteorological forcing are all important. The ensemble predictions demonstrate certain skills in estimating both ice extent and its shape, in comparison with observations. However, they do not show improved accuracy with decreasing prediction range. Assimilation of satellite ice concentration improves the predictions. Although summer 2008 has slightly larger sea ice extent than summer 2007, the ice is thinner in a large area in the western Arctic, according to PIOMAS model/data assimilation (http://psc.apl.washington.edu/zhang/IDAO/summer2007_2008_arctic_seaice.gif). Summer 2008, though having a larger ice extent, has less ice than summer 2007. Mean September 2008 ice volume is 5900 cubic kilometers, actually lower than September 2007 ice volume at 6000 cubic kilometers. So we would probably have another summer ice low next year, if not setting a new record. [19]

Note: The use of the term “ensemble” was mentioned here: “Two groups (Kauker, et al., and Zhang) ran sea ice models with an ensemble (many years) of summer weather conditions from previous years.” They run through the models the weather of past years with recent ice conditions as a starting point, then “average” the results together for the ensemble prediction.
From linked pdf (that number 19):

The September 2008 sea ice extent was predicted to be 4.5/4.6/5.1 million square km in May/June/July, based on the ensemble median of the 7 ensemble members. The NSIDC reported that the minimum summer 2008 ice extent is 4.52 million square km.

The 2008 Outlook summary report figure of 4.7 million km^2 appears to be the September mean while Zhang’s figure of 4.52 according to NSIDC appears to be the lowest extent in September.
From a 2008 outlook by Zhang found elsewhere, last updated August 1 2008:

The September 2008 arctic sea ice extent is predicted to be 5.1 million square kilometers, significantly higher than September 2007 at 4.3 million square kilometers (see the figure below).

2009 Outlook wrap-up (Pan-Arctic tab):

2009 Minimum Sea Ice Extent
The projections of the Sea Ice Outlook groups for the September 2009 mean minimum ice extent had a median value of 4.7 million square kilometers based on May data and 4.6 million square kilometers based on June data with a range of 4.2-5.2 million square kilometers (Figure 1). These estimates are below the observed value of 5.36 million square kilometers as provided by Walt Meier, NSIDC.(…)
…
Summaries of Statements Provided by Outlook Participants on the Summer 2009 Sea Ice Season:
(…)
Zhang: With the starting date of prediction being July 1st, the area of the predicted ice extent is smaller than that of the observed but the shape of the predicted ice extent is close to that of the observed. Thus the prediction is perhaps within the error tolerance in many areas, but not in other areas. The model has the tendency to under-predict September ice extent because the SAT over a large area in the Arctic Ocean is higher than the recent (2002–2008) climate during January through June. This may explain why many of the participating groups also underestimate the September ice extent when the ice conditions during January through June were heavily taken into account. When the ice conditions in spring are used as an indicator of what may happen several months later in September, there is a need to significantly raise the error bar considering the significant variability of weather and climate.

From Zhang’s summary report (link at his name):

Text:
(…) The September 2009 mean sea ice extent is predicted to be 4.5 or 5.3 million square km when the ensemble prediction starts on the first day of July or September. The NSIDC reported that the September 2009 mean ice extent is 5.4 million square km, based on satellite observations. (…)
Figure 1 caption:
(…) The ensemble median is considered to have a 50% probability of occurrence and the ensemble median ice extent for September 2009 is 4.5 million square kilometers when the prediction starts on 7/1 and 5.3 million square kilometers when the prediction starts on 9/1.

Important note: 2009 wrap-up mentions, of the monthly predictions, those based on May and June data, which corresponds to the predictions issued in June and July. Zhang mentions July and September. From the 2009 wrap-up comes these graphs showing individual predictions. Zhang predicted 4.2 million km^2 in June, 4.5 in July.
From the monthly reports, for August (Full Report tab) we find Zhang predicts 4.5 million km^2, no change from July. (Note: Figure 1 is incorrectly marked, indicating Zhang as “Heuristic” and Wang as “Modeling.” From the text it is clear that should be reversed. Both forecast 4.5 million km^2.)
The 2009 Minimum Announcement gives a 5.1 million km^2 minimum daily extent on September 12.
Section 2: Summary of Data for Zhang (PIOMAS) Predictions of September minimum extent
2008:
May…….. 4.5 million km^2
June……. 4.6
July…….. 5.1
August… 5.1
Final…… 4.7 million km^2
2009:
June………… 4.2 million km^2
July…………. 4.5
August………4.5
September.. 5.3
Final…………5.36 million km^2
Also: In his 2008 Outlook summary, Zhang predicted “So we would probably have another summer ice low next year, if not setting a new record.” However just before he was discussing ice volume so it is unclear which low was predicted, extent, volume, or both.
Section 3: Conclusions
In 2008 Zhang (PIOMAS) started out with a prediction near the final total, but as new data during the summer was used for the ensemble the predictions shifted far higher, then late season weather caused a much lower minimum than forecast in July and August. In 2009 Zhang (PIOMAS) started out with an even lower prediction than the 2008 minimum, stayed that low throughout the summer. Then at the beginning of September with less than two weeks until the daily minimum extent of the year and a greatly reduced melt rate, a dramatic revision upwards brought the prediction very close to the final figure.
Thus for these two years, the ability of Zhang (PIOMAS) to accurately predict the September minimum extent is rather low. In 2008 the prediction overshot due to unforeseen late season weather. For 2009, Zhang criticized his own model in his individual summary, citing a tendency to under-predict under certain conditions.
Section 4: Observations Concerning the Predictions
Of these publicly available predictions, from the July 2008 predictions the closest to the final figure was Pokrovsky at 4.6 million km^2, “Method: Weather and sea ice trends.” The closest prediction by a model similar to PIOMAS was from “(AWI)—Kauker, Gerdes, and Karcher” which forecast 4.5 million km^2, “Method: Numerical Model-Ensemble forced with NCEP/NCAR reanalysis data.” Note: “With a probability of 80% the minimum ice extent in 2008 will be in the range between 4.3 and 4.7 million km2.”
From the July 2009 predictions, the best prediction was the highest, from Morison and Untersteiner at 5.2 million km^2, “Heuristic.” Of those simply identified as “Model” the best was Nguyen et al at 5.0 million km^2. Morison and Untersteiner did not contribute to the June and August reports, September did not have individual predictions. For those three months, out of all the predictions, theirs was the highest and eventually the closest.
As to predicting the September minimum extent, from July 2009 in the “Key Statements from Individual Outlooks” section, from Morison and Untersteiner comes this part at the end:

Our estimate is based on the same improvement from 2008 to 2009 as occurred going from 2007 to 2008 plus 200,000 square kilometers based on gut feeling. Note: we purposely didn’t look at last month’s estimates to avoid biasing our estimate.

Note that extra 0.2 million km^2 made their prediction the best one.
For 2009, gut feeling beat modeling. Gut feeling beat PIOMAS.