By Steven Goddard
There have been a number of inaccurate claims made by commentors about Navy PIPS2 ice thickness maps. These claims have been along the lines of :
- PIPS isn’t used by the Navy any more, because it isn’t accurate enough
- PIPS maps over-represent ice because they don’t see areas of open water
- PIPS maps don’t take into account ice concentration. They consider the ice to be 100% concentrated
- PIPS is just a model. It isn’t an accurate representation of the ice.
The US Navy clearly refutes these claims –
04/06/2010 – Pamela Posey
The Polar Ice Prediction System (PIPS 2.0) is the current U.S. Navy’s operational ice forecasting system.
PIPS 2.0 forecasts ice conditions in the northern hemisphere with a horizontal grid resolution ranging from 17-33 km depending on the grid location. The system couples the Hibler ice model to the Cox ocean model and exchanges information by interfacing the top level of the ocean model with the ice model. Ice concentration fields derived from the Special Sensor Microwave/Imager (SSM/I) are assimilated into the PIPS 2.0 system along the ice edge. The system produces a 120-hour forecast of ice fields which are sent to the National Ice Center (NIC) to be used in their daily ice forecasts.
The Navy also refutes the claims that they don’t correct for concentration :
The model-derived ice thickness field and the ocean surface temperature field are then adjusted to be consistent with the concentration data.
These models are required to go through rigorous validation studies to prove their capability to produce accurate short term variability. Data assimilation plays a major role in the accuracy of these forecasts. Once operational, continuous quality control and evaluation of the products may be used to upgrade the system and improve forecast accuracy.
The video below for June 10, 2010 shows that PIPS maps accurately reproduce current ice conditions. It overlays the UIUC ice concentration map on the PIPS map.
Map sources :
http://arctic.atmos.uiuc.edu/cryosphere/NEWIMAGES/arctic.seaice.color.000.png
http://www7320.nrlssc.navy.mil/pips2/archive/pips2_thick/2010/pips2_thick.2010061100.gif
As you can see, areas of open water are shown as open water, and areas of low concentration also have lower thicknesses. The incorrect claims repeated over and over and over again by FUDsters just don’t hold any water.
PIPS2 is not perfect. Here is what the Navy says :
A recent study by a group of scientists from the NIC and NOAA (Van Woert et al., 2001), showed that although the PIPS 2.0 forecasts (48-hour) were better than persistence on average, there were still substantial biases in its prediction of the growth and decay of sea ice in the marginal ice zone. PIPS 2.0 often over-pre- dicts the amount of ice in the Barents Sea and therefore often places the ice edge too far south. In contrast, PIPS 2.0 often under-predicts the ice extent in the Labrador Sea and Hudson Bay.
This doesn’t affect my calculations, because I am only measuring regions which normally contain significant amounts of late summer ice. Also, my comparisons are relative year over year comparisons. The absolute values of ice thickness are not important to my conclusions.
Conclusion : PIPS2 maps is the best available and are used by the US Navy. They are quite accurate and they do account for ice concentration. No doubt, some commentors will continue to ignore the facts, and post instead what suits their agenda.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Peter Plail says:
June 13, 2010 at 1:15 am
R Gates
My common sense tells me that if there is such a strategic advantage to using PIPS 3.0 over V2.0 then you wouldn’t then publish the User Manual which seems to contain an explanation of all the logic and substantial parts of the code and even debugging advice.
________________
By Ms. Posey’s own admission, NAVOCEANO has been using PIPS 3.0 on a daily basis. We can only speculate as to why it is not released yet for public consumption or now is the publicly aknowledged standard. The links I gave showed one example of the differences between PIPS 3.0 and 2.0, but even that difference, as great as it was, had probably been down-scaled from the full resolution version– we know it specifically said “Unclassifed”. It is common practice in the military to release research and public versions of products that are much higher quality (GPS is a good example). Perhaps the failure of IceSat last October has delayed the full official validation of PIPS 3.0, and they’re waiting for CryoSat 2 data for further validation and testing (we know PIOMAS is). Or perhpas they’ve not completed a down-scaled version of PIPS 3.0 yet to release for research and for the general public. There are any number of reasons why, but regardless, and I’m quite confident, especially based on Ms. Posey’s statements of PIPS 3.0 being used, that for its daily operations, the Navy is using a much better version and more robust and more sophisticated sea ice model than the publicly released PIPS 2.0 that requires Internet Explorer 4.0 at a screen res of 800 by 600.
Again, anyone who believes that the PIOMAS large record anomaly is real, must also believe that we are headed for a substantial record minimum this summer. All of the top experts are forecasting a hot summer in the Arctic. Also, the current extent is lowest on record for the date.
Record thin ice, low extent and hot temperatures has got to mean a record low ice extent. Right? If people really believe in PIOMAS, they should be expecting a huge melt this summer.
For those of us who line in NW Europe, the thought of an advancing Arctic Ocean ice cap is a scary one, as it means the Gulf Stream and/or the Atlantic Ocean Conveyor is weakening. This would threaten us with a Labrador or Kamchatka type of climate.
As for PIPS 2.0 or 3.0 – there must be someone in the tens of thousands who daily access this website, who can tell us if PIPS 3.0 is actually in use or is still in the development stage.
No matter how much I bang my head against a wall, I cannot figure out why forecasts of Arctic ice thicknesses could be of such strategic military importance and therefore why PIPS 3.0 – if it is completed and in use – would not be made accessible to the general public by the US Navy.
However, September will tell us everything – but I would anticipate the Arctic Ocean in September next year will have much more ice than September 2010. By that time, the latest El Nino influence should have completely disappeared.
In closing, a flip through the alarmist websites shows these guys are having a field day over PIPS 2.0 versus PIPS 3.0 – if it exists. Why is it that alarmist websites have to be so abusive?
Peter Plail says:
June 13, 2010 at 1:15 am
R Gates
My common sense tells me that if there is such a strategic advantage to using PIPS 3.0 over V2.0 then you wouldn’t then publish the User Manual which seems to contain an explanation of all the logic and substantial parts of the code and even debugging advice.
Common sense tells me that you wouldn’t continue to publish forecasts based on an obsolete system if the people you were trying to mislead (by using data produced by an obsolete system) already knew that the system was obsolete.
Or alternatively the new system is fine for research use but isn’t capable of fast enough operation to provide forecasts? Posey addresses this in the paper (2002) cited by Steve.
“Coupled ice-ocean models, such as those used by
PIPS 2.0 and PIPS 3.0, are most often applied in academic
studies of global climate issues such as climate
change. In these studies the models are used for
decadal or longer simulations and focus on long term
changes in the overall pattern of ice thickness and ice
extent. This application of the coupled model requires
access to powerful computers and large quantities of
computer time. However, coupled ice-ocean models
used for forecasting face a different set of requirements.
As forecasts are concerned with sea ice variability
on much shorter time scales, the models must be
designed to produce the most accurate daily changes
in ice concentration, ice edge location and ice motion.
These models are required to go through rigorous
validation studies to prove their capability to produce
accurate short term variability. Data assimilation plays
a major role in the accuracy of these forecasts. Once
operational, continuous quality control and evaluation
of the products may be used to upgrade the system and
improve forecast accuracy. In addition, forecast
systems are limited by the amount of computer
resources available for each forecast as they compete
with other forecast models each day. Therefore the
combined model/assimilation system must be
designed to “fit” within these limitations. This places
restrictions on the “size” or grid resolution of the models
as well as the complexity of the model parameterizations
and the data assimilation techniques. Each of
these issues must be taken into account when developing
a new forecast system.”
stevengoddard says:
June 13, 2010 at 6:27 am
Again, anyone who believes that the PIOMAS large record anomaly is real, must also believe that we are headed for a substantial record minimum this summer. All of the top experts are forecasting a hot summer in the Arctic. Also, the current extent is lowest on record for the date.
Record thin ice, low extent and hot temperatures has got to mean a record low ice extent. Right? If people really believe in PIOMAS, they should be expecting a huge melt this summer.
———————–
Steve, I actually disagree with your statement. Record low ice volume does not necessarily mean a record low sea ice extent, though I would say it likely means continuation of low extents. This is because like many scientists have been saying (even one who has posted on here) the summer weather patterns are very important. Sure, warm temperatures will help, but what about the winds? Those are also very important as we saw in 2007. So since we can’t predict the weather, seems false to claim that those who believe the PIOMAS estimates are more accurate than PIPS2.0 also believe a new record minimum will be set this year.
Why not do the comparison I suggested? Seems that would help clear up this model vs model discussion.
jeff brown
Feel free to do whatever comparison you want. I’m not the one questioning the PIPS data.
However, the PIOMAS graph shows a huge negative anomaly, 50% greater than this date in 2007. NSIDC says that May saw a record decline and we are at a record low extent anomaly. People who trust PIOMAS have to be willing to back the logical consequences of their belief system.
Peter Miller
Let them have their fun for now. September is not very far off, and we will see if the super-secret PIPS3 database has the hidden secrets they believe in.
Apparently they must transport the PIPS3 data in convoys of government black vans with tinted windows, since it isn’t available on the Internet. ;^)
Phil. says:
June 12, 2010 at 10:28 pm
the PIOMAS model which has actually been calibrated against the satellite measurements of thickness
The PIOMAS graph is obviously wrong.
Aren’t you a professor Phil? Shouldn’t you already see PIOMAS is wrong? Or are you a different Phil and not the professor? Or, if you are that Phil, is your work as a professor connected to ‘global warming’ and that explains why you would not yourself show the clear flaw of the PIOMAS graph?
jeff brown says:
June 12, 2010 at 10:30 pm
Then we can finally move on to another topic of discussion.
It seems (IMO) the real topic here, the subtext, is Arctic ice has increased since 2008 and that is obvious from the data but the PIOMAS graph says Arctic ice has decreased since 2008. It’s easy to see from looking at graphs and images there has been an increase in North Pole ice since 2008. It is a bit bizarre that anyone would continue to promote the PIOMAS graph.
I don’t think discussion on Arctic ice is going away any time soon. Especially not until the middle of September when Arctic ice begins its yearly increase and we all see, again, how wrong PIOMAS is.
stevengoddard says:
June 13, 2010 at 6:27 am
Again, anyone who believes that the PIOMAS large record anomaly is real, must also believe that we are headed for a substantial record minimum this summer. All of the top experts are forecasting a hot summer in the Arctic. Also, the current extent is lowest on record for the date.
Record thin ice, low extent and hot temperatures has got to mean a record low ice extent. Right? If people really believe in PIOMAS, they should be expecting a huge melt this summer.
…………………………………………………………………………………………………………………….
What you are saying is obvious. But I see some doing a tap dance around the obvious.
stevengoddard says:
June 13, 2010 at 6:27 am
Record thin ice, low extent and hot temperatures has got to mean a record low ice extent. Right? If people really believe in PIOMAS, they should be expecting a huge melt this summer.
That could be why they are pushing the PIOMAS graph, because they want to convince people the ‘death spiral’ is really happening.
From reading the comments of some the real data is not mattering to them now. Even if the PIOMAS graph is shown to be absolutely wrong by the DATA in September these same people will still probably not care what the data is showing, will still believe the PIOMAS graph, and will find some way to rationalize their belief.
jeff brown says:
June 13, 2010 at 8:21 am
stevengoddard says:
June 13, 2010 at 6:27 am
Why not do the comparison I suggested? Seems that would help clear up this model vs model discussion.
You’d think so which is why I and others also suggested it. However Steve has consistently avoided addressing the suggestion, instead he comes up with posts like the following:
“stevengoddard says:
June 12, 2010 at 1:58 pm
Przemysław
I have been trying to have an intelligent discussion about the data, but the signal to noise level in the discussion forum of the last few articles has been spectacularly low. A few people seem to be intentionally and repeatedly disrupting the discussion with misinformation.”
So apparently suggesting a validation step for his calculation is ” disrupting the discussion with misinformation”.
stevengoddard says:
June 13, 2010 at 8:39 am
Peter Miller
Let them have their fun for now. September is not very far off, and we will see if the super-secret PIPS3 database has the hidden secrets they believe in.
Apparently they must transport the PIPS3 data in convoys of government black vans with tinted windows, since it isn’t available on the Internet. ;^)
________________
Each van carries one pixel of data, and the final van carries the code key that puts all the pixels in the right order. This not only helps to keep PIPS 3.0 secret, but also keeps General Motors happy as there are tens of thousands of vans involved in this secret plan with the Presidential directive that it must involve both American made pixels and American made vans.
Peter Miller said:
“However, September will tell us everything – but I would anticipate the Arctic Ocean in September next year will have much more ice than September 2010. By that time, the latest El Nino influence should have completely disappeared…”
_________
September will not tell us everything, unfortunately. We’ll know the general condition of the ice was going into the melt season, we’ll know what the summer melt season (i.e. the Arctic Weather of summer 2010) was like, and that’s about it. It’s funny that Steve now is telling us all what the PIOMAS model MUST be predicting, i.e. a record summer low, when the actual creators of that model, and those who know it best I would presume, are telling us 4.7 million sq. km., which is lower than last year, but not as low as 2007.
Proof that PIPS 3.0 exists, which is not in dispute, and that it is still in the development and validation stage, not deployed:
PIPS 3.0 site. Note how the pages have Last Modified dates in 2003.
PIPS 2.0 site, which confirms 2.0 “…is the operational model run by the Naval Oceanographic Office (NAVO) for sea ice forecasting.” There is found a pop-up window link to “Info on the future PIPS 3.0”, “The PIPS 3.0 is presently going through its final development.” Note: As previously mentioned, within the HTML code of the pop-up there’s a commented-out part, “–and will begin its adaptation for operational use in 2003 with a scheduled transition into operational use in early 2004.” This looks like an old edit. As seen at the 3.0 site, work was done in 2003, perhaps then it did look like it’d be deployed in 2004. The form of edit preserves the original content.
There is a released unclassified users manual for 3.0. R. Gates previously provided a link to a purchasable copy from a private reseller of unclassified U.S. federal government reports. Later he provided a link to a free downloadable copy. As clearly stated on the first page of the November 5, 2008 manual, “Approved for public release; distribution is unlimited.” Yes, this is a freely-available manual from off a military server (dot mil) for software that allegedly produces classified data whose existence is hidden outside the US military. No mention is made of 3.0 being deployed.
R. Gates provided a link to an unclassified pdf file. Examining the document properties reveals it was originally a MS PowerPoint presentation, has File Created and Modified dates (presumably when converted to pdf) of “Mon 12 Dec 2005 10:40:57 AM EST.” Within you’ll find an outlook on works-in-progress, as evidenced by wording on pages 15, 28, 31 and others, which could date the original presentation to 2005 or perhaps as old as 2004. Page 33 has a chart that show the first version of 3.0 (G-NCOM) was to have completed Development and Transition in 2005, then in 2006 it would replace 2.0. The second version of 3.0 (G-HYCOM) was to have D&T in 2008, then both versions would be running in 2009, with a switchover to the HYCOM version only in 2010.
R. Gates also supplied this nice map (http://www.oc.nps.edu/~pips3/pips.gif) showing the bathymetry (water depth) data inputted to 3.0. It may be found at the 3.0 site page “Model Domain and Bathymetry” (click on small image at top).
http://www.nrl.navy.mil/content_images/09_Ocean_Posey.pdf is mentioned above, which is a short article from a journal (doesn’t look like a paper) from 2009. It says: “PIPS 2.0 has been producing operational Arctic forecasts for the Navy since 1996.” It also says “PIPS 3.0 is now producing daily 48-hour forecasts at the Naval Oceanographic Office (NAVOCEANO) and will have competed its validation tests by mid-2009.” So 2.0 is the current system, 3.0 is in validation testing. Note the article uses 2.0 images throughout, not 3.0. (And it’s a shame that on the last page in Acknowledgments the URL for an archive of 2.0 images is unusable.) Of special note is with that 3.0 mention, she says NCOM is used. In the previously-mentioned PowerPoint presentation, the G-NCOM version was projected to be deployed in 2006, with G-HYCOM deployed in 2009. So as of this 2009 document, the HYCOM version is nowhere to be seen and NCOM is being finished up.
Then we come to the 2010 document linked to be Steven Goddard. PIPS 3.0 is not mentioned. It does say:
From the “Model Domain and Bathymetry” page it is noted 3.0 has a 1/12° horizontal resolution (and the vertical resolution link is bad).
For new information, using the new acronym (ACNFS) I found this abstract of a presentation at the European Geosciences Union (EGU) General Assembly 2010 conference titled “Validation of the 1/12° Arctic Cap Nowcast/Forecast System (ACNFS).” (No, I don’t know how to get the full presentation and/or paper.) From the abstract:
Summation:
PIPS 3.0 was scheduled for deployment long ago, perhaps as early as 2004. Later on there were two versions that were planned to be deployed by now, G-NCOM in 2006 with G-HYCOM in 2009. By 2009 G-NCOM was finishing validation testing, G-HYCOM not mentioned. As of 2010, the planned PIPS 2.0 upgrade is to ACNFS, a HYCOM/CICE system, which sounds amazingly like 3.0 G-HYCOM.
Standards change, hardware changes, better software becomes available. PIPS 3.0 does exist, somewhere. But as can be seen by its history of missed deployment dates, its development did not go as planned. Now ACNFS is coming, the newer and better replacement of PIPS 2.0. It appears, given this new info, that PIPS 3.0 may remain un-deployed, forever locked in a development and validation phase. 3.0 G-NCOM, the first version, didn’t show up, and G-HYCOM has now been supplanted by ACNFS. I have a good guess what “non-assimilative HYCOM/CICE system” was used for setting up ACNFS.
Sure looks like if anything does get deployed as PIPS 3.0, it will really be the new ACNFS with the 3.0 label (because they already told everyone that was the new version) rather than what was 3.0 when they started the project way back then.
(BTW, if you want to see what a great future ice thickness model looks like, check out the TOPAZ site.)
And now for something completely different… 🙂
===================
Information on PIPS 2.0
Forecast Verification of the Polar Ice Prediction System (PIPS) Sea Ice Concentration Fields
American Meteorological Society (AMS) journal. Journal of Atmospheric and Oceanic Technology 2004; 21: 944-957. By the date (submitted & accepted 2003) it involves 2.0. Long winded, with equations, many graphs, might be interesting if you’re into that sort of thing.
I shall now finally hit Post Comment, knowing the large number of links will likely automatically consign this to the spam filter, and will pray the kindly moderators safely rescue it. 😉
HR says:
June 13, 2010 at 1:02 am
Thanks for the link. According to this report dated Jan-15-2010 with Pamela Posey as an author PIPS 2.0 is the current model used and 3.0 is still in development. A quote from the report’s conclusion:
“Currently, the
operational coupled sea ice forecasting system run daily at
NAVOCEANO is the Polar Ice Prediction System (PIPS 2.0)
and has a resolution of-27 km. The HYCOM/NCODA/CICE
system is undergoing final validation testing and has a
resolution of -3.5 km in the polar region.”
So R. Gates thanks for misleading, misrepresenting, or fantasizing at the least and lying at the most to all of us about the current state and usage of PIPS 3.0. I’m ready to accept your apology and admission of at least one of the above.
R. Gates says:
June 13, 2010 at 10:02 am
September will not tell us everything, unfortunately.
Hedging your bets already?
Oh, by the way, thanks for another looong winded comment.
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.
Tom P says:
June 12, 2010 at 11:03 pm
Amino,
You say “The reason is is that the PIOMAS graph claims that Arctic ice continues to decrease since 2008. But real data shows it is increasing.”
……………………………………………………………………………………………………………….
let’s check it again
PIOMAS from 5/30/10, showing less ice on 5/30/10 than on 5/30/08
http://psc.apl.washington.edu/ArcticSeaiceVolume/images/BPIOMASIceVolumeAnomalyCurrent.png
Cryosphere Today showing more ice on 5/30/10 than on 5/30/08
http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=05&fd=30&fy=2008&sm=05&sd=30&sy=2010
So, as anyone can see PIOMAS is wrong. It’s simple. Nothing complicated.
PIOMAS is a model. Cryosphere Today is data.
………………………………………………………………………………………………………………….
Freeman Dyson on how and why modellors can be wrong:
Jeff Brown,
Have a look at the plots I’ve posted on this and previous arctic threads. I think they will answer your questions.
Freeman Dyson understands the math of modelling well. Here is some of his math, and other, background:
graduate University of Cambridge, served in WW II where he developed ‘operational research’, friend of Einstein, Fellow of the Royal Society, Honorary Fellow of Trinity College, University of Cambridge, Professor of physics at Cornell University 1951–1953, Professor of physics, Institute for Advanced Study, Princeton 1953 –1994 and Professor Emeritus 1994 to present, worked with the Institute for Energy Analysis on climate studies, current president of the Space Studies Institute, Member U.S. National Academy of Sciences, Danny Heineman Prize, Lorentz Medal in 1966, Max Planck medal in 1969, American Institute of Physics, Phi Beta Kappa Award in Science for “Infinite in All Directions”, Oersted Medal awarded by the American Association of Physics Teachers, 1991, Enrico Fermi Award given by U.S. Department of Energy, 1995, Lewis Thomas Prize, honoring the Scientist as Poet, awarded by Rockefeller University, 1996, “Selected Papers of Freeman Dyson with Commentary” published by American Mathematical Society, 1996, Templeton Prize for Progress in Religion, 2000, member of the Board of Sponsors of The Bulletin of the Atomic Scientists, 2003 Telluride Tech Festival Award of Technology, creator of “Dyson’s transform” and “Dyson series”
Phil. says:
June 12, 2010 at 10:40 pm
stevengoddard says:
June 12, 2010 at 9:51 pm
JK
You are looking at the land surface north of the Greenland Ice Sheet.
No he isn’t, he’s looking at the mess of fragmented ice and leads north of Greenland (see about 2/3rds down)
Um, yeah, it’s easy to tell land from sea ice when you zoom in a little.
http://ice-map.appspot.com/
So I still say it all breaking up there along the entire N coast of Greenland.
Amino Acids in Meteorites says:
June 13, 2010 at 11:00 am
PIOMAS from 5/30/10, showing less ice on 5/30/10 than on 5/30/08
Cryosphere Today showing more ice on 5/30/10 than on 5/30/08
So, as anyone can see PIOMAS is wrong. It’s simple. Nothing complicated.
————————————————————
Apples and oranges isn’t it? Comparing volume from Piomas with extent (and lousy low-res images at that) from Cryosphere just doesn’t make sense.
Amino Acids in Meteorites says:
June 13, 2010 at 8:45 am
Phil. says:
June 12, 2010 at 10:28 pm
“the PIOMAS model which has actually been calibrated against the satellite measurements of thickness”
The PIOMAS graph is obviously wrong.
Aren’t you a professor Phil? Shouldn’t you already see PIOMAS is wrong?
Nice job of selective quoting! The point was that the PIOMAS data has been calibrated however Steve’s calculation method has not been calibrated and he refuses to even discuss doing so. If one of my students came up with a calculation like this I’d ask the same questions, do you understand the metric you’re basing your calculation on, can you cross-check your results with a sample from the source to validate it. Steve failed to demonstrate either of those basic tenets so his numbers aren’t worth much.
Why do you think that ‘PIOMAS is wrong’? Their findings seem fairly consistent with the Canadian Ice Service Winter report which leads one to expect reduced ice thickness:
http://www.ec.gc.ca/glaces-ice/default.asp?lang=En&n=AB0DF51C-1&offset=1&toc=show
Or are you a different Phil and not the professor? Or, if you are that Phil, is your work as a professor connected to ‘global warming’ and that explains why you would not yourself show the clear flaw of the PIOMAS graph?
I don’t appreciate the implication of dishonesty, what is this “clear flaw of the PIOMAS graph” you refer to?
JK says:
June 13, 2010 at 11:30 am
Amino Acids in Meteorites says:
June 13, 2010 at 11:00 am
PIOMAS from 5/30/10, showing less ice on 5/30/10 than on 5/30/08
Cryosphere Today showing more ice on 5/30/10 than on 5/30/08
So, as anyone can see PIOMAS is wrong. It’s simple. Nothing complicated.
————————————————————
Apples and oranges isn’t it? Comparing volume from Piomas with extent (and lousy low-res images at that) from Cryosphere just doesn’t make sense.
Quite, especially as the CT data shows the two days had basically the same area, 9.92 vs. 9.99 (about one day’s average loss). Why guesstimate area from the low res images when the numerical data is available?