« ASI 2012 update 1: a new beginning | Main | 2011/2012 Winter Analysis »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Account Deleted


The most interesting graph. It looks like it shows good agreement with measurements PIOMAS.


Cheers for this, looks like we will be getting more data from them in the future and can watch this years melt in 3D even if it is only released after the melt has finished.

Be very interesting to see how this compares with PIOMAS.


The most interesting graph. It looks like it shows good agreement with measurements PIOMAS.

Do I see a maximum volume of 15,000 km3? PIOMAS has 22,000, right?

Account Deleted

Neven, I understand from the animation in a press release, they measure only the ice in the Arctic Ocean, without Hudson Bay, Bering and Okhotsk seas.

This may explain the difference in the amount of winter ice.


Thanks, Arcticicelost80, that makes sense.

That press release would be here. In that animation we also see average sea ice thickness to be well over 2 meters in April/May. According to my crude calculations based on PIOMAS volume and CT SIA average thickness was a little over 1.6 meters on April 1st 2011. But again, this difference could be due to CryoSat not taking fringe ice outside of the Arctic Ocean into account.

At the end of the press release this little nugget:

In the coming years, CryoSat data will map precise changes in sea-ice thickness year to year, furthering our understanding of the effects that climate change has on the Arctic.

Account Deleted

Summer ice is good to go. According to CryoSat-2 on October 1 6000 km3, PIOMAS - 5000 km3.


Account Deleted

Perhaps even compare with this schedule.


The average thickness of the winter 2010-2011, is approximately equal to the winter of 2007-2008.

Account Deleted

Or other schedule



Given the title of the graph, I am assuming the seven points are cryosat monthly averages. So is the line, that extends beyond Oct-April and is not a best fit line, PIOMAS data? If so, cryosat looks like it could be showing a little more ice volume particularly in November? Looks like pretty close agreement for other months.

Or are there different interpretations of the graph?

Account Deleted


The area of sea ice on October 1, 2010 about 4000 km2.

With an average thickness of 1.5 meters measure the total volume obtained by cryostat 6000 km3.

In general I am satisfied with CryoSat. In the coming months will be a lot of new data that clarify the situation in the Arctic.

Account Deleted

Сrandles, I was looking to supplement this grafics


As I understand it's not monthly averages, and the data on October 1, November 1, etc.

That explains why, in October of 6000 km 3 is obtained by volume of ice.


Sorry, I am not following. Why only post 7 points unless they are averaged over a month? If not averaged, surely they could have shown more than 7 points?

Comparisons to the PIOMAS graph are best done around minimum when least ice outside basin. PIOMAS shows less than 5k km^3 between about 7 Aug and 7 October.

First point looks nearer midpoint of graph than start of graph and I have now measured 14cm from start and 9.5cm from mid-point. So looks later than average of 2010.5 and 2011.0. Numbers are pretty close to 7:5 so seems like first point is 16th October to me.

Also line looks very close to what I would expect for PIOMAS.

Chris Reynolds

I still can't get my head around the thicknesses they're claiming.

Here's the Paris Airshow graphic, which is pretty similar to the graphic you've posted Neven.

Both graphics show typical thickness of around 3 metres.

Yet at the end of the ICESat mission thicknesses during winter in the DRA were of the order of 2 metres. Even the earlier submarine data was down at around 2.5m. If the 3m figures from Cryosat are correct then both the submarine data and ICESat look suspect, as it seems verging on the inherently improbable that ice has thickened by either 0.5 or 1m in the intervening period. As PIOMAS (and presumably NPS) was compared favourably with those data sets it puts PIOMAS into question.

Furthermore first year sea ice typically grows to around 2m thick. Yet the Paris Airshow graphic, for Jan/Feb shows 3m thick ice in areas that were ice free in 2010.

I'm unhappy with these results from Cryosat, they seem to raise more questions than answers.

Chris Reynolds

Sorry forgot to add graphic for submarine and ICESat thicknesses.

The DRA (Data Release Area) in across the Central Arctic. Right where the ice is supposed to be 3m thick.

Account Deleted

"Here's the Paris Airshow graphic, which is pretty similar to the graphic you've posted Neven."

I think they are a little clarified that map.


There is a new map from April 2011 and it looks like it is different from the old (shown in June 2011)


I'm unhappy with these results from Cryosat, they seem to raise more questions than answers.

Yes, unfortunately not more info than we already had.

I hope the presentations come online (they did after the Paris Air and Space show) so I can re-watch Duncan Wingham's presentation. Like I said, he talked a lot, and I was busy typing and cropping. But they said twice that the thickness map from Paris was made quickly to have something to show. And as Arcticicelost80 said, PIOMAS and CryoSat cover different regions, PIOMAS covers more, CryoSat only above 66N. I think.

The BBC article is interesting, with yet another thickness map (this time for April 2011) at the end. Quote:

"We can now say with good confidence that Cryosat's maps of ice thickness are correct to within 10-20cm," said Dr Seymour Laxon, from UCL's Centre for Polar Observation and Modelling (CPOM).

Tuesday's release shows a complete seasonal cycle, from October 2010, when the Arctic Ocean was beginning to freeze up following the summer melt, right through to March 2011, when the sea ice was approaching peak thickness. Cryosat found the volume (area multiplied by thickness) of sea ice in the central Arctic in March 2011 to have been 14,500 cubic kilometres.

This figure is very similar to that suggested by PIOMAS (Panarctic Ice Ocean Modeling and Assimilation System), an influential computer model that has been used to estimate Arctic sea ice volume, and which has been the basis for several predictions about when summer sea ice in the north might disappear completely.

I still hope we get more info from CryoSat, and not just post hoc, but I'm not counting on anything. They obviously need more time for the validation and calibration process. Somehow I don't think we will see daily or even weekly updates. But even then it will be great to have accurate volume/thickness data.

John Christensen


I completely agree; it will be very interesting to get a more real-time understanding of how weather events are impacting the volume/thickness.

Hypotheses I would be interested in testing out:
- When the ice pack is solid providing optimal insulation from cold air, can heat flux from warmer water below cause melting/thinning of the ice?
- Is the crushing of ice via wind or currents an important factor in building additional ice compared to a stable solid ice pack?

Chris Reynolds

Thanks for that ArcticIceLost,

The graphic shown in that arcticle is significantly thinner than the Paris Airshow graphic. This seems more reasonable and more in line with earlier work. I'm still a bit dubious but I need to cross check a few papers before saying anything further. Basically it's got me wondering again if we're near the end of the substantial volume losses.

Chris Reynolds

Neven, John,

Sorry didn't notice your posts after I'd published my last comment.

John, In answer to your hypotheses; With regards bottom melt, yes most definitely this can and is happening. With regards thinning in winter - you'd get a trade off of effects, as ice gets thinner it insulates less, so loses more heat and acretes ice on its base by freezing. I'm not sure whether thickness measurements will help make these processes more clear.

With regards crushing of ice - Maslanik shows a continuing drop of extent in ice more than four years old. While the younger ice has stabilised, and first year ice is taking up the area left by the loss f old ice.
Within the category of ice 2 years to 4 years old there is great opportunity for ice thickening by compression. Mainly due to the transpolar drift against the Canadian Arctic Archipelago (CAA). As Crandles has noted somewhere on the blog recently - the area off the CAA can be considered a safe zone, where ice can grow. Out of that zone ice flows into the Fram Strait - doomed to melt - and into the Beaufort Gyre.

Woods Hole Inst have referred to the Beaufort Gyre as a flywheel, looping ice around and ageing it before putting it back into the transpolar drift, and back into th 'safe zone'. Now that flywheel, a stabilising influence, is terminally broken. Ice that gets into Beaufort now tends to melt in the summer.

Personally I don't think Cryosat will help make this more clear as existing tools, like the Drift Age Model, or Hycom, lay the guts of the process bare.

It'd be good to see regular thickness maps, and in time, thickness animaly maps. Of course a signifcant body of opinion here, which I still may join, will see all this as too late for anything but post-mortem.


>"Yes, unfortunately not more info than we already had."

You have realised we have six pictures (instead of one) for 6 different months on the animation.


Sure, the animation is nice. But it's for last year. Wingham mentions PIOMAS, but doesn't really say anything about it (I think, I was busy). Where are high-resolution versions of the graphs shown during the presentation on the ESA website? Where are thickness maps like the one they showed last year?

In short: What is the state of the ice now, according to CryoSat? The animation is great for the media, but I think we need a bit more here.


It is really frustrating that they say all these good things about Cryostat-2 and its capabilities at ESA website but we are being served year old data. What about the 2011 melt-season data??

John Christensen

Guess I missed the boat here; had also hoped we would be getting data that was weeks old, not a full year..

Rob Dekker

I had really high hopes for Cryosat to determine Arctic sea ice volume accurately and near real-time, but after the Paris Airshow presentation of Cryosat's first ice thickness last year, much of that hope was shredded.

It was not so much that it took more than a year of 'calibration' even then to even produce the first results on sea ice thickness. It was more that the map from 2011 was simply unreal and physically unrealistic. It was so far off that even NSIDC's Julienne Stroeve noted that the thickness reported by Cryosat in various locations (including Baffin Bay) was simply impossible based on basic thermodynamic physics of ice growth. Besides that, there were other indications that the map even conflicted with in-situ measurements done by the Cryosat validation team itself.

This new "map" (more like a TV image shot) does not restore any confidence in Cryosat's ability to measure Arctic ice thickness.

For starters, there seem to be some peculiar "features" in this new sea ice growth map. Specifically, the Cryosat orbital paths seem to stand out as lines 'fanning' out from the 88 N orbital high point. And what the heck is that ridge of thick ice forming an almost perfect circle at 80 or 82 N all around the pole ?

Second, why the fuzzy pictures and video presentation ? Why not a simple and clear set of high-resolution jpg images showing ice in each of the months, with an accurate graph or table showing thickness over the various Arctic regions ?

And third, as some of you here already noted, why did it take a full year to get this data of the 2010-2011 freezing season out ?

I'm starting to thick that maybe satellite observation of ice thickness may actually simply fundamentally be inaccurate.

Think about this : To obtain an accuracy of 10 cm in sea ice thickness, the satellite needs to determine ice freeboard by 1 cm (due to Archimedes law). That accuracy can even theoretically only be achieved with a 100GHz radar and only if there is no snow cover at all. Cryosat does not operate with a 100 GHz radar, and snow cover is a factor 20 to 50 higher than that. So any estimates Cryosat puts out are going to be more a result of the snow models and pulse averaging, than actually determining the thickness of the ice.

I think I put more confidence in a physical ice growth model (like PIOMAS) which is forced by observations (NCEP/NCAR) than in a satellite system plagued by a 1:50 signal-to-noise ratio.

But that's just my opinion.

Rob Dekker

Sorry. One correction :
To obtain a 1 cm accuracy by pulse response, theoretically you need a 30GHz radar, not 100GHz.
However, Cryosat uses Interferometry (SAR), which does phase analysis, which can improve vertical resolution at the expense of spacial resolution. Since spacial resolution is in the range of km's, my argument about radar frequency is moot.

Still, the argument that Cryosat still deals with a 1:50 signal-to-noise ratio due to snow cover still stands. Cryosat results are thus mostly determined by the accuracy of their snow models. If they make even a 5% mistake in snow cover thickness and snow density, their ice thickness uncertainty estimates will exceed 20cm easily.

Rob Dekker

Here is a wild idea: one of the big uncertainties in PIOMAS model is snow cover (how much snow is deposited where and what is it's water density). Cryosat is very good at determining the altitude above sea level of snow.

So maybe if PIOMAS uses Cryosat's (before correcting for snow) altitude measurements, they can increase the accuracy of their ice-growth model...

Just an idea...

Yvan Dutil

@Rob Dekker Your analysis is wrong. You can measure distance with a precision higher than the wavelength with enough SNR. Jason altimeter only work at 5.3 and 13.6 GHz but still manage to get mm precision. In addition, GPS transmit at 1.023 MHz, 1227.60 MHz and 1575.42 MHz but can still manage mm precision with enough integration.

Daniel Bailey

Just some thoughts to help give perspective: the various satellite platforms went through many phases of error checking and calibration before acceptance of the robustness of their temperature products was given (earned is a better term). Does that mean UAH and RSS are perfect today? Far from it! They endure a continuous battery of calibrations and error checking to refine the products.

What does that mean for Cryosat-2? Likely that the true value of C2 will be in its anomaly products that will come down the road. Since the issue right now is calibration with existing ground-truths (OK, "ice-truths"), absolute accuracies will take time to achieve (and will likely occur in baby-step fashions). Relative accuracies, that is, C2 accuracies relative to its own acquired baseline data, will be initially much greater.

Thus, if the first year's data is acquired to give a baseline for context, the second year's data will give an anomaly to that baseline. With each passing year calibrations can be refined and data reworked. C2 will thus have its "eyesight" corrected over time to give more clarity to what is actually happening "on the ice".

The "ridge" mentioned as encircling the pole could be an artifact of orbital path convergence that has yet to be properly dealt with in the data. If so, it can be ignored, as it will disappear as refinements are made.


I have been wondering about

Can Cryosat 2 tell the difference or can it only find the highest point over fairly large areas?

Probably doesn't matter if the ratio of these sort of shapes remains the same, it is just a matter of sensible calibration to get the right volume?

But if thinner ice is weaker and breaks into smaller pieces, do the ratios change over time? E.g. could slabbing become more common with thinner ice that is easier to lift far enough? If the pieces are smaller are the ridges more prominent for the volume of ice?

Could such ratios change enough to matter?
Even if they can, does continuous calibration solve any problems?


It all depends on snow cover depth and as long as the C2 QA site estimates snow in Jul/Aug it is not that impressing.


Neven, many thanks for live-blogging.

Peter Ellis

Snow cover should as I understand things be irrelevant: the idea is that you get a "hard" return from the ice surface which is distinguishable from the more diffuse return from the ice cover. If I recall correctly, one of the theories (put forward by Julienne Stroeve from NSIDC) for why Cryosat had high values in Baffin bay was that there may have been icy layers within the snow cover that confounded the process. Given that the newer results show much thinner ice in Baffin, I guess we can assume they've eradicated this problem.

As for the 'fanning' lines, this is because it's a composite view of an entire month. The tracks from later in the month will necessarily have thicker ice than those from earlier in the month.

As I've posted before, anyone expecting real-time coverage from Cryosat is on a fool's errand: it simply cannot do it, because it has to use a narrow beam to detect freeboard. It's like trying to paint a wall with a felt tip pen. Cryosat will give a statistically robust picture of average ice thickness on a monthly basis. Any given swath will of course be accurate, but the composite picture will look bizarrely misleading because different parts of it were scanned on different dates.


Scroll through the data takes for each month to see what's available. Do not expect more than that.

Peter Ellis

Looking up the swath widths, it seems that Cryosat has a swath width of 250m, while AMSR-E had a swath width of 1445km. The orbital period is essentially identical, meaning that Cryosat has 1/5780 the area coverage of AMSR-E. AMSR-E "saw" the entire Arctic ocean surface multiple times every day, while Cryosat doesn't actually "see all of it even once per month. That's not what Cryosat is for.

To give a specific example: I counted the orbits for this partial data take. I counted ~200 ends = 100 tracks.

100 tracks, with an orbital period of 99.2 minutes, is just shy of a week. I probably missed few overlapping ones, so I would guess that's exactly one full week of data. Look at the holes in it. It's less than half coverage of the area - and only a quarter coverage at the latitudes where swaths cross. Remember also that the swaths in reality are only 250m wide: much thinner than shown on the diagram.

The previous month's take looks like it has full data for the entire month:

Coverage looks good (ignoring the fact that at this scale a single pixel is much more than 250m), but that ignores the fact that some areas will have been covered on day 1 and some on day 30.

Cryosat will produce good average ice thickness maps when integrated over large spatial areas and long periods of time. Not more.


Peter, regarding snow depth it is not that simple. First the radar must penetrate snow deep enough to hit solid ice and strong enough to get back to the satellite. Second even solid ice returns a diffuse echo since the surface is not perfectly polished (think ridges). Also given a swath of 250m the echo is anyway a mixture of different surfaces types (ocean, deep snow, refrozen snow, melting ponds, ice). So, a threshold is needed to distinguish ice from snow within a noisy/diffuse signal.

Once you know snow depth (e.g. from validation directly on the ice) the math is as easy as you said. But in more than 99% of all cases the surface is unknown and the accuracy of a threshold depends on the quality of your assumptions as mentioned above. Unfortunately surface conditions change often in the Arctic and I wonder whether models having a resolution below 1km even exist.

I agree in theory C2 should deliver very accurate sea ice thickness changes over years, but I doubt there is an easy solution to measure snow depth with the needed accuracy of +/-1cm.

Also the excuse that the first map was made in a hurry is not convincing, till now there was enough time to correct any known errors. However, C2 orbits as the perfect technology this planet to report about shrinking ice shields in Antarctica and Greenland and is worth every single Euro, not mentioning the growing amount of other applications.

Pete Williamson

Thanks Peter Ellis, very interesting. And I guess compared with a few submarine tracks, a few dozen aircraft flyovers and a handful of in situ ice thickness measurements every year, the Cryosat data is orders of magnitude higher. It seems important when one is criticizing the Cryosat to keep in mind what the alternative is.

Thanks for the live blogging Neven.

Rob Dekker

Peter Ellis said

Cryosat will produce good average ice thickness maps when integrated over large spatial areas and long periods of time. Not more.

Spacial and temporal integration of course increases the accuracy (by the square root of the number of measurements), and this is why other platforms, such as our good old Envisat and (as Yvan Dutil points out) Jason have been able to obtain mm accuracy of average ocean sea level.

However, the Arctic is different from sea level determination, since sea ice is always changing (growing or thawing, and moving around).

If we could just integrate spacially and temporally, we would not need Cryosat to determine ice thickness and volume. We could just use Envisat and Jason and a few more altemetry platforms.

This is where Cryosat promised to do better. With their radar altimeter called SAR Interferometric Radar Altimeter (SIRAL) Cryosat is supposed to do much better than than simple integration. They were supposed to obtain accurate and high-resolution instantanious measurements of freeboard, and thus should be able to produce near-realtime determination of ice thickness.

Now, after 2 years of caliblation, the best they come up with is a 1-year old month-to-month ice volume assessment which closely matches PIOMAS (which is updated every 30 days).

If this is the best that Cryosat can do, then it is disappointing to me. Sorry. I simply expected a lot more from this project.

Rob Dekker

On the bright side, the fact that Cryosat now produces physically realistic ice thickness neasurements that rather closely match PIOMAS numbers reinforces confidence in PIOMAS. And, as I stated before, maybe PIOMAS can use Cryosat's altemetry data to constraint a major uncertainty factor (snow cover and distribution) in their model.

Account Deleted

All who are not satisfied with the work of the CryoSat-2 will remember how many were suffering from IceSat-1.

A subsequent investigation indicated that a corrosive degradation of the pump diodes, due to an unexpected but known reaction between indium solder and gold bonding wires,[5] had possibly reduced the reliability of the lasers. Consequentially, the total operational life for the GLAS instrument was expected to be as little as less than a year as a result. After the two months of full operation in the fall of 2003, the operational plan for GLAS was changed, and it was operated for one-month periods out of every three to six months in order to extend the time series of measurements, particularly for the ice sheets.[6] The last laser failed on 11 October 2009, and following attempts to restart it, the satellite was retired in February 2010.

The unfortunate satellite could operate continuously for one year only!
The first CryoSat was lost in the derivation of the orbit.

Therefore, Cryostat-2 is already the most successful satellite mission to measure ice thickness.

Rob Dekker

Arcticicelost80, thanks for putting things in perspective.

You are right. Satellite deployment and operation is still very fragile and very costly, and many people work extremely hard to make such missions successfull.


I simply expected a lot more from this project.

Yes, maybe we were expecting too much of this. I know I was, being spoilt from all those daily updated graphs and satellite images. It also took me a while to understand that these things need val/cal (validation and calibration). I just thought: hey, make that snapshot and update the graph!

But like Rob says: if CryoSat-2 helps to affirm/improve PIOMAS, that will be a huge help in assessing the situation in the Arctic. And seeing that Wingham as well as that BBC article mentioned PIOMAS, I wouldn't be surprised if the teams from the PSC in Washington and CMOP in London are checking each other out.

But that said, I hope we do get more info. I understand they have to do these presentations every once and again to justify the money spent, but the Arctic Amateurs want more. :-)

Peter Ellis

They were supposed to obtain accurate and high-resolution instantanious measurements of freeboard, and thus should be able to produce near-realtime determination of ice thickness.

It does exactly that - but for small precise regions. Specifically, a 250m wide swath every ~90 minutes. It's still many orders of magnitude better than anything ever seen before.

However, the price of that accuracy is the fact that if you want to use only the Cryosat data, then it will take a month or so to achieve full coverage of a given region. That's what they've done for the images shown to date, which is why you can see the "fan" shape of the swaths in the images.

The real dividends, as you say, will be paid by assimilation of the daily ice thickness data tracks into models like PIOMAS, similar to the way concentration data is currently assimilated.


Imagine interpolating across the blank spaces between the tracks. You'd get a very accurate picture even from one week's data. And in fact even a single day's data would cover much the same "network" of and would hence constrain the ice thickness model (not just snow thickness) across the entire Arctic.

It's much like Earth surface temperature measurements - you can get accurate values even from a relatively sparse network of stations as long as you combine them together correctly. Cryosat, when used on a daily basis, will give a dense, evenly-spaced network of measurements across the entire Arctic. That's incredibly powerful, even though it doesn't "colour in" the whole area.

In fact, that would be a useful exercise for someone with a lot more computer skills than I: take a single day's thickness data and then fill in the blank regions by interpolation from neighbouring values using algorithms similar to the ones used for surface temperature measurements. Then display that as a map, with dashed lines superimposed to show the satellite tracks (and hence where the "true" data comes from). I think you'd be pleasantly surprised.

Yvan Dutil

Actually, the real value of Cryosat-2 will comes from the added data in the assimilation model like PIOMAS. PIOMAS does not have much measurement to constrain its physical model. Cryosat-2 will improve this aspect very much.

Chris Biscan

Maybe they can explain how their first map from Jan-Feb 2011 was so far off?

Rob Dekker

Peter Ellis,
Funny you mention, but I last year, when ESA announced that Cryosat data was open to the public, I actually tried to build such a system : Download L2 data from Cryosat, put the 'freeboard' field in a 2D map data structure, and apply an interpolation algorithm to present a 'map' with data on-line on a web site.
Near-real-time representation of Arctic thickness/volume data !

Unfortunately, it was not that simple.

I obtained username/password for Cryosat data, downloaded the BRAT user interface, and ran the first (tutorial) data tracks.
That's where the trouble started :
The only 'freeboard' data were a few pixels over Siberian lakes, and these were obviously bogus (5 meter freeboard). No freeboard over the Arctic in any of the tracks. But it got worse. The actual altimetry data was also bogus. For example, even after applying all the necessary corrections, I still obtained mountains of 800 meter over the Pacific.
Besides that, I never was able to obtain any 'up-to-date' data. Only pre-selected tracks that were 6 months old or more.

The only usefull altimetry data I could find came from Envisat, and that was actually pretty cool : sea ice realistic and a beautiful profile of Greenland ice sheet.

That's when reality hit me in the head like a 2x4. Cryosat was simply not providing usefull data yet. That was May 2011.

It seems that Cryosat did make significant improvements since then, but it's still pretty bad. Check out the 'Crossover statistics' page on UCL. Crossovers are the points where the satellite passes a point that it visited not too long ago. If that point is over land (or some other surface that does not change much), then it serves as measure of the 'drift' and thus helps to determine how accurate the measurements were in between :

Then, check out the results :
LRM (Low Resolution Mode) is not too bad :
RMS error of 0.70 meter on crossover points.
That is way too high for a reliable 'freeboard' estimate, but, hey. Its Low Resolution Mode. What did you expect?

Then go to SARIN. That's the high-resolution Interferometry SAR, and the best they have :
RMS error is 4.18 m. Many cross-over points show drifts of 10 meters or more.

That means that single-tracks are hopelessly inaccurate. You would not be able to tell the difference between ice-free and 40 meter ice.

And it's not getting much better over time. Check out how the cross-over RMS error changed over the lifetime of the mission...

And we did not even talk about the issues with snow cover (which I mentioned above). In fact, the 'snow cover' Cryosat data on UCL is actually derived from a "climate model". And 'freeboard' information Freeboard will not be calculated until launch plus one year after an initial calibration study

Which is going to be a year from now, pretty close to the end of the expected life of Cryosat 2.

This is why I have serious doubts about Cryosat being able to ever provide accurate sea ice thickness information. Realtime or post-partum.

Peter Ellis

Rob: If that were really the case, then they wouldn't have even been able to generate the maps they have to date, there must be an alternative explanation.

I note that in the "correction checks" section, the same figures for rms error are given, but this time they're stated to be in cm rather than m.

That seems much more reasonable for the SARIN mode, but a low resolution rms of 0.7 cm? Not likely.

I suspect strongly that what we actually have is a low resolution mode error of ~0.7m, and a SARIN mode error of ~4 cm.

Rob Dekker

I'm not sure what to make of the "correction checks". It seems to me a that the "rms crossover residual is calculated" rather than the actual measured difference in altitude over the crossover point. That would explain the very small rms (of 0.7 cm) in Low Resolution Mode reported from that page.

Rob: If that were really the case, then they wouldn't have even been able to generate the maps they have to date, there must be an alternative explanation.

Well, maybe that's why it took them a year to compile this latest map, and maybe that's why the first map from last year was way off.

Until they tell us exactly how they obtained either one (or both) of these maps, it would be difficult to put much trust in either one of them. Especially since the data that they released does not seem to contain info even remotely close to what they present in this press release.

Peter Ellis

Eh, the "crossover residual" is the difference in heights at the crossover point. That's what residual means. The rms crossover residual is the rms of the crossover residuals. It's exactly the same figures, just with different units. There's a mix-up of some sort - there must be a contact email address somewhere to check it out.

I think assuming bad faith on their part is disrespectful.

Rob Dekker

Peter, I did not mean to be disrespectful, not assuming bad faith.
I DO think that the Cryosat team could have been more forthcoming on how exactly they derived last year's and this recent map.

Especially since there does indeed seem to be some confusion on the data or some mix-up of some sort.
In your link :
the residual seems to be "calculated" and unrealistically small (0.7 cm or smaller for LRM) while in the link I provided
the residual seems to be "measured" and unrealistically high (4.18 meter for SARIN).

You are right. Maybe we should send off an email to the Cryosat (or UCL) team to ask for some clarification on these numbers.

Seke Rob

Spam alert

The comments to this entry are closed.