Collected from Judy Curry’s site, some well informed analysis on Ocean Heat Content (OHC) and cloud cover variation by ‘Chief Hydrologist’ which should put us in mind of Peter Berenyi’s excellent analysis published here a couple of years ago:
Chief Hydrologist | August 28, 2013 at 8:57 pm |
In ARGO from 2005 to 2010 – the increase is about 0.3E+22 J/yr. This is pretty much consistent with CERES and SORCE over the period – with the changes all in shortwave. I would suggest that earlier ocean temp data – especially to 2000m – might be a little lacking and that the splice between the old and new data looks horrendously unlikely. I suggest that the planet is again cooling with more recent increases in cloud shown in the Palle and Laken paper.
Oceans gained energy to 1998 – pretty much in line with changes in ERBS net radiant flux. Which again was all in shortwave as a result of cloud radiative forcing.
http://s1114.photobucket.com/user/Chief_Hydrologist/media/Wong2006figure7.gif.html?sort=3&o=129
The interesting bit however is the increase in cloud in the 1998/2001. Here it is in ISCCP-FD.
http://s1114.photobucket.com/user/Chief_Hydrologist/media/SWupnotes_zps3f1ab841.jpg.html?sort=3&o=34
But it shows in Project Earthshine and ERBS as well. The 1976/77 shift shows in COADS – which again is consistent with ISCCP-FD
http://s1114.photobucket.com/user/Chief_Hydrologist/media/SWupnotes_zps3f1ab841.jpg.html?sort=3&o=34
We can in fact see what happened to ocean heat content in the critical period between 1998 and 2003 in the Wong et al Fig 7 – ocean heat content here is based on annual steric sea level changes. The ocean heat content was less in 2003 than at the peak in 1998. ARGO suggests that the rebound in heat content since is entirely insufficient to make up the decline.
Found the missing heat in CERES? It is all in shortwave.
Ocean heat content increase as I said in ARGO was fairly modest as was sea level rise. This is the best data available. Results for the last decade 3 and more times larger than ARGO are questionable. I’d question the XBT/ARGO splice for a start.
But really the interesting period is the 1998/2001 climate shift. Ocean heat content peaking in 1998 and declining since. Just like the surface. All in the shortwave.
http://s1114.photobucket.com/user/Chief_Hydrologist/media/Wong2006figure7.gif.html?sort=3&o=129
This is consistent with the pattern of cloud cover in the recent and excellent Palle and Laken paper. I recommend it highly.
http://www.benlaken.com/documents/AIP_PL_13.pdf
The interesting bit is the climate shift of 1998/2001 I repeat. And it is – I repeat – covered by the Wong et al 06 paper which shows ERBS net and ocean heat content from Josh Willis.
Tallbloke's Talkshop 
Like I’ve been saying since 2007/8 I first noticed the jets turning more meridional again around 2000.
The sun shifts the jets.
The jets change the cloud amounts.
The clouds affect ocean heat content.
The oceans regulate the air temperatures.
New papers are coming out almost daily that support that hypothesis.
Stephen: Please link some of them, and your own summary on your website.
Done.
http://www.newclimatemodel.com/
Scroll down Home Page to the Natural Climate Change News Section.
Thanks Stephen.
I’ve added your site to the blog roll, a return link from your site would be appreciated.
I’ve put you in as item 12 in the Resources section.
http://www.newclimatemodel.com/resources/
If you want different wording just say so.
Stephen.
Thanks!
What I would like to know is, do they continue to collect the data as they had been before ARGO, and how do those data match ARGO?
Bart: I very much doubt it. XBT was an expensive one shot experiment every time. The lack of any overlap is crazy, unless they’re just not showing us the data…
I suspect as much. It is really poor methodology to splice measurements from differing techniques together without having any confirming evidence of agreement. It’s likely you’d get something like Figure 6 here.
I really hate it when people present unverified, extruded proxy measurements as “proof” of anything.
“I suggest that the planet is again cooling with more recent increases in cloud shown in the Palle and Laken paper.”
I would be extremely cautious in drawing conclusions from the ISCCP cloud data, particularly when the Palle and Laken paper makes it clear that they are probably corrupted:
“… several significant jumps are clearly evident in Figure 2, connected to a shift in mean cloud anomalies. This suggests that spurious changes exist within the ISCCP data that may have contributed to long-term changes, as suggested by numerous authors …. A calibration artifact origin of these changes appears to be highly likely”
And also when the ISCCP record conflicts with two other cloud data sets that cover the same period – ICOADS ocean and CRU TS 3.10 land:
Roger A: Are the ICOADS obs deck obs from ships? Where are the CRU obs from, HARRY_README? 😉
Bart: Do have a look at the post from Peter Berenyi I linked. His method shows the splice probably included a big jump. Someone as smart as yourself could probably quite quickly work out the size of jump that needs to be removed fro the splice to get a proper match from the CERES data. Data files are linked in the post.
TB:
The ICOADS observations are from ships and were presumably taken by the same people who took the SST measurements. So they must be good.
The CRU observations were taken at several thousand weather stations, presumably by the same people who took the air temperature etc. measurements. There is no record of Harry having had any trouble with them, so they must be good too. 🙂
Roger A: So on the basis of the quality of the surface temperature records compared to satellite temperature records, who shall we go with for cloud?
When did the lunar Earthshine observations start?
Brian: When did the lunar Earthshine observations start?
In fits.
Ignore the left scale in this plot, it’ an SkS lie. Use the left scale in the inset for estimating percentage cloud cover change.
Roger A: Note that the Earthshine projects early data confirms ISCCP to at least some extent.
TB:
Neither. If the data sets in the graph above were gold assays I would throw all three of them out.
I need to look into this in more detail, but based on what I’ve seen so far I would say that we don’t have any usable cloud cover data, or at least none that goes back far enough to do us any good.
Observations from individuals on ships or at specific land locations would not provide adequate coverage for global cloudiness.
Only a complete global overview from satellites strikes me as good enough and I currently favour the Earthshine data on that ground.
Rog let that prejudice. It seems to me that the last cyclone up from Japan, which has now reached over Alaska can cause about 20 strong blizzards in the central U.S.
You can see what’s happening on the forecast of the stratosphere (about 23 km).
Stephen: I agree. Shipping lanes are limited and the field of view from the deck is narrow and perspectivised. Likewise surface stations. The ISCCP people are not stupid, and did the best job they could. It ain’t perfect, but it’s the best we have. The Earthshine method is best for overall albedo, though it doesn’t give the resolution of ISCCP for differentiating high, low, tropical etc.
The earth-shine topic is an interesting one TB. Do you know of any other useful links that might have data to download?
MH: I don’t think the Earthshine data is freely available – ongoing research rules…
There’s a very long and very detailed report on problems with the ISCCP cloud data here. (It could take a while to download).
Here’s a Figure from the report comparing the ISCCP data with other satellite measurements.
And here’s the authors’ conclusion: “At present one can only conclude that global monthly mean cloud amount is constant over the last 25 years to within 2.5%, within the range of interannual variability.”
I’ll try this again if the links don’t come through.
Seems they didn’t.
The link to the figure is here:
I’ve concluded that no one is going to want to read the report, which runs to 176 pages and which takes for ever to download, so I’m not linking to it.
Roger A: knock the http and the slashes off the front and post the link please
The ISCCP website was always a nightmare to get data from and I think that’s the way it was intended to be.
All the satellites were looking at cloud through different frequency goggles, so no surprise they don’t agree.
None of these issues affect project Earthshine.
A bit torrent is all I can come up with:
http://www.google.com.mx/url?sa=t&rct=j&q=&esrc=s&source=web&cd=12&ved=0CJIBEBYwCw&url=http%3A%2F%2Fwww.wcrp-climate.org%2Fdocuments%2FGEWEX_Cloud_Assessment_2012.pdf&ei=SYlhUsSlJ4yA2QXs5IDAAw&usg=AFQjCNESiRRusLshBy6XaxOxZWbsxfm4Eg&sig2=oizqIgbm-MmDlt_7gUak3w
Here is a useful link to the Earthshine project:
Click to access Palle_etal_2006_EOS.pdf
which compares with ICCP.
If anything Earthshine shows a larger rise in global albedo since 2000.
“Since 2000, ES observations indicate an
increasing albedo [Pallé et al., 2004], whereas
Clouds and the Earth’s Radiant Energy System
(CERES) satellite data report the opposite
result [Wielicki et al., 2005].A recent intercomparison
of several albedo-related data sets
strengthens the case for an increasing global
albedo post-2000, consistent with the original
ES result [Pallé et al., 2005]”
Given that we currently have large meridional diversions of the jet stream tracks together with large slow moving mid latitude depressions with longer lines of air mass mixing I am inclined to feel that their comments fit the more general observations.
WE currently have more clouds than pre 2000 with less energy entering the oceans and El Nino fading relative to La Nina.
In my judgement the fading away of El Nino goes beyond what would have been expected just from the negative phase of PDO.
http://www.drroyspencer.com/2013/10/oceanic-cloud-decrease-since-1987-explains-13-of-ocean-heating/
The chart from SSMI and SSMIS shows an increase in global oceanic cloud water from around2000 and a decline prior.
Brief comment, very busy on software dev.
Not too large and lets the cat out of the bag, presentation by Palle, Big Bear Obs., what is not shown in formal papers. Page 3 contains a Trenberth diagram, “The Albedo, a climate driver”
The diagram and idea is wrong. Fits in with the prygoemeter con-trick too and the widespread improper application of SB.
Click to access 3_06_Palle.pdf
Spot the omissions.
Most glaring is omitting to show cloud reflecting outgoing back down, exactly what happens, hence albedo in the simplistic case cancels to unity. More cloud, less incoming and less outgoing.
Can say nothing about nightside and this does vary.
The effect across all wavelengths needs to be known, highly complex.
Tim:
Right, the whole thing is highly complex. But there are some simple tests we can perform to see whether albedo really is a “climate driver”, as Pallé claims on page 3, with the assumption being that it will have a detectable impact on basic climate variables if it is.
The first graph in the graphic below superimposes absolute ICOADS SST, which is about as basic a climate variable as you can get, on the seasonal changes in the Earthshine albedo plot shown on page 7. The seasonal changes in Earthshine albedo are a generally poor match to the seasonal changes in SST.
The second graph compares seasonal variations in SST with seasonal variations in the ICOADS ocean cloud cover data set between 1996 and 2006 (the cloud cover scale is inverted so both plots move in the same direction). We get a pretty good match between seasonal cloud cover and SST changes – certainly far better than the Earthshine match in the first graph – and it goes back at least to 1950.
So why not use the ICOADS cloud cover data set? Because it shows cloud cover increasing before 2000 and decreasing after 2000, which is the exact opposite of what conventional wisdom says we need to explain the the pre-2000 warming and the post-2000 warming pause in terms of cloud cover changes. So the ICOADS data set can’t possibly be right.
Can it?
As you said, things are highly complex ….
Roger A: We get a pretty good match between seasonal cloud cover and SST changes – certainly far better than the Earthshine match in the first graph
Well since the cloud obs and the SST readings were taken on the same ships at the same time, you can colour me unsurprised. 😉
Gunnite.
TB: Get a good night’s sleep. You obviously need it 😉
Roger A: Think about it, local readings taken at the surface reflect local conditions.
Fred and Harry do deck obs:
Fred: “It’s been bloody cloudy for the last two days”
Harry: “Aye lad, watter’s pretty cool too”
Fred: “Not like last week, when Sun were blazin’.”
Harry: “True, surface watter were warm too”
Now, add all those concomitant local readings from the thousands of ships together, and what do we find? ICOADS SST and cloud cover track each other, giving “A pretty good match”. Only problem is, the shipping lanes cover about 1% of the ocean….
Now, lets compare your ICOADS SST to satellite age SST.
Oh, we can’t, you’ve switched graphs overnight. Now you’re only showing short term stuff. Why is that? 😉
Anyway, your new plots are interesting. Particularly the 0.75C swing in global SST, colder when the Earth is closest to the Sun, with the most ocean covered hemisphere facing it. Makes you wonder where the southern tropics shift all the solar energy to, especially in the late 90’s when there was less cloud. Perhaps into the sub-surface Pacific Warm Pool, and then into the atmosphere in the ’98 El nino? Or is there a northern bias in the (raw?) data?
“Ah, Mr. Hornblower, here you are at last. You’re the Meteorological Officer of the Day, correct?”
“Yessir”
“You do remember that after you measure the temperature of the sea water in the bucket you must adjust the result to match the time of year?”
“Yessir, of course, sir”
“And that you must do the same thing with the cloud cover estimate?”
“Yessir, of course, sir”
“And why must you do these things, Mr. Hornblower?”
“Because Admiral Lord Nelson has ordered that our meteorological records should show the correct seasonal cycles, sir.”
“And why is that?”
“Because the Admiral believes it will become important in 200 years’ time, sir.”
“Excellent! And you’ve done the Fourier transforms and Monte Carlo analyses?”
“Yessir. Did them right after breakfast.”
“Very good, Mr. Hornblower, carry on.”
“Aye aye, sir.”
Roger A: 🙂
Now, answer the questions.
1) Is it raw data?
2) Does it have a northern hemisphere bias?
3) Where did the longer term plot go?
TB:
I’m still working on this stuff. It’s getting complicated, so it will take me a few minutes (or longer) to respond.
No worries, take your time.
TB:
I can now supply you with the following preliminary conclusions based on a brief analysis of the ICOADS SST and cloud cover records and other related data sets:
1) Nothing is conclusive.
2) However, the available evidence suggests a) that temperatures go up, not down, with increasing cloudiness and b) that if there was a change in cloud cover around 2000 it was from increasing to decreasing, not from decreasing to increasing.
3) See 1).
I can post volumes of backup information on this or I can just go away. State your preference. 🙂
Roger A: The ‘available evidence’ consists of several conflicting datasets which you’ve made a personal choice about without sufficiently explaining the basis for the choice.
1) Is it raw data?
2) Does it have a NH bias?
3) what percentage of Earth does it actually cover?
” if there was a change in cloud cover around 2000 it was from increasing to decreasing, not from decreasing to increasing.”
Not according to the Earthshine data you plotted, which is worldwide, and doesn’t suffer the angle of view issues of ISCCP or the limited horizon *and* angle of view problems of deck obs.
Earthshine objection stands.
SST, ah well. Cloud cover and SST are not related. Cloud is here and now, SST is about some other time and place.
Lets say there is a clear blue sky except for a 100 mile diameter cloud travelling at the same speed as the ship. Get the idea?
In addition, today maybe more commonly with aircraft but captains route according to the weather which includes wind.
We need to know about nightside anyway.
(Tim or TB: While you’re “moderating” my earlier comment you might consider junking it and replacing it with this one, which includes a graph I forgot to put in. Thx. R)
TB
I can see you want more information. Let me get the ISCCP stuff out of the way first.
Here’s the ISCCP vs. eight other satellites graph I presented earlier. It shows why the WCRP concluded that the ICSSP data tells us nothing about whether cloud cover has gone up, down or sideways over the last 25 or 30 years.
This plot also shows only some of the available data. We could add the ICOADS and CRU TS 3.10 series, which I plotted against the ISCCP data in an earlier graph:
We could add the Roy Spencer plot Stephen Wilde linked to earlier.
We could add the Earthshine data, which when we remove all the ancillary lines and points that were added to beef them up in the graph you posted are found to consist of only eight data points:
Finally we could add the FRESCO (Fast Retrieval Scheme for Clouds from the Oxygen A-band) cloud amount and cloud pressure data, which I plot in the graph below for completeness:
Then, if we like, we could plot all fourteen data sets together, but there’s no question what we would get. Spaghetti, just like tree ring proxies, and about as useful.
And to round things off here’s a plot of the ISCCP low, medium and high cloud data sets (data from KNMI). The low cloud plot is really something.
So much for ISCCP. Now on to your comments and questions.
You say I’ve made a “personal choice (of data sets) without sufficiently explaining the basis for the choice”. I hope the data I’ve already presented will have made it clear to you why I didn’t choose any of the satellite data sets, and the Earthshine record is far too short to do anything with even if we knew the data were correct. So by default I was left with two data sets to look at – ICOADS, which is based on approximately 200 million visual cloud cover observations taken from ships, and CRU TS 3.10, which is based on an unknown number of visual cloud cover observations from 4,000 weather stations on land. I chose ICOADS for the following reasons:
It covers the oceans, i.e. up to 70% of the Earth’s surface, while CRU covers only land areas.
There are a large number of other ICOADS variables I can compare ICOADS cloud cover with. There are no other variables I can compare CRU cloud cover with, or at least not directly.
The ICOADS data have been averaged into 2 degree grid squares but are otherwise “raw”. I don’t know what’s been done to the CRU observations, but if CRU hasn’t tweaked them in some fashion it would be a first.
But I didn’t choose ICOADS because I thought it was “right”. I chose it simply to see what might drop out after I took a closer look at it.
As to what dropped out, here are the global plots of mean monthly ICOADS cloud cover and mean monthly SST since 1950. The two plots track each other quite closely over the long term, indicating a positive correlation between cloud cover and temperature.
But there’s a negative correlation during the seasonal cycle, with the peaks in the cloud cover record coinciding with the troughs in the SST record. I showed a close up of this in a graph I posted earlier, and here it is again for reference (the second graph is the relevant one):
It turns out, however, that this negative correlation isn’t real. It’s an artifact of summing out-of-phase cycles from the NH and SH, as shown in the next graph. Seasonal correlations are in fact positive in both hemispheres, matching the long-term correlation seen in the graph above, except that cloud cover leads SST by about two months in the SH. (Why? Dunno. But I provisionally interpret it to mean that clouds control SST and not the other way round.)
Right now I can’t think of a way in which thousands of sailors on thousands of ships taking millions of SST and cloud cover measurements all over the world’s oceans for a period of more than 60 years (the effects persist at least as far back as 1950) could possibly have manipulated or otherwise screwed up the readings so as to manufacture a seasonal effect as regular and pervasive as this. So I think we have to go with the assumption that it’s real and that the ICOADS cloud cover record might even be – dare I say it? – right. (Although I’m still not claiming that it is. More work needed.)
And if we go with this assumption we have to consider the possibility that more clouds equate to higher, not lower temperatures, and that the change in cloud cover trend around 2000, if indeed there was one, was from increasing to decreasing clouds and not the other way round.
Tim: Your Earthshine objection doesn’t appear to be about their data, or the method, but about the interpretation Palle makes from it.
“Lets say there is a clear blue sky except for a 100 mile diameter cloud travelling at the same speed as the ship. Get the idea”
And how often does this happen?
more commonly with aircraft but captains route according to the weather which includes wind
Big difference in journey times makes tacking about in the ocean looking for following wind an unprofitable exercise.
We need to know about nightside anyway.
Sure. Not just Palle that lets us down there though.
Roger A: “It [ICOADS] covers the oceans, i.e. up to 70% of the Earth’s surface,”
It covers the shipping lanes and some rare research vessels tracks. About 1% of the Earth’s surface. Granted it’s a useful set of transects, but that’s what it is, a useful set of transects.
we have to consider the possibility that more clouds equate to higher, not lower temperatures
Or that higher temperatures equate to more clouds, as Roy Spencer contends. Given that the cloud feedback is strongly negative as he has demonstrated, this looks the more likely to me. See below for discussion of SH lag of SST behind cloud.
the change in cloud cover trend around 2000, if indeed there was one, was from increasing to decreasing clouds and not the other way round.
I agree there is high uncertainty, but pushed to make a choice, I go with ISCCP and Earthshine because of their concordance with other data as I interpret it. 😉
The ICOADS data have been averaged into 2 degree grid squares but are otherwise “raw”
So it doesn’t show a northern bias. I had a brainfart earlier. Earth closest to Sun in Jan so that’s why T is higher then. Southern oceans soak up summer heat in Jan, not July. Your separate hemisphere plots are informative, thanks for those. Seasonally,the north swings around 0.45C and the south 0.33C. Tells me the south is more efficient at storing heat subsurface, possibly due to longer more continuous latitudinal current flows.
The lead in cloud doesn’t tell us its a principle cause of SST change. On a 24 hour average the TSI at the surface swings 25 Watts over the year – 80W at noon zenith. This is a much bigger effect than any longwave cloud radiative effect. Maybe the leading cloud is due to a solar Svensmark effect, which is effective within about a week, whereas ocean surface T lags a couple of months behind the solstices due to thermal inertia (as you can see from the plot). Such a solar Svensmark effect would be more effective in the southern oceans where there is less land and wider oceans more deprived of CCN from pollen, dust etc. Or perhaps springtime warming produces relatively more cloud even though the absolute T is lower, because of humidity differences over the oceans, and/or a higher availablity of CCN early in the warm season from plankton emissions. Plankton are more numerous in the winter months before the strong solar UV gets them.
Speculation on a postcard to the usual address. 🙂
Note that when the jets and climate zones shift latitudinally many regions see less clouds as they are moved away from active areas.
You really must rely on satellites and the level of Earthshine is a good proxy.
TB:
“(ICOADS) covers the shipping lanes and some rare research vessels tracks. About 1% of the Earth’s surface. Granted it’s a useful set of transects, but that’s what it is, a useful set of transects.”
Allow me to present the following very excellent global surface temperature model replication, constructed by your good self, in which you compare your model with HadSST3, which is 100% derived from this “useful set of transects” that covers only “about 1% of the Earth’s surface”. 🙂
“I agree there is high uncertainty, but pushed to make a choice, I go with ISCCP and Earthshine because of their concordance with other data as I interpret it.”
As I showed, or tried to show, in the plots I posted earlier, ISCCP and Earthshine DON’T match the numerous other cloud cover data sets that are available over this time period. They only match each other, and that only after applying what appear to be some very large adjustments to the Earthshine data.
How large are these adjustments? Well, the seasonal change in albedo figure on page 7 of
Click to access 3_06_Palle.pdf
shows short-term albedo changes of up to 20%. There are questions as to whether these changes really are seasonal (they look more like noise to me) but the figure does give us some additional and more detailed Earthshine data to plot, and in the graph below I’ve superimposed them on the “Earth’s albedo” plot you posted earlier. Words are superfluous:
On the question of cloud/temperature lags I freely admit that I don’t understand why there should be a lag in the SH but not in the NH. My next step will be to see whether the SH lag isn’t a result of something funny happening in a specific area. Stay tuned.
Hope this speculative postcard doesn’t get spammed like the last one did 🙂
I’m not sure you can blame Palle for the variability of Earth’s cloud albedo as seen from Big Bear and CrAO via the Moon. What is interesting about it is the implication for the variability of SW absorption by cloud. The other datasets don’t see as much variability in cloud cover as Earthshine does in reflectivity.
This will partly be due to coverage, since Earthshine is currently only measured at two observatories, but it looks to me like there’s a whole heap of uncertainty around cloud absorption. The ‘from first principles’ physics of Mie scattering is off in it’s predictions of this by 20-30W/m^2, according to papers I’ve seen.
Conclusion is same as Roger A;s; More work and more comprehensive network needed.
Note that the optical depth of an atmosphere is THE parameter that determines the proportion of solar energy that reaches the surface.
That parameter incorporates the netted out effect of all the infinite variety of densities and volumes through all the different types of clouds at every level.
Simply trying to ascertain cloud amounts doesn’t really help.
Earthshine’s reflectivity measurement does achieve it.
If Earthshine’s numbers differ from the others then that is probably because of its better ability to record the netted out change in optical depth.
I much prefer the Earthshine product and hope it can be improved quickly.
Stephen, I agree with that assessment. I wonder how much success Palle is having with recruiting participant observatories. I also wonder if it could be crowdsourced by amateur astronomers, with the right sort of equipment.
More Earthshine stuff here:
5 years old but highly relevant