New Computer Fund

Wednesday, November 26, 2014

Why do the "Alarmists" Love Marcott et al.?

Because it looks like what they want.  The Marcott et al. "hockey" stick is spurious, A result of a cheesy method, not including data that was readily available and a few minor date "correcting" errors.  The authors admit that the past couple of hundreds years are "not robust" but even highly respected institutions like NOAA include the "non-robust" portion as part of a comedic ad campaign.

I don't think that self-deception is actionable since that is just part of human nature, but when the likes of NOAA join the moonbeams, it can become more than comical.  Unlike the Mann hockey stick where inconvenient (diverging) data was cutoff, Marcott et al. just didn't dig a bit deeper to find data that didn't "prove" their point.  A great example is the Indo-Pacific Warm Pool (IPWP).  Oppo et al. published a 2000 year reconstruction in 2009 and Mohtadi et al. a Holocene reconstruction of the IPWP in 2010.  If you compare the two, this is what you "see".

The sparse number of data points in Mohtadi 2010 picks up the basic rough trend, but the Oppo 2009 indicates what is likely "normal" variability.  When you leave out that "normal" variability then compare to "normal" instrumental variability, the instrumental data looks "unprecedented".

Since the lower resolution reconstructions have large, on the order of +/- 1 C of uncertainty and your cheesy method ignores the inherent uncertainty of the individual times series used, you end up with an illusion instead of a reconstruction.

The funny part is that normally intelligent folks will defend the cheesy method to the death instead looking at the limits of the method.  The end result is once trusted institutions jumping on the group think bandwagon.

The data used by Marcott et al. is available on line in xls format for the curious and NOAA paleo has most of the data in text or xls formats so it is not that difficult to verify things fer yerself.  Just do it!

My "just do it" effort so far has the tropical ocean temperatures looking like this.

The tropical oceans, which btw are the majority of the oceans, tend to follow the boring old precessional orbital cycle with few "excursions" related to other climate influencing events like Ice dams building/breaking, volcanoes spouting off and the occasional visit of asteroids wanting a new home.  That reconstruction ends in 1960 with some "real" data and some last known values so there is not so much of a "non-robust" uptick at the end.  It only includes "tropical" reconstructions and there are a few more that I might include as I find time and AD reconstructions to "finish" individual time series.

Saturday, November 22, 2014

The Problem with Changing your Frame of Reference from "Surface" Temperature to Ocean Heat Content

The few of you that follow my blog know that I love to screw with the "geniuses" aka minions of the Great and Powerful Carbon.  Thermodynamics gives you an option to select various frames of reference which is great if you do so very carefully.  Not so great if you flip flop between frames.  The minions picked up the basics of the greenhouse effect fine, but with the current pause/hiatus/slowdown of "surface" warming, they have jumped on the ocean heat content bandwagon without considering the differences that come with the switch.

When one theorizes about the Ice Ages of Glacial periods, when the maximum solar insulation is felt at the 65 north latitude, the greater energy would help melt snow and ice store on land in the higher northern latitudes.  (Update:  I must add that even the shift to maximum 65 north insolation is not always enough to end an "ice age".)  Well there is more land mass in the northern hemisphere, so with more land benefiting from greater solar, what happens to the oceans that are now getting less solar?  That is right sports fans, less ocean heat uptake.  There is a northern to southern hemisphere "seesaw" because of the variation in the land to ocean ratio between the hemispheres.

So the Minions break out something like this Holocene temperature cartoon.

Then they wax all physics-acal about What's the Hottest Temperature the Earth's Been Lately moving into how the "unprecedented" rate of Ocean Heat Uptake is directly caused by their master the Great Carbon.  Earth came from the word earth, dirt, soil, land etc.  The oceans store energy a lot better than dirt.

If you want to use ocean heat content, then you need to try and reconstruct past ocean heat content.  The tropical oceans are a pretty good proxy for ocean heat content so I put together this reconstruction using data from the NOAA Paleoclimate website.  Based on this quick and dirty reconstruction, the oceans have been warming during the Holocene and now that the maximum solar insolation is in the southern hemisphere, lots more ocean area, the warming of the oceans should be reaching its upper limit.  Over the next 11,000 years or so the situation will switch to minimum ocean heat uptake due to the solar precessional cycle.  Pretty basic stuff.

With this reconstruction, instead of trying to split hairs, I just used the Mohtadi, M., et al. 2011.
Indo-Pacific Warm Pool 40KYr SST and d18Osw Reconstructions. which only has about 22 Holocene data points to create bins for the average of the other reconstructions, Marchitto, T.M., et al. 2010.
Baja California Holocene Mg/Ca SST Reconstruction, Stott 2004 Western Tropical Pacific and the two,  Weldeab et al. 2005&6 equatorial eastern and western Atlantic reconstructions.  There are plenty more to choose from so if you don't like my quick and dirty, go for it, do yer own.  I did throw it together kinda quick so there may be a mistake or two, try to replicate.

I have been waiting for a while for a real scientist to do this a bit more "scientifically", but since it is raining outside, what the heck, might as well poke at a few of the minions.

Update:  When Marcott et al. published their reconstruction they done good by providing a spread sheet with all the data.  So this next phase is going to include more of the reconstructions used in Marcott et al. but with a twist.  Since I am focusing on the tropical oceans, Mg/Ca (G. Ruber) proxies are like the greatest thing since sliced bread.  Unfortunately not all of the reconstructions used extend back to the beginning of the Holocene.  The ones that don't will need to be augmented with a similar recon is a similar area if possible or they are going to get the boot.  So far these are the (G. Ruber) reconstructions I have on the spread sheet.

As you can start to see, the Holocene doesn't look quite the same in the tropics.  There isn't a much temperature change and some parts of the ocean are warming throughout, or almost and some are cooling.  The shorter reconstructions would tend to bump the end of the Holocene up which might not be the case.  That is my reason for giving them the boot if there are others to help take they a bit further back in time.  As for binning, I am going to try and shoot for 50 years if that doesn't require too much interpolation.  Too much, is going to be up to my available time and how well my spread sheet wants to play.  With 50 year binning I might be able to do 30 reconstructions without going freaking insane waiting for Open Office to save every time I change something.  I know, there are much better ways to do things, but I am a programming dinosaur and proud of it.

Update: After double checking the spread sheet, a few of the shorter reconstructions had been cut off due to the number of points in my lookup table.  After fixing that, the shortest series starts 8600 years before present, 1950.

That is still a bit shorter than I want but better.  The periods where there aren't enough data points tend to produce hockey sticks upright or inverted which tends to defeat the purpose.  So until I locate enough "cap" reconstructions, shorter top layer or "cap" reconstructions can bring data closer to "present" and lower frequency recons to take data points back to before the Holocene starting point, I am trying Last Known Value, instead of any fancy interpolation or curve fitting.  That just carries the last available data value to the present/past so that the averaging is less screwed up.  So don't freak, as I find better extensions I can replace the LKV with actual data.  This is what the first shot looks like.

Remember that ~600 AD to present and 6600 BC and before have fill values, but from 6500 BC to 600 AD the average shown above should be pretty close to what actual was there.  The "effective" smoothing is in the ballpark of 300 years, so the variance/standard deviation is small.  Based on rough approximation, a decade bin with real data should have a variance of around +/- 1 C.  Also when comparing SST to "surface" air temperature, land amplifies tropical temperature variations.  I haven't figured out any weighting so far that would not be questioned, but weighting the higher frequency reconstructions a bit more would increase the variation.  In any case, there is a bit of a MWP indication and possibly a bigger little ice age around 200-300 AD.

Correction, +/- 1 C variance is a bit too rough, it is closer to 0.5 C for decade smoothing ( standard deviation of ~0.21 C) in the tropics 20S-20N that I am using. For the reconstruction so far the standard deviation from 0 to 1950, which has LKV filling is 0.13 C.  So instead of 1 SD uncertainty, I think 2 SD would be more appropriate estimate of uncertainty.  I am not at that point yet, but here is a preview.

A splice of observation with decadal smoothing to the recon so far looks like that.  It's a mini-me hockey stickette about 2/3rds the size of NOAA cartoon.  The Marcott "non-robust" stick is mainly due to the limited number of reconstructions making it to the 20th century which LKV removes.

Now I am working on replacing more of the LKV fill with "real" data.  One of the reconstructions that I have both low and high frequency versions of is the Tierney et al. TEX 86 for Lake Tanganyika which has a splicing choice.;  The 1500 year recon is calibrated to a different temperature it appears than the 60ka recon.  Since this is a Holocene reconstruction, I am "adjusting" the 1500 year to match the short overlap period of the longer.  That may not be the right way, but that is how I am going to do it.  There are a few shorter, 250 to 2000 year regional reconstructions that I can use to extend a few Holocene reconstructions, but it looks like I will have to pitch a few that don't have enough overlap for rough splicing.  Here is an example of some of the issues.

This is the Oppo et al. 2009 recon of the IPWP that I use very often because it correlates extremely well with local temperatures combined with the lower resolution Mohtadi 2010 recon of the same region.  They overlap from 0AD to 1950, but there is very little correlation.  Assuming both authors knew what they were doing, there must be an issue with the natural smoothing and/or dating.  Since both are in C degrees there would be about +/- 1 C uncertainty and up to around +/-300 years dating issues.  If I wiggle and jiggle to get a "better" fit, who knows if it is really better?  If I base my uncertainty estimate on the lower frequency recon of unknown natural smoothing, I basically has nice looking crap.  So instead I will work under the assumption that the original authors knew what they were doing and just go with the flow, keeping in mind that the original recon uncertainty has to be included in the end.

With most of the reconstructions that ended prior to 1800 "capped" with shorter duration reconstructions from the same area, often by the same authors, things start looking a bit more interesting.

Instead of a rapid peak early and steady decline, there is more of a half sine wave pattern that looks like precessional cycle solar peaking around 4000 BC then starting a gradual decline.  Some of the abrupt changes, though not huge amplitude changes in the tropics, appear around where they were when I was in school.  There is a distinct Medieval Warmer Period and an obvious Little Ice Age,  I am kind of surprised the original authors of the studies have left the big media reconstructions to the newbies instead of doing it themselves.

I have updated the references and noticed one blemish, Rühlemann et al. 1999, uses an Alkenones proxy with the UK'37 calibration.  Some of the "caps" are corals since there was not that many to chose from.  Since the corals are high resolution, I had to smooth some to decade bins to work in the spreadsheet.  I am sure I have missed a reference here or there, so I will keep looking for them any any spreadsheet miscues that may remain.  Stay tuned.


Since there is a revised version here is how it compares to tropical temperatures.  I used the actual temperatures with two scales to show the offset.  The recon and observations are about 0.4 C different and of course the recon is over smoothed compared to the decade smoothed observation.  Still a mini-me hockey stick at the splice but not as bad as most reconstructions.

references: No: Author

31 Kubota et al., 2010 (32)
36 Lea et al., 2003 (36)
38 Levi et al., 2007 (39)
40 Linsley et al., 2010 (40)
5 Benway et al.,2006 (9)
41 Linsley et al., 2011 (40)
45 Mohtadi et al., 2010 (43)
60 Steinke et al., 2008 (56)
62 Stott et al., 2007 (58)
63 Stott et al., 2007 (58)
64 Sun et al., 2005 (59)
69 Weldeab et al., 2007 (65)
70 Weldeab et al., 2006 (66)
71 Weldeab et al., 2005 (67)
72 Xu et al., 2008 (68)
73 Ziegler et al., 2008 (69)

Not in the original Marcott.
10(1) with Oppo et al. 2009
36(1) with Black et al. 2007
38(1) with Newton et al. 2006
74 Rühlemann et al. 1999
75(1)Lea, D.W., et al., 2003, with Goni, M.A.; Thunell, R.C.; Woodwort, M.P; Mueller-Karger, F.E. 2006

Thursday, November 20, 2014

Why is Volcanic Forcing so Hard to Figure Out?

I have a few posts on volcanic forcing and it is very difficult to nail down what is due to what because Solar and Volcanic tend to provide mixed signals.  As far as Global Mean Surface Temperatures go, there is a short term impact some of the time that is easy to "see" but at times volcanic response tends to lead volcanic forcing which you know just cannot be true.  The problem is multifaceted.

First, there are internal ocean pseudo oscillations that are amplified in the NH and muted in the SH.  The SH really doesn't have any room on the down side to have much easy to spot volcanic response and thanks to the thermal isolation of the Antarctic, not much of an upside.  There is a response, just temperature anomaly doesn't do it justice.

However, if you normalize the anomalies by dividing by the standard deviation you can "see" that both higher latitudes, above 40 degrees and the bulk of the oceans 40S-40N have similar responses.  Since the NH has land mass choking the pole ward flow, it has a larger fluctuation, but about the same longer term trend.  Temperature in this case, especially anomaly based on "surface" temperatures doesn't accurately represent energy change.  This gets back to Zeroth Law requirements that the "system" a fair size planet, should be somewhat in equilibrium.  A simple way to see that is a 10 degree swing in polar winter temperatures would only represent about a third of the same swing in the bulk of the oceans.  Around 40 degrees to the poles, the surface has more energy advected from the bulk than provided by solar.  Since advected energy is as or more important than solar, volcanic forcing reducing solar irradiance has less impact.

Second, due to Coriolis effect and differences in land mass, you will have ocean oscillations.  I have previously mentioned that there are three major thermal basins, North Atlantic, North Pacific and Southern Hemisphere with ratios of roughly 1:2:4 which can produce a fairly complex coupled harmonic oscillation.  Since sea ice extent would vary the amplitude, especially in the NH due to land amplification of advected energy, you can just about drive yourself nuts trying to figure out cycles.  Yes, I know that ocean oscillations are a part of both of these reasons.  That is because they influence surface temperature both ways since in addition to providing energy that energy drives precipitation that can produce snow that can reverse the "surface" temperature trend.  In the first case, advected energy offset volcanic forcing and in this case, advected energy causing increase snowfall can enhance volcanic forcing.

To illustrate, this chart compares the Northern Hemisphere and northern high latitude temperature reconstruction by Kobashi et al. with the Indo-Pacific Warm Pool reconstruction by Oppo et al. that I use frequently.  While I cannot vouch for the accuracy of the reconstructions, they should reasonably reflect the general goings on.  The Kobashi reconstruction doesn't have a pronounce Medieval Warm Period since that is more a land based phenomenon, think less glacial expanse, and a less obvious but longer Little Ice Age since land based ice doesn't have much impact on ocean currents other than more sea ice/ice melt can cause impact the THC.  Solar and Volcanic impact on the ocean heat capacity would be a slow process requiring up to 300 year per degree with an average forcing of around 1 to 2 Watts per meter squared.  Impacts felt in the SH may take 30 to 100 years to migrate back to the other hemisphere producing those delightful pseudo oscillations we have all grown to love.  So "natural" variability can cause more "natural" variability since the original forcing can't be sorted out.

In the NH which is more greatly influenced by the oscillations, volcanic forcing on an upswing might have no discernible impact while on the downswing it might have a larger than expected impact.  Add to that potential impacts of volcanic ash on snow melt which would be affected by the the time of year and general atmospheric circulation trends and you have some serious job security if your employer has the patience.

Finally, the southern hemisphere temperature records generally suck big time.  That is simply because the winds and temperatures that work so well isolating the Antarctic  are not all that hospitable for folks that might want to record temperatures.  That means that early 20th century correlations are not reliable.  So there might have been more out of phase situations, the hemispheric "seesaw" effect, that would help determine actual "cause" of some of the fluctuations.

So what does this all mean?  Glacials and interglacials are due to glaciers, ice on land.  They are the number one climate driver so when there is a lot, the climate is more sensitive to things that melt ice and when there isn't much, climate is less sensitive in general.  Since evidence suggests that more ice is more common you can have a lot of climate change without much ocean energy change.  If you want to reconstruct "surface" temperature you need to reconstruct glacial extent.  If you want to reconstruct ocean temperatures, stick to the bulk of the oceans especially the tropics.  Since the Indo-Pacific Warm Pool has lower THC through flow, it would be your best spot for estimating the energy changes in the "global" oceans.

That said, the average temperature of the oceans is about 4 C degrees which happens to be about the "ideal" black body temperature of a rock in space located about Earth's distance from our sun.  Based on that temperature, "Sensitivity" to a doubling of CO2 would be about 0.8 C "all things remaining equal".  4 C by the way is roughly an ideal black body energy of 334.5 Wm-2 which is roughly equal to the best estimates of down welling longwave radiation and since Stefan-Boltzmann included that ~0.926 fudge factor, "normal" DWLR could be between 334.5 and about 360 Wm-2.

If the Oppo et al. 2009 reconstruction is reasonably accurate, then Earth's climate is about "normal" for the past 2000 years, unless you think the Little Ice Age should be "normal".  In any case, things should start slowing down now that climate is closer to the back half of the Holocene mean.

Based on comparing the NODC Ocean Heat/Temperature data, the Oppo et al. 2009 IPWP trend from ~1700 is very close to the projected trend in upper ocean (0-700 meter) average vertical temperature anomaly.  Also while I took some liberties "fitting" lag response times, the combination of solar and volcanic forcing estimated by Cowley et al and Stienhilbert et al.  (Sol y Vol) do correlate reasonably well with the Oppo et al. reconstruction.

To conclude, trying to determine "global" changes in average SST using higher latitude data tends to remove information if the hemispheric "seesaw" is not properly considered and logically the majority of ocean heat should be better represented by the bulk of the oceans of which the Indo-Pacific Warm Pool is a good representation.  To say that there is no evidence of long term persistence/memory in the vast oceans is a bit comical.  

Sunday, November 9, 2014

Cowtan and Way Make it to Climate Explorer

Since Lewis and Curry 2014 was published the fans of Cowtan and Way's satellite kriging revision to the Hadley Centers HADCRUT4.something is constantly mentioned as a L&C spoiler.  C&W does run warmer than the older HADCRUT4.somethingless.  I haven't used much of the C&W new and improved version since I like to check regions using the Climate Explorer mask capability.  Well, now C&W are on Climate Explorer.

This is one of my first looks at any data set, tropics, extended slightly to 30S-30N and the extratropics, 30 to 90.  As expected, the majority of the world is consistently warming as would be expected recovering from a period former known as the Little Ice Age and the Northern Extratropical region is warming more because that would be where ice during the period formerly known as the Little Ice Age would have been stored.  Ice/snow is a major feedback and is influences by temperature and dirtiness, albedo.  Industry, wild fires, wind blown erosion, volcanic ash, earthquakes and just about anything that disturbs the pristine whiteness of glaciers and snow fields would impact the stability of the glaciers and snow fields.  So in my opinion, the Rest of the World is a better baseline for "global" warming than the Northern Extratropical region if you want to try and attribute things to CO2, Volcanic sulfates or clouds.  Not that the Northern Extratropical aren't important, but you can see they are noisy as all hell.

Climate Explorer also allows you to use nation masks like the one above for New Zealand.  New Zealand skeptics are quite a bit miffed by the amount of adjustments used by national weather services to create the scary warming trend.  New Zealand is a large island and as such would have a maritime climate.  Since Cowtan and Way krige using satellite data which is heavily influenced by ocean temperatures, the C&W version of New Zealand looks a lot more like the skeptical Kiwis would expect.  I use three tiered 27 month smoothing and left the first two stages in so you can see how smoothing impacts the Warmest Year Evah.  The default baseline for Climate Explorer is 1980 to 2010, the satellite era which would also have an impact on Warmest Year Evah, but not the general trend.  This is a reminder that "Global" temperature reconstructions really should not be used for "local" temperature references since the interpolation, standard or krige, smears area temperatures, good for "Global" stuff but not so good for local forecasting.

 So there is a brief intro to HADCRUT4.something with new and improved smearing.  Not meaning to be a cut down of C&W, it serves a purpose, but increase smearing reduces the relationship between temperature and the energy it is supposed to represent.  Now if C&W had the "skepticism" required they would compile a "Global" energy anomaly reconstruction which would illustrate just how much error using temperature anomaly for temperature ranges from -80C to 50C can produce.  Then they would focus on the "other" forces that effect climate rather than just the CO2 done it all meme.  Since Way is Inuit, I am sure he knows about yellow snow and the variety of other than pristine white varieties.  It would be nice to see him focusing on the whole picture instead of trying to be a good CO2 band member.

Dirty Greenland Glacier courtesy NOAA