New Computer Fund

Wednesday, March 4, 2015

More on that Elusive Cause and Effect

Using the Tropics to Isolate Cause and Effect just leads to more questions.

This chart of 30 year sliding correlations with the Hot Tropics, 10S-10N points out a perturbation circa 1913 which could be due to a volcano not included in forcing estimates used for CMIP5 models runs.  The timing though is off, the perturbation leads the volcano I limited the SST regions to 60S-60N because of issues with early temperature measurements near the arctic circles.

The drop in correlation around 2005 was a bit of a surprise.  So I used the actual temperature data available for land areas, specifically the maximum annual temperature and minimum annual temperature to get a rough estimate of change in atmospheric forcing.  Lots of potential questions about how useful that might be, but variation in air temperature converted to approximate Wm-2 should produce a rough estimate of the total change in atmospheric forcing.

So I expand on that by doing the same thing with changes in sea surface temperature.

Using the same 1901-2013 baseline the energy anomaly for 40S-40N SST peaks at about 3 Wm-2 which is close to the land based forcing estimate.

I used 40S-40N for the SST version but I doubt using 50S-50N would make much difference.

Forty to 60 North is a bit more interesting, there is actually a significant difference, almost 2 Wm-2, at a couple of points.  Early in the record sould be dismissed as limited observation but since the most recent divergence is during the period of the best coverage, that part should be "real".

Forty to 60 South has the worse coverage, but should be some what reasonable after 1930.  This shows less "forcing", about 2 Wm-2, interesting, but the minimum is around 1930.

Comparing the actual temperatures recorded for the high latitude oceans shows a very interesting lag in responses.  The northern 40-60 band has a bottom from roughly 1905 to 1915 with a continuing down trend until it pops around 1985.  The SH portion starts its upward trend around 1930 with a plateau starting around 1980.  Both of these regions represent a fairly small portion of total ocean energy, but since temperature anomalies are not weighted towards energy, they could have a large impact on global mean temperature anomaly.  The northern high latitude ocean can impact more land area which tends to amplify change due to its lower specific heat capacity, so it could produce most of the combined PDO/AMO temperature variation.

Here I need to remind anyone still following that there is a huge change in the ratio of land to ocean area in the Northern Hemisphere.  The 60S-40S ocean region is nearly all ocean.  Latent heat loss from this region is much likely to return to the ocean in this region.  Water vapor from the surface would form clouds, releasing latent, but as the water falls as precipitation it would warm with the air regaining a large portion of that energy.

In the Northern Hemisphere it is much more likely that precipitation would fall on land transferring energy to that land then it would return to the oceans via river outflow with some delay.  When there is more water stored in sheds, reservoirs or ground water, the delay would be much longer.

The ratio of the 40N-60N band is 0.45 ocean, 20N-40N band 0.60 ocean and 0-20N band 0.75 ocean so the further north the more likely precipitation falls on land and is stored for some time, the further south the less likely precipitation is stored on land mass.  Above 60 degrees, in the north the stability of sea ice determines if the precipitation is stored and in the south, the Antarctic land mass would provide more consistent storage.

In a nut shell, northern hemisphere especially higher latitude, land and water use changes would have a much greater impact on climate via the hydrology cycle than in any other latitude band.  It isn't particularly easy to determine how much impact those changes actually have though without a "normal" period to be used as a reference.  It would generate lots of noise in the climate data.

 If you look at the temperature anomaly for the higher latitude bands you can see how response varies.  The 40N-60N band has less ocean area meaning a lower heat capacity, so it would have a faster response time.  These anomalies are based on the same 1901-2013 base line.

If you compare the actual temperature of the 40N-60N band with the bulk of the oceans, 40S to 40N you see about the same fast response.  Area wise, the 40N-60N is about 10% of the total area of these two and energy wise, 40N-60N provides about 8.8% of the entire area.

What all this seems to indicate that there is noise that makes any attempt to determine cause and effect pretty difficult.  The only consistent part of all these comparisons is that there has been some increase in atmospheric forcing on the order of 3 Wm-2 over the instrumental period.

Specific heat capacity is a great noise filter and since this is really an energy problem, the more specific heat capacity the better.

 Sea level rise may be the go to metric to estimate "global warming" in the long run.  Scaling the Jevrejeva et al sea level rise reconstruction to tropical 25S-25N SST indicates a long term increase with a slight acceleration over the past 150 years.  As you can see it pretty well filters out the noise.  I used a 1854 to 2002 baseline for this effort which indicates "noise" aka natural variability is likely on the order of +/-0.35 C or so at least in the tropics.  One could try to make a case for Volcanic forcing or solar, but considering the Crowley and Unterman 2013 Volcanic Aerosol Optical Depth reconstruction, there are more exceptions to the normal concept of atmospheric forcing than agreements.

So using SLR as a reference, today's temperatures may seem exceptional, but Berkeley's land temperature reconstruction seems to indicate other wise.

Sea level rise also filters out the oscillations in the Oppo et al 2009 Indo-Pacific Warm Pool reconstruction.

Just like the 30N-60N SST comparison above, Northern Hemisphere reconstructions appear to have the same fast response.  This Christian and Ljungqvist reconstruction is for 30N to 90N and uses a lot of tree ring reconstruction in very high northern latitudes.  Each of these individual reconstructions have a great deal of variability and how a multi-proxy combination ends up looking depends a lot on choice of proxies and methodology.  Spliced on the end I have GISS land only in both annual and 50 year moving average to match the C&L recon.  Their 2009 version is archived at NCDC with 71 of the reconstructions in an easy to download file with actual temperatures in most cases so you can compare the relative energy contribution as a double check of their weighting.  From that you can see most of the real signal is in the lower latitudes where there is more energy per unit anomaly.  The sigma variations they use don't give you any sense of how those recons should be weighted other than a rough area.

I picked this recon btw to show how messing with smoothing has become an art for some of the less ethical.

Skeptical Science played the loess game of hockey stick enhancement which produces those too good to be true uncertainty intervals so popular with advocates.  You can easily download the data from NCDC and instrumental from Climate Explorer to make you own comparison should you think I am some kind of lying SOB.

Using the less adulterated data I believe indicates a good deal of the "cause" of the warming is buried in multi-century scale wandering and less that representative presentation of selected facts.  Focusing on energy anomaly instead of temperature anomaly might limit some of the creativity.

No comments:

Post a Comment