New Computer Fund

Sunday, August 19, 2012

Back from the Future

More Stuff at the End
The choice of base line is critical in climate science.  By "cherry picking" a base line you can determine that something is "unprecedented" or not "unprecedented".  Welcome to non-linear dynamics

Being I am a rebel without a scientific clue, I have "cherry picked" my base line as the AQUA quality era from 2002 to 2010.  Quality because the Sea Surface channel took a dump about 2010.  My hypothesis is that the Glacial cycles are primarily influenced by Wobble.  I was told by the real scientific types I needed a theory, so that is my theory, though it does tend to blend a number of other theories.  So it is not really all that original.  Possibly the only original part is my estimate of the range of bi-stability.  I haven't figured out how to make a convincing argument yet, but there appears to be an approximately 2C separation between the low normal and high normal bi-stable SST set points controlled by the fresh water and salt water freezing points.  Based on Tierney et al. TEX86 Lake Tanganyika temperature reconstructions and the apparent peak-peak values of Neukum et al. Southern South American temperature reconstructions, the 2002 to 2010 base line appears to be close to the high normal.

There is considerable uncertainty in the paleo reconstructions and the TEX86method  is prone to have more potential issues, so high normal range is a touch iffy.  Now the paleo fans will know that the Northern hemisphere and Antarctic reconstructions of past temperature show a hellava lot more variation than this chart shows.  Looking back at the first chart there are regression lines for RSS northern hemisphere and southern hemisphers.  The regression line for the southern hemisphere is a reasonably nice match to the slope of the GISS Land and Ocean Temperature Index (LOTI).  That can be an indication of the lag of SST with land temperature change.  You can graph the RSS NH with the BEST land temperature product and find that it more closely matches land temperature change.

This is a close up of the Instrumental temperature data with the BEST volcanic forcing estimate replacing GISS LOTI.  Using the 2002 to 2010 "cherry picked" base line, there appears to be stronger sensitivity to forcing in the northern hemisphere than the southern hemisphere.  Comparing the 1983 and 1991 volcanic pulses, the sensitivity to forcing appears to decrease, especially in the southern hemisphere.  Of course, both of those "pulses" where NH volcanoes, but the 1963 "pulse" appears to have had a greater impact of SST pre satellite era.  Decreasing sensitivity to forcing "pulses" would be a characteristic of a system closer to a control set point as would the over shoot and decay patterns that if you squint hard enough, you can "eyeball" in the chart above.

Since it is likely that a good deal of the depression below the high normal set point would be natural, CO2 forcing would be most discernible at the upper or lower bi-stable set points.  Pretty much explains why Dr. Vaughan Pratt notices a 14.5 year lag in the CO2 forcing that only is apparent in the past 30 years.

Of course, being scientifically clueless, there is a fairly high probability that I am wrong.  After all, I am just a guy that has experience measuring temperatures and thermal capacities while adjusting control systems in non-linear systems with regions of instability.  It is not like I was silly enough to design them.

More Stuff

 I just added the GISS land only and the equations for the RSS trend lines.  You can compare the mean value line for the GISS land and HADSST2.

This is a close up of my "cherry picked" base line from the 1995 climate regime change.  Note the 0.008 linear trend for RSS northern hemisphere.  That would be 0.08 per decade or 0.8 per century. There have been some positive land use changes since 1995, so that would have to be considered before estimating climate sensitivity to CO2 doubling.   

Even more stuff

This is my K-Mart sequential linear regression analysis of the same time series combination.  Starting at point A, HADSST2 leads the way with the vertical linebetween A and B showing the lag.  At point B there is a shift where HADSST2 decreases but both GISS series continue.  Around 1988 there is a reorganization of sorts and all three converge in 1995, what I consider a regime change.  Block C highlights the relationships of all series around the 2002-2010 base line period.

To highlight the relationship this plot compares the ration of GISS LOTI and LAND to HADSST2.    Since 2002 to 2010 is the common base line, there would be a convergence, but the reduction in variation looks a little stringer than I would expect with the common base line.  Changing the base line to 1980 to 1988 just offset the two plots slightly.  The change in variance could be due to instrumentation improvements, but I am more suspicious of the 1995 regime change since it is consistent in all data sets.

No comments:

Post a Comment