cc: Dave Thompson <davet@atmos.colostate.edu>
date: Mon Oct 19 16:58:19 2009
from: Phil Jones <p.jones@uea.ac.uk>
subject: Re: food for thought...
to: Dave Thompson <davet@atmos.colostate.edu>, John Kennedy <john.kennedy@metoffice.gov.uk>, Mike Wallace <wallace@atmos.washington.edu>

    Dave,
       Have had a look at these one and off today.
    My suggestion would be to smooth these series then look at the results. Maybe something at
   about 5 year timescale.
    If there was a problem with the NH SSTs in 1969 it should also affect the SH - it doesn't.
    You could also correlate your red line with gridded land data. It will be noisier, but I'd
   reckon that the basic pattern would show there.
     I think what you're seeing is the aerosol cooling of the NH. It's also in the SH, but
   much lower amplitude.
    Try a comparison with some of the global dimming/brightening series. There are a number of
   papers on this my Martin Wild.
    A proxy for this is the DTR - See Fig 1 in the attached. This is global. What you want is
   separate ones for the NH and SH. The NH one will be steeper.
    Cheers
    Phil

   At 22:02 16/10/2009, Dave Thompson wrote:

     Dear Phil, John (cc Mike),
     I've spent some more time looking at the differences between NH and SH
     SSTs... and I'm increasingly convinced the global cooling in the 70s
     occurred over a discrete period, and that it likely reflects some
     residual data issues. Either that, or the climate system changed
     abruptly in one hemisphere ~1969, but not in the other ....
     So... I figured I'd bounce my latest ideas off of you folks to see
     what you think... the last couple of papers were a lot of fun...maybe
     there is another story that needs to be told...
     I've attached three pages of figures. They are based on the corrected
     SST time series John sent me, but all of the results I discuss below
     are evident in HadSST2 as well. The vertical tickmarks are 0.5 K in
     all figures.
     Figure 1 shows NH (top), SH (middle) and the difference between NH and
     SH SSTs (NH-SH). The data have not been filtered in any way. There is
     some mess around WW1 and WW2, and that's to be expected. But there is
     also - to my eye - a distinct drop in the difference between NH and SH
     SSTs ~1969 (1969 is indicated by the dashed line).
     (Note the nice thing about the difference time series is that it
     filters much of the ENSO and volcanic variability.)
     Is the drop also evident in the land data?
     Figure 2 shows NH-SH time series of SSTs (top) and land data (bottom).
     I've filtered the data for COWL, ENSO and Volcano. This makes little
     difference in the case of the SST difference time series (since ENSO
     and volcanos show up in both hemispheres). But it does make a
     difference in the land difference time series, as the COWL pattern
     shows up only in the NH. There is a weak drop in the land data ~1969.
     But nothing like the SST data.
     Looking at the NH only:
     Figure 3 shows the land (top) and SST (bottom) time series for the NH.
     Both have been filtered as per our recent paper. The drop in the NH
     SST time series is very pronounced. Dominant, I'd say. Can't believe I
     missed it earlier. It's not as discrete as the 45 drop. But it's
     certainly rapid.  To my eye, the NH land areas are doing something
     similar to the NH ocean areas ~1969. But it's hard to tell for sure.
     The drop is there in the land data, but it is not as large.
     Looking at the SH region only:
     Figure 4 shows the land (top) and SST (bottom) time series for the SH.
     Both have been filtered as per our recent paper. Both seem to warm
     throughout the century. Note the large drop in land variance ~WW2.
     There is no evidence of the drop in 1969 that is readily apparent in
     the NH SST time series in Figure 3.
     Figures 5 and 6 show one attempt to find the pattern associated with
     the drop. Figure 5 shows the NH-SH difference time series along with a
     fit to the drop in 1969 (as a red line). Figure 6 shows the
     correlations between gridded SSTs and the red line from Fig. 5 over
     the period 1960-1980. It's clear no particular region is giving rise
     to the drop in 1969. Rather, the entire NH appears to have cooled
     rapidly relative to the SH at nearly the same time (red shading
     indicates the NH cooled relative to the SH, and vice versa). I
     explored a range of slightly different analyses to try and tease out
     the source of the drop ... but the results always stayed the same.
     The bottom line...
     - To my eye, the SST difference time series in Figure 2 looks
     suspicious around 1969. The difference is almost 0.5 K (tickmarks are
     at 0.5 K).
     - I think the NH SST time series in Figure 3 is striking... the drop
     in 1969 stands out clearly above the rest of the record. A similar
     drop is not evident in the SH (Figure 4), and that is why the drop
     exists in the difference time series in Figure 2.
     - The NH land areas might be doing something interesting around 1969
     (Figure 3). But as always, they are noisy (even when filtered) and
     it's hard to tell.
     - I think the drop in the NH SST time series in Figure 3 is
     potentially more important for interpreting long-term trends in global- mean
     temperatures than the drop in 1945. It lies close to the middle
     of the period extending from WW2-now...
     So .... what do you think? To your eyes, does the drop in 1969 look
     normal or unusual? John... have you had any luck finding any data
     issues at that time? Maybe the SST data is too high in the 1960s?
     -Dave
     --------------------------------------------------------------------
     --------------------------------------------------------------------
     David W. J. Thompson
     [1]www.atmos.colostate.edu/~davet
     Dept of Atmospheric Science
     Colorado State University
     Fort Collins, CO 80523
     USA
     970-491-3338
     Dear Phil, John (cc Mike),
     I've spent some more time looking at the differences between NH and SH SSTs... and I'm
     increasingly convinced the global cooling in the 70s occurred over a discrete period,
     and that it likely reflects some residual data issues. Either that, or the climate
     system changed abruptly in one hemisphere ~1969, but not in the other ....
     So... I figured I'd bounce my latest ideas off of you folks to see what you think... the
     last couple of papers were a lot of fun...maybe there is another story that needs to be
     told...
     I've attached three pages of figures. They are based on the corrected SST time series
     John sent me, but all of the results I discuss below are evident in HadSST2 as well. The
     vertical tickmarks are 0.5 K in all figures.
     Figure 1 shows NH (top), SH (middle) and the difference between NH and SH SSTs (NH-SH).
     The data have not been filtered in any way. There is some mess around WW1 and WW2, and
     that's to be expected. But there is also - to my eye - a distinct drop in the difference
     between NH and SH SSTs ~1969 (1969 is indicated by the dashed line).
     (Note the nice thing about the difference time series is that it filters much of the
     ENSO and volcanic variability.)
     Is the drop also evident in the land data?
     Figure 2 shows NH-SH time series of SSTs (top) and land data (bottom). I've filtered the
     data for COWL, ENSO and Volcano. This makes little difference in the case of the SST
     difference time series (since ENSO and volcanos show up in both hemispheres). But it
     does make a difference in the land difference time series, as the COWL pattern shows up
     only in the NH. There is a weak drop in the land data ~1969. But nothing like the SST
     data.
     Looking at the NH only:
     Figure 3 shows the land (top) and SST (bottom) time series for the NH. Both have been
     filtered as per our recent paper. The drop in the NH SST time series is very pronounced.
     Dominant, I'd say. Can't believe I missed it earlier. It's not as discrete as the 45
     drop. But it's certainly rapid.  To my eye, the NH land areas are doing something
     similar to the NH ocean areas ~1969. But it's hard to tell for sure. The drop is there
     in the land data, but it is not as large.
     Looking at the SH region only:
     Figure 4 shows the land (top) and SST (bottom) time series for the SH. Both have been
     filtered as per our recent paper. Both seem to warm throughout the century. Note the
     large drop in land variance ~WW2. There is no evidence of the drop in 1969 that is
     readily apparent in the NH SST time series in Figure 3.
     Figures 5 and 6 show one attempt to find the pattern associated with the drop. Figure 5
     shows the NH-SH difference time series along with a fit to the drop in 1969 (as a red
     line). Figure 6 shows the correlations between gridded SSTs and the red line from Fig. 5
     over the period 1960-1980. It's clear no particular region is giving rise to the drop in
     1969. Rather, the entire NH appears to have cooled rapidly relative to the SH at nearly
     the same time (red shading indicates the NH cooled relative to the SH, and vice versa).
     I explored a range of slightly different analyses to try and tease out the source of the
     drop ... but the results always stayed the same.
     The bottom line...
     - To my eye, the SST difference time series in Figure 2 looks suspicious around 1969.
     The difference is almost 0.5 K (tickmarks are at 0.5 K).
     - I think the NH SST time series in Figure 3 is striking... the drop in 1969 stands out
     clearly above the rest of the record. A similar drop is not evident in the SH (Figure
     4), and that is why the drop exists in the difference time series in Figure 2.
     - The NH land areas might be doing something interesting around 1969 (Figure 3). But as
     always, they are noisy (even when filtered) and it's hard to tell.
     - I think the drop in the NH SST time series in Figure 3 is potentially more important
     for interpreting long-term trends in global-mean temperatures than the drop in 1945. It
     lies close to the middle of the period extending from WW2-now...
     So .... what do you think? To your eyes, does the drop in 1969 look normal or unusual?
     John... have you had any luck finding any data issues at that time? Maybe the SST data
     is too high in the 1960s?
     -Dave

     --------------------------------------------------------------------
     --------------------------------------------------------------------
     David W. J. Thompson
     [2]www.atmos.colostate.edu/~davet
     Dept of Atmospheric Science
     Colorado State University
     Fort Collins, CO 80523
     USA
     970-491-3338

   Prof. Phil Jones
   Climatic Research Unit        Telephone +44 (0) 1603 592090
   School of Environmental Sciences    Fax +44 (0) 1603 507784
   University of East Anglia
   Norwich                          Email    p.jones@uea.ac.uk
   NR4 7TJ
   UK
   ----------------------------------------------------------------------------

