The Great Stasis

(c) 2022 by Barton Paul Levenson



The poster on RealClimate who signs himself Victor (possibly Victor Grauer of my home town of Pittsburgh) is that rarity, a liberal AGW denier. Victor insists there's no correlation between carbon dioxide and mean global annual surface temperature. Of course there is--see, for example, here. Victor doesn't know his statistics. But let's look at a possible explanation for his mistake.

He insists, for example, that while CO2 rose from 1940 to 1978, temperature fell. That's not quite true--while the slope of temperature versus year for that period is negative at -0.00236 K per year, the t-statistic on the slope is insignificant: t = -1.27, for p < 0.212. So the error bars are wider than the slope, and for all we know, given the noisiness of the data, it could have actually been rising. But give him his point for the moment. Does it prove what he thinks it does?

No. CO2 is not the only thing that affects temperature. Other factors known to have effects on Earth's temperature are sunlight, volcanic aerosols, aerosols from pollution, heat cycles between the atmosphere and the ocean (such as El Niņo), variations in cloudiness, other greenhouse gases, etc., etc., etc. So some other factor might have suppressed temperature during those years. By 1978 (actually, according to statistician Grant Foster, 1975) CO2 had come to dominate everything else and global warming really took off.

Let's examine a bunch of data for the period. Here is data on temperature and several other factors for the years 1850-2019. Not all are complete, unfortunately, but we've got complete coverage from 1866 onward and 2018 backward.


YeardTCO2SNTSISOIAerosols
1850-0.239284.7126.31361.7005-0.077
1851-0.004284.91221361.5037-0.027
1852-0.137285.0102.71361.4177 0.004
1853-0.326285.1 74.11361.2921 0.110
1854-0.125285.3 391361.0975 0.237
1855-0.522285.4 12.71360.9770-0.210
1856-0.587285.6  8.21360.8772 0.036
1857-0.580285.7 43.41360.9499-0.335
1858-0.555285.9104.41361.1913-0.090
1859-0.193286.1178.31361.4779-0.051
1860-0.649286.2182.21361.6335-0.217
1861-0.516286.4146.61361.6242-0.022
1862-0.756286.5112.11361.4956-0.230
1863-0.275286.6 83.51361.3489 0.302
1864-0.825286.8 89.21361.2972-0.092
1865-0.453286.9 57.81361.2800-0.083
1866-0.440287.0 30.71361.1589-0.16 0.039
1867-0.603287.1 13.91361.0301-0.03-0.471
1868-0.424287.2 62.81361.1520-0.88-0.317
1869-0.393287.4123.61361.4661 0.55 0.057
1870-0.549287.52321361.6794-0.15-0.236
1871-0.659287.7185.31361.9733-0.15-0.087
1872-0.355287.9169.21361.7310 1.91-0.056
1873-0.384288.1110.11361.6337 0.08-0.259
1874-0.483288.4 74.51361.3396 0.59-0.342
1875-0.832288.7 28.31361.0057 0.27-0.019
1876-0.497289.0 18.91360.7805 0.44-0.024
1877-0.252289.4 20.71360.7192-1.80-0.080
1878-0.023289.8  5.71360.6067 0.17-0.127
1879-0.553290.2 101360.5930 1.27-0.334
1880-0.363290.7 53.71360.7627 0.79 0.003
1881-0.440291.2 90.51361.0932-0.59-0.152
1882-0.336291.7 991361.1539-0.60 0.065
1883-0.617292.1106.11361.1109-0.25-0.364
1884-0.724292.6105.81361.3241-0.38-0.262
1885-0.693293.0 86.31361.2171-0.68-0.138
1886-0.532293.3 42.41361.0569 0.70-0.119
1887-0.578293.6 21.81360.8577 0.47-0.058
1888-0.594293.8 11.21360.8100-1.27-0.101
1889-0.276294 10.41360.7764 0.16-0.046
1890-0.446294.2 11.81360.7650 0.56-0.191
1891-0.536294.3 59.51360.8840-0.28-0.218
1892-0.633294.5121.71361.1688 0.52-0.154
1893-0.708294.61421361.3458 0.99-0.104
1894-0.484294.71301361.4701 0.21-0.199
1895-0.574294.8106.61361.3363-0.25-0.143
1896-0.395294.9 69.41361.2281-1.55-0.217
1897-0.310295.0 43.81360.9385-0.68 0.060
1898-0.396295.2 44.41360.8914 0.57-0.062
1899-0.329295.5 20.21360.8152 0.16-0.352
1900-0.194295.8 15.71360.7290-0.65-0.337
1901-0.157296.1  4.61360.6394 0.07-0.147
1902-0.398296.5  8.51360.6319 0.10-0.067
1903-0.432296.8 40.81360.7228 0.47 0.204
1904-0.533297.2 70.11361.0143 0.39-0.353
1905-0.405297.6105.51361.0005-1.79 0.039
1906-0.184298.1 90.11361.2441 0.21-0.600
1907-0.606298.5102.81361.1264-0.20-0.295
1908-0.457298.9 80.91361.1528 0.28-0.178
1909-0.423299.3 73.21361.0208 0.29-0.398
1910-0.333299.7 30.91360.9185 1.26-0.545
1911-0.415300.1  9.51360.7133-0.64-0.294
1912-0.434300.4  61360.6625-0.98-0.454
1913-0.306300.8  2.41360.7110-0.68-0.330
1914-0.063301.1 16.11360.7851-0.93-0.144
1915-0.06301.4 791361.0964 0.16-0.046
1916-0.346301.7 951361.4054 0.64-0.105
1917-0.652302.1173.61361.5701 2.14-0.391
1918-0.417302.4134.61361.7757 0.05-0.323
1919-0.275302.7105.71361.4560-1.09-0.411
1920-0.262303.0 62.71361.1199 0.09-0.363
1921-0.099303.4 43.51360.8915 0.66-0.360
1922-0.248303.8 23.71360.7013 0.33-0.309
1923-0.263304.1  9.71360.5976-0.36-0.427
1924-0.314304.5 27.91360.6045 0.34-0.441
1925-0.212305.0 741360.7004-0.24-0.383
1926-0.064305.4106.51361.0393-0.49-0.498
1927-0.223305.8114.71361.3065 0.27-0.630
1928-0.146306.3129.71361.2413 0.42-0.231
1929-0.469306.8108.21361.1000 0.46-0.373
1930-0.108307.2 59.41361.1858 0.03-0.335
1931-0.072307.7 35.11360.9547 0.39-0.378
1932-0.076308.2 18.61360.8472-0.68-0.265
1933-0.317308.6  9.21360.8356 0.09-0.112
1934-0.025309.0 14.61360.8648-0.01-0.088
1935-0.163309.4 60.21360.9766 0.14-0.286
1936-0.117309.8132.81361.4379 0.03-0.615
1937-0.032310.0190.61361.7170 0.08-0.251
19380.152310.2182.61361.6735 0.87-0.356
1939-0.002310.31481361.6299 0.02-0.606
1940-0.041310.41131361.3999-1.52-0.446
1941-0.029310.4 79.21361.2455-1.45-0.584
1942-0.046310.3 50.81361.0801 0.02-0.801
19430.010310.2 27.11360.9052 0.36-0.635
19440.107310.1 16.11360.9080-0.28-0.801
1945-0.119310.1 55.31361.1456 0.42-0.493
1946-0.078310.1154.31361.2842-0.79-0.453
1947-0.005310.2214.71361.7197 0.16-0.410
19480.000310.31931361.9235-0.23-0.652
1949-0.108310.5190.71361.7306-0.21-0.422
1950-0.301310.7118.91361.6156 1.51-0.475
1951-0.107311.1 98.31361.1933-0.69-0.708
1952-0.063311.5 451361.1126-0.22-0.811
19530.129311.9 20.11361.0253-0.77-0.710
1954-0.145312.4  6.61360.9819 0.23-0.541
1955-0.192313.0 54.21361.1123 0.90-0.286
1956-0.426313.6200.71361.4949 1.02-0.826
1957-0.078314.2269.31362.0982-0.45-0.881
19580.041314.9261.71362.3628-0.50-0.743
19590.019315.97225.11362.0576-0.10-0.589
1960-0.068316.911591361.9256 0.29-0.656
19610.039317.64 76.41361.5795-0.01-0.826
19620.027318.45 53.41361.1854 0.39-0.909
19630.024318.99 39.91361.0138-0.32-0.903
1964-0.283319.62 151360.8779 0.54-1.001
1965-0.233320.04 221360.8301-0.95-0.762
1966-0.132321.38 66.81360.9332-0.54-0.785
1967-0.090322.16132.91361.1374 0.25-0.970
1968-0.207323.041501361.4356 0.20-1.037
1969-0.114324.62149.41361.5281-0.66-0.776
1970-0.054325.681481361.4117 0.29-1.105
1971-0.202326.32 94.41361.2392 1.07-0.974
1972-0.225327.45 97.61361.1948-0.83-1.136
19730.138329.68 54.11361.0660 0.64-1.150
1974-0.278330.18 49.21360.9792 0.96-1.000
1975-0.065331.11 22.51360.9566 1.33-1.077
1976-0.354332.04 18.41360.9440 0.06-1.322
19770.082333.83 39.31361.0500-1.14-1.507
1978-0.047335.401311361.2295-0.29-1.550
19790.039336.84220.11361.7566-0.26-1.137
19800.125338.75218.91361.6909-0.43-1.064
19810.300340.11198.91361.7126 0.07-1.239
1982-0.020341.45162.41361.4217-1.45-1.024
19830.317343.05 911361.3923-0.95-1.047
1984-0.047344.65 60.51360.9775-0.14-1.261
1985-0.042346.12 20.61360.8359-0.02-1.151
19860.106347.42 14.81360.8877-0.32-1.387
19870.264349.19 33.91360.9677-1.48-1.111
19880.361351.571231361.2840 0.74-1.254
19890.234353.12211.11361.7357 0.62-1.258
19900.490354.39191.81361.6844-0.33-1.366
19910.380355.61203.31361.5875-1.01-1.349
19920.094356.451331361.5255-1.23-1.066
19930.177357.10 76.11361.3081-1.11-1.236
19940.331358.83 44.91361.1154-1.35-1.351
19950.570360.82 25.11361.0200-0.33-1.357
19960.228362.61 11.61360.9005 0.56-1.510
19970.490363.73 28.91361.0787-1.29-1.057
19980.841366.70 88.31361.4702-0.23-1.525
19990.556368.38136.31361.7339 0.79-1.159
20000.474369.55173.91361.9522 0.72-1.078
20010.668371.14170.41361.9009-0.08-1.161
20020.762373.28163.61361.9725-0.72-1.190
20030.758375.80 99.31361.4971-0.43-1.646
20040.660377.52 65.31361.2960-0.63-1.260
20050.863379.80 45.81361.1431-0.46-1.261
20060.805381.90 24.71361.1004-0.28-1.264
20070.895383.79 12.61361.0004 0.03-1.308
20080.682385.60  4.21360.9794 1.00-1.163
20090.731387.43  4.81360.9856-0.13-1.020
20100.914389.90 24.91361.2296 0.95-1.308
20110.707391.65 80.81361.4640 1.40-1.408
20120.761393.85 84.51361.5920-0.18-1.411
20130.826396.52 941361.6117 0.35-1.265
20140.866398.65113.31361.5882-0.38-0.989
20151.151400.83 69.81361.7017-1.17-1.084
20161.292404.24 39.81361.3384-0.46-0.840
20171.120406.55 21.71361.2147 0.11-1.115
20180.941408.52  71361.2560 0.09-1.015
20191.158411.44  3.6-0.79-0.851


Let's do some math. First, a regression of dT on CO2 alone, using the R-based freeware statistics package Gretl:

Model 1: OLS, using observations 1867-2019 (T = 153)
Dependent variable: dT

             coefficient   std. error    t-ratio    p-value 
  ----------------------------------------------------------
  const      -4.17731      0.132377      -31.56    2.31e-068 ***
  CO2         0.0128167    0.000408405    31.38    4.77e-068 ***

Mean dependent var  -0.044229   S.D. dependent var   0.451814
Sum squared resid    4.124960   S.E. of regression   0.165280
R-squared            0.867060   Adjusted R-squared   0.866179
F(1, 151)            984.8466   P-value(F)           4.77e-68
Log-likelihood       59.32611   Akaike criterion    -114.6522
Schwarz criterion   -108.5913   Hannan-Quinn        -112.1902
rho                  0.360023   Durbin-Watson        1.277002

Here's what that looks like with the predicted values (blue) and the actual values (red) both graphed against time:

chart image

This may give us a clue to Victor's confusion. The CO2-based regression curve is relatively smooth, while the actual dT values have all kinds of zigzags up and down. Maybe what Victor means by "correlation" is "conformity to the original shape." That's not what the term means in statistics, of course.

Let's see what happens when we try for the best fit using the table data:

Model 2: OLS, using observations 1867-2019 (T = 153)
Dependent variable: dT

               coefficient   std. error   t-ratio    p-value 
  -----------------------------------------------------------
  const        -3.07818      0.375077     -8.207    1.05e-013 ***
  CO2           0.00965124   0.00120088    8.037    2.77e-013 ***
  SOI          -0.0479114    0.0170797    -2.805    0.0057    ***
  SOI_1        -0.0465968    0.0169148    -2.755    0.0066    ***
  Aerosols_1    0.102327     0.0465809     2.197    0.0296    **
  dT_1          0.335131     0.0746078     4.492    1.42e-05  ***

Mean dependent var  -0.044229   S.D. dependent var   0.451814
Sum squared resid    3.146518   S.E. of regression   0.146304
R-squared            0.898593   Adjusted R-squared   0.895144
F(5, 147)            260.5209   P-value(F)           3.67e-71
Log-likelihood       80.03923   Akaike criterion    -148.0785
Schwarz criterion   -129.8958   Hannan-Quinn        -140.6924
rho                  0.018642   Durbin's h           0.598684

Not much difference in variance accounted for; we're now getting 90% instead of 87%. But this one looks different. With a number of factors brought into play, the graph looks like this:

chart image

This one is more zigzagged, and comes closer to the original shape. It's especially closer during the 1940-1978 period Victor was worried about. Our hypothesis about what might be causing the discrepancy is confirmed; the warming effect of CO2 was temporarily overwhelmed by other factors.



Page created:05/01/2022
Last modified:  05/01/2022
Author:BPL