2011-07-16 16:13:42Lessons from Past Climate Predictions: William Kellogg
Dana Nuccitelli

A look at Kellogg's 1979 temperature predictions.  Not very accurate, but some interesting lessons to be learned.

Lessons from Past Climate Predictions: William Kellogg

2011-07-17 11:51:41
Rob Painting

2nd last paragraph - "if two thins" = things

Looks good, but again it reads like climate sensitivity is an input into models that can be tweaked, which certainly isn't the case. The ocean modelling was very crude back yonder, and was a major reason Hansen's earlier modelling had high climate sensitivity. 

I just think it's useful to put that misunderstanding to bed, but your call.

Thumbs up from me anyway.

2011-07-17 12:06:01
Andy S



could only ensue if two thins occurred: first, (things)


 much moreso than Broecker a few years (more so)


on a business as usual path (business-as-usual path)



You provide this quote from Kellogg (my emphasis)

Action on the part of the community of nations could only ensue if two thins occurred: first, the climatologists, economists, social scientists, and politicians must understand the future clearly enough to decide that a global warming would indeed be too costly to mankind as a whole to be "acceptable" (a value judgment in the last analysis);...

and then you say:

While the first criterion has been met....

I don't think many of our politicians truly understand the threat, or if they do, their inaction is even less excuseable. Some economists seem a little slow in getting it also.

It would be interesting to know of Kellogg's reaction to what has unfolded since his prediction but according to Wiki, he died in 2007.

It's not clear to me why you only compared the actual temperature outcomes to his "high" prediction.

Otherwise very good.

2011-07-17 13:54:16
Dana Nuccitelli

Good comments, thanks guys.  Andy, this isn't clear?

"We digitized Kellogg's "high" predictions, since CO2 has increased at a similar exponential rate to this scenario,"

2011-07-17 14:37:51
Andy S

Dana, I guess I didn't understand the distinction between Kellogg's high and low cases, assuming they depended on a range of sensitivity as well as different CO2 increase rates. But I didn't read his original paper and obviously I should. Added, and I should have read the figure 1 caption more carefully.
2011-07-18 10:55:26
Andy S


Allow me to ramble a little, feel free to ignore this.

I read through the Kellogg paper and his linear temperature predictions, if that's what they are, make little sense to me. I suspect that he just made predictions for 2050 and straight-lined the trends. If you look at the "prediction" for the high trend in 2011, you could calculate an implied climate sensitivity (based on his 2011 exponential CO2 prediction of about 390 ppm) and it would be a very high number, about 15-20C per CO2 doubling (I'm guessing, I haven't calculated it) and I doubt that's what he intended, either as an equilibium or a transient number (they seem to be the same in his book, as Dana pointed out). Essentially, if you take him at his word, the climate sensitivity for the high trend decreases with time.

If you assume a constant sensitivity over 1980-2050, then you can back-calculate the implied CO2 concentration and it would look like the red dotted line in this graph (added later: actually a properly calculated implied CO2 consvcentartion (which I did subsequently) would not be linear (due to the logarithmic CO2-temp relationship) but would be a concave-up curve sagging a little below my dotted red line).

The green box is roughly the outline of Dana's Fig 2. What this pic shows is that the exponential growth curve is closer to the Low Growth  line than it is to a linearized high growth line in 2011 (32 years after "present").

If you draw the low growth straight line on Dana's fig 2 (green) and sketch in the exponential curve (light blue dots, done free-hand) using the same sensitivity as Dana derived then you see that Kellogg's prediction is still high, but not nearly as bad as implied by the linear high case.

Anyway, this is a very roundabout way of saying that the biggest problem with Kellogg's prediction in 2011 is with the linearisation of the temperature prediction. All the other discrepancies that Dana identified are still there. My guess is that Kellogg was just trying to graphically demonstrate a range rather than predict the details of a temperature trajectory for the high case, but that's not what he said, I admit. 

2011-07-18 15:08:11
Dana Nuccitelli
Yeah Andy it's a good point that because of the linearity of the prediction, it actually gets better over time. I'll add a bit about that.
2011-07-19 03:45:51
Andy S



In your post you say:

His projected atmospheric CO2 increase (400 ppmv by 2000) was a bit more rapid than the actual CO2 increase since 1979

But look at Kellogg's Figure 3:

400 ppm in his exponential (high) case is reached in 2011/2012 not 2000. His CO2 concentration prediction for 2000 is about 360 ppm (330 ppm in 1978 plus 31 ppm). 

Using your estimate of his climate sensitivity (3.6) and his prediction of 400 ppm CO2 in 2011/2012 would lead to a predicted increase of 1 degree, lower than your blueline in your figure2 (about 1.4 deg) and  lower still than the approx 1.6 deg derived from eyeballing his linear high case prediction in your figure 1.

Perhaps I've misunderstood something, not for the first time.

PS Kellogg's musings about ice sheets and sea ice are interesting.

2011-07-19 06:07:19
Dana Nuccitelli

My bad, I wasn't clear about the distinction between CO2 and CO2 equivalent.  You're right that CO2 reaches 400 ppm around 2011 - it's CO2 equivalent that reaches 400 ppm in 2000.  The fact that he's including other GHGs in addition to CO2 but excluding aerosols increases his model sensitivity, which contributes to his overprediction of the warming rate.