August 2008 Charts for NCDC
Posted by The Diatribe Guy on August 29, 2008
Yesterday I posted my update on NCDC Temperature analysis. It was a verbal presentation, and while I’m a stats guy, my eyes also glaze over with too many numbers and formulas muddying the page. So, I’m presenting in this post the charts that represent my discussion points in the previous post, linked to above.
Starting with the overall tend since inception of the data set:
Moving on to current cooling period chart, based on raw anomlay data:
Next, I present the current 30-year trend and how that compares to the most recent peak 30-year trend:
Now, the same charts for the 20-year trends, based on raw anomaly data:
And now the same for the 10-year trends:
And now the same for the 5-year trends:
I also noted that the current slope value on the 60-month trend is the lowest since 1995. That can be seen here:
Finally, I discussed my projection model. Just for kicks, I put together a chart showing the most recent 60 months of raw anomaly data, along with the monthly projections through December 2015, based on the projection model. This is based on a model that assesses the running 60-month slopes of the raw anomaly data, and then assesses the running 14-month slopes of the 60-month trend. I then calculate the second difference to determine how those slopes change, and weight those changes based on the least squares estimation value that minimizes the difference between actual and projected anomalies, based on the projected second difference. These weighting values can then be used to project future anomalies. Obviously, the further out one goes, the less certainty there is, but it’s an interesting exercise, nonetheless. The trend lengths selected (60 and 14) represent the current tested parameters yielding the lowest least squares value. I will continue testing different parameters to see if other measures yield even better results. In any case, here are the projections:
Digital Diatribes 
John Nicklin said
Thanks for the visuals.
Flanagan said
Yes, thanks, but where did you find these data exactly? Also: while I understand your data-guy point of view, I would bet a buck on your projected anomalies in the sense that the short-term trends are often dictated by sun activity which is strangely low rightnow but should go up within the next months.
Diatribical Idiot said
From my previous post, the data is located here: ftp://ftp.ncdc.noaa.gov/pub/data/anomalies/monthly.land_and_ocean.90S.90N.df_1901-2000mean.dat
As for the projected anomalies, the weighting values and slope changes are based on the 132 previous months to accommodate anticipated changes in solar cycles, but I’m claiming no particular confidence in the model’s results, especially the further out it goes. Just having fun with the data and seeing if there’s some kind of future predictability that can be gleaned that has the ability to buck short-term trends. We shall see how that goes.
What’s sort of interesting in the model as far as your mention on solar activity is that the current activity is not only unusually low, but remarkably so. We just completed the first spotless month in nearly 100 years. The length of the current cycle, while one can only guess on final length, is a concern because it has been shown that temperatures in the years following longer cycles are cooler, and vice-versa. Dr. Hathaway has finally changed course a bit lately and admitted that the current solar (in)activity certainly points to a weaker cycle 24 – he previously predicted an active 24 with a weaker 25. Now, it appears he is concluding that 24 looks to be weaker than expected.
Direct conclusions are hard to come by, but estimates on the impact on temperatures based on changes in solar activity have a lag of anywhere from 2 to 8 years, depending on the depth of the change and the nature of it. I mention that as a point of interest only in that if there truly is a longer cycle 23, with an extended minimum, followed by a weaker 24 than expected, we can most certainly expect lower global temps. Interesting then, that the model I put together using only global temperature changes with no solar data incorporation seems to be lining up with the predictions of some others based on their obswervations of solar activity.
But we shall see. Right now, this is simply an experiement.