Digital Diatribes

A presentation of data on climate and other stuff

September 2008 Update on Global Temperature – HadCrut

Posted by The Diatribe Guy on September 28, 2008


I had such high hopes for this post. I updated the HadCrut analysis spreadsheets a few days ago and put my predictive models together, and despite numerous adjustments and iterations, the results simply were not sensible for any long-term predictions. So, I will need to continue to take a look at what kind of adjustments I need to make in the HadCrut model to get better results. It’s somewhat bizarre at this point. The other temperature measurements seemed to converge to consistent weights on past months that created a trend of somewhat sensible anomalies going into the future. HadCrut seems to have much more random noise and does not exhibit as clear behavior. I will continue to investigate this aspect of HadCrut analysis to see what I can come up with. In the meantime, here are the updates reflecting the August anomaly.

The August HadCrut anomaly came in at a value of 38.7 The predicted anomaly was between 35.9 and 37.0 as a best estimate, so as you can see, the predicted anomaly was basically spot-on. You may be wondering why I was working to refine such an awesome model, but while the model seems to do a pretty reasonable job in the near-term, it falls apart to unreasonable values in the longer term, and that is the part I am addressing.

The information is found here.

The August anomaly is 38.7 (in terms of 0.01 degree Celsius).

*It is 0.025 degrees colder than July 2008
*It is 0.017 degrees warmer than August 2007 (HadCrut deviates from all other models in this regard)
*It is the 2nd coldest August anomaly since 2000, trailing the 2007 value
*The furthest back we can go to find an anomaly at least as large is January 1983. There are no anomalies previous to that where the value is at least 38.7.

*9th warmest (150th coldest) August anomaly out of 159 data points since 1850
*85th warmest (1,820th coldest) anomaly out of the total 1,904 monthly observations
*More recently, it is the 4th warmest anomaly of the last 11 anomalies which is again a departure from the other measures, where August was one of the cooler anomalies of the last 3+ years.

*The latest 12-month average is now 30.0, wich is the second consecutive increase of 0.1. This is yet another departure from other measures, where the current 12-month average is the lowest in some time.

*This is the 2nd consecutive month with a year-over-year anomaly increase
*3 of the last 6 months show an anomaly increase over previous year, rebounding from a previous run of declines over previous year

*Overall trend line since inception is presented below. This represents warming of 0.44 degrees per Century. Proving that HadCrut is an equal-opportunity confuser, this is the lowest overall trend line of all the measures.
Overall trend of HadCrut data since inception - 0.44 degrees warming per Century

*Current running negative slope extends back to April 1997, or 136 months (graph below)
*The current cooling trend line added one data point with the new month, and added one data point at the front end versus the July month-end data
*Strangely enough, while HadCrut paints a warmer recent picture than other measures, the cooling trendline extends back further than any of the other measures at the same time.
Current cooling trend dating back to April 1997

*Current 60-month slope is -0.2944, not quite as steep as the measure at previous month-end
*Current slope is still on par with the steepest negative slope since December 1985
Current 60-month slope based on raw data
Chart showing how the slope has declined over the last few years
Chart showing the varying levels of 60-month slopes over time

*Current running 120-month slope is 0.0621
*This slope value has increased over the last few months after a previous decline. This is due to some higher front-end anomalies dropping off, lowering the front part of the line.
*This increase should continue next month, and then level off for a few months, before starting to decline again.
*The current value is the highest value since the period ending March 2007.

*180-month slope is 0.1412, continuing to decrease as time goes on.
*Lowest value since the period ending February 1998

*240-month slope is currently at 0.1543, continuing a consistent decline
*This is the lowest 20-year trend value since the period ending November 2001

Current 240-month trend on raw anomalies
Changing 240-month trend line over recent history

*300-month trend is at 0.1580, which is in the same ballpark as the value over the last few months

*Current 360-month trend is 0.1352, which continues a decline which has been occurring for a few years
*The current 30-year slope value is the lowest since the period ending July 2001
Current 360-month trend on raw anomalies
Changing 360-month trend line over recent history

Hopefully next month I will have progress to report on the greater model. I am also hoping to finally do a better correlation analysis between all the temperature measures, as well as Fourier analysis over the next few months. I also have a grand plan for incorporating an iterative minimum-bias approach considering outside variables, but that will take time to put together.


2 Responses to “September 2008 Update on Global Temperature – HadCrut”

  1. jeff id said

    I know it might be a bit of work but would you mind putting the source data in the graph titles. i.e. HadCrut, Giss.

    It would make it more clear to experienced readers and make them easier to reference.

    Thanks for the great work.

  2. Diatribical Idiot said

    That shouldn’t be much additional work. I’ll try to remember to take care of that with the next updates. Good suggestion.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: