Thank you, Richard. That’s exactly what I’ve been looking for. ]]>

I left a few posts over there. I’ll see if there are any response in the next day or two, after which I’ll have probably forgotten all about it.

]]>summary(fit.lm)$coef[2,”Std. Error”]# where fit.lm is the name of your first order least squares model. ]]>

Earle,

this isn’t smoothed temperature anomaly, it’s the **rate** of warming (given in °C/a) as derived from the reconstruction data Steve pulled from the web.

That is for every single reconstruction, for every year, I calculated the average trend (steepness, slope) of temperature raise (or decline) as derived by linear regression for the previous 50/100 years until that very year. There has been no smoothing applied to these results.

Of course there is a smoothing effect since to fit the regression line a window of certain width is neccessary. I consider 50+ years a reasonable period with respect to the changes in climate under discussion.

]]>Wolfgang Flamme,

Would you be so kind as to post the unsmoothed data? I’m curious what the unfiltered curves look like.

Thanks,

Earle

How easy it seems to screw things 180 degrees out of order. Oh well, being attacked by desmogdogmatism is almost like a badge of honour.

]]>I created two graphs from this, showing the rate of warming within a 50/100yr sliding window.

According to some reconstructions the current rate of warming is not unprecedented.

Btw:

Need help with this: Given that in r-project I have obtained a linear model for every point of these curves drawn, how to determine this rate’s confidence interval from the model?

After all there are chances that the warming rate derived from the 50/100yr window is more or less steep than the best fit from the model so how to obtain these confidence intervals?

I’m asking about the ‘red noise’ data & analysis.

]]>