Climatic Data More Random Than It Appears

On Penny Wong's "Day of Reckoning" for climate change, the climate blog wattsupwiththat presents some very interesting data that suggests that despite first impressions the actual temperature data resembles a "random walk". The idea is that if you go on a walk and toss a coin at every intersection to determine which way to go, there may be quite long periods where it might appear that you are walking in a particular direction even though each choice is determined randomly.

Here are the latest climate data with the long-term trend drawn as a stragiht line:



Fair enough temperatures are generally increasing. But notice that the data actually spends extended periods of time both below and above the trend line. If the trend was a genuine trend with random variation about the trend line, you would expect there to be an above average figure followed usually by a below average figure or reasonably short runs below or above the trend line.

What if the month to month figures are skewed by seasonal effects? We could remove those effects to determine the real trend by comparing each month to the same month a year earlier. This then gives you the underlying annual rate of change. When we do that we get the graph below:



Not that looks rather like a random variation to me.

I won't bore you with the details, but you can use statistical analysis to work out whether it is truly random variation or whether it is moving towards or away from a mean. It turns out that this data set shows absolutely NO sign of anything other than true random variability.

I'm not a statistician so I can't pledge for the validity of this analysis, but it seems to make sense.

And we are spending trillions in transforming to a "low carbon economy" for this?

Full article