Dig and Delve Part 2: Running Detrended Data

In my last post, Dig and Delve Part1, I showed how running trends are useful for showing the linear trend to any point in a dataseries, and that trends in different regions can be compared, after the first 10 to 15 years (120 to 180 datapoints).

In this post I use running trends to derive running detrended data, and analyse data for various Land regions using UAH anomalies.  Firstly, the following figure shows global mean UAH to November 2016 detrended using the current linear trend:

Fig. 1:  Global data detrended from linear trend

This is the usual way to show detrended data.  It clearly shows where temperatures have been above or below the trend, however with each new data point the detrended data changes.  Running detrended data is calculated for each point as actual data minus  the running trend multiplied by the count of data points plus the running intercept.  Here is a plot of running detrended data and ordinary detrended data for Global UAH.

Fig. 2:  Global running detrended data and standard detrended data.

After about 1987 they differ by less than +/- 0.1C, and UAH is accurate to +/- 0.1C anyway.   Note also that when temperatures are rising, running detrended data are greater and when falling, they are lower than standard detrended data.

The benefit of running detrended data is that they never have to be recalculated.  Historic detrended values are preserved.  As well, the running trend per month is conveniently not different from zero (which is why I showed it in Part 1 as degrees per 100 years by multiplying by 1,200), so it is easy to show the detrended data with a zero line.  The detrended value is also a measure of how much each month’s data has contributed to the trend- positive values indicate warming influence, negative values indicate cooling influence on the trend.

I now turn to using running detrended data to analyse what has been happening with Land data recently.

The following plots show running detrended data for Land TLT in the Northern Hemisphere.

Fig. 3:  Northern Hemisphere Land data detrended from running trend

The large spike in February was followed by a plunge to October, but values in November have returned to trend.

Fig. 4:  Northern Extra Tropics Land data detrended from running trend

Similar to the Northern Hemisphere.

Fig. 5:  North Polar Land data detrended from running trend

Note how much North Polar data fluctuates.  However the Poles only contribute less than 7% to the Global mean.

Summing up:

• Running detrended data never have to be recalculated, and historic values are preserved.
• The detrended data show how much each month’s data has contributed to the trend at that point.
• The recent plunge in Northern Hemisphere Land anomalies is nothing to get excited about as it is the recovery from a huge spike. The November value is still on trend.

The next in this series will use an estimate of discrete Extra Tropics (20-60 North and South) to look at trends and detrended data.

Tags: climate