DMI Arctic temperature data does show increasing temperature trend

September 9, 2009

I would like to make a few comments about the DMI arctic temperature data.  I made the numeric version of this data available yesterday by digitizing the yearly graphs available from DMI.  I have had a chance to look at the data and draw some initial conclusions.

Anthony Watts, who I have the greatest respect for, presented an animation of the 52 yearly DMI Arctic temperature plots form 1958 to 2009.  He said “See if you can spot the temperature spikes or the “…cooling trend reversed in the mid-1990s.”  His animation appears below…

Anthony Watts' animation of DMI temperature plots.

As Anthony Watts implied, I found it difficult to detect a trend over time by viewing the animation.  So I created a  simpler version that shows only five frames, each consisting of an overlay of 10 years, 1958 to 1967, 1968 to 1977, …, 1998 to 2007.  The problem is that when I view this simpler animation, I do see a trend, with temperatures rising in the freezing season on the the far left and right sides of the graph in the last frame (1998-2007).  My animation is below (double click is the image does not appear animated)…

Moriarty's animation of 10 year composites of DMI Arctic temperature graphs.

The data that I made available yesterday gives the DMI Arctic temperature for each half day from 1958 to 2009. This set of data allows a plot of the Arctic temperature for a particular day of the year as a function of the 52 years covered. For example the following two graphs show the temperature on September 1st and October 16th. Notice how the temperature trend seems to increase after 1995 for the October 16th data but not for the September 1st data.

September 1st temperature as a function of year.

.

October 16th temperature as a function of year.

If we are interested in a change of trend after the mid-1990s, then the trend before up to 1995 and the trend after 1995 for every day of the year can be compared in the same way they are compared for October 16th in the above image. The following image shows the temperature trend for each day of the year for 1958 to 1995 and from 1995 to the present.

Arctic temperature trends for each day of the year.

.
The total trend for each day for all data from 1958 to present is shown in the following graph.
.

Arctic temperature trend vs. day of the year for all data from 1958 to 2009.

So, the DMI data, presented in the crude fashion that I have used, lends support to the idea that the Arctic has been heating more rapidly since the mid-1990s than before.  Those of you who have read my blog in the past know where I stand on the probability of the Arctic ice melting in the near future, and I stand by my previous posts.  But I think this data must be presented as part of the scientific pursuit of truth.

I would be very happy to hear the opinions of people smarter than me on the significance of this data.

Numerical version of DMI Arctic temperature data

September 8, 2009

I am providing the DMI arctic temperature data in numerical format as a public service.  This data is available in graphical format at the DMI Centre for Ocean and Ice,  but I could not figure out a simple  way to get the numerical data that was used to make the graphs.  So I have extracted it from the graphs myself.

Here it is, for the entire set of DMI images from 1958 to the present.  You must go through a slightly convoluted procedure to use it.  First, open this word document.   Then, select all the text and copy it.  Finally, paste it into a spreadsheet.

The left column gives the day of the year.  Notice that the days are given in half day increments.  The top row shows the year, and each column represents the temperature data for the coresponding year.

You can now plot, average, and compare trends to your heart’s content.

Explanation of Data Handling

This data was acquired by copying all of the yearly graphs from DMI and writing image handling code in LabView to extract the temperature data.  The 365 days of data in each graph is distributed through 518 columns of pixels, so about 1.4 columns of pixels per day.  I decided to interpolate the data to half days.  Since leap years and non-leap years both consisted of 518 columns,  I have treated all years as 365.25 days.

All of the data suffers from the uncertainties of my extraction process.  So if anybody sees problems with the data please feel free to post your observations as comments here.

I will be showing some results of my analysis in subsequent posts.

Sample comparison:

Here is the image from DMI for the year 2008, followed by a plot of my version of the data for 2008.

DMI Arctic temperature graph for 2008

2008 Arctic data as extracted from DMI image