[Coloring (verb). Definition: to cause to appear different from the reality: Example: In order to influence the jury, he colored his account of what had happened]
NCDC issued their February 2014 climate report, it looked like this:
The colored US map (that gets top billing in the PR) can be viewed full size here. http://www.ncdc.noaa.gov/file/february-2014-us-divisional-temperature-ranks-mapgif
But what happens when we start showing actual temperature data instead of colored areas? Joe D’Aleo sent this map over today, that shows the same map with temperature anomalies for many major cities in February 2014 he added. And with it, is a surprise.
For example. NCDC shows Connecticut as “Near Average” according to the color assigned, but Hartford has a -5.5°F anomaly, but in California, Los Angeles is +2.6F and is shown as “Much Above Average”. You can draw other similar comparisons.
At first glance, it seems that NCDC has a clear warm bias, but I’ll also point out that these divisional ranks are made up from data from dozens to hundreds of weather stations. As we know, some weather stations are good, some are bad in the way the are maintained and produce data, so the city anomalies, while interesting and suggestive that there might be a warm bias, isn’t definitive. Apples/oranges and all that.
Now, have a look at this map which shows temperature anomalies by division. It is from NCDC also, and you can create it yourself using the Climate at a Glance plotter here: http://gis.ncdc.noaa.gov/map/cag/#app=cdo
That certainly looks different than the rank maps shown above at the top of the NCDC press release, but that is to be expected. Of course, temperature anomalies are a different animal than ranks. Ranks are statistical constructs, removed from the actual measurements by at least two calculation passes. Anomalies require generally only one calculation pass, so they are closer to the measurements, but they have the advantage of showing where the departures from “normal” are. Ranks seem to be less effective at this, IMHO.
Now here is the divisional rank map as plotted by NCDC as the companion map to the one above using the same tool at http://gis.ncdc.noaa.gov/map/cag/#app=cdo
The divisions are ranked exactly the same as the map used in the February NCDC press release, but do you notice how the colors are significantly different? The blue shade is darker over much of the country in the CAG plotter output than the one in the press release at the top.
Here is that map again for easy comparison. The difference is obvious. The colors are much lighter on the map below, the one used in the NCDC February 2014 press release.
And just to be complete, here are the average temperatures for those divisions, what NCDC calls “values” on the selector tool. It looks a bit more natural for what we’d expect in February, doesn’t it?
My point to all this? Perceived CONUS temperature is in the eye of the map maker and the beholder.
You can create different impressions by choosing color schemes and what maps you put front and center in reports such as NCDC does.
Some people got bent out of shape because we ran a post from Harold Ambler who had criticized the odd way the Divisional Temperature Ranks map was presented in the February press release, which made February look warmer than measurements. Clearly, as I have demonstrated, NCDC seems to have different color schemes for the same Divisional Temperature Ranks map product; one for press releases, and another for the Climate at a Glance plotter. In a place that supposedly prides itself on standards and accuracy, I find that sloppy. Of course the maxim “close enough for government work” also comes to mind.
Most people clearly associate darker blue colors with cold, lighter blue colors, not so much, and yellows, oranges, and reds with warmth. Obviously, in the CAG plotter output of Divisional Temperature Ranks, there’s no washed out colors, and the temperature delineation by color is clear. The question is: why does NCDC need two color schemes to tell us the same thing?
If NCDC wants to avoid criticisms on map perceptions related to color, I suggest they use a single color scheme across the board, so that there won’t be any confusion or perception issues between maps used for press release and maps provided for research.
After all, do they really want people “seeing red” when they really should be seeing blue?
UPDATE: reader KenF writes:
Besides mapping representation (full disclousure, I am a professional GIS mapper), there is another issue about NCDC on presenting February temperatures that worth addressing: the way they present the section of “February 2014 Winter Cold: Historical Perspective”.
If you check their link from NCDC
(http://www.ncdc.noaa.gov/sotc/national/2014/2/supplemental/page-3/) they will provide you the (a) day-by-day temperature from Dec 2013 to Feb 2014 compared with 1981-2010 normal and (b) the coldest day of the winter compared with cold outbreaks from the station’s deeper history, by comparing the number of days with temperatures as cold or colder than the threshold occurred each year (in NCDC words).
I have particular issue with (b). Since we know what make this winter so cold is not about the record low temperatures, but rather the frequent (and prolonged) cold wave and the well-below normal AVERAGE temperautre from Dec 2013-Feb 2014. NCDC definitely aware of this and use the more fuzzy approach to represent part (a) and cherry pick the threshold on part (b).
To illustrate my point, I choose two cities from midwest: Fort Wayne IN and Minneapolis-St. Paul MN. According to Northern Indiana WFO, Fort Wayne has the 6th coldest winter (20.7F from Dec 2013 to Feb 201, or 6.7 degrees below normal) and the 19 days with low temperatures below 0F (compared with 1981-2010 normal of 6.5 days)
Then what does NCDC do? On representing average temperatures (part a), instead of showing tables and figures how they ranked with “station’s deeper history, they just do a fuzzy bar diagram in order to show “hey, we know this winter is cold, but there are still some warm days”.
Part b is ABSOLUTELY ATROCIOUS: instead of using the standard Zero F that normally used, they pick -15F as the threshold, then show there are “only” 2 days in winters with temperatures below-15F, which attempt to make it a “normal winter”!
Same thing can be applied to MSP. This is the 9th coldest winter for twin cities (9.7F) in history all the way from 1876 (http://www.dnr.state.mn.us/climate/journal/coldest_winters.html). According to NWS regional office, there are 50 days of below Zero F days in twin Cities from Dec 2013 to Feb 2014 (53 days if you include Zero days itself, and this figure did NOT include March), which is ranked 5th in history.
Then what does NCDC treat those facts? They pick the threshold at -20F! It effectively reduced the days below -20F to 2 days and the chart show it almost as if it is a warmer-than-normal winter!
Welcome to this world, as cherry picking the thresholds and moving the goalposts are NCDC expertise. We must not let their malpractice off the hook!