In this post I show how to map NASA GISS’s 2×2 degree temperature anomaly data using R mapping tools. Rather than rely on a single value to reflect monthly global temperature anomaly, this map shows the anomalies in each of the 16,200 cells in a 2 degree lon/lat grid. This lets us see the details that make up the global mean, we can see which areas are warmer and which are cooler. I provide a link to my RClimate script and data file so that interested R users can make their own maps.
Here’s my R Climate map of NASA’s July 2010 2×2 degree data set. (Click map to check out the enlargement )
Users can use NASA’s interactive tool to make a comparable map at this link. Here’s what NASA’s map version looks like.
NASA’s map is very good, so why bother making my version in R? I had several reasons:
- I am not happy with NASA’s color scheme, I wanted to see if I could improve it
- Northern Hemisphere polar amplification is an important issue that I think gets masked with NASA’s color scheme because so a large area shows up “reddish”, Does better color separation help to show NH polar amplification?
- I don’t like NASA’s color bar scale, notice how the intervals vary by location along the scale axis. Interval span range from a low of 0.1 to a high of 1.8 . I’m sure there’s a good reason, but I wanted to see how other scales look.
- Size, I wanted to make a larger image so that I could magnify the zoom farther to see specific areas in more detail.
- I wanted to learn how to map climate data myself.
RClimate Map Features
Here are the features I was trying to incorporate into my map:
- Larger size (1200 * 1000) – feel free to zoom in
- Increased color separation to distinguish really cool/warm areas
- Uniform 1 degree C scale bar intervals
- Missing data shown in white
- Can show entire globe or map only a portion of the globe by adjusting the “bounding box”
RClimate Script Details
Here are the data and RClimate Script links:
Sorry Kelly, where I can find the script? I am new in R
Please look at bottom of post.There is a link to data source and R script.
I’d like to suggest that in addition to a white region around zero, that you pick lightly saturated tones for the next few bands out. Save the strongly saturated colors for the extreme values.
Also, this map projection hugely overstates both the importance of and knowledge of the poles. Pick a projection that doesn’t have such a large area error. This is a hard problem, and there are no perfect answers.
An underlying problem is the use of a 2×2 grid cell in the various data series. That is probably acceptable in the tropics (but even so the boxes are not equal in area). But it is just wrong for covering an entire sphere. There are much better meshes that could be used for surface data sets, where the individual mesh cells are constrained to equal (or nearly equal) area.
One thing I think NASA does do right is the white area. This represents near normal temperatures. When you have small fluctuations from normal different colours for negative and positive values, large areas slightly above or below normal by an insignificant amount can give a very misleading visual impression.
swiftright
I agree with your point about having a white area for near zero areas. I’m working on a revised color scheme for August, 2010 anomalies.
link
Hello,
thanks, very nice!
I found two little errors in your R script:
* the url to the data is wrong
* the separator in the read.table command should be ‘,’
Stefan
Thank you for catching these and letting me know so that others won’t have the same problem you did.
It should be working properly now. Please let me know if you still have any problems with my script.