For my research I needed to download gridded weather data from ERA-Interim, which is a big dataset generated by the ECMWF. Getting long term data through their website is very time consuming and requires a lot of clicks. Thankfuly, I came accross the nifty ecmwfr R package that allowed me to do it with ease. One of the great things about open source is that users can also be collaborators, so I made a few suggestions and offered some code.
tl;dr: The functionality shown in this post is now on the ggnewscale package! 📦. You can find the original code in this gist. A somewhat common annoyance for some ggplot2 users is the lack of support for multiple colour and fill scales. Perusing StackOverflow you can find many questions relating to this issue: Unfortunately, this deluge of questions is met with a shortage of conclusive answers, most of them being some variation of “you can’t, but here’s how to hack it or visualise the data differently”.
As an atmospheric scientists, a lot of my research consists on plotting and looking at global fields of atmospheric variables like pressure, temperature and the like. Since our planet is a sphere (well, almost), it is unbound and so longitude is a periodic dimension. That is, to the right of 180°E you go back to 180°W. But ggplot2 and other plotting systems, for the most part, assume linear dimensions.
For a while now I’ve been thinking that, yes, ggplot2 is awesome and offers a lot of geoms and stats, but it would be great if it could be extended with new user-generated geoms and stats. Then I learnt that ggplot2 actually has a pretty great extension system so I could create my own geoms I needed for my work or just for fun. But still, creating a geom from scratch is an involved process that doesn’t lend itself to simple transformations.
While trying to build a circular colour scale to plot angles and wind direction, I stumbled upon an easy way to make shaded reliefs in R. You known, when you look at cool maps of mountain areas where peaks and valleys are easily distinguishable from their shadows like this: What I accidentally discovered is that one way of approximating this look is by taking the directional derivatives of height and then plotting the cosine of its angle from the sun.
- NEWER POSTS
- page 2 of 2