I usually teach one course per semester on topics related to natral hazards, remote sensing, data mining and data visualization. My courses are usually offered to M.S. and Ph.D. students only.
Although I do not teach programming courses, I use my computer scientist background to teach how to use computer languages to automate different processes related to the download, analysis and distribution of data. I like to teach R, because it is easy to learn both for people who have a computer programming background, and also for those who just start learning formal languages. It is possible to use R to solve many problems, but most important is its ability to produce publication quality graphs and maps.
I always try to include at least one lecture in symblic machine learning and data mining in general to show how modern tools can be used to discover patterns in massive amount of data. Remote sensing data tend to be massive, and they cannot be analized easily. Furthermore the spatial and temporal component of remote sensing data require complex techniques to cope not only with the massive volume, but also with the spatio/temporal relayopnship.
Usually I invite my friend Rafael Amellier of Stormcenter Communications to give one lecture towards the end of the class on the use of remote sensing in media. Rafael gives a unique lecture from an angle which is often neglected in University classes: How scientific results are shown to the general public through standard media, in particular TV,
I supervise several students both for Ph.D. and M.S. degree.