Modis R: Package tutorial
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
The MODIS package for R
For people who work in GIS in R there has been a bit of challenge in working with data from Modis. I’ll use an example from my own work on UHI to illustrate. A while back I decided that I wanted to look at albedo data from MODIS. If you want to get MODIS data there are basically two approaches. You can use the web services like REVERB which is basically a web interface that allows you to construct queries for a wide variety of data ( or similar services shown here ) or you can download files directly from the datapool ftp.
For my Albedo project I decided to download directly from ftp because I could write a simple program in R to collect the entries in the ftp directories and then download them over the course of a few hours. It’s essentially the same process I use in my study of SUHI for small cities. Getting the data locally is not a big problem, except for the storage issue. You can imagine what it would take to download 90 days of LST data for the US. with 10-14 tiles of data required that ends up being a nice chunk of data.
The next challenge is getting MODIS data out of its native format into a format that is friend for R GIS packages. In the case of Albedo that data was stored in a format known as “HDF” or HDF-EOS. I work with the raster package and raster ( along with rgdal) have no way of reading HDF directly. This meant I had to find a way to transform HDF into a more common format like “geoTiff”. There was also the issue of tiling multiple hdf files into a single raster. Unlike other datasets that might give you a monolithic file of the whole world, the MODIS collections are distributed as tiles using their tiling system.
To solve this problem there are a couple tools: MRT or the modis reprojection tool. or GDAL, the Geospatial Data abstraction Library. For Albedo I decided to use GDAL and I used the distribution of GDAL provided by FWTools. With a few hundred albedo files I had to write a script to load every Hdf file and transform it to a ‘geotiff’ file. That worked fine, but now I had a set of R code for doing my statistics and a “bat” file for controlling Gdal. With my SUHI project I decided to use MRT. In that case I used the GUI they provide to select all the files, mosaic them, crop them, resample and reproject. I quickly learned that duplicating the same steps with a GUI is not always easy. Further, if I am trying to create work that is reproduceable, it makes it hard. Plus I could not see using that tool to create 365 maps for every day in the year.
What’s needed is an R package that allows us to write R code and achieve the same results as we get with GDAL or MRT. We want to write R code that controls these external applications and tools. That’s what MODIS R does.
The team: The MODIS R team consists of Matteo Mattiuzzi, Jan Verbesselt, Forrest Stevens, Tomislav Hengl, Anja Klisch, Bradley Evans and Agustin Lobo. Some of the seminal work was done by Tomislav at his wonderful page
My Tutorial. I had some discussions with Matteo who is the package maintainer and suggested that I might do a tutorial to help people get started with the package. At this stage the package is stable, so I decided to start a tutorial. The way I work is pretty simple. As I try new things I write them up. For now the tutorial starts with the basics of R, R studio and preparing your system for MODIS R. That includes downloading a the tools and testing them out and finally downloading and building a WINDOWS version of MODIS R.
Then I will start to go through the API and test and explain all the calls in more depth than the manual and write some sample code.
A note. If you have questions about what I am doing go ahead and leave comments. I can collect up questions and those I cant answer I can pass on to the maintainers.
![](https://i0.wp.com/stats.wordpress.com/b.gif?resize=1%2C1)
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.