Google Summer of Code – A GeoTiff reader for GeoTrellis

Google Summer of Code – A GeoTiff reader for GeoTrellis

Applying to Google Summer of Code

I first heard of Google Summer of Code (from here on GSOC) when a
former student at my university in Stockholm told our class how he
nailed a job at Google. He said that he performed very well in a
competitive programming tournament and that he also had done GSOC for
the Python Software Foundation. I was already trying out competitive
programming and had zero experience of working with open source.

When the GSOC 2014 season started and the accepted organizations were
announced I decided to do something that few people would apply to,
i.e. not submitting to the Twitter Open Source Organization, to
increase my own chances to join the program. I submitted 3 proposals,
one was writing a GUI test library for OWASP ZAP, a desktop program
written in Java for testing attacks on a web server, one was a graph
format exporter for Bio4j and the last one was the GeoTiff reader for

The first one, the OWASP ZAP GUI test library, seemed the most boring
one and was in Java, but the guys maintaining it was very
friendly. The second one was supposed to be in Scala but was changed
to Java. I really wanted to learn more about Scala and Functional
Programming in General and when I got accepted to all 3 proposals I
talked to Rob Emanuele, who later has been my mentor during GSOC,
which instantly told me that if I want to do Scala this summer,
GeoTrellis was the way to go!


It was both exciting and a little bit frightening to work with Scala
when I haven’t written anything in Scala before, and truth be told, it
isn’t as easy as Java at all. Rob recommended the Coursera course for
Scala and I did the whole thing, it was great. I had a lot of stuff in
school so I actually didn’t prepare too much for the actual project,
except exploring the GeoTrellis source code for a bit. I also found
some specifications for Tiff and GeoTiff and tried to read those, but
I didn’t understand too much. I also got a book from Azavea, which was
about Rasters and Map Algebra, which was a very good read for this

Start to Midterms

GeoTiffs are essentially Tiff files with a few add-ons; GeoTiffs are a
superset of Tiffs. I started reading the Tiff 6.0 specification, and
since that was written in 1992 it felt a bit outdated and hard to
interpret. But I worked hard and tried to read in all the tags (Tiff =
Tagged Image File Format) and all the extra stuff that GeoTiff brought
in to the picture. It went pretty slow because I was both learning to
use Scala and getting familiar with working in a larger group of
developers, with a rather big codebase. I got a lot of help by my
mentor Rob and he read and commented my code on Github, making stuff a
million times easier.

Midterms to End

After the midterms I had tried to do some decompressions and I also
did a pull request for fixing a locale bug (the dreaded comma vs dot)
in the whole of GeoTrellis. From here on stuff got more easy and with
the help of Rob I really started to get things done. Today the reader
supports all of the Tiff 6.0 specification decompressions except JPEG
and also works fine with ZLib. The reader is now used in other parts
of GeoTrellis and it is really nice to see that something I have
created is used by others.


I will continue after the GSOC 2014 season is over to work with
GeoTrellis and further improve the reader and also create a GeoTiff
writer. I look very much forward to doing this and I’m very grateful
for both the program, the people at Azavea and my mentor throughout
the program.