Manaaki Whenua researcher Pierre Roudier is enjoying being both author and contributor on different software packages that are providing ways to accelerate soil research.
Both packages were developed as extensions to R - one of the leading languages for statistical analysis and graphics. One package, CLHS (Conditioned Latin Hypercube Sampling), was born from a selfish need to find an easier way to identify soil sampling points, while the other, AQP (Algorithm for Quantitative Pedology), was discovered while looking for an easier way to handle soil data.
R, as an open-source software, allows researchers to extend the capabilities of the program through an extension. It enables people to collaborate and contribute ideas to a package. You never start from scratch, but a bit like Lego, you build on work people are doing. I started working on implementing the CLHS several years ago (2012 FWIW). The challenge I was finding in the context of soil mapping, was that it was hard to identify where to collect samples in the first place. Because you can’t sample widely, you have to be smart about where you sample.
That’s where the CLHS algorithm comes in. It’s trying to find the locations that best represent the environmental variations. Even as the program fulfilled his needs, it was obviously satisfying an unmet need for other researchers. To date, the package has been downloaded 700,000 times. And scientists have also been contributing to enhancing CLHS. Most recently a student in Canada approached me with an idea to increase the speed of the programme.
He found the AQP package while searching for a better way to analyse soil data. The package was developed by US-based researcher, Dylan Beaudette, as part of his PhD project. The researchers met online nearly 11 years ago, and still collaborate today. The software is used by the US Department of Agriculture’s Natural Resources Conservation Service for its soil surveys, and I am eager to connect this massive database to New Zealand’s National Soil Data Repository.
I’m interested to see how we can keep connected to the emerging ways we have to store and broadcast soil data. It’s this collaborative ethos I enjoy most about working on R plug ins. I have had a lot of feedback from people at conferences, and CLHS has also been cited often in other scientists’ research papers. The AQP development team that has now grown from initially just Dylan and I, has been invited to contribute a book chapter on the software.
This has been an important way for me to give back to the community. I feel the work I have done on these two packages has been far more impactful than any paper I’ve ever written.