Last week

Tropical Moist Forest Evaluation Methodology Implementation

Not much new on the TMFEMI front from me. I updated the stopping criteria scores I generated to add a few more projects as Patrick continues to push projects through the pipeline. This seemed to shift the balance from about 2/3rds stopping within a hundred iterations down to only about 1/2. Tom pointed out that I can tweak the code to get it to print out how close from the 5% stability it was after 100 iterations, so we know if the projects were close or far apart.

I also did a small bug fix, whereby we had some tiles with odd empty features in that we then passed to code that generated an SQL query which fails. I patched that in the TMF code, but I still need to submit a fix to biomass_recovery to get it to do something sensible there.

Biodiversity

Chatted with Simon, and I’m going to attempt to mirror the Redlist data in our postGIS instance. It looks like unpacked it’s about 125 GB of data for the full DB. This is way more than we need for the STAR work, but I figure having a local copy that people could access is going to be a generally good resource.

I did a small amount of Quality of Life work on the AoH code to add some progress indicators for individual species runs.

Immutable OSs

I spent a day to so exploring the Vanilla OS 2 beta, which is a linux distribution that lets you have different personalities (or I guess, flavours, given the name): users can have their shell and package manager be like ubuntu or fedora etc.

The way they achieve this is the base OS is “immutable” and then your interactive shell just runs in an OCI container, and it’s that container where you install tools and do work, and if you want to switch you can just switch in another terminal and have multiple personalities side-by-side. A neat trick, though the edifice falls apart quickly if you stray off the beaten path :)

Still, I find this very exciting as it is a lower level version of what I’ve been trying to achieve in part with FSArk, and so there’s an opportunity here I can leverage this as part of my “reproducible environments for doing data-science” train of thoughts.

To save a bunch of time, I’ll not go into more details here, I need to do a proper write up of how this works and the opportunities it affords as a bigger blog post.

This week

• Try setting up the redlist database
• I need to get the GEDI data off our NAS to the other machines
• Write up Vanilla OS learning
• Work out how to do a run of the AoH data for Simon
• Write an EMFCamp talk abstract with Patrick