Discussion on multidisciplinary PROPL-work
summary
Group discussion foregrounds concrete pain points and fixes: counterfactual deforestation analysis relies on spatial matching but fails when undocumented dataset versions change between downloads; biodiversity pipeline speeds from months to days by handling 150 GB maps and optimizing I/O for 40,000 species; tooling includes lightweight Fortran static analysis deployed at the Met Office; recurring problems are missing provenance, floating‑point nondeterminism, incompatible model coupling, and lack of training and CI'd pipelines to make results reproducible.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:06Computer scientists & ecologists on deforestation
- 7:47Biodiversity analysis challenges
- 11:41Fortran static analysis in climate science
- 14:10Climate & environmental initiatives
- 18:41Climate science education
- 20:27Coupling scientific models
- 23:15Data provenance struggles
- 24:20Floating point errors & language barriers
- 25:47Carbon footprint modeling
- 27:11Diverse skills for climate action
- 28:55PROPL engineering insights
- 30:01Building interdisciplinary pipelines
- 31:48Wrap-up
related talks
01h 11mProgramming for the planet
24mManaging Heterogeneous Spatial Data with Analysis Ready Cloud Optimized formats
39mPanel Discussion
24mProgramming Opportunities for the Global Biodiversity Observation Network
44mPanel Discussion: How Open Source Helps DataOps
33mPanel Discussion
33mPanel discussion - Melbourne
39mDoing Data: The Critical Process of Data Preparation (...)