Contact person: R Spatial Model Debugging & Uncertainty

Phone:Show

Email:Show

Location: Tokyo, Netherlands

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I am close to releasing an R-based spatial emissions model, but a final sweep is required to make sure the entire multi-script workflow is stable and, most importantly, that the uncertainty and error metrics are reported rigorously. You will collaborate directly with our internal coordinator, who already runs the pipeline and prepares the annex, so your time can be spent purely on the technical side.

The codebase is built around sf, raster and terra objects and currently executes on a cloud VM. Most functions already run in parallel through future, foreach and doParallel, yet a few stubborn bottlenecks and occasional path-related errors still appear. Once those are ironed out, the main focus will be to quantify data accuracy, precision and error propagation throughout the workflow, then surface those metrics in a concise report that can be dropped straight into the annex.

Deliverables
• Clean, runnable set of R scripts with all remaining bugs fixed
• Integrated functions (or a separate module) that calculate and summarise uncertainty and error metrics for every spatial output layer
• Short write-up (markdown or RMarkdown) explaining methods, assumptions and key results, ready to be copied into the annex

If you have a track record in environmental or emissions modelling, you’ll feel right at home, but a solid grasp of large spatial data handling and parallel processing in R is the real prerequisite. I can grant SSH access to the VM immediately so we can start turning these last loose ends into a polished, defensible model." (client-provided description)