ARR Project Pipeline
Documentation for the ARR Project Pipeline in LandOS.
Overview
The ARR Project Pipeline automates the sequential execution of the core computations required for an ARR prefeasibility assessment. Rather than running each computation independently, the pipeline manages execution order, dependency, and data passing between steps, reducing manual workflow overhead and ensuring that downstream computations receive necessary inputs. These ARR Project Pipelines are for native forestation projects that restore forest to a closed canopy, secondary forest. Agroforestry models are coming soon.
Two pipeline configurations are available, differentiated by how carbon credit estimates are derived:
| Pipeline | Tier | Land Eligibility Input to Carbon Model |
|---|---|---|
| ARR Project with Eligible Area Pipeline | Pro | Eligible area from Land Eligibility computation |
| ARR Project Pipeline | Free | Project area from General Polygon only |
Pipeline Steps
Both pipeline configurations execute the same four computations. The dependency structure — which steps must complete before others begin — differs between tiers.
Step 1 — Biophysical Characterization
Characterizes the biophysical context of the project area: biomass, climate, terrain, ecozones, and ecoregions. Runs immediately on pipeline start with no dependencies.
Step 2 — Land Eligibility
Assesses which portions of the project area qualify for ARR crediting under VM0047. Automatically selects between the Terra Vista and Forest Loop methods based on geographic coverage. Runs immediately on pipeline start with no dependencies.
Step 3 — Native Restoration Carbon Removals
Projects ex-ante carbon removals using the Chapman-Richards growth model. The inputs to this step depend on the pipeline tier:
- Pro tier: Requires both General Polygon and Land Eligibility to complete. Carbon projections are derived from the eligible area identified by Land Eligibility, producing a more constrained and methodologically defensible estimate.
- Free tier: Requires only General Polygon to complete. Carbon projections are derived from the full project area, without applying a land eligibility filter.
Step 4 — Project Development Overview
Outputs the carbon project development overview and reference materials. Runs immediately on pipeline start with no dependencies.
Execution Model
Steps with no dependencies (General Polygon, Land Eligibility, and Project Development Overview) are enqueued simultaneously when the pipeline starts. The Native Restoration Carbon Removals step is held until its required dependencies reach a COMPLETED state.
Each step transitions through the following lifecycle states:
PENDING → QUEUED → PROCESSING → COMPLETED
↘ FAILED
Steps are executed as independent computation tasks. If a step fails, dependent steps will not be enqueued. The pipeline as a whole does not automatically retry failed steps; rerun functionality is available at the pipeline level.
Dependency IDs
When starting a pipeline, users may supply dependency_ids, a mapping of step
keys to existing computation IDs. This lets the user select a specific
version of a computation if multiple variations have been run.
Steps with pre-supplied computation IDs are marked COMPLETED immediately and
skipped during execution. Their associated computation assets are preserved if the pipeline is deleted.
This mechanism allows users to reuse prior computation results (for example, a previously completed Land Eligibility assessment) without re-running the full pipeline.
Rerun
The pipeline supports full rerun from the start. On rerun, existing computation records are reset rather than replaced, preserving computation IDs across runs. This ensures consistency when referencing pipeline outputs from external tools.