Downscale management scenarios to field scale

[dlebauer]

Objective

Generate field level estimates of AGB and SOC for each annual crop field in the LandIQ dataset, for each of six management scenarios.

Approach

Update R scripts 030, 031, and 040 in the scripts/ folder of the downscaling repository.

Until ccmmf/downscaling#5 is merged, develop on the mixed_downscaling branch (PRs targeting this branch will automatically be redirected after the branch is merged).

There are six scenarios defined in the scenario repository. Files used to specify different management scenarios are in the data/ folder of https://github.com/ccmmf/scenarios.

Scenario	Changes from baseline
baseline	Following conventional practices
compost	Adds compost (0.25 kg C/m2, 0.015 kg N/m2); remove mineral N fertilizer
reduced_till	Reduce tillage intensity to 0.10.
zero_till	No tillage with tillage_eff_0to1 = 0.00.
reduced_irrig_drip	Replace canopy irrigation schedule with 40 drip events (8 mm each, soil method).
stacked	Combine compost, reduced_till, and reduced_irrig treatments.

SIPNET outputs for these management scenarios are on geo.bu.edu at /projectnb2/dietzelab/ccmmf/modelout/ccmmf_phase_3_scenarios_20251210/.

These outputs should be treated as fixed inputs for the task

Considerations / Context:

For context, we should

• Maintain backward compatibility, and the resulting scripts should support:
  • multiple PFTs (from phase 2, implemented in Downscaling mixed systems downscaling#5)
  • multiple management scenarios (being implemented here)
• Consider that in the future we will support
  • multiple PFTs x scenarios
  • future ensemble based management scenarios (Generate ensembles of management scenarios #160)

Acceptance criteria

1. The following files:
   • downscaled_preds_metadata.json
   • downscaled_deltas.csv
   • downscaled_preds.csv
   • These outputs should be converted to parquet if/when size becomes unwieldy
2. Updated documentation

[divine7022]

addressed in ccmmf/downscaling#7