Baseline Matching and Diagnostics Assessment

Purpose

The Baseline Matching and Diagnostics Assessment tool applies the Spatially Explicit Matched Dynamic Baseline (SEMDB) approach to prepare a defensible baseline package before Carbon Verification Audit by screening donor candidates, validating matching quality, and emitting trajectory plus uncertainty artifacts for audit preparation.

Typical Questions This Tool Helps Answer

• Is the donor pool sufficiently comparable to project units before verification packaging?
• Do matching diagnostics pass configured SMD and trend-consistency thresholds?
• What baseline trajectory and uncertainty outputs should be carried into downstream carbon audit reporting?

Background

The Baseline Matching and Diagnostics Assessment tool is an upstream baseline-defensibility workflow, not a replacement for downstream verification packaging. It helps teams reduce audit risk by enforcing donor eligibility and comparability diagnostics before claim interpretation.

This workflow is most useful when regulator or verifier scrutiny requires explicit evidence that baseline assumptions are stable and reproducible.

Inputs

Parameters

• donor_pool_exclusion_distance_km (optional): leakage-control exclusion distance from project boundary in km; default 5.0.
• matching_mode (optional): mahalanobis or propensity_score; default mahalanobis.
• neighbors (optional): nearest-neighbor count per treated cell; default 1.
• with_replacement (optional): donor reuse toggle; default true.
• tenure_field (required when tenure_status is provided): attribute field name in the tenure layer carrying the legal-status class value.
• calliper_elevation_m (optional): hard calliper for elevation in absolute metres (SEMDB spec default: 200 m).
• calliper_slope_deg (optional): hard calliper for slope in absolute degrees (SEMDB spec default: 10°).
• calliper_soc_pct (optional): hard calliper for SOC as a percentage of the project-area mean (SEMDB spec default: 10%).
• calliper_ndvi_pct (optional): hard calliper for mean pre-period NDVI as a percentage of the project-area mean (SEMDB spec default: 10%).
• calliper_road_distance_km (optional): hard calliper for road distance in absolute kilometres (SEMDB spec default: 1 km).
• exact_match_soil_texture_class (optional): enforce exact class matching; default true.
• random_seed (optional): deterministic seed for reproducibility; default 0.
• smd_threshold (optional): maximum accepted SMD across all covariates; default 0.1.
• parallel_trend_tolerance (optional): trend-slope tolerance; default 0.01.
• pre_period_min_years (optional): minimum pre-period depth; default 5.
• output_prefix (required): output basename for all artifacts.

Matching Covariates

Six covariates are used in Mahalanobis distance matching. A separate SMD is reported for each in smd_diagnostics:

NDVI Stack Band Ordering

Bands in pre_period_ndvi_stack must be in chronological order, oldest first (band 1 = year −5, band 5 = year −1). There is no embedded year labelling — the tool uses band position. To create the stack:

gdal_merge.py -separate -o ndvi_stack.tif ndvi_y-5.tif ndvi_y-4.tif ndvi_y-3.tif ndvi_y-2.tif ndvi_y-1.tif

In QGIS: Raster → Miscellaneous → Merge, enable "Place each input file into a separate band", add files chronologically.

Spatial Alignment

All inputs are aligned to the year0_landcover grid before any computation:

• Raster inputs are reprojected and resampled to match year0_landcover CRS, cell size, and extent.
• Vector inputs (roads, tenure_status, project_boundary, ecoregion_layer) are reprojected to the year0_landcover CRS if needed.
• Road and tenure vectors are rasterised directly onto the year0_landcover grid — you do not need to pre-rasterise them.

Outputs

QA and Acceptance Criteria

Minimum acceptance before downstream Carbon Verification Audit:

1. Required inputs validated and aligned.
2. Donor pool remains operationally adequate after filters.
3. SMD diagnostics pass agreed threshold policy.
4. Parallel trend diagnostics pass agreed tolerance policy.
5. Baseline trajectory/uncertainty artifacts are complete.

Troubleshooting

• Sparse donor pool: review exclusion distance and eligibility constraints.
• SMD failures: tighten callipers and inspect covariate outliers.
• Trend failures: increase pre-period depth and review NDVI stack quality.
• Reproducibility mismatch: lock random_seed and verify unchanged inputs.

For additional operator support, share these artifacts with Whitebox support:

• summary
• preflight
• smd_diagnostics
• parallel_trends

Example

import whitebox_workflows as wbw

env = wbw.WbEnvironment(include_pro=True, tier="pro")

result = env.baseline_matching_and_diagnostics_assessment(
    project_boundary="data/project_boundary.gpkg",
    ecoregion_layer="data/ecoregions.gpkg",
    year0_landcover="data/landcover_y0.tif",
    pre_period_ndvi_stack="data/ndvi_pre_stack.tif",  # bands in order: yr-5, yr-4, yr-3, yr-2, yr-1
    elevation="data/elevation.tif",
    precipitation="data/precip.tif",
    soil_organic_carbon="data/soc.tif",
    soil_texture_class="data/soil_texture.tif",
    roads="data/roads.gpkg",
    tenure_status="data/tenure.gpkg",     # optional
    tenure_field="legal_status",           # required when tenure_status is provided
    donor_pool_exclusion_distance_km=5.0,
    matching_mode="mahalanobis",
    neighbors=1,
    calliper_elevation_m=200.0,
    calliper_slope_deg=10.0,
    calliper_soc_pct=10.0,
    calliper_ndvi_pct=10.0,
    calliper_road_distance_km=1.0,
    smd_threshold=0.1,
    parallel_trend_tolerance=0.01,
    output_prefix="output/semdb_baseline",
)

print(result)

References

• Tool implementation: wbtools_pro/src/tools/workflow_products/baseline_matching_and_diagnostics_assessment.rs

Advanced Operational Guidance

• Use fixed-seed reruns for verifier-facing reproducibility.
• Archive summary, preflight, smd_diagnostics, and parallel_trends together.
• Keep threshold profiles stable within a reporting cycle.

Positioning vs Carbon Verification Audit

• Baseline Matching and Diagnostics Assessment: baseline construction and comparability diagnostics.
• Carbon Verification Audit: downstream verification packaging and audit-ready reporting.

Use SEMDB first, then Carbon Verification Audit once baseline diagnostics are acceptable.

When To Use This Workflow

Use Baseline Matching and Diagnostics Assessment when baseline defensibility is under scrutiny and you need explicit comparability diagnostics before audit-stage reporting.

Results Delivery Checklist

1. Input provenance and temporal scope documented.
2. Threshold policy values recorded (smd_threshold, parallel_trend_tolerance).
3. SMD and trend outputs reviewed and accepted.
4. Baseline trajectory plus uncertainty outputs attached to handoff package.

Common Questions

Q: Can we skip SEMDB and go directly to Carbon Verification Audit? A: You can, but you lose explicit baseline-defensibility diagnostics that many reviewers request.

Q: What is the most common SEMDB review blocker? A: Imbalance or trend diagnostics that fail policy thresholds.

Q: Does SEMDB issue certified credits? A: No. It is an upstream baseline diagnostics workflow that supports defensible verification preparation.