Glossary & decision log

Plain-language meaning of the method codes referenced across the report

The report keeps a full audit trail: every method choice has a numbered decision (DEC-NNN) in the decision log, and the contracts live in docs/. Those codes signal rigor but are not something a reader needs to follow the findings. This page translates the ones the report cites into one plain sentence each, so the narrative pages can stay readable while the provenance stays one click away.

Decision codes

Code What it means in plain language
DEC-005 The control-area view (RQ3) is descriptive only — it maps where damage falls, never claims one administration fared better or caused anything.
DEC-007 The validation gate is closed by a human domain expert, never by an automated agent. Tests passing is not the same as correctness.
DEC-008 The report is a static, reproducible website (no live server); every figure is rebuilt from committed, validated data at render time.
DEC-015 Cropland is measured two ways — UNION of two land-cover datasets (the headline) and their INTERSECTION (a conservative low bound).
DEC-017 The 2025 rainfall deficit is the Nov–May season total minus the 1991–2020 climatological normal.
DEC-019 The 2025 ~1.2 Mt national cereal floor is split across governorates by cropland-area share (a uniform-yield assumption, stated as a limitation).
DEC-023 Flood extent is radar (SAR) open-water detection. Dry or harvested June fields can mimic open water in radar — the source of the Hasakah-June flag.
DEC-031 A burn scar counts as fire damage only where a VIIRS active-fire detection confirms it within 375 m — this separates fire from ordinary harvest/ploughing and makes the estimate a lower bound.
DEC-032 Fire damage is carried as a single union headline per governorate, with the union/intersection range in a companion table.
DEC-033 The food-security layer reads validated records only, normalises everything to the union headline, and uses each flood AOI’s peak single event-date so standing water is not double-counted across dates. Loss = hectares × the conservative 2025 drought-floor yield.
DEC-037 The fire pipeline runs nationally (all governorates with 2026 cropland fire) and frames fire as a drought-and-heat agricultural hazard, not a conflict story.
DEC-038 Latakia and Damascus City were checked with the full method and found to have negligible cropland fire — verified-excluded, not silently dropped.
DEC-039 This is a first-half-2026 snapshot (floods Mar–Jun; fires May 1 – Jun 12). The summer harvest/fire peak is unobserved, so every figure is a lower bound.

Document & section references

Code What it points to
§3.3 The 2025 drought baseline the 2026 shock is measured against.
§3.4 The food-security join: damaged hectares → production loss → indicative pressure.
§5 The locked research questions (RQ1–RQ3).
§6 The validation tiers; Tier-2 outputs (damage hectares) require human sign-off.
§9 The proportionate-claims discipline: causal claims kept within the evidence.
S6 / S13 The flood / fire pipeline work sessions whose Tier-2 gates a human closed.

Full text: the decision log · PRODUCT.md · STRUCTURE.md.