OCHA-DAP/ds-aa-uga-flooding

Anticipatory action for flooding in Uganda — trigger design across four zones (Teso/Kyoga riverine, Mt Elgon, Karamoja, Adjumani/Albert Nile) # ds-aa-uga-flooding Trigger design for **anticipatory action (AA) for flooding in Uganda** — a multi-zone mechanism for the OCHA/CERF framework under development (Sep 2026). Builds on the exploratory work in `ds-seas5-skill` (Uganda drought & flood analysis, OND 2026 flood-trigger design options, FloodScan recurrence layers, GloFAS skill verification) — that repo keeps the country-team readouts; this one holds the trigger analysis proper. ## The four zones | zone | regime | indicator under design | validation chain | |---|---|---|---| | **Teso / Lake Kyoga** — Akokoro river | riverine | GloFAS reporting point **G5196** (Akokoro at Uganda gauge) reforecast/reanalysis | discharge → FloodScan extent per district → impact; first task: which districts the point covers | | **Mount Elgon** | flash floods + landslides | rainfall forecast (CHIRPS-GEFS, then ECMWF) over the massif, antecedent-wetness qualifier | forecast rain → observed rain (IMERG) → observed flooding/landslide impact | | **Karamoja** | flash floods | rainfall forecast, likely sub-zoned by basin | same chain as Elgon | | **Adjumani / Albert Nile** | riverine / lake-backwater | undecided: GloFAS Albert Nile points, Lake Albert level (altimetry), or rainfall | history of what actually flooded there comes first | Plus an **observational backstop** in every zone (FloodScan flood extent, report-based where the satellite is blind) so a forecast miss still activates. Zone membership (core vs candidate districts) is in `src/constants.py`; the coverage map is `outputs/zones_coverage_map.png` (`pipeline/zones_map.py`), which also shows other organisations' flood AA coverage (`src/frameworks.py`). ## Layout - `src/constants.py` — blob prefix, zones, GloFAS point, seasons - `src/zones.py` — CODAB resolution of zones; `src/zonal.py` — windowed COG reads + exactextract - `src/datasources/` — `glofas.py` (EWDS reanalysis/reforecast), `chirps_gefs.py` (CHC rainfall forecasts), `impact.py` (EM-DAT + curated events → …