Kenya Avocado Atlas — Kandara Sub-County pilot

Kenya Avocado Atlas Kandara Sub-County pilot

Kandara drone COGs · Sentinel-2 optional

View extent

Murang'a County outline

Kenya national outline

Kandara imagery loading Sentinel-2…

Optional 10 m context over all six wards. Moisture, EVI, and clouds are Sentinel-2-only. High-resolution drone COGs are in the panel below.

Show live Sentinel-2 layer

Searching AWS Earth Search for the clearest recent scenes over Kandara…

Time-travel (last 5 years, quarterly)

Live — most recent clear scenes

Kandara drone COGs ready

Ward / mosaic

Opacity

Show drone COG layer

low NDVI high

Served via same-origin /cog (TiTiler) from public processed COGs under kandara-subcounty/, date 2026-10-02. No API keys in the browser.

Downloads

For the ward and layer selected above. Public S3 links only — no farmer names, no API keys.

Download GeoTIFF (source COG)HTTPS S3 object Ward layers manifest (CSV)kandara-subcounty/ward_layers_manifest.csv

Ward boundaries

Show Kandara ward outlines

Shade wards by NDVI (Sentinel-2)

Kandara ward comparison

no data yet

Ward	Imagery	Latest NDVI	Quarters

Sourced from Sentinel-2 (10m, quarterly) via src/pipeline/sentinel2_timeseries.py — free public imagery covering all six wards since 2021, independent of the drone COG tiles. Not yet generated for this repo; see data/derived/README.md. Click a ward on the map for its summary once data exists.

NDVI scale

-1 (bare/water)+1 (dense canopy)

Vegetation canopy status

No canopy< 0.20

Establishing0.20 – 0.45

Developing0.45 – 0.65

Closed canopy> 0.65

Tree-crop-tuned thresholds, not generic vegetation bins — see docs/technical/vegetation-status-scheme.md. "Closed canopy" means healthy-looking, not confirmed healthy: NDVI saturates at high canopy density and can miss early stress.

Area analysis

Draw a polygon over the map to get real NDVI statistics for that area — mean, range, and a canopy-status call, computed from the actual capture data.

Area—

Mean NDVI—

Range—

Sample cells—

—

⚠ Estimate — remote sensing only, not field-verified