Open pest early-warning pipeline for tomato leafminer and whitefly - Kirinyaga County, Kenya
# Pest Risk Kenya
An open pest early-warning pipeline for tomato leafminer (*Tuta absoluta*) and
whitefly (*Bemisia tabaci*), built for the Mwea tomato belt in Kirinyaga County,
Kenya — and written so that the risk map answers a food security question, not
just an entomological one.
**Status: methodology and code, not an operational service.** The repository
ships with synthetic demonstration data. It has not been validated against
field observations from Mwea.
## What this does
Satellite climate data goes in. Three things come out:
1. **A degree-day phenology model** — how many pest generations the accumulated
heat supports, and when the next one emerges. Mechanistic, parameterised from
published laboratory studies, with every constant cited and configurable.
2. **A validated statistical model** — pest presence against environmental
covariates, cross-validated *spatially* rather than randomly, which is the
difference between a real accuracy figure and a flattering one.
3. **A targeting layer** — hazard combined with exposure and household
vulnerability, because the places where the pest thrives and the places where
an outbreak does the most damage are not the same places.
The third is the part that makes this a livelihoods tool rather than a map.
## Why Mwea
Kirinyaga is Kenya's leading tomato-producing county, accounting for roughly
17% of national output, ahead of Kajiado (11.8%) and Taita Taveta (8.5%)
(Ndung'u et al. 2022). *Tuta
absoluta* arrived in Kenya in 2014 and reached most tomato-growing regions
within two years; icipe made the first African field release of the parasitoid
*Dolichogenidea gelechiidivoris* in Mwea East in 2020. Uncontrolled, the pest
causes 80–100% yield loss.
So the crop matters, the pest is present and documented, and there is an active
IPM programme an early-warning tool could plug into.
## Architecture: Earth Engine *and* R
These are not alternatives, and the question "GEE or R?" has a boring answer:
both, doing different …