Satellite crop condition and fertilizer demand data for Kenya, Tanzania, Uganda and Rwanda

How the crop condition anomaly is read

No unit here is measured against another. Each of the 243 is compared with itself: this week’s cropland vegetation against what the same weeks looked like over the 8 years before, so a county that simply runs greener than its neighbours only counts as above normal once it beats its own record. A live 10 m product supplies the fresh cropland-only reading; a 300 m product with 8 years behind it supplies the climatology they’re read against. The platform calibrates the 2 against each other per unit, and where they disagree past a set tolerance it serves the lagged reading and says so in the response rather than picking one silently. Thresholds of ±0.02 and ±0.05 sort the anomaly from well below to well above normal, and the outlook itself reads rising, steady or softening off the anomaly and its trend together.

The 3 crop condition datasets

The satellite demand indices carry the raw values: 4 optical indices per unit and week, plus the 300 m NDVI and soil-water series behind them. The rainfall anomaly states season rainfall as a departure from the long-term normal, by unit and season, for the whole continent. The agricultural stress layer is a monthly bulletin: warning levels for cropland and rangeland, plus an analyst’s paragraph naming the actual places and causes.

The fertilizer demand board

The board ranks every unit in a country by demand pull, which is anomaly against the unit’s own normal blended with cropland intensity, and in Kenya with fertilizer affordability on top. A low-intensity unit having a green spell reads as a small fertilizer market catching a lucky week; the blend keeps it from outranking a breadbasket county that is merely on normal. The board returns 1 row per unit the country has.

At this snapshot the top units were Nandi, Nyamira and Lamu in Kenya; Bundibugyo, Kiryandongo and Masindi in Uganda; Morogoro and Singida in Tanzania; Nyaruguru in Rwanda, while most of the rest of the region was still climbing out of a dry July. 3 sources tell the same story: the July bulletin flagged Kenya as a hotspot and named Kakamega, Bungoma, Uasin Gishu, Elgeyo Marakwet and Trans Nzoia; the news signals named the same counties a week later. Satellite, bulletin and press: 3 datasets, 1 shared county vocabulary.

Refresh cadence of the satellite layers

The 10 m product refreshes weekly across all 243 units. The 300 m product runs 10-daily and publishes about 40 days behind. The rainfall anomaly updates weekly, the bulletin monthly. A unit with no clean satellite composite in its window simply keeps whatever value it had before.