A season that looks normal on paper can still fail a farmer, according to an analysis of five years of monthly rainfall data for Kisumu, Kenya, published by the Alliance of Bioversity International and CIAT. The study, by Agnes Njambi Wanjau, Anastasia Wahome, Chris Miyinzi Mwungu, Benson Kenduiywo, Felix Otieno, Aniruddha Ghosh and Pedro Anglaze Chilambe, underpins irrigation advisory work under the iSPARK project.

The researchers used monthly rainfall estimates for 2018-2022 from CHIRPS (Climate Hazards Center InfraRed Precipitation with Station), a satellite-and-station product openly published through the Digital Earth Africa CHIRPS registry. Time-series decomposition separated the record into trend, seasonality and remainder, to tell a real climate signal from ordinary noise.

Kisumu's seasonal shape is consistent: rains in March-May and July-August, and drier months from September to January. But the month-to-month swings are large. September is the most predictable month, reliably dry with little variation between years. May is the hardest: on average the second-wettest month, yet its rainfall ranged from under 50 mm to over 350 mm across the five years, right in the planting window.

May 2021 shows the problem. Statistically it was a normal May, but rainfall collapsed to below 50 mm in the middle of the month, when maize was at its most water-sensitive growth stage. Kenya's Ministry of Agriculture recorded an unusually wet April that year, at 134% of normal, followed by the abrupt stop, a sequence that a seasonal average hides entirely.

For farmers, the gap between what is normal and what actually happens is the difference between a good season and a lost one. A month that is usually wet can fail without warning, and by the time the failure is obvious the planting window has closed.

The findings line up with T-CSIA, the Targeted Climate-Smart Irrigation Advisory Using Integrated Climate Indicators, which identifies eastern Kisumu as an area of persistent irrigation-water stress. The authors conclude that advisory systems have to track rainfall as the season unfolds rather than rely on historical averages, which is the approach the iSPARK irrigation advisories take.

Source: CGIAR