Integrating Meteorological Forecasts and Scenario Simulation to Support Smallholder Maize Farming Decisions in Kakamega County, Kenya

Abstract

Seasonal rainfall outlooks can support advance planning in rainfed maize systems, but farm operations depend not only on seasonal rainfall totals but also on rainfall timing, continuity and short dry periods. This study examined Kenya Meteorological Department (KMD) March-May seasonal outlooks for Kakamega County from 2021 to 2025 and used a transparent stochastic scenario model to explore rainfall indicators relevant to farm decision-making. Official outlook messages were reviewed separately from retrospective observations. Three illustrative rainfall scenarios, each comprising 10,000 synthetic 92-day sequences, represented sparse, intermediate and frequent rainfall conditions. A candidate planting trigger was defined as at least 20 mm of rainfall accumulated over three consecutive days and completed by day 30. Among sequences reaching the trigger, a subsequent seven-day dry spell within 21 days occurred in 54.6%, 11.0% and 0.7% of the sparse-, intermediate- and frequent-rain scenarios, respectively. A separate three-day dry run within 30 days occurred in 100.0%, 95.3% and 58.5% of the same scenarios. Sensitivity analysis showed that rainfall occurrence and persistence could substantially alter post-trigger dry-spell exposure even when expected seasonal rainfall was held constant. The results demonstrate that seasonal rainfall totals alone do not capture decision-relevant rainfall sequences. A practical climate-service approach should therefore combine seasonal contingency planning with short-range updates and field verification. The simulated proportions are illustrative and should not be interpreted as calibrated probabilities of crop failure, yield loss or other farm outcomes in Kakamega.