AquaCrop model performance evaluation and centennial simulations in a rainfed dryland agroecosystem

Rainfed smallholder farming systems in semi-arid sub-Saharan Africa (SSA) are vulnerable to intra-seasonal rainfall variability, prolonged dry spells, and high evaporative demand that constrain crop productivity. AquaCrop—a water driven crop growth model, has been widely applied to assess crop performance and water use in water-limited environments. However, long-term, multivariable evaluations spanning multiple growing seasons remain scarce in SSA dryland.

Thus, this study was conducted to evaluate the capacity of AquaCrop to reproduce maize yield components, crop evapotranspiration (ETc), soil water storage (SWS) and plant growth dynamics. We assessed these variables across multiple growing seasons spanning contrasting hydroclimatic years and quantified how rainfall characteristics relate to yield in a typical semi-arid agrosystem in Africa.

Field experiments in Maktau, Kenya, spanned six growing seasons (2019–2024), covering both long (LR) and short (SR) rains. AquaCrop was calibrated for SR2023 and validated for LR2024, with additional growing seasons adopted for testing. LARS-WG—a stochastic weather generator was utilized to generate a 100–year weather data for Maktau by utilising in-situ meteorological data (2013–2024). This weather data was then employed to quantify how plant–available soil water relate to yield.

Simulated maize final biomass and yields generally tracked observed data, with percent errors for final aboveground biomass and grain yield ranging from −23% to 33% and −19% to 7%, respectively, under total seasonal rainfall of 176–489mm. The model showed satisfactory performance for ETc (R² = 0.41–0.81), mixed performance for SWS across growing seasons (R² = 0.15–0.69) and accurately captured canopy cover (CC) dynamics (R² ≥ 0.91). In the centennial analysis, maize grain yield variability was strongly associated with total seasonal rainfall (R² = 0.44), with grain yield ranging from crop failure to 3.8 t/ha under total seasonal rainfall of 37–592mm.

Crop failure and low yields were associated with lower rainfall in May, which coincided with the tasselling–flowering stage of the dryland maize variety DH02 planted in early March as is typical in Maktau. Soil water deficit in the tasselling–flowering stage disproportionally impacted maize yield.

Beyond reaffirming the established rainfall and yield relationship, the findings provide clear, actionable insights for smallholder dryland systems in Kenya and similar dryland agrosystems: (i) timing of rainfall, particularly during tasselling–silking, is a critical determinant of yield loss or realization of yield, suggesting the value of matching cultivar maturity and sowing windows to the temporal distribution of rainfall; (ii) supplemental irrigation targeted to critical maize growth stages; and (iii) selection of maize varieties with accelerated growth cycles to minimises the exposure to extended periods of no rainfall that leads to soil water deficits as the crops are able to complete critical growth stages (tasselling–flowering) faster averting the risk of crop failure.