A Dynamic Framework for Optimizing Crop Yield Under Climate Variability

Thesis Abstract

Climate variability poses a substantial threat to stable crop production, constraining yield by altering temperature regimes, precipitation patterns, and water availability. This study develops a dynamic framework to optimize crop yield under climate variability by integrating biophysical crop models with probabilistic climate scenario analysis and decision-support tools. The aim is to generate adaptive management strategies that maintain yield resilience while minimizing resource use and environmental impact. Specific objectives include (i) quantifying the responsiveness of major cereal crops to stochastic climate drivers using a process-based model and long-term observational data, (ii) developing a dynamic optimization framework that links phenological development, irrigation scheduling, and nutrient management to temporal climate risk, (iii) incorporating farmer-specified risk preferences through a utility function and exploring robust and stochastic optimization under multi-scenario climate futures, and (iv) validating the framework across diverse agro-ecological zones with an emphasis on water-limited and heat-stressed conditions. The methodology adopts a mixed-methods design anchored in a systems-based, model-driven approach. A process-based crop model (e.g., DSSAT or APSIM) will be calibrated and validated with field trial data comprising 12 site-years of multi-environment trials for wheat and maize, including 200+ on-farm observations. Climate inputs will be drawn from downscaled CMIP6 projections under Representative Concentration Pathways RCP4.5 and RCP8.5, generating 30-year hindcast and 20-year forward scenarios. The population consists of commercial and smallholder farming units across three contrasting agro-ecological zones, with a stratified random sample yielding 150 farm households and 20 extension agents as key informants. Data collection instruments include standardized crop-phenology sensors, soil moisture probes, irrigation and fertilizer records, and structured interview guides for stakeholders. Instrument validity will be established through content validity with agronomy experts and pilot testing in two districts, while reliability will be assessed via test-retest procedures and calculation of Cronbach’s alpha for attitudinal indicators. Analytical methods entail a three-tier modeling approach. First, a process-based crop model will simulate yield responses to climate variables and management practices, enabling sensitivity analyses to identify critical drivers. Second, a dynamic optimization module, employing stochastic programming and robust optimization, will determine adaptive irrigation and nutrient regimes that maximize expected yield subject to risk constraints, parameterized by farmer risk aversion. Third, a scenario analysis will compare strategies across climate futures and zones, with results synthesized through meta-regression to identify generalizable policy-relevant patterns. Statistical techniques will include linear and non-linear regression, ANOVA for treatment effects, and Bayesian updating to refine parameter estimates as new data accrue. Theoretical anchors include the Yield-Risk Optimization Theory and the Climate-Resilient Agriculture framework, complemented by the Theory of Planned Behavior to interpret adoption intentions among farmers. A conceptual model will illustrate the interaction between climate variability, crop response, management decisions, and yield outcomes. Expected findings include (i) quantification of the marginal yield gains achievable through adaptive management under variable climates, (ii) identification of threshold climate conditions beyond which conventional practices lose efficacy, (iii) robust guidelines for irrigation and fertilization that balance yield, water-use efficiency, and cost, and (iv) insights into barriers to adoption and the influence of risk preferences on management choices. The study contributes to knowledge by offering a dynamic, integrative framework that unites process-based crop modeling with optimization under uncertainty, enabling evidence-based, climate-responsive agricultural decisions. It will inform policy on water allocation, extension messaging, and investment in climate-smart infrastructure. The main conclusion is that adaptive, scenario-informed management markedly improves resilience and yield stability under climate variability; recommendations include deploying decision-support tools at district levels, tailoring recommendations to zone-specific risk profiles, and promoting farmer training to operationalize probabilistic management of climate risk.