Growing global food demand and freshwater scarcity are exacerbating pressure on agricultural systems, particularly in arid and semi-arid regions. While there is growing interest in sustainable land management, few studies have comprehensively addressed the trade-offs and synergies among water use, agricultural productivity, and ecological services. This study develops the MOWAE_CAO model-a multi-objective optimization framework that integrates water resource efficiency, net economic benefits, and ecosystem service values. By constructing four scenarios: S1 (water efficiency priority), S2 (economic priority), S3 (ecological priority), S4 (integrated optimization), and coupling the Non-dominated Sorting Genetic Algorithm II (NSGA-II) with entropy-weighted TOPSIS, the model identifies and evaluates Pareto cropping structure optimization under competing policy goals within the water-agriculture-ecology nexus. Applied to the Shiyang River Basin in northwestern China, the model revealed scenario-specific land allocation strategies. Using the actual 2021 cropping pattern as a baseline, Scenario 4 (S4) achieved the most balanced outcome, simultaneously improving all three sustainability criteria and integrated multi-objective solution. Compared to the baseline, S4 enhanced the net economic benefit by 4.05%, increased ecosystem service value by 3.67%, and improved crop water productivity for both maize and spring wheat. These findings underscore the potential of integrated, data-driven optimization to manage complex trade-offs and promote synergistic outcomes, offering actionable insights for sustainable land-use planning in ecologically vulnerable and water-scarce regions.