Adaptation simulation and planning for crop yield under climate change: Integrating AquaCrop and DSSAT to project drought-induced yield risks in the Sanjiang Plain

Shehakk Muneer Baluch, Luchen Wang, Muhammad Abrar Faiz, Haiyan Li, Yingshan Chen, Lijuan Wang and Mo Li

Agricultural Water Management, 2025, vol. 319, issue C

Abstract: The Sanjiang Plain is used as a case study to demonstrate a novel fusion of mechanistic crop modeling and machine-learning for enhanced yield prediction under climate change, focusing on mid-century (2021–2060) conditions. We introduce a Gaussian-process (GP) ensemble framework that integrates simulated outputs from AquaCrop and DSSAT with multi-source environmental covariates to leverage both process-based realism and data-driven flexibility. Applied to maize, rice, soybean, and wheat under rain-fed and irrigated regimes, this ensemble improves out-of-sample accuracy by 10–15 % relative to either model alone, with performance of R² = 0.85–0.98 for DSSAT and R² = 0.52–0.78 for AquaCrop. To deconstruct prediction uncertainty, SHAP (SHapley Additive exPlanations) is applied to the GP outputs, transparently attributing variance to irrigation depth, in-season rainfall, and multi-scale SPEI (Standardized Precipitation Evapotranspiration Index). This analysis reveals that irrigation parameters accounts for over 60 % of yield variability across all crops, substantially outweighing climate-stress factors, and identifies maize’s tasseling to grain-filling, rice’s panicle initiation to grain-filling, soybean’s flowering to pod-filling, and wheat’s jointing to grain-filling stages for targeted water management. Projecting mid-century yields under Shared Socio-Economic Pathways (SSP) SSP1–2.6, SSP2–4.5, and SSP5–8.5, we quantify steep rain-fed declines in maize (–42 %), rice (–8 %), soybean (–15 %), and wheat (–12 %) and generate high-resolution maps of 30th- and 70th-percentile shortfall probabilities. Under SSP5–8.5, the median probability of ≥ 30 % wheat loss reaches 80 % in rain-fed fields, pinpointing the central and eastern belts as urgent adaptation hotspots. However, these projections are still constrained by model settings, data quality, structural differences between the models, historical calibration, and uncertainty in future climate. Overall, this study provides a transferable blueprint for climate-resilient agriculture on the Sanjiang Plain and beyond.

Keywords: Machine learning; Shared Socio-Economic Pathways (SSPs); Drought indices (SPEI); Irrigation management; SHAP attribution; Spatial risk mapping (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0378377425005323
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:agiwat:v:319:y:2025:i:c:s0378377425005323

DOI: 10.1016/j.agwat.2025.109818

Access Statistics for this article

Agricultural Water Management is currently edited by B.E. Clothier, W. Dierickx, J. Oster and D. Wichelns

More articles in Agricultural Water Management from Elsevier
Bibliographic data for series maintained by Catherine Liu ().