Irrigation Scheduling Recommendations and Dataset Construction (LLM-based Meta-analysis)

Overview

Irrigation scheduling strongly influences crop yield and water productivity, yet practical adoption of “smart” scheduling methods remains limited because method performance varies across crops, climates, soils, and farm constraints. Prior work is dominated by localized case studies, making it difficult for growers to select an appropriate method for their context. This study developed a large language model (LLM)-assisted mining pipeline to extract and normalize reported irrigation scheduling methods and outcomes across published studies. The final processed dataset contains 1083 samples from 328 papers, covering 1037 unique irrigation scheduling methods, 345 plant types, and 415 geographical locations. To illustrate potential usage, we built a yield prediction model that, for a given scenario, predicts plant yield for candidate scheduling methods. Six different irrigation methods, including those based on soil moisture, evapotranspiration, and deficit irrigation, were used in this analysis. XGBoost achieved the best performance with an R^2 score of 0.867. Finally, we utilize the dataset to build a recommendation system that predicts optimal irrigation methods under various conditions. It is evaluated by comparison against selected research papers that assessed multiple irrigation systems. Therefore, the unique contributions of this project are a LLM-assisted literature mining pipeline, the first known dataset of its kind containing irrigation scheduling methods, a demonstrated example of how ML models can be trained on this dataset and used for personalized irrigation method effectiveness prediction and recommendation at scale. A research paper on this work has been accepted to the IEEE Conference on Technologies for Sustainability 2026.

Competition history

• CSEF 2026 Environmental Engineering (Track 2) (Senior Division) · Entry S-12-12

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Environmental Engineering projects
• CSEF Environmental Engineering projects
• Every CSEF 2026 project