Development of a hybrid crop model for rice using AI-assisted coding

Date

5-2026

Adviser

John Carlo L. Navasero

Principal

Buela, Mabel S.

Abstract

Crop models simulate how plants grow and produce yields based on data on environmental conditions (Oteng-Darko et al., 2013). These models help create crop-planting schedules, determine the land required for crop cultivation, and support a wide range of activities related to planning production, conducting research on experimental projects, commercial uses, and making decisions about agricultural policy (Le et al. 2023). With rice being one of the world's staple foods, providing over 50% of the daily calorie intake for billions of people, climate-induced temperature stress is emerging as a major threat to rice production, especially in rice-producing countries such as the Philippines. In addition to improving agricultural efficiency and securing food supplies, the learning capabilities and adaptability of artificial intelligence can help farmers make informed decisions (Zha, 2020). Therefore, the primary goal of this research was to build an AI-assisted crop model for rice that could simulate the growth, development, and productivity of rice plants. More specifically, the researchers developed modules representing processes during the life cycle of rice plants, trained the model to predict biomass, and validated its predictions of leaf, stem, and grain weights using DSSAT-generated outputs. Additionally, the researchers used DSSAT datasets that include information for growing rice, including soils, cultivars, weather conditions, and experiment files from two experiments conducted on IR58 and Basmati 385. ChatGPT Codex supported the researchers in developing modules, calibrating parameters, and analyzing data files. The model achieved "excellent" accuracy for leaf and grain biomass, but "good" for stem biomass. For dataset 2 (Basmati 385), RRMSE values were higher (11.68% leaf, 14.22% stem, 10.43% grain), with "good" predictions across all components. Overall, this study demonstrated that the model could predict vegetative development and grain production across the crop cycle, with predicted-versus-actual trends remaining consistent with DSSAT.

Language

English

Location

UP Rural High School

Document Type

Capstone

This document is currently not available here.

Share

COinS