Optimizing Irrigation through Deep Reinforcement Learning for Cotton Crops
AGU Fall Meeting Abstracts, 2024 · pp. B01-81
Abstract
Cotton cultivation in the southeastern United States significantly impacts regional freshwater resources due to its substantial irrigation requirements. Current irrigation practices, largely dependent on soil moisture sensors and grower experience, fail to optimize water use by not considering dynamic factors such as soil texture, prevailing weather conditions, and the crop's phenological stage. This paper proposes an innovative approach to enhance irrigation efficiency through the integration of a Deep Reinforcement Learning (DRL) system with a crop simulation emulator. This model serves as a simulation environment to devise optimal irrigation strategies tailored specifically for cotton crops. The DRL system dynamically adjusts irrigation amount by continuously learning from environmental inputs and crop feedback to maximize yield while minimizing water use. Preliminary results suggest that our model can significantly conserve water without compromising crop health or yield, presenting a promising solution for sustainable agriculture practices in water-intensive crops. This research aims to contribute to the development of precision agriculture technologies that ensure resource efficiency and sustainability in cotton production.
Cite this work
Krishna Panthi, Vidya Samadi, and Carlos Toxtli-Hernández. 2024. Optimizing Irrigation through Deep Reinforcement Learning for Cotton Crops. AGU Fall Meeting Abstracts.
@inproceedings{Panthi2024Optimizing,
title = {Optimizing Irrigation through Deep Reinforcement Learning for Cotton Crops},
author = {Panthi, Krishna and Samadi, Vidya and Toxtli, Carlos},
booktitle = {AGU Fall Meeting Abstracts},
volume = {2024},
pages = {B01-81},
year = {2024},
note = {Conference abstract},
url = {https://ui.adsabs.harvard.edu/abs/2024AGUFM.B01...81P/abstract}
}Related