Home // ADVCOMP 2025, The Nineteenth International Conference on Advanced Engineering Computing and Applications in Sciences // View article

Artificial Intelligence-Based Local Weather Forecasting for Agricultural Digital Twins

Authors:
Miguel Zaragoza-Esquerdo
Alberto Ivars-Palomares
Irene Eiros-Fonseca
Sandra Sendra
Jaime Lloret

Keywords: Digital twin; local weather forecasting; machine learning; deep learning; agriculture; Gradient Boosting.

Abstract:
This work presents the development of a local weather forecasting system integrated into an agricultural digital twin, leveraging classical machine learning. Data were collected from ESP32-based weather stations equipped with temperature, relative humidity, and atmospheric pressure sensors. The acquired measurements were processed through a Node.js server and used to train predictive models, including Random Forest, Gradient Boosting, Ridge Regression, Lasso Regression and K-Nearest Neighbors. A sliding window approach was applied to structure the input data for short-term forecasting. Experimental results show that Gradient Boosting achieved the best performance among classical methods for atmospheric pressure but exhibited overfitting for temperature and humidity. These findings highlight the potential of Artificial Intelligence (AI)-powered digital twins to enhance precision agriculture by providing accurate, localized, and up-to-date weather forecasts.

Pages: 20 to 25

Copyright: Copyright (c) IARIA, 2025

Publication date: September 28, 2025

Published in: conference

ISSN: 2308-4499

ISBN: 978-1-68558-289-0

Location: Lisbon, Portugal

Dates: from September 28, 2025 to October 2, 2025