ScholarWorks@경상국립대학교

초록

This study aimed to develop a regression model to accurately estimate napa cabbage fresh weight using UAV-based multispectral imagery, incorporating accumulated temperature (AT) to improve prediction accuracy under varying environmental conditions. Growth data and multispectral images were collected for two cultivars, Cheongmyeonggael and Bulam No.3, during the 2022 and 2023 growing seasons, and ten vegetation indices (VIs) were calculated. Both linear regression models (Multiple Linear Regression, Ridge, Lasso) and nonlinear models (Support Vector Regression, K-Nearest Neighbors) were applied, and their performance was evaluated using K-Fold Cross Validation. As a result, Ridge Regression showed the highest prediction accuracy in cultivar-specific models, while Multiple Linear Regression performed best in the integrated model. NDRE and TCARI were the most influential variables selected in the Ridge Regression models of Cheongmyeonggael and Bulam No.3, respectively. Furthermore, the inclusion of accumulated temperature significantly improved model performance, confirming its potential to reflect environmental growth conditions. This study presents the potential of integrating remote sensing imagery with climate data to enhance crop biomass estimation and suggests the feasibility of applying this precision agriculture-based yield prediction model under diverse environmental conditions.

키워드