The identification of water stress in plants using hyperspectral imaging is challenging due to the substantial volume of data involved. This study proposes an approach for detecting plant water deficit status by integrating Principal Component Analysis (PCA) with a One-Dimensional Convolutional Neural Network (1D-CNN). The 1D-CNN model, featuring three convolutional layers and two fully connected layers, takes the PCA-reduced hyperspectral data as input and provides a binary classification output, indicating whether the plant is experiencing water stress. The proposed 1D-CNN was evaluated using hyperspectral images of maize plants cultivated in a high-throughput plant phenotyping facility in a greenhouse setting. The dataset encompassed 10 healthy plants and 10 water-stressed plants monitored over nine consecutive days, totaling 180 hyperspectral images. Our results reveal that the proposed 1D-CNN, when applied to PCA-transformed spectra, achieved an accuracy of 84.3% on the validation data, surpassing the accuracy of 78.1% obtained when applying the 1D-CNN directly to the spectra without PCA transformation. Furthermore, our trained model demonstrated the ability to detect early signs of water stress in maize plants induced by water deficit stress conditions, as early as three days after stress induction. This underscores the efficacy of our proposed method in early water stress detection, even in cases where visible symptoms are not apparent to the naked eye.