<p>The rapid growth of tourism activities has intensified ecological pressures in many scenic areas, making spatial monitoring and timely assessment of tourism ecological environments increasingly important for sustainable management. With the advancement of remote sensing, large-scale spatial data provide new opportunities for dynamic environmental monitoring. However, traditional assessment approaches often rely on limited ecological indicators or static observations, which restrict their ability to capture complex spatial–temporal interactions between tourism activities and ecological systems. This study proposes a remote sensing framework for assessing tourism ecological environment quality by combining spatial feature extraction and temporal modeling techniques. First, multi-source remote sensing imagery is used to detect key ecological and anthropogenic features in tourist areas through an improved ResNeXt-YOLOv5s target detection model, enabling accurate identification of vegetation, water bodies, infrastructure, and waste accumulation. Second, deep spatial features are extracted from remote sensing images using a ResNeXt-based feature extraction module, allowing the characterization of ecological quality indicators beyond traditional spectral indices. Third, a long short-term memory model is employed to capture temporal dynamics and delayed effects of tourism activities on ecological conditions. Based on these components, a spatial–temporal evaluation model for tourism ecological environment quality is constructed. Experimental results based on multi-source satellite data demonstrate that the proposed framework achieves an overall assessment accuracy of 92.8%, with mean squared error and mean absolute error values of 0.17 and 0.18, respectively. The model maintains high stability under noise interference and performs consistently across multiple tourism scenarios, including peak tourist seasons and post-extreme-weather conditions. The results indicate that integrating remote sensing data, spatial analysis, and deep learning models can significantly improve the reliability and timeliness of tourism ecological environment assessment. This research provides a practical spatial monitoring approach for sustainable tourism management and ecological protection. The proposed framework contributes to the integration of remote sensing techniques in tourism-environment studies, offering decision support for ecological governance and sustainable development of tourist destinations.</p>

Remote sensing assessment of tourism ecological environment quality using a ResNeXt-YOLOv5s-LSTM framework

期刊论文
发表日期：

2026年7月23日

查看全文

Ling Yang

摘要

The rapid growth of tourism activities has intensified ecological pressures in many scenic areas, making spatial monitoring and timely assessment of tourism ecological environments increasingly important for sustainable management. With the advancement of remote sensing, large-scale spatial data provide new opportunities for dynamic environmental monitoring. However, traditional assessment approaches often rely on limited ecological indicators or static observations, which restrict their ability to capture complex spatial–temporal interactions between tourism activities and ecological systems. This study proposes a remote sensing framework for assessing tourism ecological environment quality by combining spatial feature extraction and temporal modeling techniques. First, multi-source remote sensing imagery is used to detect key ecological and anthropogenic features in tourist areas through an improved ResNeXt-YOLOv5s target detection model, enabling accurate identification of vegetation, water bodies, infrastructure, and waste accumulation. Second, deep spatial features are extracted from remote sensing images using a ResNeXt-based feature extraction module, allowing the characterization of ecological quality indicators beyond traditional spectral indices. Third, a long short-term memory model is employed to capture temporal dynamics and delayed effects of tourism activities on ecological conditions. Based on these components, a spatial–temporal evaluation model for tourism ecological environment quality is constructed. Experimental results based on multi-source satellite data demonstrate that the proposed framework achieves an overall assessment accuracy of 92.8%, with mean squared error and mean absolute error values of 0.17 and 0.18, respectively. The model maintains high stability under noise interference and performs consistently across multiple tourism scenarios, including peak tourist seasons and post-extreme-weather conditions. The results indicate that integrating remote sensing data, spatial analysis, and deep learning models can significantly improve the reliability and timeliness of tourism ecological environment assessment. This research provides a practical spatial monitoring approach for sustainable tourism management and ecological protection. The proposed framework contributes to the integration of remote sensing techniques in tourism-environment studies, offering decision support for ecological governance and sustainable development of tourist destinations.