基于深度学习的重大铁路工程施工期生态环境质量评价

doi:10.3799/dqkx.2025.296

基于深度学习的重大铁路工程施工期生态环境质量评价

doi: 10.3799/dqkx.2025.296

姬翠翠^1,2,3,4,
万胜¹,
黄早阳¹,
张晓超^2,3,5,
罗璟^3,5,
杨恒聪^1, ,,
裴向军^2,3, ,

1. 重庆交通大学智慧城市学院, 重庆 400074
2. 成都理工大学生态环境学院, 四川成都 610059
3. 成都理工大学地质灾害防治与地质环境保护国家重点实验室, 四川成都 610059
4. 成都理工大学国家环境保护水土污染协同控制与联合修复重点实验室, 四川成都 610059
5. 天府永兴实验室, 四川成都 610213

基金项目:

国家重点研发计划项目(NO. 2023YFC3007103)

西藏自治区重点研发项目（NO. XZ202401ZY0091）；天府永兴重大项目 (NO. 2023KJGG05)

国家青年基金(NO. 42301459)

地质灾害防治与地质环境保护国家重点实验室开放基金资助项目(NO. SKLGP2022K028)；国家环境保护水土污染协同控制与联合修复重点实验室开放基金(NO. GHBK-2023-04).

详细信息

作者简介:
姬翠翠（1984-），副教授，博士，硕士生导师，研究方向为植被参量定量反演、森林火灾风险评估和区域生态环境遥感监测。ORCID：0000-0001-7764-6422. Email：cuicuiji@whu.edu.cn

通讯作者:
杨恒聪,E-mail:yanghc@mails.cqjtu.edu.cn

裴向军,E-mail:pxj@cdut.edu.cn

中图分类号: X82
计量
- 文章访问数: 61
- HTML全文浏览量: 0
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2025-09-01
- 网络出版日期: 2026-01-05

Assessment of Eco-environmental Quality of Major Railway Projects during Construction by Deep Network

Cuicui Ji^1,2,3,4,
Sheng Wan¹,
Zaoyang Huang¹,
Xiaochao Zhang^2,3,5,
Jing Luo^3,5,
Hengcong Yang^{1
, ,},
Xiangjun Pei^{2,3
, ,}

1. School of Smart City, Chongqing Jiaotong University, Chongqing 400074
2. School of Ecology and Environment, Chengdu University of Technology, Sichuan Chengdu 610059
3. State Key Laboratory of Geohazard Prevention and Geo-environment, Chengdu University of Technology, Chengdu 610059
4. State Environmental Protection Key Laboratory of Synergetic Control and Joint Remediation for Soil & Water Pollution, Chengdu University of Technology, Chengdu 610059
5. Tianfu Yongxing Laboratory, Chengdu 610213

摘要

摘要: 针对西南山区重大铁路工程施工期生态环境扰动监测空间精度低的问题，本研究提出“深度网络耦合遥感光谱特征”框架进行生态环境质量动态评价。利用2019-2024年Sentinel-2的10 m分辨率时序影像，构建由影像光谱波段、光谱指数和纹理特征组成的输入特征集，采用U-Net语义分割模型，对比全卷积网络（FCN）与LinkNet，实现多年雅安至林芝段重大铁路工程施工期生态环境质量动态评价。结果表明：U-Net测试集精度达91.65 %，显著优于FCN（82.95 %）和LinkNet（90.23 %），该网络模型结合光谱波段和植被指数特征更鲁棒可靠。其次，发现2019-2024年优等区面积锐减28.5 %，精准识别出2021-2022年铁路主体施工期为生态退化关键阶段，生态质量指数下降0.18。再者，结合变化分析揭示了“一般→较差”与“良→一般”为主的生态退化路径，并量化了扰动时空异质性，发现以弃渣场，交通便道为主体的持续退化热点区域。总体区域上生态质量呈“东西高、中间低”格局，优等区主要分布在雅安—康定段与林芝段，差等区集中在昌都—波密段。研究证明，深度学习网络耦合遥感光谱特征方法为铁路工程施工期生态环境质量评价新范式，可实现复杂山区交通廊道生态质量的高精度、快速监测，为重大铁路工程生态风险精准防控与恢复策略制定提供科学依据。
- 重大铁路工程 /
- 生态环境评价 /
- 深度网络 /
- 光谱特征
Abstract: To address the challenges of low spatial accuracy and poor spatio-temporal continuity in monitoring ecological disturbances during the construction of major railway projects in mountainous regions of Southwest China, this study proposes a dynamic evaluation framework that integrates deep networks with remote sensing spectral features. Using Sentinel-2 10 m resolution time-series imagery from 2019 to 2024, we developed input feature datasets comprising bands of remote sensing images, spectral indices and textural features. A U-Net-based semantic segmentation model was developed for assessing ecological quality, with comparative analyses conducted against FCN and LinkNet. This approach enabled dynamic evaluation of ecological quality during the construction of the Ya'an–Nyingchi railway section over multiple years. The results show that the U-Net model achieved a test accuracy of 91.65 %, significantly outperforming FCN (82.95 %) and LinkNet (90.23 %), demonstrating its robustness when coupled with spectral features. From 2019 to 2024, the area of high-quality zones decreased sharply by 28.5 %, with the main construction period of 2021-2022 identified as the most critical phase of ecological degradation, during which the ecological quality index dropped by 0.18. Change trajectory analysis further revealed dominant degradation pathways, such as "moderate → poor" and "good → moderate," and quantified the spatio-temporal heterogeneity of disturbances. Persistent degradation hotspots were primarily associated with waste dump sites and construction access roads. Otherwise, the region exhibited a spatial pattern of “high quality in the east and west, low in the center”—high-quality areas were mainly distributed in the Ya'an–Kangding and Nyingchi sections, while low-quality areas were concentrated in the Qamdo–Bomi section. This study demonstrates that integrating deep learning with remote sensing spectral features provides a novel paradigm for ecological quality assessment during railway construction. It enables high-precision, rapid monitoring of ecological quality in complex mountainous corridors, offering a scientific basis for precise ecological risk management and restoration strategies in major railway projects.
- Major railway projects /
- Ecological environment assessment /
- Deep learning networks /
- Spectral characteristics

HTML全文

参考文献(46)

Aanins R, 2020. Land Cover Classification Using Very High Spatial Resolution Remote Sensing Data and Deep Learning. Latvian Journal of Physics and Technical Sciences, 57(1-2): 71-77.
Akhtar A M, Qazi W A, Ahmad S R, et al., 2020. Integration of High-Resolution Optical and SAR Satellite Remote Sensing Datasets for Aboveground Biomass Estimation in Subtropical Pine Forest, Pakistan. Environmental Monitoring and Assessment, 192(9): 584. doi: 10.1007/s10661-020-08546-1
Arshad F, Iqbal M, Riaz A, et al., 2024. Road Corridors Vegetation in the Semi-Arid Region: Functional Trait Diversity and Dynamics. Scientific Reports, 14(1): 25212. doi: 10.1038/s41598-024-76484-w
Chaurasia A , Culurciello E , 2017. LinkNet: Exploiting encoder representations for efficient semantic segmentation[J].IEEE, 2017.DOI: 10.1109/VCIP.2017.8305148.
El-Husseini H, Chowyuk A N, Gustafson R R, et al., 2023. Roadside Vegetated Filter Strips to Simultaneously Lower Stormwater Pollution Loadings and Improve Economics of Biorefinery Feedstocks. Journal of Environmental Management, 347: 119168. doi: 10.1016/j.jenvman.2023.119168
Gao B.-c.1996. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sensing of Environment, 58(3), 257-266. doi: 10.1016/S0034-4257(96)00067-3
Genç ÇÖ,Altunel AO, 2025. Monitoring the operational changes in surface reflectances after logging, based on popular indices over Sentinel-2, Landsat-8, and ASTER imageries. Environ Monit Assess. 2025;197 (1):120. doi: 10.1007/s10661-024-13526-w
Gitelson, A, & Merzlyak M N. 1994. Spectral reflectance changes associated with autumn senescence of Aesculus hippocastanum L. and Acer platanoides L. leaves. Spectral features and relation to chlorophyll estimation. 143(3), 286-292.
Guo B, Yuan X, 2022. Dynamic Monitoring of the Ecological Vulnerability for Multi-Type Ecological Functional Areas during 2000–2018. Sustainability, 14: 15987. doi: 10.3390/su142315987
He X,Zhang F,Cai Y, et al., 2023. Spatio-temporal changes in fractional vegetation cover and the driving forces during 2001-2020 in the northern slopes of the Tianshan Mountains, China. Environ Sci Pollut Res Int. 2023;30 (30):75511-75531. doi: 10.1007/s11356-023-27702-x
Jin S, & Sader S A. 2005. Comparison of time series tasseled cap wetness and the normalized difference moisture index in detecting forest disturbances. Remote Sensing of Environment, 94(3), 364-372. doi: 10.1016/j.rse.2004.10.012
Kaloyianni M, Feidantsis K, Nteli I, et al., 2019. Biochemical and Molecular Responses of Cyprinids in Two Mediterranean Lacustrine Ecosystems: Opportunities for Ecological Assessment and Biomonitoring. Aquatic Toxicology, 211: 105-115. doi: 10.1016/j.aquatox.2019.03.021
Karisa JF,Obura DO,Chen CA, 2020. Spatial heterogeneity of coral reef benthic communities in Kenya. PLoS One. doi: 10.1371/journal.pone.0237397
Khosravi V,Gholizadeh A,Saberioon M, 2022. Soil toxic elements determination using integration of Sentinel-2 and Landsat-8 images: Effect of fusion techniques on model performance. Environ Pollut. 2022;310:119828. doi: 10.1016/j.envpol.2022.119828
24-12801-0
Li Z, Jiang X, Wang J, et al. 2019. Multiple facets of stream macroinvertebrate alpha diversity are driven by different ecological factors across an extensive altitudinal gradient. Ecol Evol. 2019;9(3):1306-1322. doi: 10.1002/ece3.4841
Long,Jonathan,Shelhamer,et al., 2017. Fully Convolutional Networks for Semantic Segmentation[J].IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017.DOI: 10.1109/TPAMI.2016.2572683.
Matyukira C, Mhangara P, 2024. Advances in Vegetation Mapping through Remote Sensing and Machine Learning Techniques: A Scientometric Review. European Journal of Remote Sensing, 57(1):1. doi: 10.1080/22797254.2024.
Magidi J,Nhamo L,Mpandeli S, et al., 2021. Application of the Random Forest Classifier to Map Irrigated Areas Using Google Earth Engine. Remote Sens (Basel). 2021;13 (5):876. doi: 10.3390/rs13050876
Mohammadimanesh F, Salehi B, Mahdianpari M, et al., 2019. A New Fully Convolutional Neural Network for Semantic Segmentation of Polarimetric SAR Imagery in Complex Land Cover Ecosystem. ISPRS Journal of Photogrammetry and Remote Sensing, 151: 223-236. doi: 10.1016/j.isprsjprs.2019.04.015
Namazi F,Ezoji M,Parmehr EG, 2023. Paddy Rice mapping in fragmented lands by improved phenology curve and correlation measurements on Sentinel-2 imagery in Google earth engine. Environ Monit Assess. 2023;195 (10):1220. doi: 10.1007/s10661-023-11808-3
Qi J, Chehbouni A, Huete AR, et al., 2019. A modified soil adjusted vegetation index. Remote Sensing of Environment , 48:119-26.
Ronneberger, O., Fischer, P., Brox, T., 2015. U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Navab, N., Hornegger, J., Wells, W., Frangi, A., eds., Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. Lecture Notes in Computer Science, 9351: 234-241. doi: 10.1007/978-3-319-24574-4_28
Rouse J W, Haas R H, Schell J A, et al., 1973. Monitoring vegetation systems in the great plains with ERTS.
Santana F C, Francelino M R, Siqueira R G, et al., 2025. Sentinel-2 Imagery Coupled with Machine Learning to Model Water Turbidity in the Doce River Basin, Brazil. Environmental Monitoring and Assessment, 197(4): —. doi: 10.1007/s10661-025-13918-6
Salomonson V V, & Appel I. 2004. Estimating fractional snow cover from MODIS using the normalized difference snow index. Remote Sensing of Environment, 89(3), 351-360. doi: 10.1016/j.rse.2003.10.016
Shen, Y.J., Chen, X., Peng, J.B., et al., 2024. Background Characteristics of Ecological Geological Environment System in Qinling Mountains and Assumption of Its Theoretical System. Earth Science, 49(6): 2103-2119. doi: 10.3799/dqkx.2023.210 (in Chinese with English abstract)
Song S, Wang S, Gong Y, et al., 2024. The Past and Future Dynamics of Ecological Resilience and Its Spatial Response Analysis to Natural and Anthropogenic Factors in Southwest China with Typical Karst. Scientific Reports, 14: 19166. doi: 10.1038/s41598-024-67849-x
Tsegaye L, Bharti R, 2022. The impacts of LULC and climate change scenarios on the hydrology and sediment yield of Rib watershed, Ethiopia. Environ Monit Assess. 2022;194 (10):717. doi: 10.1007/s10661-022-10391-3
Wang Bin, Fan Donglin, 2019. Research Advances in the Application of Deep Learning for Remote Sensing Image Classification and Identification: A Review. Bulletin of Surveying and Mapping, (02): 99-102+136. (in Chinese with English abstract)
Wang DZ, Qiu PH, Fang YM, 2015. Scale effect of Li-Xiang Railway construction impact on landscape pattern and its ecological risk. Ying Yong Sheng Tai Xue Bao. 2015;26 (8):2493-503.
Wang Q, Yu G A, Li Z, et al. 2025. Assessing the ecohydrological impact of hydropower-induced flow regulation on fish habitats in the Jinsha River. J Environ Manage. 2025;381:125303. doi: 10.1016/j.jenvman.2025.125303
Wen Y, Cai H, Han D, 2024. Driving Factors Analysis of Spatial-Temporal Evolution of Vegetation Ecosystem in Rocky Desertification Restoration Area of Guizhou Province, China. Environmental Science and Pollution Research International, 31(9): 13122-13140. doi: 10.1007/s11356-024-31934-w
Xiao Y, Guo Y, Yin G, et al., 2022. UAV Multispectral Image-Based Urban River Water Quality Monitoring Using Stacked Ensemble Machine Learning Algorithms—A Case Study of the Zhanghe River, China. Remote Sensing, 14: 3272. doi: 10.3390/rs14133272
Xuan, B.X., Wei, X.Y., Mao, X.F., et al., 2025. Research on Ecosystem Services in the Xining-Haidong Metropolitan Area under Different Land Use Scenarios: Based on the GMOP-PLUS-InVEST Model. Earth Science. doi: 10.3799/dqkx.2025.114 (in Chinese with English abstract)
Yadav PK, Jha P, Joy MS, et al., 2024. Ecosystem health assessment of East Kolkata Wetlands, India: Implications for environmental sustainability. J Environ Manage. 2024;366:121809. doi: 10.1016/j.jenvman.2024.121809
Yang S, Zhang L, Zhu G, 2023. Effects of transport infrastructures and climate change on ecosystem services in the integrated transport corridor region of the Qinghai-Tibet Plateau. Sci Total Environ. 2023;885:163961. doi: 10.1016/j.scitotenv.2023.163961
Zha Y, Gao J, & Ni S. 2003. Use of normalized difference built-up index in automatically mapping urban areas from TM imagery. International Journal of Remote Sensing, 24(3), 583-594. doi: 10.1080/01431160304987
Zhang Bing, 2018. Remote Sensing Big Data Era and Intelligent Information Extraction. Geomatics and Information Science of Wuhan University, 43(12): 1861-1871. (in Chinese with English abstract)
Zhang Y, Cong J, Lu H, et al. 2015. Soil bacterial diversity patterns and drivers along an elevational gradient on Shennongjia Mountain, China. Microb Biotechnol. 2015;8(4):739-746. doi: 10.1111/1751-7915.12288
Zhang Z, Zhang G, Song S, et al. 2020. Spatial Heterogeneity Influences of Environmental Control and Informal Regulation on Air Pollutant Emissions in China. Int J Environ Res Public Health, 2020;17 (13):null. doi: 10.3390/ijerph17134857
Zhou D, Song W, 2021. Identifying Ecological Corridors and Networks in Mountainous Areas. Int J Environ Res Public Health. 2021;18(9):. doi: 10.3390/ijerph18094797
王斌, 范冬林, 2019. 深度学习在遥感影像分类与识别中的研究进展综述[J]. 测绘通报, (02): 99-102+136.
张兵, 2018. 遥感大数据时代与智能信息提取[J]. 武汉大学学报(信息科学版), 43(12): 1861-1871.
申艳军, 陈兴, 彭建兵, 等, 2024. 秦岭生态地质环境系统本底特征及研究体系初步构想. 地球科学, 49(6): 2103-2119. doi: 10.3799/dqkx.2023.210
宣炳旭, 魏晓燕, 毛旭锋, 等, 2025. 西宁—海东都市圈不同土地利用情景下生态系统服务模拟：基于GMOP-PLUSInVEST模型. 地球科学. doi: 10.3799/dqkx.2025.114

施引文献

资源附件(0)

访问统计

点击查看大图

计量

文章访问数: 61
HTML全文浏览量: 0
PDF下载量: 0
被引次数: 0

姓名
邮箱
手机号码
标题
留言内容
验证码

留言板

基于深度学习的重大铁路工程施工期生态环境质量评价

doi: 10.3799/dqkx.2025.296

作者简介:
姬翠翠（1984-），副教授，博士，硕士生导师，研究方向为植被参量定量反演、森林火灾风险评估和区域生态环境遥感监测。ORCID：0000-0001-7764-6422. Email：cuicuiji@whu.edu.cn

通讯作者:
杨恒聪,E-mail:yanghc@mails.cqjtu.edu.cn

裴向军,E-mail:pxj@cdut.edu.cn

计量

Assessment of Eco-environmental Quality of Major Railway Projects during Construction by Deep Network

计量

目录

留言板

基于深度学习的重大铁路工程施工期生态环境质量评价

doi: 10.3799/dqkx.2025.296

作者简介: 姬翠翠（1984-），副教授，博士，硕士生导师，研究方向为植被参量定量反演、森林火灾风险评估和区域生态环境遥感监测。ORCID：0000-0001-7764-6422. Email：cuicuiji@whu.edu.cn

通讯作者: 杨恒聪,E-mail:yanghc@mails.cqjtu.edu.cn 裴向军,E-mail:pxj@cdut.edu.cn

计量

出版历程

Assessment of Eco-environmental Quality of Major Railway Projects during Construction by Deep Network

计量

出版历程

目录

作者简介:
姬翠翠（1984-），副教授，博士，硕士生导师，研究方向为植被参量定量反演、森林火灾风险评估和区域生态环境遥感监测。ORCID：0000-0001-7764-6422. Email：cuicuiji@whu.edu.cn

通讯作者:
杨恒聪,E-mail:yanghc@mails.cqjtu.edu.cn

裴向军,E-mail:pxj@cdut.edu.cn