Chuanhao Pu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Chuanhao Pu has a total of 62 co-authors. Here are the top co-authors by number of shared publications:
- Qiang Xu
State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu 610059, China
6 shared publications Qiang Xu
4 shared publications- Huajin Li
School of Architecture and Civil Engineering, Chengdu University, Chengdu 610106, China
3 shared publications Fanshu Xu
2 shared publications- Kuanyao Zhao
College of Architecture and Civil Engineering, Huanghuai University, Zhumadian 463000, China
2 shared publications Pinglang Kou
2 shared publications
Latest Publications
- Open Access
- 13 August 2026
- Published by MDPI AG in Remote Sensing
- Vol. 18 (16) , 2727
Encoder Choice Outweighs Modular Refinement in U-Net Architectures for Globally Distributed Coseismic Landslide Segmentation
Rapid post-earthquake landslide mapping is critical for emergency response, yet the quantitative hierarchy of architectural choices in U-Net-based segmentation remains unresolved. Using the Globally Distributed Coseismic Landslide Dataset (GDCLD; nine earthquakes, four continents), we conducted controlled ablations to isolate the contributions of encoder backbones (EfficientNet-b4, ResNet34, DenseNet121) and task-specific modules (RFB, CBAM, PANet). EfficientNet-b4 outperformed ResNet34 by 11.02 pp IoU; per-image IoU differed significantly across the three evaluated encoders (two-sided Kruskal–Wallis test, p < 0.0001), exceeding the cumulative gain of all downstream modules (RFB +0.34 pp, CBAM +0.54 pp, PANet +0.75 pp). EfficientNet regression metrics (MAE, MSE, R2) showed no statistically significant differences across module ablations (p > 0.26), suggesting that gains arose primarily from decision-boundary sharpening rather than large probability-map changes. The final model achieved 82.79% Dice, 71.15% IoU, and 86.86% Recall, with consistent performance across diverse geomorphological and imaging conditions within the present GDCLD evaluation. Independent event-held-out and external-dataset tests are still required to establish broader geographic generalization. These results demonstrate performance within the present GDCLD evaluation; independent event-held-out testing remains necessary for definitive cross-region validation. These findings establish a two-tier design hierarchy—encoder backbones provide the dominant performance capacity within the evaluated U-Net framework, while modules refine decisions via a “see–purify–delineate” mechanism—and propose a dual-metric diagnostic protocol for future architectural evaluation. This hierarchy is limited to the architectures and experimental settings evaluated in this study.
- 13 August 2026
- Published by MDPI AG in Remote Sensing
- Vol. 18 (16) , 2727
Open AccessEncoder Choice Outweighs Modular Refinement in U-Net Architectures for Globally Distributed Coseismic Landslide Segmentation
Rapid post-earthquake landslide mapping is critical for emergency response, yet the quantitative hierarchy of architectural choices in U-Net-based segmentation remains unresolved. Using the Globally Distributed Coseismic Landslide Dataset (GDCLD; nine earthquakes, four continents), we conducted controlled ablations to isolate the contributions of encoder backbones (EfficientNet-b4, ResNet34, DenseNet121) and task-specific modules (RFB, CBAM, PANet). EfficientNet-b4 outperformed ResNet34 by 11.02 pp IoU; per-image IoU differed significantly across the three evaluated encoders (two-sided Kruskal–Wallis test, p < 0.0001), exceeding the cumulative gain of all downstream modules (RFB +0.34 pp, CBAM +0.54 pp, PANet +0.75 pp). EfficientNet regression metrics (MAE, MSE, R2) showed no statistically significant differences across module ablations (p > 0.26), suggesting that gains arose primarily from decision-boundary sharpening rather than large probability-map changes. The final model achieved 82.79% Dice, 71.15% IoU, and 86.86% Recall, with consistent performance across diverse geomorphological and imaging conditions within the present GDCLD evaluation. Independent event-held-out and external-dataset tests are still required to establish broader geographic generalization. These results demonstrate performance within the present GDCLD evaluation; independent event-held-out testing remains necessary for definitive cross-region validation. These findings establish a two-tier design hierarchy—encoder backbones provide the dominant performance capacity within the evaluated U-Net framework, while modules refine decisions via a “see–purify–delineate” mechanism—and propose a dual-metric diagnostic protocol for future architectural evaluation. This hierarchy is limited to the architectures and experimental settings evaluated in this study.
- 11 February 2026
- Published by Springer Nature in Landslides
- Vol. 23 (5) , 1247-1264
The catastrophic 2025 Junlian landslide in Sichuan, China: insights revealed by seismic signals
- 11 February 2026
- Published by Springer Nature in Landslides
- Vol. 23 (5) , 1247-1264
The catastrophic 2025 Junlian landslide in Sichuan, China: insights revealed by seismic signals
- 21 October 2025
- Published by Elsevier in Engineering Geology
- Vol. 358
Multi-timescale responses of ancient rock landslides to reservoir fluctuations and precipitation: a comparative study
- 21 October 2025
- Published by Elsevier in Engineering Geology
- Vol. 358
Multi-timescale responses of ancient rock landslides to reservoir fluctuations and precipitation: a comparative study
- Open Access
- 20 July 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-17
Establishing radar-derived rainfall thresholds for a landslide early warning system: a case study in the Sichuan Basin, Southwest China
Rainfall-induced landslides often result in significant human and property losses, and reliable rainfall thresholds can effectively mitigate the hazards associated with them. However, constructing reliable rainfall thresholds in mountainous areas with sparse rain gauge stations is challenging. This study aims to establish reliable empirical rainfall thresholds for the landslide early warning systems (LEWSs) in the study area, utilizing radar-derived rainfall data processed by deep learning. Firstly, the accuracy of radar-derived rainfall data was verified based on the data with rain gauge measurements. Subsequently, utilizing frequency theory and Bayesian probability analysis methods, in conjunction with the collected landslide data and radar-derived rainfall data, various exceedance probability thresholds for rainfall-induced landslides were determined. Furthermore, the influence of cumulative effective antecedent rainfall on the initiation of landslides was investigated. The proposed threshold equations and the effect of antecedent rainfall on landslides are intended to aid in enhancing the LEWSs for this region. The findings provide valuable insights for managing rainfall-induced landslides, and can be applied to other areas with sparse rainfall data, offering a scientific basis for improved landslide prediction and risk management.
- 20 July 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-17
Open AccessEstablishing radar-derived rainfall thresholds for a landslide early warning system: a case study in the Sichuan Basin, Southwest China
Rainfall-induced landslides often result in significant human and property losses, and reliable rainfall thresholds can effectively mitigate the hazards associated with them. However, constructing reliable rainfall thresholds in mountainous areas with sparse rain gauge stations is challenging. This study aims to establish reliable empirical rainfall thresholds for the landslide early warning systems (LEWSs) in the study area, utilizing radar-derived rainfall data processed by deep learning. Firstly, the accuracy of radar-derived rainfall data was verified based on the data with rain gauge measurements. Subsequently, utilizing frequency theory and Bayesian probability analysis methods, in conjunction with the collected landslide data and radar-derived rainfall data, various exceedance probability thresholds for rainfall-induced landslides were determined. Furthermore, the influence of cumulative effective antecedent rainfall on the initiation of landslides was investigated. The proposed threshold equations and the effect of antecedent rainfall on landslides are intended to aid in enhancing the LEWSs for this region. The findings provide valuable insights for managing rainfall-induced landslides, and can be applied to other areas with sparse rainfall data, offering a scientific basis for improved landslide prediction and risk management.
- Open Access
- 17 June 2025
- Published by MDPI AG in Remote Sensing
- Vol. 17 (12) , 2083
Assessing the Influence of Environmental Factors on Landslide Frequency and Intensity in Northwestern Sichuan, SW China, Using Multi-Temporal Satellite Imagery
Landslides are a significant geological hazard with substantial socio-economic and environmental consequences, particularly in northwestern Sichuan, SW China, where complex geological and climatic conditions contribute to their occurrence. This study examines 1629 recorded landslide events, including 240 active landslides that have undergone substantial changes over the past two decades. By analyzing multi-temporal satellite imagery, this research investigates the relationship between landslide occurrence and key environmental factors such as annual rainfall and the Normalized Difference Vegetation Index (NDVI). The results reveal that landslides are most frequent on southwest-, south-, east-, and southeast-facing slopes, where the Föhn effect interacts with rainfall and vegetation patterns, thereby increasing landslide susceptibility. Rainfall intensity is identified as a critical factor, with landslide areas expanding significantly when annual rainfall exceeds 650 mm, while minimal changes are observed when rainfall is below 550 mm. The relationship between the NDVI and landslide occurrence is non-linear; higher vegetation cover does not necessarily correlate with reduced landslide frequency. Notably, landslide expansion is more pronounced when NDVI values are below 0.82, with a suppression effect occurring beyond this threshold. A threshold model based on the interaction between the NDVI and rainfall provides valuable insights into landslide dynamics, offering a framework for improved risk management. Slope characteristics are crucial in landslide evolution, with steeper slopes leading to greater vertical drops and more frequent events, making slope zone identification key for predicting future expansion.
- 17 June 2025
- Published by MDPI AG in Remote Sensing
- Vol. 17 (12) , 2083
Open AccessAssessing the Influence of Environmental Factors on Landslide Frequency and Intensity in Northwestern Sichuan, SW China, Using Multi-Temporal Satellite Imagery
Landslides are a significant geological hazard with substantial socio-economic and environmental consequences, particularly in northwestern Sichuan, SW China, where complex geological and climatic conditions contribute to their occurrence. This study examines 1629 recorded landslide events, including 240 active landslides that have undergone substantial changes over the past two decades. By analyzing multi-temporal satellite imagery, this research investigates the relationship between landslide occurrence and key environmental factors such as annual rainfall and the Normalized Difference Vegetation Index (NDVI). The results reveal that landslides are most frequent on southwest-, south-, east-, and southeast-facing slopes, where the Föhn effect interacts with rainfall and vegetation patterns, thereby increasing landslide susceptibility. Rainfall intensity is identified as a critical factor, with landslide areas expanding significantly when annual rainfall exceeds 650 mm, while minimal changes are observed when rainfall is below 550 mm. The relationship between the NDVI and landslide occurrence is non-linear; higher vegetation cover does not necessarily correlate with reduced landslide frequency. Notably, landslide expansion is more pronounced when NDVI values are below 0.82, with a suppression effect occurring beyond this threshold. A threshold model based on the interaction between the NDVI and rainfall provides valuable insights into landslide dynamics, offering a framework for improved risk management. Slope characteristics are crucial in landslide evolution, with steeper slopes leading to greater vertical drops and more frequent events, making slope zone identification key for predicting future expansion.
- Open Access
- 10 April 2025
- Published by Elsevier in Journal of Rock Mechanics and Geotechnical Engineering
- Vol. 17 (12) , 7864-7881
Can different machine learning methods have consistent interpretations of DEM-based factors in shallow landslide susceptibility assessments?
- 10 April 2025
- Published by Elsevier in Journal of Rock Mechanics and Geotechnical Engineering
- Vol. 17 (12) , 7864-7881
Open AccessCan different machine learning methods have consistent interpretations of DEM-based factors in shallow landslide susceptibility assessments?
- 1 February 2025
- Published by Elsevier in Engineering Geology
- Vol. 347
Revealing terrestrial uplift in large-scale land creation areas on the Loess Plateau using InSAR time series data
- 1 February 2025
- Published by Elsevier in Engineering Geology
- Vol. 347
Revealing terrestrial uplift in large-scale land creation areas on the Loess Plateau using InSAR time series data
- 25 June 2024
- Published by Elsevier in Science of The Total Environment
- Vol. 946, 174256
Vegetation response to large-scale mountain excavation and city construction projects on the Loess Plateau of China
- 25 June 2024
- Published by Elsevier in Science of The Total Environment
- Vol. 946, 174256
Vegetation response to large-scale mountain excavation and city construction projects on the Loess Plateau of China