Prediction Models for Post-Fire Debris Flow Susceptibility and Debris Flow Volume in Yajiang County, Sichuan, China
-
摘要: 2024年3月15日, 雅江县发生大型森林火灾, 并在首个雨季引发数百起泥石流, 为西南山区火后泥石流研究提供了充足样本.基于野外调查、无人机影像、卫星遥感和降雨数据, 以2024年“3·15”雅江县城森林火灾火烧迹地火后泥石流暴发数据库构建了易发性评估和一次冲出规模预测模型为依据, 开展了2024年“12·09”雅江县城厢村和2025年“02·02”木泽西村两处火烧迹地火后泥石流成灾预测.结果表明, (1)最优随机森林易发性模型AUC为0.905, 精度为0.950, 城厢村和木泽西村极高和高易发流域分别为10个和22个, 占各自火烧迹地流域总数的40.57%和73.68%.(2)最优体积预测模型的因子包含小时雨强、坡度大于30°面积占比、土壤粘粒含量、沟壑密度、植被归一化指数和中-重度火烧面积, 体积预测模型R2为0.65.在3种降雨场景下, 城厢村火烧迹地流域体积规模200 m3以上的占比分别为2.86%、25.72%和34.29%, 木泽西村火烧迹地流域体积规模1 000 m3以上流域占比分别为0, 15.79%和63.16%.城厢村火烧迹地泥石流体积普遍较小, 而保护对象密集, 其中CX05、CX08、CX13和CX25流域危害较大; 木泽西村火烧迹地泥石流体积相对较大, 其中MZX02和MZX04流域危害较大.Abstract: On March 15, 2024, an extensive forest fire occurred in Yajiang County, Sichuan Province. In the first post-fire rainy season, hundreds of post-fire debris flows (PFDFs) were triggered, providing a valuable dataset for studying PFDFs in southwestern mountainous regions. This study developed PFDF susceptibility and volume prediction models based on field investigations, UAV imagery, satellite remote sensing, and rainfall data. The models were constructed using the PFDF event database from the burned area of Yajiang County on March 15, 2024, and were subsequently applied to hazard prediction for two burned areas: Chengxiang Village (December 9, 2024) and Muzexi Village (February 2, 2025). The results show follows. (1) The optimal Random Forest susceptibility model achieved an AUC of 0.905 and an accuracy of 0.950. For the Chengxiang and Muzexi burned areas, 10 and 22 catchments, respectively, were classified as extremely high or high susceptibility, accounting for 48.57% and 73.68% of their total watersheds. (2) The optimal factor combination for the volume prediction model included hourly rainfall intensity, percentage of catchment area with slopes exceeding 30°, soil clay content, gully density, normalized difference vegetation index (NDVI), and the moderate and severe burned area. The generalized additive model for volume prediction achieved an R2 of 0.65. Under Q25%, Q75%, and P20% rainfall scenarios, the proportion of catchments in Chengxiang with debris flow volumes exceeding 200 m3 was 2.86%, 25.72%, and 34.29%, respectively, while in Muzexi, the proportion of catchments with volumes exceeding 1 000 m3 was 0%, 15.79%, and 63.16%, respectively. Debris flow volumes in Chengxiang are generally smaller, but the area contains a high density of vulnerable elements, with catchments CX05, CX08, CX13, and CX25 posing significant hazards. In contrast, debris flows in Muzexi tend to have larger volumes, with catchments MZX02 and MZX04 identified as high-risk areas.
-
Key words:
- Yajiang /
- post-fire debris flow /
- susceptibility /
- debris flow volume /
- prediction model /
- engineering geology
-
图 1 研究区区域位置与火烈度划分
Fig. 1. Regional location and classification of burn severity of the study area
图 3 “3·15”雅江县城火烧迹地首个雨季火后泥石流的暴发数据统计
Fig. 3. Statistics of post-fire debris flow events during the first rainy season following the "3·15" Yajiang fire
图 4 典型泥石流沟暴发特征
图a~c分别为土窝沟泥石流暴发前、暴发后和雨季结束后的沟口堆积特征.图d~f分别为布色亚龙沟泥石流暴发前、暴发后和雨季结束后的沟口堆积特征
Fig. 4. Characteristics of typical debris flow outbursts
图 5 基于RF模型的易发性结果验证及因子重要性排序
Fig. 5. Verification of susceptibility results based on random forest model and factor importance
图 6 城厢村(a)和木泽西村(b)火烧迹地火后泥石流易发性制图
Fig. 6. The susceptibility mapping of post-fire debris flow in Chengxiang Village (a) and Muzexi Village (b)
图 7 6个变量对冲出规模影响的样条函数曲线
阴影部分为95%置信区间, 纵坐标表示平滑函数值, 数字是定义的自由度
Fig. 7. The spline curves of six variables affecting the debris flow volume
图 8 2024年“3·15”雅江县城火烧迹地冲出规模预测模型预测值和实测泥石流冲出规模值对比
红色虚线表示预测和实测冲出规模零残差.黑色实线代表预测和实测冲出规模残差在一个数量级内的包络线
Fig. 8. Comparison between the predicted using the volume prediction model and measured debris flow volumes following the "3·15" Yajiang fire in 2024
图 9 3种降雨场景下的雅江“12·9”城厢村(a, c, e)和“02·02”木泽西村(b, d, f)火烧区泥石流冲出规模预测
Fig. 9. The volume prediction of debris flow under three rainfall scenarios in the "12·9" Chengxiang Village (a, c, e) and "02·02" Muzexi Village (b, d, f) in Yajiang
表 1 3个火烧迹地过火情况统计
Table 1. Statistics of burned severity in the three burned area of Yajiang County
火烧迹地 轻度火烧面积(km2) 中度火烧面积(km2) 重度火烧面积(km2) 过火面积(km2) “3·15”雅江县城 84.00 73.90 142.10 278.80 “12·09”城厢村 1.00 0.90 1.77 3.46 “02·02”木泽西村 7.72 10.62 4.51 22.85 
下载: 导出CSV
表 3 规模预测模型的11个解释变量数据分析
Table 3. Data analysis of the 11 explanatory variables in the volume prediction model
解释变量 平均值mean 标准差std 显著性P 斯皮尔曼相关系数$ \rho $ i60 6.01 2.96 0.000* 0.38 S30 0.68 0.15 0.001* -0.26 WS 0.39 0.13 0.297 0.21 SPI 3.79 0.33 0.432 0.01 TWI 4.79 0.40 0.079 0.12 GD 5.34 3.93 0.186 -0.55 CC 173.26 16.75 0.102 0.56 STI 46.19 55.68 0.455 0.30 NDVI 0.23 0.07 0.167 0.24 MSBA 0.93 1.64 0.000* 0.70 MSBI 0.74 0.21 0.666 0.05 注: *表示该变量与泥石流规模具有显著性.
下载: 导出CSV
-
Austin, M. P., 2002. Spatial Prediction of Species Distribution: An Interface between Ecological Theory and Statistical Modelling. Ecological Modelling, 157(2/3): 101-118. https://doi.org/10.1016/S0304-3800(02)00205-3 Bordoni, M., Galanti, Y., Bartelletti, C., et al., 2020. The Influence of the Inventory on the Determination of the Rainfall-Induced Shallow Landslides Susceptibility Using Generalized Additive Models. CATENA, 193: 104630. https://doi.org/10.1016/j.catena.2020.104630 Cannon, S. H., Gartner, J. E., Rupert, M. G., et al., 2010. Predicting the Probability and Volume of Postwildfire Debris Flows in the Intermountain Western United States. Geological Society of America Bulletin, 122(1-2): 127-144. https://doi.org/10.1130/B26459.1 Cannon, S. H., Gartner, J. E., Wilson, R. C., et al., 2008. Storm Rainfall Conditions for Floods and Debris Flows from Recently Burned Areas in Southwestern Colorado and Southern California. Geomorphology, 96(3-4): 250-269. https://doi.org/10.1016/j.geomorph.2007.03.019 Cannon, S. H., Powers, P. S., Savage, W. Z., 1998. Fire-Related Hyperconcentrated and Debris Flows on Storm King Mountain, Glenwood Springs, Colorado, USA. Environmental Geology, 35(2): 210-218. https://doi.org/10.1007/s002540050307 Chen, S. Q., Sun, Y. H., Ding, G. Q., et al., 2025. Holocene Dynamics of Vegetation Cover and Their Driving Mechanisms in Asian Drylands. Journal of Earth Science, 36(2): 839-842. https://doi.org/10.1007/s12583-025-0173-x de Jesus Pugliese Viloria, A., Folini, A., Carrion, D., et al., 2024. Hazard Susceptibility Mapping with Machine and Deep Learning: A Literature Review. Remote Sensing, 16(18): 3374. https://doi.org/10.3390/rs16183374 DeGraff, J. V., Cannon, S. H., Gartner, J. E., 2015. The Timing of Susceptibility to Post-Fire Debris Flows in the Western United States. Environmental & Engineering Geoscience, 21(4): 277-292. https://doi.org/10.2113/gseegeosci.21.4.277 Dormann, C. F., Elith, J., Bacher, S., et al., 2013. Collinearity: A Review of Methods to Deal with It and a Simulation Study Evaluating Their Performance. Ecography, 36(1): 27-46. https://doi.org/10.1111/j.1600-0587.2012.07348.x Esposito, G., Gariano, S. L., Masi, R., et al., 2023. Rainfall Conditions Leading to Runoff-Initiated Post-Fire Debris Flows in Campania, Southern Italy. Geomorphology, 423: 108557. https://doi.org/10.1016/j.geomorph.2022.108557 Fang, Z. C., Wang, Y., van Westen, C., et al., 2024. Landslide Hazard Spatiotemporal Prediction Based on Data-Driven Models: Estimating Where, When and How Large Landslide may Be. International Journal of Applied Earth Observation and Geoinformation, 126: 103631. https://doi.org/10.1016/j.jag.2023.103631 Gartner, J. E., Cannon, S. H., Santi, P. M., 2014. Empirical Models for Predicting Volumes of Sediment Deposited by Debris Flows and Sediment-Laden Floods in the Transverse Ranges of Southern California. Engineering Geology, 176: 45-56. https://doi.org/10.1016/j.enggeo.2014.04.008 Gartner, J. E., Cannon, S. H., Santi, P. M., et al., 2008. Empirical Models to Predict the Volumes of Debris Flows Generated by Recently Burned Basins in the Western U. S.. Geomorphology, 96(3-4): 339-354. https://doi.org/10.1016/j.geomorph.2007.02.033 Gorr, A., McGuire, L., Youberg, A., 2024. Empirical Models for Postfire Debris-Flow Volume in the Southwest United States. Journal of Geophysical Research: Earth Surface, 129(11): e2024JF007825. https://doi.org/10.1029/2024jf007825 Green, S. B., 1991. How Many Subjects does It Take to do a Regression Analysis. Multivariate Behavioral Research, 26(3): 499-510. https://doi.org/10.1207/s15327906mbr2603_7 He, K., Hu, X. W., Wu, Z. L., et al., 2024. Preliminary Analysis of the Wildfire on March 15, 2024, and the Following Post-Fire Debris Flows in Yajiang County, Sichuan, China. Landslides, 21(12): 3179-3189. https://doi.org/10.1007/s10346-024-02364-z Hou, Z. F., Li, J. J., Song, C. H., et al., 2014. Understanding Miocene Climate Evolution in Northeastern Tibet: Stable Carbon and Oxygen Isotope Records from the Western Tianshui Basin, China. Journal of Earth Science, 25(2): 357-365. https://doi.org/10.1007/s12583-014-0416-8 Hu, X. W., Jin, T., Yin, W. Q., et al., 2020. The Characteristics of Forest Fire Burned Area and Susceptibility Assessment of Post-Fire Debris Flow in Jingjiu Township, Xichang City. Journal of Engineering Geology, 28(4): 762-771(in Chinese with English abstract). Hu, X. W., Wang, Y., Yang, Y., 2018. Research Actuality and Evolution Mechanism of Post-Fire Debris Flow. Journal of Engineering Geology, 26(6): 1562-1573(in Chinese with English abstract). Hu, X. W., Zhou, Y. H., He, K., et al., 2024. Post-Fire Debris Flow Mechanisms in Different Lithological Zones. Mountain Research, 42(4): 535-545(in Chinese with English abstract). Jin, T., Hu, X. W., Liu, B., et al., 2022. Susceptibility Prediction of Post-Fire Debris Flows in Xichang, China, Using a Logistic Regression Model from a Spatiotemporal Perspective. Remote Sensing, 14(6): 1306. https://doi.org/10.3390/rs14061306 Kean, J. W., Staley, D. M., Cannon, S. H., 2011. In Situ Measurements of Post-Fire Debris Flows in Southern California: Comparisons of the Timing and Magnitude of 24 Debris-Flow Events with Rainfall and Soil Moisture Conditions. Journal of Geophysical Research, 116(F4): F04019. https://doi.org/10.1029/2011jf002005 Kern, A. N., Addison, P., Oommen, T., et al., 2017. Machine Learning Based Predictive Modeling of Debris Flow Probability Following Wildfire in the Intermountain Western United States. Mathematical Geosciences, 49(6): 717-735. https://doi.org/10.1007/s11004-017-9681-2 Lombardo, L., Tanyas, H., Huser, R., et al., 2021. Landslide Size Matters: A New Data-Driven, Spatial Prototype. Engineering Geology, 293: 106288. https://doi.org/10.1016/j.enggeo.2021.106288 Long, Y. J., Li, W. L., Huang, R. Q., et al., 2023. A Comparative Study of Supervised Classification Methods for Investigating Landslide Evolution in the Mianyuan River Basin, China. Journal of Earth Science, 34(2): 316-329. https://doi.org/10.1007/s12583-021-1525-9 Merghadi, A., Yunus, A. P., Dou, J., et al., 2020. Machine Learning Methods for Landslide Susceptibility Studies: A Comparative Overview of Algorithm Performance. Earth-Science Reviews, 207: 103225. https://doi.org/10.1016/j.earscirev.2020.103225 Nikolopoulos, E. I., Destro, E., Bhuiyan, M. A. E., et al., 2018. Evaluation of Predictive Models for Post-Fire Debris Flow Occurrence in the Western United States. Natural Hazards and Earth System Sciences, 18(9): 2331-2343. https://doi.org/10.5194/nhess-18-2331-2018 Pourghasemi, H. R., Kariminejad, N., Amiri, M., et al., 2020. Assessing and Mapping Multi-Hazard Risk Susceptibility Using a Machine Learning Technique. Scientific Reports, 10: 3203. https://doi.org/10.1038/s41598-020-60191-3 Riley, K. L., Bendick, R., Hyde, K. D., et al., 2013. Frequency-Magnitude Distribution of Debris Flows Compiled from Global Data, and Comparison with Post-Fire Debris Flows in the Western U. S.. Geomorphology, 191: 118-128. https://doi.org/10.1016/j.geomorph.2013.03.008 Galiano, V. R., Castillo, M. S., Olmo, M. C., et al., 2015. Machine Learning Predictive Models for Mineral Prospectivity: An Evaluation of Neural Networks, Random Forest, Regression Trees and Support Vector Machines. Ore Geology Reviews, 71: 804-818. https://doi.org/10.1016/j.oregeorev.2015.01.001 Shen, T., Deng, C. Y., Ding, X. H., et al., 2023. Analysis of Influencing Factors of Shanghai University Students' Self-Study Efficiency Based on Multiple Linear Stepwise Regression. Statistics and Application, 12(1): 100-109(in Chinese with English abstract). doi: 10.12677/SA.2023.121012 Staley, D. M., Negri, J. A., Kean, J. W., et al., 2017. Prediction of Spatially Explicit Rainfall Intensity-Duration Thresholds for Post-Fire Debris-Flow Generation in the Western United States. Geomorphology, 278: 149-162. https://doi.org/10.1016/j.geomorph.2016.10.019 Tan, L., Zhang, L. L., Wei, X., et al., 2025. Study on Regional Landslide Susceptibility Assessment Method Based on U-Net Semantic Segmentation Network and Its Cross-Generalization Ability. China Civil Engineering Journal, 58(6): 103-116 (in Chinese with English abstract). Thomas, M. A., Kean, J. W., McCoy, S. W., et al., 2023. Postfire Hydrologic Response along the Central California (USA) Coast: Insights for the Emergency Assessment of Postfire Debris-Flow Hazards. Landslides, 20(11): 2421-2436. https://doi.org/10.1007/s10346-023-02106-7 Wall, S. A., Roering, J. J., Rengers, F. K., 2020. Runoff-Initiated Post-Fire Debris Flow Western Cascades, Oregon. Landslides, 17(7): 1649-1661. https://doi.org/10.1007/s10346-020-01376-9 Wall, S., Murphy, B. P., Belmont, P., et al., 2023. Predicting Post-Fire Debris Flow Grain Sizes and Depositional Volumes in the Intermountain West, United States. Earth Surface Processes and Landforms, 48(1): 179-197. https://doi.org/10.1002/esp.5480 Wang, L., Zhang, Y. K., Shu, B., et al., 2023. Improved Method for Fusion of Loess Landslide Monitoring Data Based on Feature Selection and Stepwise Regression. Journal of Earth Sciences and Environment, 45(3): 511-521 (in Chinese with English abstract). Wang, Y., Hu, X. W., Wu, L. J., et al., 2022. Evolutionary History of Post-Fire Debris Flows in Ren'e Yong Valley in Sichuan Province of China. Landslides, 19(6): 1479-1490. https://doi.org/10.1007/s10346-022-01867-x Wang, Y. H., Shen, H. W., Xie, W. Y., et al., 2024. Analysis of the Rainfall Threshold for Post-Fire Debris Flow Initiation: A Case Study of the Debris Flow at Ren'eyong Gully in Xiangcheng County, Sichuan Province. The Chinese Journal of Geological Hazard and Control, 35(1): 108-115 (in Chinese with English abstract). Weiss, K., Khoshgoftaar, T. M., Wang, D. D., 2016. A Survey of Transfer Learning. Journal of Big Data, 3(1): 9. https://doi.org/10.1186/s40537-016-0043-6 Yang, Y., Hu, X. W., Han, M., et al., 2022. Post-Fire Temporal Trends in Soil Properties and Revegetation: Insights from Different Wildfire Severities in the Hengduan Mountains, Southwestern China. CATENA, 213: 106160. https://doi.org/10.1016/j.catena.2022.106160 Zhou, R. C., Hu, X. W., Jin, T., et al., 2024. Characteristics of Burned Area and Susceptibility Assessment of Post-Fire Debris Flow of Chongqing Wildfire in August, 2022. Hydrogeology & Engineering Geology, 51(5): 150-160 (in Chinese with English abstract). 胡卸文, 金涛, 殷万清, 等, 2020. 西昌市经久乡森林火灾火烧区特点及火后泥石流易发性评价. 工程地质学报, 28(4): 762-771. 胡卸文, 王严, 杨瀛, 2018. 火后泥石流成灾特点及研究现状. 工程地质学报, 26(6): 1562-1573. 胡卸文, 周永豪, 何坤, 等, 2024. 不同岩性区火后泥石流成灾机理. 山地学报, 42(4): 535-545. 沈婷, 邓辰钰, 丁小荷, 等, 2023. 基于多元线性逐步回归的上海大学学生自习效率影响因素分析. 统计学与应用, 12(1): 100-109. 谭林, 张璐璐, 魏鑫, 等, 2025. 基于U-Net语义分割网络的区域滑坡易发性评价方法和跨地区泛化能力研究. 土木工程学报, 58(6): 103-116. 王利, 张懿恺, 舒宝, 等, 2023. 基于特征优选和逐步回归的黄土滑坡监测数据融合改进方法. 地球科学与环境学报, 45(3): 511-521. 王元欢, 沈昊文, 谢万银, 等, 2024. 火后泥石流启动降雨阈值分析: 以四川乡城县仁额拥沟泥石流为例. 中国地质灾害与防治学报, 35(1): 108-115. 周瑞宸, 胡卸文, 金涛, 等, 2024. 重庆市2022年8月森林火灾火烧区特点及火后泥石流易发性评价. 水文地质工程地质, 51(5): 150-160. -
点击查看大图
图(9) / 表(2)
计量
- 文章访问数: 101
- HTML全文浏览量: 11
- PDF下载量: 4
- 被引次数: 0
