一种沟谷型潜在泥石流危险性评价方法：基于多源数据融合的卷积神经网络

doi:10.12082/dqxxkx.2023.220514

地球信息科学学报 ›› 2023, Vol. 25 ›› Issue (3): 588-605.doi: 10.12082/dqxxkx.2023.220514

一种沟谷型潜在泥石流危险性评价方法：基于多源数据融合的卷积神经网络

徐繁树¹(), 王保云²^,³^,^*(), 韩俊¹

1.云南师范大学信息学院，昆明 650500
2.云南师范大学数学学院，昆明 650500
3.云南省高校复杂系统建模及应用重点实验室，昆明 650500

收稿日期:2022-07-14 修回日期:2022-08-22 出版日期:2023-03-25 发布日期:2023-04-19
通讯作者: * 王保云（1977— ），男，云南人，副教授，主要从事机器学习及图像处理研究。E-mail: wspbmly@163.com
作者简介:徐繁树（1997— ），男，上海人，硕士生，研究方向为泥石流灾害识别与机器学习。E-mail: xufanshu@user.ynnu.edu.cn
基金资助:
国家自然科学基金项目(61966040)

A Potential Gully Debris Flows Hazard Assessment Method: A CNN Model based on Multi-source Data Fusion

XU Fanshu¹(), WANG Baoyun²^,³^,^*(), HAN Jun¹

1. School of Information, Yunnan Normal University, Kunming 650500, China
2. School of Mathematics, Yunnan Normal University, Kunming 650500, China
3. Key Laboratory of Modeling and Application of Complex Systems in Universities of Yunnan Province, Kunming 650500, China

Received:2022-07-14 Revised:2022-08-22 Online:2023-03-25 Published:2023-04-19
Contact: WANG Baoyun
Supported by:
National Natural Science Foundation of China(61966040)

摘要/Abstract

摘要：

山区多发沟谷型泥石流，而由于山区地形崎岖，导致无法开展大面积的泥石流危险性评价工作。本文使用遥感数据、DEM (Digital Elevation Model)数据以及岩性、土壤、植被数据，构建了一个基于多源数据，能快速进行大面积排查工作的卷积神经网络模型RSDNet (Residual-Shuffle-Dense residual Net)。该模型首先使用最大池化改进的残差结构对各类不同数据进行浅层特征提取，然后使用通道重排以加强各类数据底层特征间的关联性，接着使用密集残差结构对底层特征作进一步的特征提取，学习各类特征间的相互作用对潜在泥石流危险性的影响，最后根据待评价沟谷与已发生过泥石流沟谷的相似度给出沟谷的潜在危险性。在训练过程中，使用了交叉熵和基于焦点损失改进的联合损失函数，使模型能更好地区分各类沟谷的形态特征和致灾特征。RSDNet在沟谷分类任务上可达到89.7%的精确率。在对怒江州全境沟谷进行潜在危险性评价的任务中，132条历史泥石流沟谷有122条被模型判断为高危险或极危险。结果表明模型性能良好，为沟谷泥石流的危险性评价提供了新思路。

关键词: 泥石流, 灾害预测, 山区沟谷, 多源数据融合, 怒江州, 卷积神经网络, 残差模块, 焦点损失

Abstract:

Gully debris flows frequently occur in mountainous regions. Hazard investigation in large area is always hampered by the rugged mountains. In this paper, a convolutional neural network named Residual-Shuffle-Dense residual Net (RSDNet) based on remote sensing, DEM, soil, lithology, and vegetation data was proposed to conduct large-scale territorial surveys. First, shallow features were extracted by modified residual structure using maximize pooling. Then, feature fusion was conducted to strengthen the correlation between the underlying features of various data. Next, dense residual structure was applied to make further feature extraction from underlying features and identify the impact of interaction between various features on potential debris flow hazard. Finally, the potential hazard level of a valley was given. During the training process, a joint loss function based on cross entropy loss and modified focal loss was used to make the model better distinguish the morphological and disaster-causing characteristics of various valleys. In this study, the valley classification achieved a precision rate of 0.92 by using RSDNet. In the potential hazard assessment of all valleys in Nujiang Prefecture, 122 out of 132 historical debris flow valleys were judged to be dangerous or very dangerous. Results indicate that the proposed model performs well, and this work would offer new ideas for potential debris flows hazard assessment.

Key words: debris flow, hazard prediction, mountainous valleys, multisource data fusion, Nujiang Prefecture, convolutional neural network, residual block, focal loss

徐繁树, 王保云, 韩俊. 一种沟谷型潜在泥石流危险性评价方法：基于多源数据融合的卷积神经网络[J]. 地球信息科学学报, 2023, 25(3): 588-605.DOI:10.12082/dqxxkx.2023.220514

XU Fanshu, WANG Baoyun, HAN Jun. A Potential Gully Debris Flows Hazard Assessment Method: A CNN Model based on Multi-source Data Fusion[J]. Journal of Geo-information Science, 2023, 25(3): 588-605.DOI:10.12082/dqxxkx.2023.220514

图/表 17

图1

图2

图3

图4

图5

图6

图7

表1

图8

表2

图9

图10

图11

图12

图13

表3

RSDNet对东月各沟谷的潜在危险性评价结果

类别	0	1	2	3	4	5	S
相似度( $x i$ )	0.001	0.489	0.503	0.0	0.0	0.07	0.993

表3

图14

参考文献 38

[1]	刘传正. 中国崩塌滑坡泥石流灾害成因类型[J]. 地质论评, 2014, 60(4):858-868.
	[Liu C Z. Genetic types of landslide and debris flow disasters in China[J]. Geological Review, 2014, 60(4):858-868.] DOI:10.16509/j.georeview.2014.04.017 doi: 10.16509/j.georeview.2014.04.017
[2]	程维明, 夏遥, 曹玉尧, 等. 区域泥石流孕灾环境危险性评价——以北京军都山区为例[J]. 地理研究, 2013, 32(4):595-606.
	[Cheng W M, Xia Y, Cao Y Y, et al. Regional hazard assessment of disaster environment for debris flows: Taking Jundu Mountain, Beijing as an example[J]. Geographical Research, 2013, 32(4):595-606.]
[3]	魏斌斌, 赵其华, 韩刚, 等. 基于灰色关联法的地震灾区泥石流危险性评价——以北川县泥石流为例[J]. 工程地质学报, 2013, 21(4):525-533.
	[Wei B B, Zhao Q H, Han G, et al. Grey correlation method based hazard assessment of debris flow in quake-hit area—Taking debris flows in Beichuan as an example[J]. Journal of Engineering Geology, 2013, 21(4):525-533.]
[4]	任光明, 张涛. 基于层次分析法与可拓法的新疆某泥石流危险性评价[J]. 水电能源科学, 2013, 31(9):144-147.
	[Ren G M, Zhang T. Risk assessment of debris flow in Xinjiang based on AHP and extension method[J]. Water Resources and Power, 2013, 31(9):144-147.]
[5]	杨涛, 唐川, 朱金勇, 等. 四川省汶川县绵虒镇小流域泥石流危险性评价[J]. 长江科学院院报, 2018, 35(10):82-87.
	[Yang T, Tang C, Zhu J Y, et al. Hazard assessment of debris flow in small watershed of miansi town, Wenchuan, Sichuan Province[J]. Journal of Yangtze River Scientific Research Institute, 2018, 35(10):82-87.] DOI:10.11988/ckyyb.20170411 doi: 10.11988/ckyyb.20170411
[6]	王毅, 唐川, 李为乐, 等. 基于GIS的模糊数学模型在泥石流敏感性评价中的应用[J]. 自然灾害学报, 2017, 26(1):19-26.
	[Wang Y, Tang C, Li W L, et al. Application of GIS-based fuzzy mathematics model to sensitivity evaluation of debris flow[J]. Journal of Natural Disasters, 2017, 26(1):19-26.] DOI:10.13577/j.jnd.2017.0103 doi: 10.13577/j.jnd.2017.0103
[7]	张浩韦, 刘福臻, 王军朝, 等. 基于FLO-2D数值模拟的工布江达县城泥石流灾害危险性评价[J]. 地质力学学报, 2022, 28(2):306-318.
	[Zhang H W, Liu F Z, Wang J C, et al. Hazard assessment of debris flows in Kongpo Gyamda, Tibet based on FLO-2D numerical simulation[J]. Journal of Geomechanics, 2022, 28(2):306-318.]
[8]	乔渊, 刘铁骥, 陈亮, 等. 基于Massflow模型的甘肃省岷县二马沟泥石流危险性评价[J]. 水利水电技术, 2020, 51(4):184-192.
	[Qiao Y, Liu T J, Chen L, et al. Massflow Model-based hazard assessment on Erma Gully Debris Flow in Minxian County of Gansu Province[J]. Water Resources and Hydropower Engineering, 2020, 51(4):184-192.] DOI:10.13928/j.cnki.wrahe.2020.04.022 doi: 10.13928/j.cnki.wrahe.2020.04.022
[9]	Kim M, Lee S, Kwon T H, et al. Sensitivity analysis of influencing parameters on slit-type barrier performance against debris flow using 3D-based numerical approach[J]. International Journal of Sediment Research, 2021, 36(1):50-62. DOI:10.1016/j.ijsrc.2020.04.005 doi: 10.1016/j.ijsrc.2020.04.005
[10]	Choi S K, Kwon T H. A case study on the closed-type barrier effect on debris flows at Mt. woomyeon, Korea in 2011 via a numerical approach[J]. Energies, 2021, 14(23):7890. DOI:10.3390/en14237890 doi: 10.3390/en14237890
[11]	Hsu Y C, Liu K F. Combining TRIGRS and DEBRIS-2D models for the simulation of a rainfall infiltration induced shallow landslide and subsequent debris flow[J]. Water, 2019, 11(5):890. DOI:10.3390/w11050890 doi: 10.3390/w11050890
[12]	陈刚, 何政伟, 杨斌, 等. 遗传BP神经网络在泥石流危险性评价中的应用[J]. 计算机工程与应用, 2010, 46(3):228-231.
	[Chen G, He Z W, Yang B, et al. Application of genetic BP Neural Network on risk assessment of debris flow[J]. Computer Engineering and Applications, 2010, 46(3):228-231.] DOI:10.3778/j.issn.1002-8331.2010.03.070 doi: 10.3778/j.issn.1002-8331.2010.03.070
[13]	李秀珍, 孔纪名, 李朝凤. 多分类支持向量机在泥石流危险性区划中的应用[J]. 水土保持通报, 2010, 30(5):128-133,157.
	[Li X Z, Kong J M, Li C F. Application of multi-classification support vector machine in regionalization of debris flow hazards[J]. Bulletin of Soil and Water Conservation, 2010, 30(5):128-133,157.] DOI:10.13961/j.cnki.stbctb.2010.05.010 doi: 10.13961/j.cnki.stbctb.2010.05.010
[14]	刘永垚, 第宝锋, 詹宇, 等. 基于随机森林模型的泥石流易发性评价——以汶川地震重灾区为例[J]. 山地学报, 2018, 36(5):765-773.
	[Liu Y Y, Di B F, Zhan Y, et al. Debris flows susceptibility assessment in Wenchuan earthquake areas based on random forest algorithm model[J]. Mountain Research, 2018, 36(5):765-773.] DOI:10.16089/j.cnki.1008-2786.000372 doi: 10.16089/j.cnki.1008-2786.000372
[15]	刘佳, 伍宇明, 高星, 等. 基于GEE和U-net模型的同震滑坡识别方法[J]. 地球信息科学学报, 2022, 24(7):1275-1285. doi: 10.12082/dqxxkx.2022.210704
	[Liu J, Wu Y M, Gao X, et al. Image recognition of co-seismic landslide based on GEE and U-net neural network[J]. Journal of Geo-Information Science, 2022, 24(7):1275-1285.]
[16]	武雪玲, 杨经宇, 牛瑞卿. 一种结合SMOTE和卷积神经网络的滑坡易发性评价方法[J]. 武汉大学学报·信息科学版, 2020, 45(8):1223-1232.
	[Wu X L, Yang J Y, Niu R Q. A landslide susceptibility assessment method using SMOTE and convolutional neural network[J]. Geomatics and Information Science of Wuhan University, 2020, 45(8):1223-1232.] DOI:10.13203/j.whugis20200127 doi: 10.13203/j.whugis20200127
[17]	Hakim W L, Rezaie F, Nur A S, et al. Convolutional neural network (CNN) with metaheuristic optimization algorithms for landslide susceptibility mapping in Icheon, South Korea[J]. Journal of Environmental Management, 2022, 305:114367. DOI:10.1016/j.jenvman.2021.114367 doi: 10.1016/j.jenvman.2021.114367
[18]	Youssef A M, Pradhan B, Dikshit A, et al. Landslide susceptibility mapping using CNN-1D and 2D deep learning algorithms: Comparison of their performance at Asir Region, KSA[J]. Bulletin of Engineering Geology and the Environment, 2022, 81(4):1-22. DOI:10.1007/s10064-022-02657-4 doi: 10.1007/s10064-022-02657-4
[19]	位栋梁, 王延伟, 王自法, 等. 基于卷积神经网络的地震震级快速估算方法[J]. 地震学报, 2022, 44(2):316-326.
	[Wei D L, Wang Y W, Wang Z F, et al. A fast estimation method of earthquake magnitude based on convolutional neural networks[J]. Acta Seismologica Sinica, 2022, 44(2):316-326.]
[20]	Zhang X Y, Zhou X Y, Lin M X, et al. ShuffleNet: an extremely efficient convolutional neural network for mobile devices[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018: 6848-6856. DOI:10.1109/CVPR.2018.00716 doi: 10.1109/CVPR.2018.00716
[21]	Iandola F, Moskewicz M, Karayev S, et al. DenseNet: implementing efficient ConvNet descriptor Pyramids[EB/OL]. 2014: arXiv: 1404. 1869. https://arxiv.org/abs/1404.1869
[22]	Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]// 2017 IEEE International Conference on Computer Vision. 2017: 2999-3007. DOI:10.1109/ICCV.2017.324 doi: 10.1109/ICCV.2017.324
[23]	He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. 2016:770-778. DOI:10.1109/CVPR.2016.90 doi: 10.1109/CVPR.2016.90
[24]	樊辉, 何大明. 怒江流域气候特征及其变化趋势[J]. 地理学报, 2012, 67(5):621-630. doi: 10.11821/xb201205005
	[Fan H, He D M. Regional climate and its change in the nujiang river basin[J]. Acta Geographica Sinica, 2012, 67(5):621-630.] DOI:10.11821/xb201205005 doi: 10.11821/xb201205005
[25]	李益敏, 谢亚亚, 蒋德明, 等. 怒江州斜坡地质灾害孕灾环境因素敏感性研究[J]. 水土保持研究, 2018, 25(5):300-305.
	[Li Y M, Xie Y Y, Jiang D M, et al. Study on sensitivity in disaster-pregnant environmental factors of slope geological hazards in nujiang prefecture[J]. Research of Soil and Water Conservation, 2018, 25(5):300-305.] DOI:10.13869/j.cnki.rswc.2018.05.043 doi: 10.13869/j.cnki.rswc.2018.05.043
[26]	阳友奎, 周迎庆, 姜瑞琪. 坡面地质灾害柔性防护的理论与实践[M]. 北京: 科学出版社, 2005.
	[Yang Y K. Zhou Y Q, Jiang R Q. Theory and practice of flexible protection against slope geological disasters[M]. Beijing: Science Press, 2005.]
[27]	朱渊, 余斌, 亓星, 等. 地形条件对泥石流发育的影响——以岷江流域上游为例[J]. 吉林大学学报(地球科学版), 2014, 44(1):268-277.
	[Zhu Y, Yu B, Qi X, et al. Topographical factors in the formation of gully type debris flows in the upper reaches of Minjiang River[J]. Journal of Jilin University (Earth Science Edition), 2014, 44(1):268-277.] DOI:10.13278/j.cnki.jjuese.201401204 doi: 10.13278/j.cnki.jjuese.201401204
[28]	Strahler A N. Quantitative analysis of watershed geomorphology[J]. Eos, Transactions American Geophysical Union, 1957, 38(6):913-920. DOI:10.1029/TR038i006p00913 doi: 10.1029/TR038i006p00913
[29]	姚振国, 刘建周, 牛贝贝, 等. 流域面积对沟道泥石流发育的影响分析[J]. 资源环境与工程, 2019, 33(2):217-219.
	[Yao Z G, Liu J Z, Niu B B, et al. Influence of drainage area on development of debris flow in gully[J]. Resources Environment & Engineering, 2019, 33(2):217-219.] DOI:10.16536/j.cnki.issn.1671-1211.2019.02.014 doi: 10.16536/j.cnki.issn.1671-1211.2019.02.014
[30]	胡凯衡, 游勇, 庄建琦, 等. 北川地震重灾区泥石流特征与减灾对策[J]. 地理科学, 2010, 30(4):566-570.
	[Hu K H, You Y, Zhuang J Q, et al. Characteristics and countermeasures of debris flows in beichuan's meizoseismal area[J]. Scientia Geographica Sinica, 2010, 30(4):566-570.] DOI:10.13249/j.cnki.sgs.2010.04.012 doi: 10.13249/j.cnki.sgs.2010.04.012
[31]	He K M, Zhang X Y, Ren S Q, et al. Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification[C]// 2015 IEEE International Conference on Computer Vision. 2015: 1026-1034. DOI:10.1109/ICCV.2015.123 doi: 10.1109/ICCV.2015.123
[32]	Xie S N, Girshick R, Dollár P, et al. Aggregated residual transformations for deep neural networks[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. 2017: 5987-5995. DOI:10.1109/CVPR.2017.634 doi: 10.1109/CVPR.2017.634
[33]	Howard A G, Zhu M L, Chen B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications[EB/OL]. 2017: arXiv: 1704.04861. https://arxiv.org/abs/1704.04861
[34]	Tan M, Le Q. Efficientnet: Rethinking model scaling for convolutional neural networks[C]. International Conference on Machine Learning, 2019:6105-6114
[35]	Szegedy C, Liu W, Jia Y Q, et al. Going deeper with convolutions[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition. 2015: 1-9. DOI:10.1109/CVPR.2015.7298594 doi: 10.1109/CVPR.2015.7298594
[36]	唐邦兴, 周必凡, 吴积善, 等. 中国泥石流[M]. 北京: 商务印书馆, 2000.
	[Tang B X, Zhou B F, Wu J S, et al. Debris Flows in China[M]. Beijing: The Commercial Press, 2000.]
[37]	Moore I D, Grayson R B, Ladson A R. Digital terrain modelling: A review of hydrological, geomorphological, and biological applications[J]. Hydrological Processes, 1991, 5(1):3-30. DOI:10.1002/hyp.3360050103 doi: 10.1002/hyp.3360050103
[38]	Beven K J, Kirkby M J. A physically based, variable contributing area model of basin hydrology/Un modèle à base physique de zone d'appel variable de l'hydrologie du bassin versant[J]. Hydrological Sciences Bulletin, 1979, 24(1):43-69. DOI:10.1080/02626667909491834 doi: 10.1080/02626667909491834

所属类别	类别编号	条数/条	高程差/km	主沟长度/km	流域面积/km²
正样本	0	51	(1.1, 2.9)	(2.3, 15.4)	(1.0, 24.0)
	1	22	(1.5, 3.2)	(9.8, 18.8)	(26.0, 64.0)
	2	9	(2.5, 3.1)	(18.0, 27.0)	(69.0, 95.1)
负样本	3	42	(0.3, 2.4)	(2.3, 7.1)	(1.0, 7.4)
	4	29	(0.7, 2.9)	(4.9, 15.6)	(7.5, 19.2)
	5	14	(1.3, 3.2)	(9.1, 16.9)	(21.7, 46.0)

网络名称	OA/%	AA/%	P/%	R/%	kappa
RSDNet	84.7	83.6	89.7	80.6	0.82
ResNet18^[23]	74.7	74.6	76.8	74.8	0.71
ResNet34^[23]	72.7	72.9	72.8	77.1	0.67
ShuffleNet^[20]	72.9	73.0	73.3	74.5	0.68
DenseNet^[21]	76.2	77.1	75.7	76.6	0.72
ResNext^[32]	75.0	74.2	76.4	77.2	0.70
MobileNet^[33]	70.3	74.4	75.1	69.5	0.69
EfficientNet^[34]	70.7	70.2	77.4	71.5	0.66
InceptionV1^[35]	73.3	73.5	72.9	75.0	0.68

一种沟谷型潜在泥石流危险性评价方法：基于多源数据融合的卷积神经网络

A Potential Gully Debris Flows Hazard Assessment Method: A CNN Model based on Multi-source Data Fusion

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 17

参考文献 38

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	邢子瑶, 董芯蕊, 昝糈莉, 杨帅, 黄梓焓, 刘哲, 张晓东. 融合VGI和遥感等多源数据的洪涝范围提取与模拟方法[J]. 地球信息科学学报, 2023, 25(9): 1869-1881.
[2]	令振飞, 刘涛, 杜萍, 赵丹, 陈朴一, 马天恩. 一种支持建筑群组相似模式检索的变分图卷积自编码模型[J]. 地球信息科学学报, 2023, 25(7): 1405-1417.
[3]	刘潇, 刘智, 林雨准, 王淑香, 左溪冰. 面向遥感影像场景分类的类中心知识蒸馏方法[J]. 地球信息科学学报, 2023, 25(5): 1050-1063.
[4]	侯慧太, 蓝朝桢, 徐青. 基于卫星影像全局和局部深度学习特征检索的无人机绝对定位方法[J]. 地球信息科学学报, 2023, 25(5): 1064-1074.
[5]	陈科, 管海燕, 雷相达, 曹爽. 基于特征增强核点卷积网络的多光谱LiDAR点云分类方法[J]. 地球信息科学学报, 2023, 25(5): 1075-1087.
[6]	衡雪彪, 许捍卫, 唐璐, 汤恒, 许怡蕾. 基于改进全卷积神经网络模型的土地覆盖分类方法研究[J]. 地球信息科学学报, 2023, 25(3): 495-509.
[7]	高鹏飞, 曹雪峰, 李科, 游雄. 融合多元稀疏特征与阶层深度特征的遥感影像目标检测[J]. 地球信息科学学报, 2023, 25(3): 638-653.
[8]	黄帅元, 董有福, 李海鹏. 黄土高原区SRTM1 DEM高程误差校正模型构建及对比分析[J]. 地球信息科学学报, 2023, 25(3): 669-681.
[9]	黄昕, 毛政元. 基于时空多图卷积网络的网约车乘客需求预测[J]. 地球信息科学学报, 2023, 25(2): 311-323.
[10]	刘洋, 康健, 管海燕, 汪汉云. 基于双注意力残差网络的高分遥感影像道路提取模型[J]. 地球信息科学学报, 2023, 25(2): 396-408.
[11]	何海清, 周福阳, 陈敏, 陈婷, 官云兰, 曾怀恩, 魏燕. 耦合卷积神经网络与注意力机制的无人机摄影测量果树树冠分割方法[J]. 地球信息科学学报, 2023, 25(12): 2387-2401.
[12]	贺日兴, 唐宗棣, 姜超, 林艳, 陆宇梅, 李欣然, 龙伟, 邓悦. 顾及道路权重的图卷积犯罪时空预测模型[J]. 地球信息科学学报, 2023, 25(10): 1986-1999.
[13]	吴新辉, 毛政元, 翁谦, 施文灶. 利用基于残差多注意力和ACON激活函数的神经网络提取建筑物[J]. 地球信息科学学报, 2022, 24(4): 792-801.
[14]	焦洋洋, 刘平芝, 刘爱龙, 刘松林. AlexNet支持下的地图建筑物形状分类方法[J]. 地球信息科学学报, 2022, 24(12): 2333-2341.
[15]	施群山, 蓝朝桢, 徐青, 周杨, 胡校飞. 面向卫星遥感影像检索定位的深度学习全局表征模型评估与分析[J]. 地球信息科学学报, 2022, 24(11): 2245-2263.