应用迁移学习的林火烈度初始评估研究
Initial assessment of burn severity using the transfer learning model
- 2022年26卷第10期 页码：2001-2013
纸质出版日期： 2022-10-07
DOI： 10.11834/jrs.20210156
扫 描 看 全 文
浏览全部资源
扫码关注微信
摘要
林火发生后，开展森林生态系统烈度信息的初始评估，能够为灾后生态修复管理措施的快速实施提供定量依据。为了改善传统林火烈度评估模型的时效性，本研究利用历史过火区域的实地调查数据，构建基于迁移学习的烈度评估模型，并将其应用于2020年3月30日发生的西昌泸山森林大火烈度初始评估研究中。研究结果表明：迁移学习算法能够将源区域和目标区域的遥感影像光谱转换为多个新的特征变量，在这些新特征变量构成的投影空间中，源区域和目标区域样本具有相似的特征分布。在此基础上，基于源区域历史实地调查数据构建的烈度评估模型，能够迁移应用于目标区域的烈度评估。在本研究林火烈度的初始评估中，基于迁移学习的烈度评估模型精度较高，总体精度为71.20%，Kappa系数为0.64。与该模型对比，未进行迁移学习的支持向量回归模型精度较低，其总体精度为58.00%，Kappa系数为0.48。同时，基于dNDVI、dLST和dNBR指数的经验回归模型精度最低，其总体精度分别为：20.80%、34.8%和24.80%，Kappa系数分别为：0.01、0.19和0.06。本研究可为林火灾后管理措施的快速响应，提供一种新的思路和参考。
Abstract
In recent years
forest fires occur frequently around the world
which severely damage the structure and function of the forest ecosystem. The initial assessment of burn severity could provide a quantitative basis for rapid implementations of post-fire restoration measures. In the last decades
remote sensing-based models have become an appropriate choice to assess burn severity
which generally require a certain amount of field survey data. However
this requirement could not be sufficiently satisfied in the first moments after fire
since the field survey work would cost a substantial amount of time and labor. The absence of field survey data in the initial assessment of burn severity would largely limit the efficient application of remote sensing technologies. In this study
a transfer learning algorithm (i.e.
SSTCA
semi-supervised Transfer Component Analysis) was employed to propose an initial assessment model of burn severity to improve the time-efficiency of traditional remote sensing-based models. Firstly
the SSTCA algorithm was applied to project a series of new features from original spectral features of remotely sensed data. Based on these projected features
a Support Vector Regression (SVR) model was then trained using historical field survey data from source areas (i.e.
Bear fire on June 27
2002 and Mule fire on July 11
2002). Thereafter
the SSTCA-SVR model was transferred to the initial assessment of burn severity of a target area (i.e.
Lushan fire on March 30
2020). Finally
the performance of this proposed model was quantitatively compared with those of some traditional models (i.e.
dNDVI-
dLST-
dNBR-
and SVR-based models). Results showed that original spectral features of remote sensing images over source and target areas were quite different. After the SSTCA projection
projected features of source and target samples have a similar distribution pattern in the new features-based space. Meanwhile
in the initial assessment of burn severity
dNDVI- and dNBR-based models have overestimated burn severity levels with low accuracies (i.e.
overall accuracy was from 20.80% to 24.80% and Kappa value was between 0.01 and 0.06). Compared with them
the dLST-based model has a better performance with an overall accuracy of 34.80% and a Kappa value of 0.19. Although SVR-based model has shown a promising performance with an overall accuracy of 58.00% and a Kappa value of 0.48
this model has overestimated the burn severity levels in some regions of burned areas. The assessment results of burn severity levels using SSTCA-SVR model has the best performance with an overall accuracy of 71.20% and a Kappa value of 0.64. We conclude that the application of a transferring learning algorithm would be helpful for building an assessment model of burn severity with a good transferring ability. In this way
more accurate results could be obtained in the initial assessment of burn severity
and the response of post-fire management might be accelerated after forest fires.
关键词
林火烈度迁移学习初始评估Landsat森林火灾泸山
Keywords
burn severitytransfer learninginitial assessmentLandsatforest fireLushan
references
Chen X F, Liu L, Li J G, Ou W H and Zhang Y H. 2020. Application and research progress of fire monitoring using satellite remote sensing. Journal of Remote Sensing, 24(5): 531-542
陈兴峰, 刘李, 李家国, 欧文浩, 张玉环. 2020. 卫星遥感火点监测应用和研究进展. 遥感学报, 24(5): 531-542 [DOI: 10.11834/jrs.20209118http://dx.doi.org/10.11834/jrs.20209118]
Chen X X, Vogelmann J E, Rollins M, Ohlen D, Key C H, Yang L M, Huang C Q and Shi H. 2011. Detecting post-fire burn severity and vegetation recovery using multitemporal remote sensing spectral indices and field-collected composite burn index data in a ponderosa pine forest. International Journal of Remote Sensing, 32(23): 7905-7927 [DOI: 10.1080/01431161.2010.524678http://dx.doi.org/10.1080/01431161.2010.524678]
Cherkassky V and Ma Y Q. 2004. Practical selection of SVM parameters and noise estimation for SVM regression. Neural Networks, 17(1): 113-126 [DOI: 10.1016/S0893-6080(03)00169-2http://dx.doi.org/10.1016/S0893-6080(03)00169-2]
Chuvieco E, Riaño D, Danson F M and Martin P. 2006. Use of a radiative transfer model to simulate the postfire spectral response to burn severity. Journal of Geophysical Research, 111(G4): G04S09 [DOI: 10.1029/2005jg000143http://dx.doi.org/10.1029/2005jg000143]
De Santis A and Chuvieco E. 2007. Burn severity estimation from remotely sensed data: performance of simulation versus empirical models. Remote Sensing of Environment, 108(4): 422-435 [DOI: 10.1016/j.rse.2006.11.022http://dx.doi.org/10.1016/j.rse.2006.11.022]
De Santis A and Chuvieco E. 2009. GeoCBI: a modified version of the Composite Burn Index for the initial assessment of the short-term burn severity from remotely sensed data. Remote Sensing of Environment, 113(3): 554-562 [DOI: 10.1016/j.rse.2008.10.011http://dx.doi.org/10.1016/j.rse.2008.10.011]
Keeley J E. 2009. Fire intensity, fire severity and burn severity: a brief review and suggested usage. International Journal of Wildland Fire, 18(1): 116-126 [DOI: 10.1071/WF07049http://dx.doi.org/10.1071/WF07049]
Key C H and Benson N C. 2006. Landscape Assessment (LA): Sampling and Analysis Methods. Fort Collins: USDA Rocky Mountain Research Station: 1-55. [2016-03-17]. https://www.fs.usda.gov/rm/pubs_series/rmrs/gtr/rmrs_gtr164/rmrs_gtr164_13_land_assess.pdfhttps://www.fs.usda.gov/rm/pubs_series/rmrs/gtr/rmrs_gtr164/rmrs_gtr164_13_land_assess.pdf
Lentile L B, Holden Z A, Smith A M S, Falkowski M J, Hudak A T, Morgan P, Lewis S A, Gessler P E and Benson N C. 2006. Remote sensing techniques to assess active fire characteristics and post-fire effects. International Journal of Wildland Fire, 15(3): 319-345 [DOI: 10.1071/WF05097http://dx.doi.org/10.1071/WF05097]
Liu S C, Li X T, Qin X L, Sun G F and Liu Q. 2020. Adaptive threshold method for active fire identification based on GF-4 PMI data. Journal of Remote Sensing, 24(3): 215-225
刘树超, 李晓彤, 覃先林, 孙桂芬, 刘倩. 2020. GF-4 PMI影像着火点自适应阈值分割. 遥感学报, 24(3): 215-225 [DOI: 10.11834/jrs.20208297http://dx.doi.org/10.11834/jrs.20208297]
Loboda T V, French N H F, Hight-Harf C, Jenkins L and Miller M E. 2013. Mapping fire extent and burn severity in Alaskan tussock tundra: an analysis of the spectral response of tundra vegetation to wildland fire. Remote Sensing of Environment, 134: 194-209 [DOI: 10.1016/j.rse.2013.03.003http://dx.doi.org/10.1016/j.rse.2013.03.003]
Matasci G, Volpi M, Kanevski M, Bruzzone L and Tuia D. 2015. Semisupervised transfer component analysis for domain adaptation in remote sensing image classification. IEEE Transactions on Geoscience and Remote Sensing, 53(7): 3550-3564 [DOI: 10.1109/TGRS.2014.2377785http://dx.doi.org/10.1109/TGRS.2014.2377785]
Miller J D and Thode A E. 2007. Quantifying burn severity in a heterogeneous landscape with a relative version of the delta Normalized Burn Ratio (dNBR). Remote Sensing of Environment, 109(1): 66-80 [DOI: 10.1016/j.rse.2006.12.006http://dx.doi.org/10.1016/j.rse.2006.12.006]
Pan S J, Tsang I W, Kwok J T and Yang Q. 2011. Domain adaptation via transfer component analysis. IEEE Transactions on Neural Networks, 22(2): 199-210 [DOI: 10.1109/TNN.2010.2091281http://dx.doi.org/10.1109/TNN.2010.2091281]
Parks S A, Dillon G K and Miller C. 2014. A new metric for quantifying burn severity: The relativized burn ratio. Remote Sensing, 6(3): 1827-1844 [DOI: 10.3390/rs6031827http://dx.doi.org/10.3390/rs6031827]
Patterson M W and Yool S R. 1998. Mapping fire-induced vegetation mortality using Landsat Thematic Mapper data: a comparison of linear transformation techniques. Remote Sensing of Environment, 65(2): 132-142 [DOI: 10.1016/S0034-4257(98)00018-2http://dx.doi.org/10.1016/S0034-4257(98)00018-2]
Pu D C, Zhang Z M, Long T F, Niu X F, He G J, Wang G Z, Sun J Y, Tang C and Wei M Y. 2020. GABAM2010 accuracy assessment using stratified random sampling. Journal of Remote Sensing, 24(5): 550-558
蒲东川, 张兆明, 龙腾飞, 牛雪峰, 何国金, 王桂周, 孙嘉悦, 唐朝, 魏明月. 2020. 分层随机抽样下全球30 m火烧迹地产品验证. 遥感学报, 24(5): 550-558 [DOI: 10.11834/jrs.20209171http://dx.doi.org/10.11834/jrs.20209171]
Qin X L, Li X T, Liu S C, Liu Q and Li Z Y. 2020. Forest fire early warning and monitoring techniques using satellite remote sensing in China. Journal of Remote Sensing, 24(5): 511-520
覃先林, 李晓彤, 刘树超, 刘倩, 李增元. 2020. 中国林火卫星遥感预警监测技术研究进展. 遥感学报, 24(5): 511-520 [DOI: 10.11834/jrs.20209135http://dx.doi.org/10.11834/jrs.20209135]
Quintano C, Fernández-Manso A, Calvo L, Marcos E and Valbuena L. 2015. Land surface temperature as potential indicator of burn severity in forest Mediterranean ecosystems. International Journal of Applied Earth Observation and Geoinformation, 36: 1-12 [DOI: 10.1016/j.jag.2014.10.015http://dx.doi.org/10.1016/j.jag.2014.10.015]
Quintano C, Fernández-Manso A and Roberts D A. 2013. Multiple Endmember Spectral Mixture Analysis (MESMA) to map burn severity levels from Landsat images in Mediterranean countries. Remote Sensing of Environment, 136: 76-88 [DOI: 10.1016/j.rse.2013.04.017http://dx.doi.org/10.1016/j.rse.2013.04.017]
Rao Y M, Wang C and Huang H G. 2020. Forest fire monitoring based on multisensor remote sensing techniques in Muli County, Sichuan Province. Journal of Remote Sensing, 24(5): 559-570
饶月明, 王川, 黄华国. 2020. 联合多源遥感数据监测四川木里县森林火灾. 遥感学报, 24(5): 559-570 [DOI: 10.11834/jrs.20209125http://dx.doi.org/10.11834/jrs.20209125]
Soverel N O, Coops N C, Perrakis D D B, Daniels L D, and Gergel S E. 2011. The transferability of a dnbr-derived model to predict burn severity across 10 wildland fires in Western Canada. International Journal of Wildland Fire, 20(4): 518-531 [DOI: 10.1071/WF10081http://dx.doi.org/10.1071/WF10081]
Tan L X, Zeng Y N and Zheng Z. 2016. An adaptability analysis of remote sensing indices in evaluating fire severity. Remote Sensing for Land and Resources, 28(2): 84-90
谭柳霞, 曾永年, 郑忠. 2016. 林火烈度遥感评估指数适应性分析. 国土资源遥感, 28(2): 84-90 [DOI: 10.6046/gtzyyg.2016.02.14http://dx.doi.org/10.6046/gtzyyg.2016.02.14]
Tang Y, Wang L J, Deng C, Gan Y Q and Zhao J. 2021. Research on the emergency response of forest fires in Sichuan with the help of high-definition remote sensing technology: An example of emergency monitoring of forest fires in Mianning “4·20”. National Remote Sensing Bulletin, 25(9): 2015-2026
唐尧, 王立娟, 邓琮, 甘玉泉, 赵娟. 2021. 高分遥感技术助力森林火灾应急扑救及隐患预判—以冕宁“4·20”森林火灾为例. 遥感学报, 25(9): 2015-2026 [DOI: 10.11834/jrs.20211352http://dx.doi.org/10.11834/jrs.20211352]
Veraverbeke S, Hook S and Hulley G. 2012. An alternative spectral index for rapid fire severity assessments. Remote Sensing of Environment, 123: 72-80 [DOI: 10.1016/j.rse.2012.02.025http://dx.doi.org/10.1016/j.rse.2012.02.025]
Wang M, Meng H W, Huang L P, Sun Q F, Zhang H C and Shen C M. 2020. Vegetation succession and forest fires over the past 13000 years in the catchment of Yangzonghai Lake, Yunnan[J]. Quaternary Sciences, 40(1): 175-189
王敏, 蒙红卫, 黄林培, 孙启发, 张虎才, 沈才明. 2020. 云南阳宗海流域过去13000年植被演替与森林火灾. 第四纪研究, 40(1): 175-189 [DOI: 10.11928/j.issn.1001-7410.2020.01.17http://dx.doi.org/10.11928/j.issn.1001-7410.2020.01.17]
Wang X L, Wang W J, Chang Y, Feng Y T, Chen H W, Hu Y M and Chi J G. 2013. Fire severity of burnt area in Huzhong forest region of Great Xing’an Mountains, Northeast China based on normalized burn ratio analysis. Chinese Journal of Applied Ecology, 24(4): 967-974
王晓莉, 王文娟, 常禹, 冯玉婷, 陈宏伟, 胡远满, 池建国. 2013. 基于NBR指数分析大兴安岭呼中森林过火区的林火烈度. 应用生态学报, 24(4): 967-974 [DOI: 10.13287/j.1001-9332.2013.0250http://dx.doi.org/10.13287/j.1001-9332.2013.0250]
Xu B B, Wang W Y, Chen L F, Tao J H, Ji X Y, Zhang C J and Fan M. 2022. Forest fire spread simulation based on VIIRS active fire data and FARSITE model. National Remote Sensing Bulletin, 26(8): 1575-1588
徐奔奔, 王炜烨, 陈良富, 陶金花, 纪轩禹, 张成杰,范萌. 2022. 基于VIIRS火点数据和FARSITE系统的森林火灾蔓延模拟.遥感学报, 26(8): 1575-1588 [DOI: 10.11834/jrs.20219427http://dx.doi.org/10.11834/jrs.20219427]
Yan K, Kou L and Zhang D. 2018. Learning domain-invariant subspace using domain features and independence maximization. IEEE Transactions on Cybernetics, 48(1): 288-299 [DOI: 10.1109/TCYB.2016.2633306http://dx.doi.org/10.1109/TCYB.2016.2633306]
Zheng Z, Huang W, Li S N and Zeng Y N. 2017. Forest fire spread simulating model using cellular automaton with extreme learning machine. Ecological Modelling, 348: 33-43 [DOI: 10.1016/j.ecolmodel.2016.12.022http://dx.doi.org/10.1016/j.ecolmodel.2016.12.022]
Zheng Z, Wang J F, Shan B, He Y J, Liao C H, Gao Y H and Yang S Q. 2020. A new model for transfer learning-based mapping of burn severity. Remote Sensing, 12(4): 708 [DOI: 10.3390/rs12040708http://dx.doi.org/10.3390/rs12040708]
Zheng Z, Zeng Y N, Li S N and Huang W. 2016. A new burn severity index based on land surface temperature and enhanced vegetation index. International Journal of Applied Earth Observation and Geoinformation, 45: 84-94 [DOI: 10.1016/j.jag.2015.11.002http://dx.doi.org/10.1016/j.jag.2015.11.002]
Zheng Z, Zeng Y N, Li S N and Huang W. 2018. Mapping burn severity of forest fires in small sample size scenarios. Forests, 9(10): 608 [DOI: 10.3390/f9100608http://dx.doi.org/10.3390/f9100608]
文章被引用时，请邮件提醒。
提交
相关作者
暂无数据
相关机构
中国科学院大学
自然资源部国土整治中心
中国科学院大学 电子电气与通信工程学院
中国科学院空天信息创新研究院 遥感科学国家重点实验室
河南大学 黄河中下游数字地理技术教育部重点实验室
选择翻译语言
- 地址：北京市海淀区北四环西路19号 《遥感学报》编辑部 邮编：100190
- 联系电话：010-58887052 Email：nrsb@aircas.ac.cn
- 技术支持由北京北大方正电子有限公司提供 京ICP备20021838号-9
京公网安备11010802024621 - 本系统建议在Chrome、 IE9+ 以上版本浏览器阅读本站内容，360浏览器请切换至极速模式
- Cookies帮助我们提供服务并提供个性化体验。使用本网站，即表示您同意我们使用Cookies
0
回到顶部