基于中低分辨率遥感影像的分类方法对比研究——以鄱阳湖水鸟栖息地遥感分类为例

doi:10.11870/cjlyzyyhj201511005

摘要/Abstract

摘要： 选取中国鄱阳湖南矶山湿地为研究范围,通过分析候鸟在越冬期的栖息特征,构建适宜于提取候鸟栖息地的湿地分类系统。以Landsat8卫星OLI遥感影像为数据源,采用面向对象分类方法,通过多尺度分割、特征提取和决策树建立等关键步骤,实现湿地信息的快速提取;通过与传统像元法的分类结果对比,系统分析了面向对象方法在基于中低分辨率遥感影像的湿地信息提取中的有效性。研究表明:在面向对象湿地信息提取中,构建不同等级的分割尺度,在多个尺度上提取同一地物,可以更好识别复杂的湿地景观类型;相比仅依据像元光谱特征进行分类的传统像元方法,面向对象方法综合利用光谱、空间、形状和纹理等多种特征,因此可以获得更高的精度(总体分类精度达到87.64%,Kappa系数为 0.855 2);基于中低分辨率遥感影像,采用面向对象分类方法,能够获得较高精度的湿地景观分布,是一种成本较低且行之有效的技术手段。

关键词: 湿地, 鄱阳湖, 候鸟栖息地, 面向对象分类方法, 中低分辨率遥感影像, 信息提取

Abstract: Poyang Lake is an important flood mediation lake of the middle and lower reaches of Yangtze River. It is also recognized as an important international wetland which provides key wintering habitats for migratory birds along the East Asian-Australian Flyway. The phenomenon of lower water levels and longer dry season persisting in Poyang Lake is very common in recent years. In this context, the proposal for construction of the Poyang Lake hydraulic project is being made, and it has aroused a broad attention on the habitats from ecologists and environment protection organizations. Wetland information extraction is important for the protection of habitats, and we found that the appliation of object-oriented classification method applying medium-low resolution remote sensing image in wetland classification is not sufficient. In this paper, we classified wetlands according to the migratory habitat preference in wintering period by taking Poyang-Nanji National Wetland Nature Reserve as study area. The main data were extracted from Landsat8 OLI images by object-oriented approaches such as muti-scale segmentation, feature selection and decision tree establishment. Meanwhile, NDVI and NDWI were applied for classification, and characterizing the bands and band combinations were constructed. This paper systematically analyzed the validity of object-oriented approaches in medium-low resolution remote-sensing images by comparing with traditional pixel-based approaches including maximum likehood method, support vector machines method and artificial neural net method. The results show that extracting certain object from multi-segmentation scale levels better recognizes complex wetland landscape and works more efficiently. Compared with spectral characteristics method, object-oriented approach gains higher precision by integrating spectral, spatial, shape and texture features (Total precision 87.64%;Kappa coefficient 0.8552). In conclusion, the medium-low resolution Landsat8 OLI images can meet the needs of wetland information extraction and can be used as the main data source of wetland classification. The object-oriented approach is relatively low-costing and efficient to achieve high-precision wetland landscape information from medium-low remote-sensing images. What's more, this paper provides a new way for other areas of wetland information extraction.

Key words: wetland, Poyang Lake, migratory bird habitat, object-based classification, medium-low resolution remote sensing image, information extraction

中图分类号:

TP79

陈彬, 于秀波, 刘宇. 基于中低分辨率遥感影像的分类方法对比研究——以鄱阳湖水鸟栖息地遥感分类为例[J]. 长江流域资源与环境, 2015, 24(11): 1834-1842.

CHEN Bin, YU Xiu-bo, LIU Yu. COMPARISON RESEARCH ON CLASSIFICATION METHODS BASED ON MEDIUM-LOW RESOLUTION REMOTE SENSING IMAGE ——A CASE STUDY OF HABITATS EXTRACTION IN POYANG LAKE[J]. RESOURCES AND ENVIRONMENT IN THE YANGTZE BASIN, 2015, 24(11): 1834-1842.

参考文献

[1] 刘红玉.中国湿地资源特征、现状与生态安全[J].资源科学,2005,27(3):54-60.
[2] 王学雷,许厚泽,蔡述明.长江中下游湿地保护与流域生态管理[J].长江流域资源与环境,2006,15(5):564-568.
[3] 孙志高,牟晓杰,陈小兵,等.黄河三角洲湿地保护与恢复的现状、问题与建议[J].湿地科学,2011,9(2):107-115.
[4] JIAO L. Scientists line up against dam that would alter protected wetlands[J]. Science, 2009, 326(5952): 508-509.
[5] 李玉凤,刘红玉.湿地分类和湿地景观分类研究进展[J].湿地科学,2014,12(1):102-108.
[6] 张方方,齐述华,廖富强,等.鄱阳湖湿地出露草洲分布特征的遥感研究[J].长江流域资源与环境,2011,20(11):1361-1367.
[7] 陈有明,刘同庆,黄燕,等.长江流域湿地现状与变化遥感研究[J].长江流域资源与环境,2014,23(6):801-808.
[8] MELACK J M, HESS L L. Remote sensing of the distribution and extent of wetlands in the Amazon basin[M]// Amazonian Floodplain Forests. Netherlands: Springer, 2011: 43-59.
[9] LIU G L, ZHANG L C, ZHANG Q, et al. Spatio-temporal dynamics of wetland landscape patterns based on remote sensing in Yellow River Delta, China[J]. Wetlands, 2014, 34(4): 787-801.
[10] 韦玉春,汤国安.遥感数字图像处理教程[M].北京:科学出版社,2007.
[11] 王传辉,郭振亚,高超,等.基于遥感的巢湖流域景观格局变化研究[J].长江流域资源与环境,2013,22(12):1586-1592.
[12] 林川,宫兆宁,赵文吉.基于中分辨率TM数据的湿地水生植被提取[J].生态学报,2010,30(23):6460-6469.
[13] 乔婷,张怀清,陈永富,等.基于NDVI分割与面向对象的东洞庭湖湿地植被信息提取技术[J].西北林学院学报,2013,28(4):170-175.
[14] 谢静,王宗明,毛德华,等.基于面向对象方法和多时相HJ-1影像的湿地遥感分类——以完达山以北三江平原为例[J].湿地科学,2012,10(4):429-438.
[15] 陈忠.高分辨率遥感图像分类技术研究[D].北京: 中国科学院研究生院(遥感应用研究所),2006.
[16] 郭英荣.鄱阳湖湿地越冬候鸟群落特征与生态保护对策[D].南京: 南京林业大学,2005.
[17] 涂晓斌,马建华,宋玉赞,等.鄱阳湖越冬水禽资源现状及保护对策[J].野生动物,2000(4):6-7.
[18] 纪伟涛,吴建东,易武生,等.鄱阳湖国家级自然保护区涉禽调查报告[J].江西林业科技,2001(2):29-31,34.
[19] 崔鹏,夏少霞,刘观华,等.鄱阳湖越冬水鸟种群变化动态[J].四川动物,2013,32(2):292-296.
[20] 齐述华,刘影,于秀波,等."堑秋湖"对鄱阳湖越冬候鸟栖息地功能影响的辨析[J].长江流域资源与环境,2011,20(S1):18-21.
[21] 刘红玉,吕宪国.三江平原湿地景观生态制图分类系统研究[J].地理科学,1999,19(5):432-436.
[22] 罗宏宇,黄方,张养贞.辽河三角洲沼泽湿地时空变化及其生态效应[J].东北师大学报自然科学版,2003,35(2):100-105.
[23] 杨敏,刘世梁,孙涛,等.黄河三角洲湿地景观边界变化及其对土壤性质的影响[J].湿地科学,2009,7(1):67-74.
[24] 夏少霞,于秀波,范娜.鄱阳湖越冬季候鸟栖息地面积与水位变化的关系[J].资源科学,2010,32(11):2072-2078.
[25] DESGRANGES J L, INGRAM J, DROLET B, et al. Modelling wetland bird response to water level changes in the Lake Ontario-St. Lawrence River hydrosystem[J]. Environmental Monitoring and Assessment, 2006, 113(1/3): 329-365.
[26] BANCROFT G T, GAWLIK D E, RUTCHEY K. Distribution of wading birds relative to vegetation and water depths in the northern Everglades of Florida, USA[J]. Waterbirds, 2002, 25(3): 265-277.
[27] TAYLOR P J, WALKER G R, HODGSON G, et al. Testing of a GIS model of Eucalyptus Iargiflorens health on a semiarid, saline floodplain[J]. Environmental Management, 1996, 20(4): 553-564.
[28] 唐小平,黄桂林.中国湿地分类系统的研究[J].林业科学研究,2003,16(5):531-539.
[29] 初庆伟,张洪群,吴业炜,等.Landsat-8卫星数据应用探讨[J].遥感信息,2013,28(4):110-114.
[30] 张俊,于庆国,侯家槐.面向对象的高分辨率影像分类与信息提取[J].遥感技术与应用,2010,25(1):112-117.
[31] 谭衢霖,刘正军,沈伟.一种面向对象的遥感影像多尺度分割方法[J].北京交通大学学报,2007,31(4):111-114,119.
[32] 陈春雷,武刚.面向对象的遥感影像最优分割尺度评价[J].遥感技术与应用,2011,26(1):96-102.

相关文章 15

[1]	李云良, 姚静, 张小琳, 张奇. 鄱阳湖水体垂向分层状况调查研究[J]. 长江流域资源与环境, 2017, 26(06): 915-924.
[2]	张小琳, 张奇, 王晓龙. 洪泛湖泊水位-流量关系的非线性特征分析[J]. 长江流域资源与环境, 2017, 26(05): 723-729.
[3]	李金前, 王吉, 刘亚军, 邹锋, 马燕天, 吴兰. 水位高程变化对湿地土壤微生物代谢功能的影响研究——以蚌湖为例[J]. 长江流域资源与环境, 2017, 26(05): 730-737.
[4]	侯立春, 林振山, 琚胜利, 赖正清, 吴连霞, 张志荣. 环鄱阳湖旅游圈旅游经济联系与区域发展策略[J]. 长江流域资源与环境, 2017, 26(04): 508-518.
[5]	齐凌艳, 黄佳聪, 高俊峰, 郭玉银. 鄱阳湖枯水水位及流速时空分布模拟[J]. 长江流域资源与环境, 2017, 26(04): 572-584.
[6]	赵志刚, 余德, 韩成云, 王凯荣. 2008~2016年鄱阳湖生态经济区生态系统服务价值的时空变化研究[J]. 长江流域资源与环境, 2017, 26(02): 198-208.
[7]	李冰, 杨桂山, 万荣荣, 刘宝贵, 戴雪, 许晨. 鄱阳湖出流水质2004~2014年变化及其对水位变化的响应:对水质监测频率的启示[J]. 长江流域资源与环境, 2017, 26(02): 289-296.
[8]	布乃顺, 胡悦, 杨骁, 张雪, 王俭, 李博, 方长明, 宋有涛. 互花米草入侵对长江河口湿地土壤理化性质的影响[J]. 长江流域资源与环境, 2017, 26(01): 100-109.
[9]	戴雪, 何征, 万荣荣, 杨桂山. 近35 a长江中游大型通江湖泊季节性水情变化规律研究[J]. 长江流域资源与环境, 2017, 26(01): 118-125.
[10]	张范平, 方少文, 周祖昊, 温天福, 张梅红. 鄱阳湖水位多时间尺度动态变化特性分析[J]. 长江流域资源与环境, 2017, 26(01): 126-133.
[11]	朱婧瑄, 齐述华, 刘贵花, 王点, 熊梦雅. 2000~2013年鄱阳湖流域蒸散量时空变化[J]. 长江流域资源与环境, 2016, 25(Z1): 9-16.
[12]	汪丹, 王点, 齐述华. 鄱阳湖水位-淹水面积关系不确定性的分析[J]. 长江流域资源与环境, 2016, 25(Z1): 95-102.
[13]	李云良, 张小琳, 赵贵章, 姚静, 张奇. 鄱阳湖区地下水位动态及其与湖水侧向水力联系分析[J]. 长江流域资源与环境, 2016, 25(12): 1894-1902.
[14]	李艳, 高艳娜, 戚志伟, 姜楠, 仲启铖, 姜姗, 王开运, 张超. 滨海芦苇湿地土壤微生物数量对长期模拟增温的响应[J]. 长江流域资源与环境, 2016, 25(11): 1738-1747.
[15]	李云良, 姚静, 李梦凡, 张奇. 鄱阳湖水流运动与污染物迁移路径的粒子示踪研究[J]. 长江流域资源与环境, 2016, 25(11): 1748-1758.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed