EN中文

地理研究 >

2016 , Vol. 35 >Issue 8: 1433 - 1446

DOI: https://doi.org/10.11821/dlyj201608003

研究论文

首套全球30 m分辨率土地覆被产品区域尺度精度评价——以河南省为例

  • 黄亚博 , 1, 2, 3 ,
  • 廖顺宝 , 1, 3, 4
展开
  • 1. 河南大学环境与规划学院,开封 475004
  • 2. 河南大学计算机与信息工程学院,开封 475004
  • 3. 黄河中下游数字地理技术教育部重点实验室,开封 475004
  • 4. 中国科学院地理科学与资源研究所,北京 100101
通讯作者：廖顺宝（1966- ）,男,四川德阳人,博士,教授,博士生导师,主要从事遥感与GIS应用以及地学信息产品加工与质量评价等方面的研究与教学工作。E-mail:

作者简介：黄亚博（1982- ）,女,河南开封人,博士研究生,讲师,主要从事遥感地学应用研究。E-mail:

收稿日期: 2016-03-23

  要求修回日期: 2016-05-27

  网络出版日期: 2016-08-29

基金资助

国家科技基础性工作专项（2012FY111800）

河南省高等学校重点科研项目（16A520081）

中国科学院战略性先导科技专项（XDA05050000）

收起

Regional accuracy assessments of the first global land cover dataset at 30-meter resolution: A case study of Henan province

  • HUANG Yabo , 1, 2, 3 ,
  • LIAO Shunbao , 1, 3, 4
Expand
  • 1. College of Environment and Planning, Henan University, Kaifeng 475004, Henan, China
  • 2. College of Computer and Information Engineering, Henan University, Kaifeng 475004, Henan, China
  • 3. Key Laboratory of Geospatial Technology for Middle & Lower Yellow River Regions of Ministry of Education, Kaifeng 475004, Henan, China
  • 4. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China

Received date: 2016-03-23

  Request revised date: 2016-05-27

  Online published: 2016-08-29

Copyright

《地理研究》编辑部
Fold

摘要

以河南省为研究区,对全球首套30 m分辨率土地覆盖产品GlobleLand30进行区域尺度精度评价。首先,以中国1∶10万土地利用数据（CHINA-2010）为参考,分析两种产品的空间一致性;而后,通过Google Earth样本分析GlobleLand30在空间不一致区域的制图精度;最后,利用野外实地考察样本对GlobleLand30进行总体精度评价,并从土地覆被复杂度、高程等方面分析影响精度的原因,结果表明：① GlobleLand30与CHINA-2010空间一致性达80.20%。两种产品对耕地、林地、人工地面一致性高,对草地、水体、灌木、湿地、未利用土地的一致性低。② 在空间不一致区域,GlobleLand30的总体分类正确率略低于CHINA-2010,但两者对不同地类的优势不同。③ 经野外实地考察验证可知,GlobleLand30的总体精度达83.33%。④GlobleLand30与CHINA-2010的空间一致性随土地覆被复杂度的增加而降低,并在高程过渡带较低。

关键词： 土地利用/土地覆被; 精度评价; 空间一致性; 30 m分辨率; GlobleLand30

本文引用格式

黄亚博 , 廖顺宝 . 首套全球30 m分辨率土地覆被产品区域尺度精度评价——以河南省为例[J]. 地理研究, 2016 , 35(8) : 1433 -1446 . DOI: 10.11821/dlyj201608003

Abstract

Accurate global land cover datasets are key environmental information for research on global changes and regional sustainable development, such as global climate change, ecosystem assessment, biodiversity conservation. However, there are still some shortcomings in showing spatial and thematic details for the existing global land cover products at medium and low resolutions (300-1000 m). Recently, GlobeLand30, a global land cover product based on Landsat images at a 30-m resolution, was produced and released by China. The product made a breakthrough in enhancing spatial resolution of global land cover products, and has a wide range of application prospects. As with any land cover products, it is necessary to carry out accuracy evaluation of GlobeLand30 in global, regional or local scales before it is widely applied. This study aims at assessing the accuracy of GlobleLand30 in Henan province with cross-comparison analyses and sampling methods respectively. Firstly, the national land use database of China at a 1:100000 scale (CHINA-2010) was used as a reference data to analyze the spatial consistency between these two products. Secondly, validation samples collected from Google Earth were used to assess the producer's accuracy in spatially inconsistent areas. Finally, the overall accuracy of GlobleLand30 was assessed by field survey. Moreover, several factors influencing the accuracy of GlobleLand30 were also taken into consideration, such as land cover complexity and elevation. The following conclusions were drawn: (a) Spatial consistency between GlobleLand30 and CHINA-2010 reached up to 80.20%. There was high spatial consistency for croplands, forests and artificial surfaces, and low spatial consistency for grasslands, water bodies, shrubs, wetlands and unused lands. (b) The overall accuracy of GlobleLand30 was slightly lower than that of CHINA-2010 in spatially inconsistent areas. As for forests, producer's accuracy of GlobleLand30 was relatively higher than that of CHINA-2010. (c) It was found by field survey that the overall accuracy of GlobleLand30 was 83.33%. (d) On the one hand, with the increase of the complexity of land cover types, spatial consistency between GlobleLand30 and CHINA-2010 decreased; on the other hand, with the increase of elevation, spatial consistency between the two products decreased first and then rose, and became lower in the elevation transition zones.

Key words： land use/land cover; accuracy assessment; spatial consistency; 30 m resolution; GlobleLand30

1 引言

土地利用/土地覆被是地表各种物质特征及属性的综合体,是全球环境变化的重要影响因素之一[1,2]。因此,精确的土地覆被数据在提高碳循环、水文循环、气候变化监测、土壤退化等多种科学研究的准确性方面起着至关重要的作用[3-5]。
目前,常见的全球土地覆被数据库分辨率在300~1000 m,这些中低分辨率产品在中国区域精度低,空间一致性在53%~57%,空间细节不足,不能满足全球及区域气候变化模拟等应用的需要[6]。2014年中国适时推出了首套全球30 m分辨率土地覆被数据产品GlobleLand30[7-9],该产品将全球土地覆被产品的空间分辨率提高了一个数量级,具有空间范围广、分辨率高、数据时相新、以自动化分类为主等优势,受到研究者的重视[10]。据不完全统计,来自90多个国家和地区的5000多名用户下载了该数据[11]。
Chen等对该产品在全球范围进行了精度评价[12];Brovelli等对该产品在意大利地区的精度进行了评价[13];孟雯等利用样本评价法研究了GlobleLand30在陕西省的精度[14]。可以看到,以上精度检验主要来源自生产方针对其他国家和中国部分局部区域的自评估结果,GlobleLand30在中国其他局部区域的精度还有待第三方进行更为详细的检验。
土地覆被产品精度评估方法主要有样本评价和比较评价两种。样本评价法具有较高的置信度[15],Clark等利用实地考查或目视解译样本评价了产品精度[14,16-18]。但由于目前标准、可共享的样本难以获得,样本评价法存在成本高,样本数量不足的问题。比较评价法则通过分析待评价产品和已有土地覆被产品的空间一致性,衡量制图精度,空间一致性越高则制图精度越高。该方法具有自动快速、成本低的优点被研究者广泛使用[19-21]。但这种方法仅能获得产品的相对精度,当待评价数据与参考数据分类不一致时,简单的认为待评价数据分类错误,无法进一步评价不同土地覆被产品在空间不一致区域的优势与不足[22]。
为全面反映首套全球土地覆被产品GlobleLand30在中国区域的特点,克服比较评价法的不足,以河南省为研究区,采用比较评价与样本评价相结合的方式,从三方面对以2010年为基准年的GlobleLand30进行评价。第一,空间一致性分析。利用比较评价法以精度较高的CHINA-2010[17]为参考,分析GlobleLand30与参考数据的空间一致性。第二,空间不一致区域精度评价。利用Google Earth样本,进一步分析GlobleLand30在空间不一致区域的制图精度。第三,总体精度分析。根据野外实地考察数据,分析GlobleLand30的总体绝对精度。此外,从土地覆被类别空间分布、数据源、数据时相、制图方法、分类系统等多角度深入分析探讨地类混淆的原因,并对土地覆被复杂度与产品精度、高程与产品精度的关系进行探讨。对GlobleLand30的精度评价不仅为使用者提供选择依据,并且为数据研制者提高产品精度提供参考及建议。

2 研究区概况与数据来源

2.1 研究区概况

河南省地处北亚热带和暖温带过渡地区,110°21′E~116°39′E,31°23′N~36°22′N,面积约16.5万km2,地势自西向东逐级急剧降低,平原、山地、丘陵、盆地等地貌类型俱全,黄淮两大水系贯穿其中。由于阶梯地形和气温降水的多种组合,土地覆被类别丰富多样,是研究中国区域尺度土地覆被产品的典型地区。

2.2 数据来源

2.2.1 GlobleLand30数据 GlobleLand30是由国家基础地理信息中心、中国科学院遥感应用研究所、国家气候中心等有关单位共同制作的世界首套30米分辨率全球土地覆被数据集。该数据集包含2000年和2010年两期,于2014年制作完成,并捐献给联合国组织。本研究对2010年的GlobleLand30进行精度评价。该产品采用“像素分类—对象提取—知识检核”方案研制而成[7,12],以2009-2012年间的9907景Landsat TM/ETM+影像为主要数据源,并使用2640景中国环境减灾卫星（HJ-1）影像数据进行补充[9]。
2.2.2 参考及样本数据 2010年1 ∶ 10万土地利用数据是由全国各地区遥感专家通过人机交互、目视解译制作完成。两产品数据源基本相同,但在分类方法方面,CHINA-2010采用目视解译方法,而GlobleLand30以自动化分类为主、人工核验为辅。野外实地调查验证表明,CHINA-2010全国尺度精度达94.3%以上[17]。GlobleLand30和CHINA-2010的特征如表1所示。
Tab. 1 Characteristics of GlobleLand30 and CHINA-2010

表1 GlobleLand30和CHINA-2010的特征

产品特征 GlobleLand30 CHINA-2010
输入数据 Landsat TM/ETM+、HJ-1、生态区划等 Landsat TM/ETM+、HJ-1、地形图等
数据时相 2010基准年 2010基准年
分类方法 像素分类—对象提取—知识检核 目视解译
数据格式 栅格 矢量
空间尺度 30 m分辨率 1∶10万比例尺
空间范围 全球 中国
投影 WGS84坐标系、UTM投影 GCS Krasovsky 1940 坐标系、Albers 投影
制图单位 国家基础地理信息中心等 中国科学院资源环境科学数据中心
为克服比较评价仅能获得相对精度的缺陷,选取Google Earth影像样本233个对空间不一致区域进行精度评价,野外实地考察样本114个对研究区总体精度进行评价。

2.3 数据预处理

GlobleLand30与CHINA-2010的存储格式、空间范围、空间分辨率、投影方式、分类系统均不相同,因此,在进行比较评价前需要进行以下数据预处理。
首先,为统一数据存储格式和空间分辨率,将CHINA-2010从矢量数据转换为栅格数据,并重采样为30 m分辨率。而后,为保证土地覆被数据无面积变形和投影一致性,选择阿尔伯特等积投影作为基准投影,进行投影变换。
统一的分类体系是土地覆被数据比较评价的前提。GlobleLand30分为10个一级类,未分二级类。CHINA-2010则将土地划分为6个一级类,25个二级类[17],两者的分类体系存在一定差异。例如,GlobleLand30包含“裸地”和“苔原”两个类别,而CHINA-2010则没有区分“裸地”和“苔原”。为解决类似问题,将土地覆被类别聚合为9类。不同土地覆被分类系统间类别的对应关系如表2所示。
Tab. 2 Corresponding relationships of types in different land cover classification system

表2 不同土地覆被分类系统间的类别对应关系

统一类别 GlobleLand30 CHINA-2010
1 耕地 10 农业、园艺用地 11 水田 12 旱地
2 林地 20 覆盖度>30%和10%~30%的林地 21 有林地 23 疏林地 24 其他林地
3 草地 30 覆盖度>10%,以自然草地覆盖为主 31 高覆盖度草地 32 中覆盖度草地 33 低覆盖度草地
4 灌木 40 覆盖度>30%的灌木林 22 灌木林
5 湿地 50 湿地植物和湿地水体 64 沼泽地 45 滩涂 46 滩地
6 水体 60 内陆水体 41 河渠 42 湖泊 43 水库坑塘
7 未利用土地 70 苔原90 裸地 61 沙地 62 戈壁 63 盐碱地65 裸土地 66 裸岩石砾地 67 其他
8 人工地面 80 由人类活动造成的土地改变 51 城镇用地 52 农村居民点 53 其他建设用地
9 永久冰雪 100 终年积雪、冰川 44 永久性冰川雪地
利用表2进行分类系统转换后,GlobleLand30和CHINA-2010的土地覆被空间分布如图1所示。由于在研究区并未出现永久冰雪,因此该地物类型不参与精度评价。
Fig. 1 GlobleLand30 and CHINA-2010 with a common land cover classification system

图1 采用统一分类系统的GlobleLand30与CHINA-2010

3 研究方法

3.1 比较评价

3.1.1 空间一致性评价 在将GlobleLand30与CHINA-2010按地类进行面积汇总比较的基础上,利用一致性比率定量评价两种产品的空间一致性。一致性比率越高,则GlobleLand30与参考数据越相似,相对精度越高。两种产品不同地类的一致性比率Ai及总体一致性比率B定义如下：
A i = T i N i + M i 2 × 100 % （1）
B = ∑ 1 k T i S × 100 % （2）
式中：Mi、Ni分别为土地覆被产品M和土地覆被产品N中第i种土地覆被类别的像元数;Ti为在相同位置两种产品都为土地覆被类别i的像元数;由于参与比较评价的两种数据空间范围及分辨率相同,因此S为任一土地覆被数据的总像元数;k为不同土地覆被类别个数,此处k=8。
3.1.2 空间混淆评价 尽管一致性比率可以从统计角度评估GlobleLand30与CHINA-2010的空间一致性,但无法描述两种土地覆被产品类别的混淆程度及空间分布。在进行土地覆被空间混淆评价时,首先,以CHINA-2010为参考数据计算混淆矩阵,定量分析两种产品不同地类的混淆程度。而后,将GlobleLand30和CHINA-2010叠合分析,以确定地类混淆的空间位置,并从数据源、制图方法、分类系统、光谱特征等多方面深入研究地类混淆原因。

3.2 样本精度评价

在上述评价方法中,通过GlobleLand30与CHINA-2010的空间一致性来描述GlobleLand30的精度。尽管研究表明CHINA-2010的总体精度较高,满足参考数据的精度要求。但由于在目视解译的过程中,不同的解译专家可能会对相同的影像分析获得不同的解译结果,人工操作也会引入一定误差,因此CHINA-2010与实际地物类别仍有一定 差别。
为进一步分析GlobleLand30的总体绝对精度,2015年在研究区范围进行野外实地考察。野外考察采样过程包含了选点、坐标定位、照片拍摄、记录和修正五部分。采样过程根据土地覆被类别构成,随机选取实地考察样本。同时,考虑到GlobleLand30数据基准年为2010年,与考察时间具有一定的时间间隔。因此,利用时间轴在2009-2015年的Google Earth影像对考察样本进行核验,以降低由于样本质量造成的评价误差。通过以上步骤,共获得野外考察样本114个。
此外,针对比较评价法在待评价数据与参考数据分类不一致时,无法为数据使用者提供精确选择依据的不足,结合样本评价法,进一步评估产品在空间不一致区域的制图精度。由于空间不一致区域地形复杂,实地野外考察采样受到交通因素、成本因素的限制较多,在针对空间不一致区域进行精度评价时,选择Google Earth影像作为样本。Google Earth影像由于其定位准确、时相丰富、分辨率高、获取方便、覆被广泛等优势,是精度评价的主要数据源之一[14-16,23]。为减少定位误差及解译误差对样本质量带来的不利影响,在进行样本选择及解译时需遵循以下原则：① 由于Google Earth影像样本的定位误差在15 m左右,GlobleLand30的空间分辨率为30 m,为减少定位误差的影响,样本点选择区域大小为7像元×7像元即210 m×210 m均质区域的中心点。② 为减少数据时相带来的解译误差,主要使用和待评价数据时相一致的2009-2011年间的遥感影像,并参考其他年间的多时相数据。③ 对于部分解译难度较大样本,结合参考其他信息辅助解译。例如,结合其他地图信息区分裸地和人工地面,利用志愿地理信息平台Geo-wiki[24]辅助解译等。④ 采用多人独立解译方式,并在解译结果经商议无法统一时,舍弃该样本。在遵循以上原则的基础上,针对空间不一致区域,按类别混淆面积大小,分层随机采样,确定样本位置,并由不同专家利用Google Earth进行目视解译。野外实地考察及Google Earth样本分布情况如图2所示。
Fig. 2 The distribution of validation samples from field survey and Google Earth

图2 野外考察样本及Google Earth样本分布

在样本评价过程中以总体精度和制图精度为指标衡量产品的质量,公式[25]如下：
OA = T X × 100 % （3）
P A i = T i X i × 100 % （4）
式中：OA为总体精度,是所有被正确识别的样本数T与总样本数X的比值; P A i 为某地类i的制图精度;Ti为被正确识别为第i类地物的样本数;Xi为实际为第i类地物的样本数。判断样本分类是否正确时,为降低分类系统转换误差,认为土地具有多类型特征,当分类结果具有土地某一特征时,即认为分类正确。例如,GlobleLand30定义“苗圃”为“耕地（10）”,CHINA-2010定义“苗圃”为“其他林地（24）”,当实际地类为“苗圃”时,若GlobleLand30分类为“耕地”即分类正确,CHINA-2010分类为“林地”也同样正确。

3.3 高程分带

高程、地表起伏度等地形因素是土地覆被的主要驱动因素之一[26]。在SRTM DEM 90 m数据的基础上,结合中国地貌相关划分标准[27],同时考虑到研究区内平原、盆地的高程多在200 m以下[28],丘陵的高程多在500 m以下,高程>2000 m的地区面积不足全区0.01%的实际地貌情况,将研究区划分为 ≤ 200 m、200~500 m（含500 m）、500~1000 m（含1000 m）和>1000 m四个高程带,研究区高程带分布如图3所示。通过统计两种产品空间一致性在不同高程带的变化,分析高程与GlobleLand 30分类精度之间的关系。
Fig. 3 Zoning map of elevation in the study area

图3 研究区高程分带图

4 结果分析

4.1 比较评价

4.1.1 空间一致性评价 图4描述了GlobleLand30与CHINA-2010在研究区的土地覆被类别面积构成及空间一致性。从全局分析,GlobleLand30与CHINA-2010面积构成相似,总体一致性较高。第一,两土地覆被产品面积构成基本相同,均能够刻画河南省以耕地为主,林地、人工地面、草地为辅,其他土地覆被类别较少的格局。第二,公式（2）计算结果表明,两种产品总体一致性比率为80.20%,具有高度一致性。
Fig. 4 Area ratio and spatial consistency ratio of land cover types

图4 土地覆被类别面积比率与空间一致性比率

尽管GlobleLand30与CHINA-2010总体空间一致性较高,但是从细节来看两种产品在不同土地覆被类别的空间一致性差别很大。其中,耕地一致性比率88.88%,空间一致性最高,并且两种产品对耕地的识别面积几乎相同,约占全省面积的64%;林地一致性比率为74.96%,其中,GlobleLand30林地面积较大,占19%,CHINA-2010林地面积较小,占14%;人工地面的一致性比率为69.92%,但两种产品对人工地面的识别面积几乎相同,这说明漏分和错分现象同时存在;产品对水体和草地的空间一致性较低,一致性比率分别为56.24%和28.34%;对湿地、灌木和未利用土地而言,空间一致性极差,一致性比率分别为8.75%、1.56%和0.01%,但这些类别面积比率仅在0.1%~3%,因此,并未对总体的空间一致性造成明显影响。
综上所述,GlobleLand30与CHINA-2010总体空间一致性高,但不同地类的空间一致性差别较大。其中,核心土地覆被类别耕地、林地、人工地面空间一致性较高,水体、草地次之,面积构成比率小的灌木、湿地、未利用土地空间一致性低。
4.1.2 混淆矩阵评价 利用叠加分析法,获得两种产品在相同位置的土地覆被类别对应关系,并在对叠加结果统计汇总的基础上,建立GlobleLand30与参考数据的土地覆被类别混淆矩阵,获得地类混淆情况如图5所示。在图5中,横坐标代表参考数据的地物类别,纵坐标代表参考数据中属于某个地类的像元在待评价数据GlobleLand30中被标记为各地类的百分比。当横坐标描述的参考数据地物类别与纵坐标标示的土地覆被分类类别不一致时,即认为出现了地类混淆。由图5可以看到,在参考数据中识别为耕地的像元,约有10%被混分为林地、草地及人工地面。在参考数据中识别为水体的像元,约有40%在GlobleLand30中被混分为耕地,少部分被混分为林地、草地、湿地和人工地面。在参考数据中识别为人工地面的像元,约有30%在GlobleLand30被混分为耕地。对于湿地、灌木和未利用土地,GlobleLand30混分情况严重,几乎全部被混分为耕地、林地、草地等主要地类。结合不同土地类别所占面积可知,耕地—林地—草地混分,人工地面—耕地混分,水体—耕地混分现象是影响GlobleLand30精度的主要原因。
Fig. 5 Confusion of land cover types between GlobleLand30 and CHINA-2010

图5 GlobleLand30和CHINA-2010土地覆被类别混淆

4.2 样本精度评价

4.2.1 总体精度评价 野外实地考察精度评价共选取野外实地考察样本114个,其中耕地51个,林地31个,人工地面11个,其他土地覆被类别21个。野外实地考察评价结果表明,GlobleLand30正确分类的样本数目95个,由公式（3）可知,样本正确识别比率即总体精度为83.33%。
4.2.2 空间不一致区域精度评价 在对空间不一致区域进行精度评价时,为保证样本数据质量,依据样本选择原则,按混淆地类面积大小,分层随机选取符合条件的Google Earth验证样本233个。各种混淆地类的样本情况如表3所示。其中,评价区域总面积约为31212 km2,占空间不一致区域总面积的90%以上,是类别混淆的关键区域。此外,由于湿地和未利用土地存在分类系统转换误差,且所占面积比率小,不再对其针对空间不一致区域做精度评价。
Tab. 3 Google Earth validation samples in spatially inconsistent areas

表3 空间不一致区域Google Earth验证样本

混淆地类名称 面积(km2) 样本个数
耕地—人工地面 10973.66 82
耕地—林地 6192.77 46
耕地—草地 3945.84 30
草地—林地 5006.21 37
耕地—水体 1617.98 12
灌木—林地;灌木—耕地;
灌木—草地		 3475.79 26
合计 31212.25 233
利用公式（3）和公式（4）获得空间不一致区域制图精度和总体精度结果如图6所示。整体而言,GlobleLand30在空间不一致区域的总体精度略低于CHINA-2010,其中,GlobleLand30在空间不一致区域的总体精度为45%,CHINA-2010在空间不一致区域的总体精度为50%。对单一地类而言,两种产品在不同土地覆被类别的识别能力上各有优势。其中,GlobleLand30对林地的制图精度较高,为70%,CHINA-2010对林地的制图精度较低,仅为21%。但GlobleLand30在灌木、水体、草地、人工地面和耕地的制图精度分别为17%、18%、27%、34%和46%,均低于CHINA-2010,仍有精度提升空间。
Fig. 6 Producer's accuracy and overall accuracyin spatially inconsistent areas

图6 空间不一致区域的制图精度与总体精度

由比较评价可知,两种产品空间一致性区域占80.20%,若假定空间一致则分类正确,结合空间不一致区域的精度分析结果可知,GlobleLand30主要地类的分类精度仅略低于CHINA-2010。此外,GlobleLand30在林地的制图精度高于CHINA-2010,更适用于森林变化监测领域。

4.3 空间混淆分析

在精度分析的基础上,分别从类别混淆空间分布情况、土地覆被复杂度、高程分带等多方面分析类别混淆出现的原因。
4.3.1 类别混淆情况分析 图7直观描述了研究区内GlobleLand30与CHINA-2010主要土地覆被类别混淆空间分布情况。结合土地覆被空间分布情况（图1）可以看到,GlobleLand30和CHINA-2010均能描述分布在平原地区的耕地,分布在山地丘陵地区的林地,流经省内的黄河、淮河、汝河等多条干流和零散分布在全省各地的人工地面。但两种产品存在不可忽略的差异,例如,CHINA-2010的水体脉络明显,而GlobleLand30的水体较少;在CHINA-2010中灌木主要分布于研究区东南边界及西部山地丘陵地区,而在GlobleLand30中则主要分布于研究区西北地区,空间位置存在巨大差异;植被类别混分现象主要分布在平原与山地的过渡区域;人工地面与耕地的混分区域则零星分布在研究区东部平原地区及南阳盆地。对主要地类混淆,分析原因如下：
Fig. 7 Spatial distributions of confusion for main land cover types

图7 主要土地覆被类别混淆的空间分布

（1）耕地、林地与草地的混淆情况如图7a所示。可以看出,在耕地广泛分布的平原和盆地、林地广泛分布的山地丘陵地区,空间一致性较好。但在山地与平原、山地与盆地的过渡地带,存在明显的耕地、林地、草地混分现象,混分面积约15152.39 km2,占研究区总面积的9.15%。分析原因,该区域多见耕地、林地、草地交错现象,三者在光谱上又具有一定的相似性,增加了识别难度。
（2）水体混淆情况空间分布如图7b所示。参考数据识别水体3505.75 km2,GlobleLand30仅识别1943.30 km2,水体漏分现象明显。水体的土地覆被类别混淆主要出现在水体和耕地之间。从空间分布来看,水体识别一致性区域主要分布于河流干流和大型水库,识别不一致地区,主要分布于河流支流。在1 ∶ 10万比例尺下查看水体混分情况空间分布图谱,并结合Google Earth影像进行细节分析发现,水体—耕地类别混淆原因有以下两方面：一方面,参考数据由目视解译完成,结合了大量专家知识,因此河流脉络完整,水体识别能力强。而GlobleLand30受单时相数据源影响显著,只反映某一特定时间的水体状况[29]。若选择的影像数据正处于枯水期,会降低水体识别能力。另一方面,GlobleLand30更能反映由于环境影响造成的水资源流失情况。对不一致区域抽样分析发现,存在部分区域,水资源流失严重,尽管河道存在,但水体变窄甚至消失,而这些现象在CHINA-2010中表现不突出。综上,水体受年际、季节变化影响较大,但理想多时相30 m分辨率Landsat影像获取困难,需要进一步结合多源数据提高制图精度。
（3）人工地面混淆情况空间分布如图7c所示。该地类混分现象零星分布于平原和盆地,主要出现在人工地面和耕地之间,约占研究区总面积的6.63%,混分现象明显。对斑块较小的村庄、城市边界区的识别错误是造成分类误差的主要原因。
（4）灌木与其他土地覆被类别混分情况如图7d所示。两种产品对灌木空间一致性极差,其中,参考数据识别灌木3218.54 km2,而GlobleLand30仅识别346.19 km2。地类混分主要以灌木—林地混分为主,也存在灌木—耕地、灌木—草地的混分现象。灌木与其他植被混分的主要原因有产品分类判定标准差异和灌木—林地—草地之间光谱相似两方面。第一,GlobleLand30和CHINA-2010在灌木的郁闭度和高度定义上存在一定的差异,这会造成分类标准有所不同。对土地覆被类别定义的理解差异也造成了分类结果的不同。例如,研究区南部山区存在部分区域,GlobeLand30识别为耕地,CHINA-2010识别为灌木,结合Google Earth影像进行目视解译发现,该区域主要地类为灌木茶园;西北部太行山区丘陵地带存在一较大斑块,GlobleLand30识别为灌木,CHINA-2010识别为草地,经目视解译和野外实地调查采样可知,该区域以灌木、草地交错分布为主。第二,灌木—林地、灌木—草地的光谱较为相似,存在“异物同谱”的现象,单纯利用光谱信息区分,容易产生错分现象,需结合物候信息、地形信息、纹理信息进一步提升分类精度。
（5）研究区内湿地面积较小,两种土地覆被产品湿地面积仅占全省面积的0.1%~0.3%,一致性比率仅为8.75%。首先,湿地生态系统复杂,且界定无统一标准规范,造成遥感湿地分类难度较高。CHINA-2010中并未定义单独的湿地类型,文中将和湿地相关的沼泽地、滩涂和滩地合并为湿地,而在GlobleLand30中湿地不仅包括以上三种还包括森林/灌木湿地等,两者定义有所不同,分类标准亦不统一。再者,湿地位于陆地和水域的交汇处,土壤湿润,部分有表层积水,在进行土地覆被分类时难以与水体区分。目前,利用遥感信息进行自动湿地分类的精度较低,仍有较大提高空间。
（6）未利用土地的识别一致性几乎为0%。除了分类定义不完全一致的原因外,未利用土地斑块小,面积不足河南省面积的0.1%,也是未利用土地分类混淆的主要原因。
4.3.2 土地覆被复杂度对类别混淆影响分析 利用ArcGIS 10.2的焦点统计功能,以矩形分析窗口内不同土地覆被类别的个数,为该窗口中心像元的土地覆被复杂度。研究区共8种不同地类,土地覆被复杂度在1~8。根据土地覆被复杂度进行分级划分,当某像元的土地覆被复杂度为n时,代表以该像元为中心的矩形区域内,有n种不同的土地覆被类别,即该像元属于n级区域。矩形分析窗口大小不同,土地覆被复杂度的结果亦不相同。为全面分析土地覆被复杂度对空间一致性的影响,分别以多种分析窗口计算GlobleLand30产品的土地覆被复杂度（300 m×300 m,600 m×600 m,900 m×900 m,1200 m×1200 m,1500 m×1500 m）。当窗口大小为1200 m×1200 m时,研究区土地覆被复杂度分级情况如图8所示。
Fig. 8 Hierarchical map of land cover

图8 土地覆被复杂度分级图(1200 m×1200 m)

对土地覆被复杂度专题图与土地覆被类别混淆空间分布图叠加分析,依据式（2）计算不同土地覆被复杂度级别区域的空间一致性比率（图9）。在不同的分析窗口下,GlobleLand30和CHINA-2010的空间一致性,均随土地覆被复杂度的增加而下降。在土地覆被复杂度为1级的均质区域,空间一致性比率高达90%,而随着土地覆被复杂度的增高,空间一致性比率下降至10%~52%。这表明在某个区域土地覆被类别越丰富,复杂度越高,识别难度越高,空间一致性越低。即土地覆被复杂度是影响精度的重要因素之一。在后续的分类研究中,可着重针对土地覆被复杂度较高的区域,进一步提升分类精度。
Fig. 9 Spatial consistency at different land cover complexity levels

图9 不同土地覆被复杂度级别的空间一致性

4.3.3 高程对类别混淆影响分析 两种产品空间一致性在不同高程带的分布如表4所示。① 在高程 ≤ 200 m的区域,一致性比率达85.56%（分类一致面积占该高程带总面积比例,下同）。该地带地势平坦,光谱可分性强、分类一致的耕地（72.57%）和人工地面（10.83%）占绝对优势,但由于GlobleLand30对部分乡村居民点的漏分和两产品在城镇边界分类不一致问题造成了约8.53%的耕地—人工地面混淆。② 在200~500 m高程带,一致性比率仅为63.81%。该区域耕地分类一致性较高（37.79%）,林地（17.58%）次之。混淆地类主要包括：耕地—林地混淆（9.00%）,草地—林地混淆（8.29%）,耕地—草地混淆（7.32%）,灌木—其他植被混淆（5.07%）和耕地—人工地面混淆（4.43%）。③ 在500~1000 m高程带,一致性比率亦较低,为69.69%。该区域分类一致的林地上升至48.06%,耕地比率下降为17.35%,耕地—林地—草地—灌木之间的植被混淆情况和200~500 m高程带较为相似,共占该区域的28.38%。④ 在高程>1000 m的高程带,一致性比率较高,为85.57%。分类一致的林地高达81.50%,混淆主要包括草地—林地混淆（7.07%）,耕地—林地混淆（2.96%）和灌木—林地混淆（2.35%）。该区域海拔较高,人类影响较少,土地覆被类别单一,因此空间一致性较高。
Tab. 4 Spatial consistency at different elevation zones

表4 不同高程带的空间一致性

高程带（m） ≤200 200~500 500~1000 >1000
面积（km2） 111794 27927 17596 7996
一致性比率（%） 85.56 63.81 69.69 85.57
优势地类 耕地>人工地面>水体 耕地>林地>草地>灌木 林地>耕地>草地>灌木 林地为主
主要混淆地类 耕地—人工地面 耕地—林地—草地—灌木
耕地—人工地面		 耕地—林地—草地—灌木 林地—草地
研究区内200~500 m、500~1000 m高程带主要地貌为低海拔丘陵,小起伏山地和较少部分的中起伏低山组成,处于人工生态和自然生态组合区域,土地覆被复杂度高,林地、耕地、草地、灌木交错分布,且四种土地类型光谱较为相似,因此空间一致性低。在 ≤ 200 m和>1000 m的高程带,优势地类较少且地类间光谱可分性强,因此空间一致性较高。综上所述,研究区内空间一致性随高程增高呈现先降低后提升的变化趋势,且在200~1000 m高程过渡地带较低。

5 结论与讨论

为给数据使用者合理使用GlobleLand30提供科学依据,以河南省为研究区,利用精度较高的CHINA-2010对GlobleLand30进行空间一致性交互验证,并采用Google Earth样本数据对不一致地区做进一步分析,最后利用野外实地考察样本对GlobleLand30进行总体精度评价。虽然,本文的精度评价是基于区域尺度的,但该评价方法对全国尺度同样适用。综合研究区评价结果,主要结论如下：
（1）GlobleLand30在区域尺度具有较高的总体精度。通过比较评价可知,GlobleLand30与CHINA-2010的空间一致性为80.20%,均能描述研究区以耕地为主,林地、人工地面、草地为辅,其他土地覆被类别较少的格局。通过野外实地考察样本验证可知,GlobleLand30在研究区总体精度高达83.33%。就全局而言,无论是采用比较评价还是通过野外实地考察样本评价,结果均表明,GlobleLand30在研究区的总体精度在80%以上。
（2）在总体分析的基础上,将研究区划分为分类一致区域和分类不一致区域。由空间一致性分析可知,GlobeLand30与CHINA-2010在不同土地覆被类别的空间一致性差异较大。其中,耕地一致性高达88.88%,林地、人工地面和水体一致性依次减少在56.24%~74.96%,草地、灌木、湿地、未利用土地的一致性最差在0.01%~28.34%。为明确GlobleLand30的优势与不足,对空间不一致区域进一步分析。结果表明,当两种产品分类结果不一致时,GlobleLand30的总体制图精度（45%）仅略低于CHINA-2010（50%）,且在林地的制图精度（70%）上高于CHINA-2010（21%）,但GlobleLand30在耕地、草地、人工地面、水体和灌木的制图精度均低于CHINA-2010。由于GlobleLand30对不同地物的精度差异较大,在使用单一土地覆被类别数据时,应根据精度科学选择。
（3）GlobleLand30的精度随土地覆被复杂度的增强而迅速下降并且在高程过渡区精度较低。但高程仅是影响土地覆被分类的地形因素之一,在后续的研究工作中可进一步关注坡度、坡向、高程起伏度等其他地形特征对土地覆被分类精度的影响,为土地覆被分类过程中的特征选择提供参考。
综上所述,GlobleLand30在中国省级区域尺度具有较高精度,仅略低于CHINA-2010,可在区域尺度为气候变化等相关研究领域提供基础数据。但遥感土地覆被制图是长期的工作,在未来的高分辨率土地覆被产品建设时,可从以下几个方面进一步提高制图精度。第一,针对光谱相似带来的林地—草地—灌木混分问题,建议在结合地形特征的基础上,引入区域植被物候特征数据,以减少“异物同谱”带来的误差。第二,针对水体、耕地混分问题,考虑引入多源、多时相数据,以改善由于Landsat多时相数据不理想带来的分类误差。例如,引入重复观测周期短的中分辨率MODIS时序数据,以增加水体丰水期和植被关键物候期的信息。第三,针对高程过渡区域精度较低问题,考虑高程、地形起伏度、坡度、坡向等地形特征,提升识别能力。第四,利用数据分析与融合技术,在保留现有产品精度较高部分的基础上,对空间不一致区域进行有针对性的改进。
需要指出的是：① 比较评价法利用空间一致性描述产品的相对精度,具有快速、低成本、逐像元检验的特点。但参考产品与真实地表覆被仍有不同,尽管利用Google Earth样本对不一致区域进行了补充分析。然而两产品分类一致,仅能以较高置信度描述产品的准确性,和野外考察获取的绝对精度仍有一定差距。在未来的研究中,建立具有合理分类体系、标准统一、可重用、可共享的训练样本和评价样本数据库,是提高中国土地覆被分类精度及评价置信度的重要工作。② 在比较评价过程中,根据评价产品和参考数据的分类系统定义,进行了映射转换和类别归并,但由于GlobleLand30和CHINA-2010在灌木、湿地和未利用土地的定义存在差距和模糊性,会给一致性评价带来一定影响。

The authors have declared that no competing interests exist.

参考文献

原文顺序 | 文献年度倒序 | 文中引用次数倒序
[1]
Wulder M A, Coops N C.Make earth observations open access. Nature, 2014, 513(7516): 30-31.

[2]
Giri C, Pengra B, Long J, et al.Next generation of global land cover characterization, mapping, and monitoring. International Journal of Applied Earth Observation and Geoinformation, 2013, 25(4): 30-37.Land cover change is increasingly affecting the biophysics, biogeochemistry, and biogeography of the Earth's surface and the atmosphere, with far-reaching consequences to human well-being. However, our scientific understanding of the distribution and dynamics of land cover and land cover change (LCLCC) is limited. Previous global land cover assessments performed using coarse spatial resolution (30002m–102km) satellite data did not provide enough thematic detail or change information for global change studies and for resource management. High resolution (653002m) land cover characterization and monitoring is needed that permits detection of land change at the scale of most human activity and offers the increased flexibility of environmental model parameterization needed for global change studies. However, there are a number of challenges to overcome before producing such data sets including unavailability of consistent global coverage of satellite data, sheer volume of data, unavailability of timely and accurate training and validation data, difficulties in preparing image mosaics, and high performance computing requirements. Integration of remote sensing and information technology is needed for process automation and high-performance computing needs. Recent developments in these areas have created an opportunity for operational high resolution land cover mapping, and monitoring of the world. Here, we report and discuss these advancements and opportunities in producing the next generations of global land cover characterization, mapping, and monitoring at 30-m spatial resolution primarily in the context of United States, Group on Earth Observations Global 3002m land cover initiative (UGLC).

DOI

[3]
Fritz S, See L, You L, et al.The need for improved maps of global cropland. Eos Transactions, 2013, 94(3): 31-32.Food security is a key global concern. By 2050, the global population will exceed 9 billion, and a 50% increase in annual agricultural output will be required to keep up with demand. There are significant additional pressures on existing agricultural land through increased competition from the biofuel sector and the need to elevate feed production, which is being driven by higher levels of meat consumption in low- and middle-income countries

DOI

[4]
王慧敏, 曾永年. 青海高原东部土地利用的低碳优化模拟: 以海东市为例. 地理研究, 2015, 34(7): 1270-1284.<p>人类活动导致的土地利用变化及其碳效应在区域及全球碳循环研究中具有举足轻重的作用。以位于青海高原东部的海东市为研究区,在探讨土地利用碳循环系数的基础上,预测海东市2020年低碳情景下的土地利用结构,并运用CLUE-S模型模拟了海东市土地利用低碳优化空间格局,然后对比分析低碳情景与规划情景下,海东市2020年土地利用格局的差异。研究表明：① 尽管低碳情景下研究区域水浇地和旱地持续减少,但相对于规划情景,优质耕地的流失量减少。② 有林地稳定增加,疏林草地缓慢增加,区域固碳能力逐步提升。③ 建设用地适度扩张,增加幅度小于规划情景,其中城镇用地相对于规划情景碳排放减少14.03万t。低碳视角的区域土地利用结构调整与空间布局优化,为区域土地利用科学规划与管理提供决策支持。</p>

DOI

[Wan Huimin, Zeng Yongnian.Land use optimization simulation based on low-carbon emissionsin eastern part of Qinghai Plateau. Geographical Research, 2015, 34(7): 1270-1284.]

[5]
廖顺宝, 秦耀辰. 草地理论载畜量调查数据空间化方法及应用. 地理研究, 2014, 33(1): 179-190.以20世纪80年代全国草地资源调查的理论载畜量数据和同时期的全国1：25 万土地覆被数据为基础，按省区构建理论载畜量与不同草地类型面积之间的关系模型。根据属性数据空间化思路，利用模型计算与残差修正相结合的技术方法，将80年代的全国草地理论载畜量调查数据从市县单元转换到公里网格尺度上，同时计算2005年全国公里网格尺度的草地理论载畜量。在此基础上，从公里网格和省级行政区划两个尺度上分析全国草地理论载畜量在近30年间（20世纪80年代-2005年）的变化。结果表明，近30年来中国草地生态系统正在退化、萎缩，草地总生产力下降。为实现中国社会经济的可持续发展，必须加强对草地生态系统的保护，保持草地生态系统的健康、良性发展和草地资源的合理开发利用。

DOI

[Liao Shunbao, Qin Yaochen.A spatialization method for survey data of theoretical stock-carrying capacity of grassland in China and its application. Geographical Research, 2014, 33(1): 179-190.]

[6]
冉有华, 李新, 卢玲. 四种常用的全球1 km土地覆盖数据中国区域的精度评价. 冰川冻土, 2009, 31(3): 490-500.lt;FONT face=Verdana>精确的全球及区域土地覆盖数据是陆地表层过程研究的重要基础. 定量的评价已有数据的质量将有助于未来更好的开展土地覆盖制图. 基于一个新的分类系统（森林、灌木草地、农田、裸地、城市、湿地和水体），以中国1：10万土地利用数据为参考数据，从类型面积一致性、空间一致性两方面对4各全球土地覆盖数据集在中国区域的分类精度进行了评价，包括美国地质调查局为国际地圈-生物圈计划的全球土地覆盖数据集（IGBPDISCover）；美国马里兰大学的全球土地覆盖数据集（UMd）；欧盟联合研究中心（JRC）空间应用研究所（SAI）的2000年全球土地覆盖数据产品（GLC2000）；MODIS　2000年的土地覆盖数据产品（MOD12Q1）. 并对4种土地覆盖产品误差的空间和类型分布进行了分析. 结果表明: 在4种土地覆盖分类产品中，GLC2000和MODIS土地覆盖数据有更高的整体分类精度，IGBP数据的整体分类精度次之，但是3种数据在局部都存在明显的分类错误；UMd的分类精度整体最低. 通过对4种数据分类精度的空间和类型分布规律的分析，认为4种数据都不能很好的满足陆地表层过程模拟的需要. 建议发展土地覆盖类型决策融合方法，将现存多源土地覆盖分类信息融合起来，制备更高精度的中国土地覆盖分类图. </FONT>

DOI

[Ran Youhua, Li Xin, Lu Ling.Accuracy evaluation of the four remote sensing based land cover products over China. Journal of Glaciology and Geocryology, 2009, 31(3): 490-500.]

[7]
Chen Jun, Ban Yifang, Li Songnian.Open access to earth land-cover map. Nature, 2014, 514(7523): 434-434.

[8]
Han Gang, Chen Jun, He Chaoying, et al.A web-based system for supporting global land cover data production. ISPRS Journal of Photogrammetry and Remote Sensing, 2015, 103(5): 66-80.This web-based system not only facilitates the integration of heterogeneous data and services required by GLC data production, but also provides online access, visualization and analysis of the images, ancillary data and interim 3002m global land-cover maps. The system further supports online collaborative quality check and verification workflows. It has been successfully applied to China’s 30-m resolution GLC mapping project, and has improved significantly the efficiency of GLC data production and verification. The concepts developed through this study should also benefit other GLC or regional land-cover data production efforts.

DOI

[9]
陈军, 陈晋, 廖安平, 等. 全球30 m地表覆盖遥感制图的总体技术. 测绘学报, 2014, 43(6): 551-557.lt;p>本文针对全球30米分辨率地表覆盖遥感制图这一世界性难题，提出了以多源影像最优化处理、参考资料服务化整合、覆盖类型精细化提取、产品质量多元化检核为主线的总体研究思路，研发了影像几何与辐射重建、异质异构服务化集成、对象化分层分类、知识化检核处理等主体技术方法；用于制定了相应数据产品规范、生产技术规范，研发了多项生产型软件，用于研制了2000和2010两个基准年的全球）30米地表覆盖数据产品，将空间分辨率提高了1个数量级。</p>

[Chen Jun, Chen Jin, Liao Anping, et al.Concepts and key techniques for 30 m global land cover mapping. Acta Geodaetica et Cartographica Sinica, 2014, 43(6): 551-557.]

[10]
冉有华, 李新. 全球第一个综合高分辨率土地覆盖图: 中国30 m分辨率全球土地覆盖图评述. 中国科学: 地球科学, 2015, 45(8): 1243-1244.正土地覆盖是研究人类活动与气候变化复杂的交互作用的重要基础.遥感土地覆盖制图是一个传统但仍然非常活跃的领域.在过去的45年中,全世界大约有3万多篇关于土地覆盖制图的文章发表,其中包括中文论文约1万篇.然而,这些研究绝大多数聚焦在局地尺度,极少数的研究关注全球土地覆盖制图,而且其空间分辨率较粗(300~1000 m).

[Ran Youhua, Li Xin.The world's first comprehensive high resolution land cover map: A survey of china 30 m resolution global land cover map. Scientia Sinica Terrae, 2015, 45(8): 1243-1244.]

[11]
廖安平, 彭舒, 武昊, 等. 30m全球地表覆盖遥感制图生产体系与实践. 测绘通报, 2015, (10): 4-8.在以"多源影像最优化处理、参考资料服务化整合、覆盖类型精细化提取、产品质量多元检核"为主线的总体研究基础上,依托生产技术规范体系、全过程质量控制手段和支持环境,通过30m地表覆盖产品和技术设计、多源影像资料收集整合处理、分区按类型地表覆盖数据提取组织实施及数据产品集成与优化,构建了工程化的30m全球地表覆盖遥感制图生产体系,实现了预期的产品指标,完成了2000和2010两个基准年的30m地表覆盖数据产品研制。通过精度评价,该套数据产品分类精度达到80%以上。该生产体系的构建为开展较高分辨率全球地表覆盖数据产品研制、细化、更新奠定了基础,为开展大规模遥感影像信息提取、表达和应用起到了示范作用。

DOI

[Liao Anping, Peng Shu, Wu Hao, et al.The production system of 30 m global land cover mapping and its application. Bulletin of Surveying and Mapping, 2015, (10): 4-8.]

[12]
Chen Jun, Chen Jin, Liao Anping, et al.Global land cover mapping at 30 m resolution: A POK-based operational approach. ISPRS Journal of Photogrammetry and Remote Sensing, 2015, 103(5): 7-27.Over 10,000 Landsat-like satellite images are required to cover the entire Earth at 30 m resolution. To derive a GLC map from such a large volume of data necessitates the development of effective, efficient, economic and operational approaches. Automated approaches usually provide higher efficiency and thus more economic solutions, yet existing automated classification has been deemed ineffective because of the low classification accuracy achievable (typically below 65%) at global scale at 30 m resolution. As a result, an approach based on the integration of pixel- and object-based methods with knowledge (POK-based) has been developed. To handle the classification process of 10 land cover types, a split-and-merge strategy was employed, i.e. firstly each class identified in a prioritized sequence and then results are merged together. For the identification of each class, a robust integration of pixel-and object-based classification was developed. To improve the quality of the classification results, a knowledge-based interactive verification procedure was developed with the support of web service technology. The performance of the POK-based approach was tested using eight selected areas with differing landscapes from five different continents. An overall classification accuracy of over 80% was achieved. This indicates that the developed POK-based approach is effective and feasible for operational GLC mapping at 30 m resolution.

DOI

[13]
Brovelli M, Molinari M, Hussein E, et al.The First comprehensive accuracy assessment of GlobeLand30 at a national level: Methodology and results. Remote Sensing, 2015, 7(4): 4191-4212.As result of the lobal Land Cover Mapping at Finer Resolution project led by National Geomatics Center of China (NGCC), one of the first global land cover datasets at 30-meters resolution (GlobeLand30) has been produced for the years 2000 and 2010. The first comprehensive accuracy assessment at a national level of these data (excluding some comparisons in China) has been performed on the Italian area by means of a benchmarking with the more detailed land cover datasets available for some Italian regions. The accuracy evaluation was based on the cell-by-cell comparison between Italian maps and the GlobeLand30 in order to obtain the confusion matrix and its derived statistics (overall accuracy, allocation and quantity disagreements, user and producer accuracy), which help to understand the classification quality. This paper illustrates the adopted methodology and procedures for assessing GlobeLand30 and reports the obtained statistics. The analysis has been performed in eight regions across Italy and shows very good results: the comparison of the datasets according to the first level of Corine Land Cover nomenclature highlights overall accuracy values generally higher than 80%.

DOI

[14]
孟雯, 童小华, 谢欢, 等. 基于空间抽样的区域地表覆盖遥感制图产品精度评估: 以中国陕西省为例. 地球信息科学学报, 2015, 17(6): 742-749.地表覆盖数据是研究气候变化、生态环境、地理国情和人文经济等方面不可或缺的基础信息,因此其质量的优劣将直接影响相关决策的可靠性。本文针对我国研制的首套30 m分辨率全球地表覆盖数据产品,通过分析其海量、多维、非均质等空间特点,对传统的制图产品精度评估方法提出了改进:以地表类型进行分层抽样,样本量的计算采用优化模型并以地类所占面积比为权重逐层分配,样本的布设则考虑层内对象之间的空间相关性,在地表数据自然分布的基础上,通过分析空间相关性指数来提高样本的代表性和精度评估结果的可靠性。因此,本文提出针对区域地表覆盖遥感制图产品的空间分层抽样方法,其将精度评估分成抽样方案(样本的定量估计)和布设方案(样本的空间布设)2个部分,并以中国陕西省地表覆盖产品为例进行区域精度评估实验分析,从全区7大类地表类型数据中抽取1467个样本,经过样本检验与精度计算,该区域产品的总体精度为79.96%,Kappa系数为0.74。实验结果表明,本文提出的基于空间抽样的精度评估方法可行可靠,实验区域产品质量较好,并为后续针对全球范围的地表覆盖产品精度评估方法提供了参考。

DOI

[Meng Wen, Tong Xiaohua, Xie Huan, et al.Accuracy assessment for regional land cover remote sensing mapping product based on spatial sampling: A case study of Shanxi province. Journal of Geoinformation Science, 2015, 17(6): 742-749.]

[15]
Zhao Yuanyuan, Gong Peng, Yu Le, et al.Towards a common validation sample set for global land-cover mapping. International Journal of Remote Sensing, 2014, 35(13): 4795-4814.Validating land-cover maps at the global scale is a significant challenge. We built a global validation data-set based on interpreting Landsat Thematic Mapper (TM) and Enhanced TM Plus (ETM+) images for a total of 38,664 sample units pre-determined with an equal-area stratified sampling scheme. This was supplemented by MODIS enhanced vegetation index (EVI) time series data and other high-resolution imagery on Google Earth. Initially designed for validating 30 m-resolution global land-cover maps in the Finer Resolution Observation and Monitoring of Global Land Cover (FROM-GLC) project, the data-set has been carefully improved through several rounds of interpretation and verification by different image interpreters, and checked by one quality controller. Independent test interpretation indicated that the quality control correctness level reached 90% at level 1 classes using selected interpretation keys from various parts of the USA. Fifty-nine per cent of the samples have been verified with high-resolution images on Google Earth. Uncertainty in interpretation was measured by the interpreter's perceived confidence. Only less than 7% of the sample was perceived as low confidence at level 1 by interpreters. Nearly 42% of the sample units located within a homogeneous area could be applied to validating global land-cover maps whose resolution is 500 m or finer. Forty-six per cent of the sample whose EVI values are high or with little seasonal variation throughout the year can be applied to validate land-cover products produced from data acquired in different phenological stages, while approximately 76% of the remaining sample whose EVI values have obvious seasonal variation was interpreted from images acquired within the growing season. While the improvement is under way, some of the homogeneous sample units in the data-set have already been used in assessing other classification results or as training data for land-cover mapping with coarser-resolution data.

DOI

[16]
Clark M L, Aide T M, Grau H R, et al.A scalable approach to mapping annual land cover at 250 m using MODIS time series data: A case study in the dry Chaco ecoregion of south America. Remote Sensing of Environment, 2010, 114(11): 2816-2832.Land use and land cover (LULC) maps from remote sensing are vital for monitoring, understanding and predicting the effects of complex human ature interactions that span local, regional and global scales. We present a method to map annual LULC at a regional spatial scale with source data and processing techniques that permit scaling to broader spatial and temporal scales, while maintaining a consistent classification scheme and accuracy. Using the Dry Chaco ecoregion in Argentina, Bolivia and Paraguay as a test site, we derived a suite of predictor variables from 2001 to 2007 from the MODIS 250 m vegetation index product (MOD13Q1). These variables included: annual statistics of red, near infrared, and enhanced vegetation index (EVI), phenological metrics derived from EVI time series data, and slope and elevation. For reference data, we visually interpreted percent cover of eight classes at locations with high-resolution QuickBird imagery in Google Earth. An adjustable majority cover threshold was used to assign samples to a dominant class. When compared to field data, we found this imagery to have georeferencing error < 5% the length of a MODIS pixel, while most class interpretation error was related to confusion between agriculture and herbaceous vegetation. We used the Random Forests classifier to identify the best sets of predictor variables and percent cover thresholds for discriminating our LULC classes. The best variable set included all predictor variables and a cover threshold of 80%. This optimal Random Forests was used to map LULC for each year between 2001 and 2007, followed by a per-pixel, 3-year temporal filter to remove disallowed LULC transitions. Our sequence of maps had an overall accuracy of 79.3%, producer accuracy from 51.4% (plantation) to 95.8% (woody vegetation), and user accuracy from 58.9% (herbaceous vegetation) to 100.0% (water). We attributed map class confusion to limited spectral information, sub-pixel spectral mixing, georeferencing error and human error in interpreting reference samples. We used our maps to assess woody vegetation change in the Dry Chaco from 2002 to 2006, which was characterized by rapid deforestation related to soybean and planted pasture expansion. This method can be easily applied to other regions or continents to produce spatially and temporally consistent information on annual LULC.

DOI

[17]
刘纪远, 匡文慧, 张增祥, 等. 20世纪80年代以来中国土地利用变化的基本特征与空间格局. 地理学报, 2014, 24(2): 3-14.

DOI

[Liu Jiyuan, Kuang Wenhui, Zhang Zengxiang, et al.Spatiotemporal characteristics, patterns and causes of land use changes in China since the late 1980s. Acta Geographica Sinca, 2014, 24(2): 3-14.]

[18]
侯鹏敏, 廖顺宝, 姬广兴, 等. 中比例尺土地覆盖遥感制图质量的野外验证与评价: 以河南省为例. 河南大学学报: 自然科学版, 2016, 46(2): 139-148.精确的土地覆盖数据是进行资源环境研究的重要基础,而土地覆盖数据的精度分析是数据应用的前提与保证.通过对河南省的实地考察,选择114个野外考察样点,将样点实际土地覆盖类型与制图中相应土地覆盖类型进行对照,采用分级计分方法,评价全国1∶25万土地覆盖遥感制图在该地区的制图质量.评价结果表明该数据集在河南省范围内的总体制图精度为74.56％.除森林、草地大类精度略低于60％以外,农田、聚落及湿地水体制图精度都较高,其中农田正确率达到90％.实地考察发现,2005年左右的全国1∶25万土地覆盖遥感制图基本反映了河南省土地覆盖的区域分布状况,在大中尺度区域性研究中具有较高的应用价值.总结了基于中比例尺土地覆盖制图考察时需要考虑的不同情况,以期为今后相似研究工作提供参考.

DOI

[Hou Pengmin, Liao Shunbao, Ji Guangxing et al. Field validation and evaluation for accuracy of medium scale land cover remote sensing mapping in Henan province. Journal of Henan University: Natural Science, 2016, 46(2): 139-148.]

[19]
Zeng Tian, Zhang Zengxiang, Zhao Xiaoli, et al.Evaluation of the 2010 MODIS collection 5.1 land cover type product over China. Remote Sensing, 2015, 7(2): 1981-2006.Although the MODIS Collection 5.1 Land Cover Type (MODIS v5.1 LCT) product is one of the most recent global land cover datasets and has the shortest updating cycle, evaluations regarding this collection have not been reported. Given the importance of evaluating global land cover data for producers and potential users, the 2010 MODIS v5.1 LCT product IGBP (International Geosphere-Biosphere Programme) layer was evaluated based on two grid maps at scales of 100-m and 500-m,which were derived by rasterizing the 2010 data from the national land use/cover database of China (NLUD-C). This comparison was conducted based on a new legend consisting of nine classes constructed based on the definitions of classes in the IGBP and NLUD-C legends. The overall accuracies of the aggregated classification data were 64.62% and 66.42% at the sub-pixel and pixel scales, respectively. These accuracies differed significantly in different regions. Specifically, high-quality data were obtained more easily for regions with a single land cover type, such as Xinjiang province and the northeast plain of China. The lowest accuracies were obtained for the middle of China, including Ningxia, Shaanxi, Chongqing, Yunnan and Guizhou. At the sub-pixel scale, relatively high producer and user accuracies were obtained for cropland, grass and barren regions; the highest producer accuracy was obtained for forests, and the highest user accuracy was obtained for water bodies. Shrublands and wetlands were associated with low producer and user accuracies at the sub-pixel and pixel scales, of less than 10%. Based on dominant-type reference data, the errors were classified as mixed-pixel errors and labeling errors. Labeling errors primarily originated from misclassification between grassland and barren lands. Mixed pixel errors increased as the pixel diversity increased and as the percentage of dominant-type sub-pixels decreased. Overall, mixed pixels were sources of error for most land cover types other than grassland and barren lands; whereas labeling errors were more prevalent than mixed pixel errors when considering all of the land cover data over China, due to the large amount of misclassification between the pure pixels of grassland and barren lands. Next, the accuracy of cropland/natural vegetation mosaics was assessed based on the qualitative (a mosaic of croplands, forests, shrublands, and grasslands) and quantitative (no single component composes more than 60% of the landscape) parts in the definition, which resulted in accuracies of 91.43% and less than 19.26%, respectively. These results are summarized with their implications for the development of the next generation of MCD12Q1 data and with suggestions for potential users of MCD12Q1 v5.1.

DOI

[20]
Perez-Hoyos A, Javier Garcia-Haro F, Valcarcel N. Incorporating sub-dominant classes in the accuracy assessment of large-area land cover products: Application to GlobCover, MODISLC, GLC2000 and CORINE in Spain. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(1): 187-205.Various global land cover (LC) datasets have been produced from remote sensing data in response to the need for information about LC. Nevertheless, the potential use of global LC products is often very limited by the lack of detailed accuracy information at regional to national level. This paper proposes a methodology for performing accuracy assessment of large-area LC products, which takes into account a number of factors arising from intrinsic characteristics of LC, such as thematic uncertainty that results from the partial overlap in legend definitions and lack of homogeneity within reference and classification data.<br/>The approach compares the LC pixel label not only with the dominant reference label but also with sub-dominant LC types within the extent of the sampling unit. The methodology was illustrated in Spain using four LC datasets (GlobCover, MODIS land cover (MODISLC), GLC2000 and CORINE). The variety of reference label data offered by a detailed national database, namely SIOSE, supported several different fuzzy agreement definitions in order to derive unbiased estimates of accuracy measures. CORINE followed by GLC2000 showed the highest accuracy scores, whereas GlobCover and MODISLC showed the lowest scores. Nevertheless, the fuzzy approach revealed that a great amount of disagreement in GlobCover and MODISLC datasets does not actually correspond to classification errors, but it can be associated to legend ambiguity and mixed coverage pixels.

DOI

[21]
Herold M, Mayaux P, Woodcock C E, et al.Some challenges in global land cover mapping: An assessment of agreement and accuracy in existing 1 km datasets. Remote Sensing of Environment, 2008, 112(5): 2538-2556.Many investigators need and use global land cover maps for a wide variety of purposes. Ironically, after many years of very limited availability, there are now multiple global land cover maps and it is not readily apparent (1) which is most useful for particular applications or (2) how to combine the different maps to provide an improved dataset. The existing global land cover maps at 1km spatial resolution have arisen from different initiatives and are based on different remote sensing data and employed different methodologies. Perhaps more significantly, they have different legends. As a result, comparison of the different land cover maps is difficult and information about their relative utility is limited. In an attempt to compare the datasets and assess their strengths and weaknesses we harmonized the thematic legends of four available coarse-resolution global land cover maps (IGBP DISCover, UMD, MODIS 1-km, and GLC2000) using the LCCS-based land cover legend translation protocols. Analysis of the agreement among the global land cover maps and existing validation information highlights general patterns of agreement, inconsistencies and uncertainties. The thematic classes of Evergreen broadleaf trees, Snow and Ice, and Barren show high producer and user accuracy and good agreement among the datasets, while classes of mixed tree types show high commission errors. Overall, the results show a limited ability of the four global products to discriminate mixed classes characterized by a mosaic of trees, shrubs, and herbaceous vegetation. There is a strong relationship between class accuracy, spatial agreement among the datasets, and the heterogeneity of landscapes. Suggestions for future mapping projects include careful definition of mixed unit classes, and improvement in mapping heterogeneous landscapes.

DOI

[22]
Foody G M.Assessing the accuracy of land cover change with imperfect ground reference data. Remote Sensing of Environment, 2010, 114(10): 2271-2285.The ground data used as a reference in the validation of land cover change products are often not an ideal gold standard but degraded by error. The effects of ground reference data error on the accuracy of land cover change detection and the accuracy of estimates of the extent of change were evaluated. Twelve data sets were simulated to allow the exploration of the impacts of a spectrum of ground data imperfections on the estimation of the producer's and user's accuracy of change as well as of change extent. Simulated data were used since this ensured that the actual properties of the data were known and to exclude effects due to other sources of ground reference data error; although the impacts of simulated reference data error on two real confusion matrices are also illustrated. The imperfections evaluated ranged from the inclusion of small amounts of known error into the ground reference data through to the extreme situation in which ground data were absent. The results show that even small amounts of error in the ground reference data can introduce large error into studies of land cover change by remote sensing and reinforce the desire to avoid the expression ground truth as this might imply that the data are a gold standard reference. The effect of reference data imperfections was dependent on the degree of association between the errors in the cross-tabulated data sets. For example, in the scenarios investigated, a 10% error in the reference data set introduced an underestimation of the producer's accuracy of 18.5% if the errors were independent but an over-estimation of the producer's accuracy of 12.3% if the errors were correlated. The magnitude of the mis-estimation of the producer's accuracy was also a function of the amount of change and greatest at low levels of change. The amount of land cover change estimated also varied greatly as a function of ground reference data error. Some possible methods to reduce or even remove the impacts of ground reference data error were illustrated. These ranged from simple algebraic means to estimate the actual values of accuracy and change extent if the imperfections were known through to a latent class analysis that allowed the assessment of classification accuracy and estimation of change extent without the use of ground reference data if the underlying model is defined appropriately.

DOI

[23]
杨永可, 肖鹏峰, 冯学智, 等. 大尺度土地覆盖数据集在中国及周边区域的精度评价. 遥感学报, 2014, 18(2): 453-475.大尺度土地覆盖数据是全球陆地表层过程研究、生态系统评估、环境建模等科学研究的重要基础，研究现有数据集的特点对数据使用者及生产新的数据集都具有指导意义。本研究以中国及周边区域为研究区，根据不同分类体系对地物的定义，研究不同分类体系中对应地物的相关系数，并将所有分类体系转换为IGBP分类体系；然后，从定性和定量两方面分析现有5种土地覆盖数据集（IGBP DISCover、UMD、GLC2000、MOD12Q1和GlobCover 2005）的空间一致性；并利用Google Earth高分影像选取两期验证样本评价5种土地覆盖数据集的精度。结果表明：同种地物在不同土地覆盖数据集之间的空间分布格局差异较大，且不同土地覆盖数据集之间的总体一致性系数较低；5种土地覆盖数据集中，GLC2000的总体精度和Kappa系数均最高，GlobCover 2005的总体精度和Kappa系数均最低。

DOI

[Yang Yongke, Xiao Pengfeng, Feng Xuezhi, et al.Comparison and assessment of large-scale land cover datasets in China and adjacent regions. Journal of Remote Sensing, 2014, 18(2): 453-475.]

[24]
Fritz S, McCallum I, Schill C, et al. Geo-Wiki: An online platform for improving global land cover. Environmental Modelling & Software, 2012, 31(7):110-123.Land cover derived from remotely sensed products is an important input to a number of different global, regional and national scale applications including resource assessments and economic land use models. During the last decade three global land cover datasets have been created, i.e. the GLC-2000, MODIS and GlobCover, but comparison studies have shown that there are large spatial discrepancies between these three products. One of the reasons for these discrepancies is the lack of sufficient in-situ data for the development of these products. To address this issue, a crowdsourcing tool called Geo-Wiki has been developed. Geo-Wiki has two main aims: to increase the amount of in-situ land cover data available for training, calibration and validation, and to create a hybrid global land cover map that provides more accurate land cover information than any current individual product. This paper outlines the components that comprise Geo-Wiki and how they are integrated in the architectural design. An overview of the main functionality of Geo-Wiki is then provided along with the current usage statistics and the lessons learned to date, in particular the need to add a mechanism for feedback and interaction as part of community building, and the need to address issues of data quality. The tool is located at geo-wiki.org .

DOI

[25]
Olofsson P, Foody G M, Herold M, et al.Good practices for estimating area and assessing accuracy of land change. Remote Sensing of Environment, 2014, 148(5): 42-57.The remote sensing science and application communities have developed increasingly reliable, consistent, and robust approaches for capturing land dynamics to meet a range of information needs. Statistically robust and transparent approaches for assessing accuracy and estimating area of change are critical to ensure the integrity of land change information. We provide practitioners with a set of “good practice” recommendations for designing and implementing an accuracy assessment of a change map and estimating area based on the reference sample data. The good practice recommendations address the three major components: sampling design, response design and analysis. The primary good practice recommendations for assessing accuracy and estimating area are: (i) implement a probability sampling design that is chosen to achieve the priority objectives of accuracy and area estimation while also satisfying practical constraints such as cost and available sources of reference data; (ii) implement a response design protocol that is based on reference data sources that provide sufficient spatial and temporal representation to accurately label each unit in the sample (i.e., the “reference classification” will be considerably more accurate than the map classification being evaluated); (iii) implement an analysis that is consistent with the sampling design and response design protocols; (iv) summarize the accuracy assessment by reporting the estimated error matrix in terms of proportion of area and estimates of overall accuracy, user's accuracy (or commission error), and producer's accuracy (or omission error); (v) estimate area of classes (e.g., types of change such as wetland loss or types of persistence such as stable forest) based on the reference classification of the sample units; (vi) quantify uncertainty by reporting confidence intervals for accuracy and area parameters; (vii) evaluate variability and potential error in the reference classification; and (viii) document deviations from good practice that may substantially affect the results. An example application is provided to illustrate the recommended process.

DOI

[26]
哈凯, 丁庆龙, 门明新,等. 山地丘陵区土地利用分布及其与地形因子关系: 以河北省怀来县为例. 地理研究, 2015, 34(5): 909-921.<p>为探究山地丘陵地区不同地形因子对土地利用空间格局的影响,以河北省怀来县为例,研究了各土地利用类型在不同地形上的分布特征。同时,定性和定量分析了土地利用景观格局与地形环境的相关性。研究表明：土地利用景观格局分布存在明显的地形梯度特征,较高地形级别主要分布着林地和草地,其分布指数在较高地形级别占据优势位;而受人类活动干扰影响较大的城乡建设用地、采矿用地、耕地、园地和水域的优势位分布在较低地形级别上。移动窗口法得到的景观格局指数的分布定性显示出土地利用景观格局随地形变化的梯度特征,典范对应分析结果定量显示出典范对应分析排序的前4个轴的累计土地利用景观指数&#x02014;地形解释变量为96.98%,地形因子第1排序轴与土地利用变化景观格局指数特征第一排序轴的相关系数达到0.671,其土地利用景观格局指数与地形因子存在显著的相关性。由此可见,高程、坡度和地形起伏度这三类地形因子对山地丘陵区的土地利用类型的结构和空间分布起着重要的影响作用。</p>

DOI

[Ha Kai, Ding Qinglong, Men Mingxin, et al.Spatial distribution of land use and its relationship with terrain factors in hilly area. Geographical Research, 2015, 34(5): 909-921.]

[27]
李炳元, 潘保田, 韩嘉福. 中国陆地基本地貌类型及其划分指标探讨. 第四纪研究, 2008, 28(4): 535-543.<p>平原、台地、丘陵、山地等是地表最基本的地貌形态,这些名称不仅为专业领域广泛引用,也为普通人所知晓。近100年来多种地貌分类方案中都涉及这些地貌类型名称,有的称其为地貌&quot;基本形态&quot;。由于每种地貌形态都不仅包含形态特征,而且还有一定的成因意义,因此应称其为基本地貌类型。通过对已有的基本地貌分类及其划分指标进行系统分析和评估,认为中国陆地基本地貌类型按照起伏高度和海拔高度两个分级指标组合来划分的原则符合起伏复杂、多台阶中国地貌的基本特点。在传统的平原、台地、丘陵和山地分类的基础上,按起伏高度对山地进一步细分,即划分平原、台地、丘陵（&lt;200m）、小起伏山地（200～500m）、中起伏山地（500～1000m）、大起伏山地（1000～2500m）和极大起伏山地（&gt;2500m）等7个基本地貌&quot;形态&quot;。本文对前人以现代雪线、多年冻土下线和森林上线高度为依据确定地貌面海拔高度的分级指标进行了全面分析,由于它们的海拔高度在全国各地存在巨大差异,我们认为海拔高度等级指标并不符合中国实际。通过全国重点地区1∶500000地形图山地顶面海拔高度分布和1∶1000000国家数字高程模型（DTM）数据库编制的中国地面高程分布图进行较系统的分析,我们提出了应以1000m,2000m,4000m和6000m作为划分低海拔（&lt;1000m）、中海拔（1000～2000m）、亚高海拔（2000～4000m）、高海拔（4000～6000m）和极高海拔（&gt;6000m）地貌海拔高度分级指标。根据7个地貌起伏高度形态和5个海拔高度等级,将全国组合成从低海拔平原至极大起伏极高山28个基本地貌类型。</p>

[Li Bingyuan, Pan Baotian, Han Jiafu.Basic terrestrial geomorphological types in China and their circumscriptions. Quaternary Sciences, 2008, 28(4): 535-543.]

[28]
张光业. 河南省地貌区划. 河南大学学报: 社会科学版, 1964, 9(1): 124-137.

[Zhang Guangye.Geomorphologic zonalization of Henan. Journal of Henan University: Social Science, 1964, 9(1): 124-137.]

[29]
廖安平, 陈利军, 陈军, 等. 全球陆表水体高辨率遥感制图. 中国科学:地球科学, 2014, 40(8): 1634-1645.陆表水体是地表覆盖的重要组成之一,是气候变化研究、生态环境评估和宏观调控分析等不可或缺的重要基础信息.本文介绍了全球地表覆盖遥感制图项目中陆表水体总体研制情况.项目通过收集、处理美国陆地卫星Landsat TM/ETM+、国产环境减灾星(HJ-1)等影像,实现了2000年、2010年两个基准年度的全球30 m分辨率多光谱影像的有效覆盖,影像纠正精度满足1:20万制图要求,两期影像配准中误差控制在1个像元以内.根据30 m分辨率尺度的水体光谱特征与几何形态,合理地设计提取指标,结合基于像元分类法简单易操作、面向对象分类法可综合利用各种规则知识的优势,开展水体信息的精细化提取,最后利用人机交互来进一步优化完善分类结果,实现全球水体的高精度遥感制图.完成的全球陆表水体数据成果GlobalLand30-water 2000和GlobalLand30-water 2010,属目前全球尺度下最高分辨率的分类成果,自评估总体精度为96%.该数据是开展全球陆表水体空间分布格局分析、揭示地域差异、研究时空波动规律以及进行生态环境健康诊断等相关研究工作的重要基础数据.

[Liao Anping, Chen Lijun, Chen Jun, et al.High-resolution remote sensing mapping of global land water. Scientia Sinica Terrae, 2014, 40(8):1634-645.]

Options
文章导航

/

〈 〉