Journal of Geo-information Science >
Fine Classification Method Study of Large-scale Land Use/Cover Based onGeoscience Knowledge
Received date: 2016-07-01
Request revised date: 2016-09-23
Online published: 2017-01-13
Copyright
Human activities have significant impacts on ecosystems. As the most direct characterization of human activity, large-scale land use/cover change is used to analyze the impacts of human activities on ecosystems. Therefore, scientists have paid much attention on the classification and extraction methods of land use/cover products. It was suggested that GlobCover (2005/2006) product was precise enough for the scientific study. However, the product has some limitations. In order to improve the quality of this product, this study developed new method for mapping and monitoring national land cover information in Brazil. The new Brazilian land use/cover data in 2005 were developed by using human-computer interactive discrimination at per-cell level based on GlobCover (2005/2006) data and the combination of geographic knowledge and the major data source of Landsat TM/ETM images. The results indicated that data accuracy and cost-efficiency were both improved by the developed method. The classification accuracy was improved from 67.17% in the GlobCover to 93.39% in our new dataset. Kappa coefficient was also improved from 0.58 to 0.91. Evergreen broadleaf forest area in Brazil was the highest among all the land cover types, with an area ratio of 45.67%. Farmland/natural vegetation mosaic area followed with an area ratio of 19.19%. The third largest land cover type was closed shrub with an area ratio of 12.34%. Modification ratio of agricultural land/natural vegetation mosaic and shrub and grassland was the largest. Among them, the proportion of mixed pixels of land class decreased 3.54%, while shrub and grassland increased 3.81%. As a result, the new developed method was proved to be more efficient and accurate. It can be used for large-scale land use/cover classification and analysis in further study.
Key words: Brazil; land use/cover change; GlobCover; fine classification; precision evaluation
DU Guoming , LIU Mei , MENG Fanhao , CHUN Xiang , FENG Yue . Fine Classification Method Study of Large-scale Land Use/Cover Based onGeoscience Knowledge[J]. Journal of Geo-information Science, 2017 , 19(1) : 91 -100 . DOI: 10.3724/SP.J.1047.2017.00091
Tab. 1 Existing global land cover datasets表1 全球现有土地覆盖数据集 |
数据集名称 | 作者 | 数据源与分辨率 | 分类方法 | 分类系统 | 数据精度 |
---|---|---|---|---|---|
DISCover | U.S. Geological Survey | 1992-1993年AVHRR数据合成的NDVI、1 km | 基于聚类和人工解译、编辑 | IGBP 17类分类系统 | 基于少数样点进行精度评价,总体精度为66.9% |
UMD | University of Maryland | 同上、1 km | 不同的分类树算法 | Simplified IGBP (14 classes) | 同上 |
MODIS 1km | Boston University | 2000-2001年的MODIS 数据、1 km | 监督分类 | IGBP 17类分类系统 | 基于少数样点进行精度评价,总体精度为78.3% |
GLC2000 | European Commission Joint Research Center | 1999年11月-2000年12月的VEGETATION数据、1 km | 由19个区域人员用不同类型算法制作完成 | FAO 的地表覆盖22类分类系统(LCCS) | 总体精度为68.6% |
GlobCover 2 套产品 | European Commission Joint Research Center | 2004年12月-2006月6日的ENVISAT/ MERIS数据2009年数据产品、300 m | 全球分为22个生态气候区,各区采用不同多维迭代聚类方法进行分类 | 同上 | 16位专家在全球3000个点进行了验证,总体精度为73% |
GlobeLand 30-2010 | 国家基础地理信息中心等7个部门的18家单位 | HJ-1星CCD影像数据2000年和2010年数据产品、30 m | 逐类型层次提取方法 | 9大类分类系统 | 选取9类超过15万个样本进行精度评估,总体精度为83.51%,Kappa系数为0.78 |
Fig. 1 Typical errors existed in the GlobCover datasets图1 GlobCover数据集存在的典型问题 |
Fig. 2 Years′ distribution of Landsat TM/ETM data source and GlobCover land cover data of Brazil图2 Landsat TM/ETM数据源年份分布图与GlobCover数据巴西土地覆盖图 |
Tab. 2 Land use/cover classification system transformation表2 土地利用/覆盖分类体系转换表 |
一级类 | 代码 | 本研究土地覆盖 分类系统 | IGBP土地覆盖 分类系统 | 代码 | GlobCover土地覆盖 分类系统 | 定义 |
---|---|---|---|---|---|---|
农地 | 11 | 农地(简单或多种作物 系统) | 农地(简单或多种作 物系统) | 11 | 水淹或灌溉农地 | 雨养农地与耕作农地,简单或多种作物系统 |
14 | 雨养农地 | |||||
12 | 农地/自然植被镶嵌(农地、森林、灌丛、草地;单一覆盖不超过60%) | 农地/自然植被镶嵌(农地、森林、灌丛、草地;单一覆盖不超过60%) | 20 | 耕作(50%~70%)/其他自然植被(20%~50%)镶嵌 | 农、草、林、灌混合镶嵌,其中单一覆盖度应小于60% | |
30 | 耕作(20%~50%)/其他自然植物(50%~70%)镶嵌 | |||||
林地 | 21 | 常绿阔叶林 | 常绿阔叶林 | 40 | 郁闭或敞开(>15%)常绿阔叶或半落叶阔叶林(>5 m) | 是热带至温带之间的过渡性质的森林类型。其群落外貌终年常绿,一般呈暗绿色而略闪烁反光,林相整齐,由于树冠浑圆,林冠呈微波状起伏 |
50 | 郁闭(>40%)常绿阔叶林 (>5 m) | |||||
22 | 落叶阔叶林 | 落叶阔叶林 | 60 | 敞开(15%~40%)落叶阔叶林 (>5 m) | 温带、暖温带地区地带性的森林类型。因其冬季落叶、夏季葱绿,又称夏绿林 | |
23 | 常绿针叶林 | 常绿针叶林 | 70 | 郁闭(>40%)常绿针叶林 (>5 m) | 1.暖温性常绿针叶林,乔木植被类型。树种以油松和华山松为主。多生于海拔1700-2100 m之间的地带2.寒温带常绿针叶林,由常绿的云杉、冷杉、松和圆柏所组成。云杉、冷杉林具有较强的耐阴性,林冠稠密、郁闭,林下光照很弱,又称“暗针叶林” | |
24 | 落叶针叶林 | 落叶针叶林 | 90 | 敞开(15%~40%)常绿针叶或落叶针叶林(>5 m) | 由落叶松柏类为主的针叶树所构成的森林。寒温带的地带性植被类型,并常由落叶松组成,林下有灌木层、草本层和地面苔藓层,落叶松林呈鲜绿色,树冠尖塔形,较喜阳,林下明亮成明亮针叶林 | |
25 | 混交林(没有主导类型超过60%覆盖) | 混交林(没有主导类型超过60%覆盖) | 100 | 郁闭或敞开(>15%)针阔混交林(>5 m) | 上述四种林地镶嵌,并且没有主导类型超过60%覆盖 | |
灌丛和草地 | 31 | 有(森林)林草原(树林冠层覆盖30%~60%,高度超过2 m) | 有(森林)林草原(树林冠层覆盖30%~60%,高度超过2 m) | 110 | 草地(20%~50%)/森林/灌丛(50%~70%)镶嵌 | 树林冠层覆盖30%~60%,高度超过2 m的草原 |
32 | 稀树草原(树林冠层覆盖10%~30%,高度超过2 m) | 稀树草原(树林冠层覆盖10%~30%,高度超过2 m) | 120 | 草地(50%~70%)/森林/灌丛(20%~50%)镶嵌 | 树林冠层覆盖10%~30%,高度超过2 m的草原 | |
33 | 封闭灌丛(灌丛覆盖度高于60%;高度低于 2 m,常绿或落叶) | 封闭灌丛(灌丛覆盖度高于60%;高度低于 2 m,常绿或落叶) | 130 | 冠层敞开或封闭(>15%)灌丛(<5 m) | 灌丛覆盖度高于60%,高度低于2 m,常绿或落叶均可 | |
34 | 敞开灌丛(灌丛覆盖率10%~60%高度低于2 m,常绿或落叶) | 敞开灌丛(灌丛覆盖率10%~60%高度低于 2 m,常绿或落叶) | 灌丛覆盖率10%~60%,高度低于2 m,常绿或落叶均可 | |||
35 | 草地或禾本植物(树冠密度低于10%) | 草地或禾本植物(树冠密度低于10%) | 140 | 冠层敞开或封闭(>15%) 草地 | 树冠密度低于10%的草地,或禾本科 植物 | |
水域 | 41 | 水体 | 水体 | 210 | 水体 | 所有湖泊、河流、滨海、人工水体等水 体的集合 |
42 | 森林湿地 | 永久湿地(水/禾本植物/有林地) | 160 | 郁闭或敞开(>15%)各种有规律水淹或长期水浸阔叶森林 | 其是泛滥平原森林,季节性的洪水一 般会和森林湿地一起出现 | |
170 | 郁闭(>40%)永久盐水水淹阔叶林或灌丛 | |||||
43 | 沼泽湿地 | 180 | 郁闭或敞开(>15%)各种有规律水淹或长期水浸草地 | 地表过湿或有薄层积水,土壤水分几达饱和,并有泥炭堆积,生长着喜湿性和喜水性沼生植物的地段 | ||
城市和建成区 | 51 | 城市和建成区 | 城市和建成区 | 190 | 人工地表或附属区域 | 城镇居民点与工矿等建设用地 |
裸地和冰雪 | 61 | 裸地或稀疏植被(植被覆盖低于10%) | 裸地或稀疏植被(植被覆盖低于10%) | 150 | 稀疏植被(<15%) | 植被覆盖低于10%或裸地 |
200 | 裸地 | |||||
62 | 雪/冰 | 雪/冰 | 220 | 永久雪/冰 | 覆盖永久冰、雪 |
Fig. 3 Technology process diagram图3 技术流程图 |
Tab. 3 Distribution of sample points表3 样本点分配表 |
地类代码 | GlobCover数据 | 本研究数据 | |||
---|---|---|---|---|---|
A | B | 面积比/% | 样本数 | 面积比/% | 样本数 |
14 | 11 | 8.33 | 250 | 9.07 | 300 |
20 | 12 | 10.73 | 320 | 19.30 | 580 |
30 | 11.61 | 340 | |||
40 | 21 | 43.90 | 1300 | 45.81 | 1300 |
50 | 3.11 | 90 | |||
60 | 22 | 0.86 | 30 | 0.82 | 30 |
100 | 25 | 0.0005 | 30 | 0.001 | 30 |
110 | 31 | 1.13 | 33 | 0.90 | 30 |
130 | 33 | 15.37 | 450 | 12.33 | 350 |
34 | 6.19 | 200 | |||
120 | 32 | 0.10 | 30 | 0.11 | 30 |
140 | 35 | 0.19 | 30 | 1.05 | 30 |
160 | 42 | 1.88 | 50 | 1.95 | 50 |
170 | 0.08 | 30 | |||
180 | 43 | 1.21 | 36 | 0.74 | 30 |
190 | 51 | 0.05 | 30 | 0.24 | 30 |
150 | 61 | 0.02 | 30 | 0.07 | 30 |
200 | 0.05 | 30 | |||
210 | 41 | 1.38 | 41 | 1.43 | 36 |
220 | 62 | 0.006 | 30 | 0 | 0 |
注:A代表GlobCover数据;B代表本研究数据 |
Fig. 4 Samoling program图4 采样方案 |
Tab. 4 Matrix of accuracy evaluation表4 精度评价矩阵表 |
参考数据 | 产品数据 | ||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
14 | 20 | 30 | 40 | 50 | 60 | 100 | 110 | 120 | 130 | 140 | 150 | 160 | 170 | 180 | 190 | 200 | 210 | 220 | 总计 | 生产者 精度/% | |
14 | 156 | 18 | 14 | 5 | 29 | 1 | 4 | 1 | 10 | 238 | 65.55 | ||||||||||
20 | 19 | 162 | 90 | 15 | 53 | 339 | 47.79 | ||||||||||||||
30 | 33 | 42 | 170 | 16 | 10 | 1 | 1 | 24 | 1 | 298 | 57.05 | ||||||||||
40 | 14 | 29 | 14 | 1142 | 2 | 4 | 1 | 47 | 9 | 2 | 27 | 12 | 1303 | 87.64 | |||||||
50 | 8 | 3 | 53 | 7 | 17 | 88 | 60.23 | ||||||||||||||
60 | 8 | 1 | 1 | 17 | 21 | 2 | 7 | 57 | 36.84 | ||||||||||||
100 | 1 | 14 | 1 | 16 | 87.50 | ||||||||||||||||
110 | 5 | 2 | 5 | 18 | 7 | 11 | 2 | 50 | 36.00 | ||||||||||||
120 | 2 | 4 | 5 | 17 | 1 | 2 | 31 | 54.84 | |||||||||||||
130 | 13 | 58 | 49 | 95 | 10 | 2 | 244 | 4 | 1 | 5 | 2 | 4 | 1 | 12 | 20 | 520 | 46.92 | ||||
140 | 2 | 8 | 11 | 5 | 3 | 29 | 37.93 | ||||||||||||||
150 | 1 | 3 | 4 | 75.00 | |||||||||||||||||
160 | 4 | 1 | 15 | 2 | 22 | 68.18 | |||||||||||||||
170 | 2 | 14 | 7 | 23 | 60.87 | ||||||||||||||||
180 | 10 | 2 | 3 | 1 | 14 | 30 | 46.67 | ||||||||||||||
190 | 1 | 2 | 1 | 2 | 5 | 2 | 30 | 2 | 45 | 66.67 | |||||||||||
200 | 1 | 1 | 3 | 1 | 2 | 24 | 4 | 36 | 66.67 | ||||||||||||
210 | 1 | 1 | 1 | 2 | 1 | 13 | 1 | 1 | 2 | 25 | 48 | 52.08 | |||||||||
220 | 3 | 3 | 100.00 | ||||||||||||||||||
总计 | 250 | 320 | 340 | 1300 | 90 | 30 | 30 | 33 | 30 | 450 | 30 | 30 | 50 | 30 | 36 | 30 | 30 | 41 | 30 | 3180 | |
用户精度/% | 62.40 | 50.63 | 50.00 | 87.85 | 58.89 | 70.00 | 46.67 | 54.55 | 56.67 | 54.22 | 36.67 | 10.00 | 30.00 | 46.67 | 38.89 | 100.0 | 80.00 | 60.98 | 10.00 | 67.17 |
Tab. 5 Statistics of land use/cover data modification表5 土地利用/覆盖数据修改统计表 |
地类名称 | 地类代码 | 修改图斑数/块 | 修改比例/% |
---|---|---|---|
简单农地 | 11 | 2 497 095 | 21.52 |
农地/自然植被镶嵌 | 12 | 1 203 876 | 10.38 |
常绿阔叶林 | 21 | 199 895 | 1.72 |
落叶阔叶林 | 22 | 1897 | 0.02 |
混交林 | 25 | 1201 | 0.01 |
有(森林)林草原 | 31 | 6886 | 0.06 |
稀树草原 | 32 | 11252 | 0.10 |
封闭灌丛 | 33 | 3 318 585 | 28.60 |
敞开灌丛 | 34 | 3 250 248 | 28.02 |
草地或禾本植物 | 35 | 796 474 | 6.87 |
水体 | 41 | 98 869 | 0.85 |
森林湿地 | 42 | 16 075 | 0.14 |
沼泽湿地 | 43 | 10 227 | 0.09 |
城市和建成区 | 51 | 181 430 | 1.56 |
裸地或稀疏植被 | 61 | 7549 | 0.07 |
Tab. 6 Evaluation of accuracy verification表6 精度验证评价表 |
参考数据 | 产品数据 | ||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
11 | 12 | 21 | 22 | 25 | 31 | 32 | 33 | 34 | 35 | 41 | 42 | 43 | 51 | 61 | 总计 | 生产者精度/% | |
11 | 264 | 16 | 6 | 1 | 1 | 288 | 91.67 | ||||||||||
12 | 12 | 544 | 10 | 3 | 1 | 1 | 571 | 95.27 | |||||||||
21 | 2 | 4 | 1236 | 3 | 4 | 2 | 1 | 2 | 7 | 4 | 1265 | 97.71 | |||||
22 | 2 | 27 | 1 | 30 | 90.00 | ||||||||||||
25 | 23 | 23 | 100.00 | ||||||||||||||
31 | 3 | 26 | 1 | 1 | 31 | 83.87 | |||||||||||
32 | 25 | 25 | 100.00 | ||||||||||||||
33 | 4 | 8 | 24 | 3 | 1 | 1 | 344 | 10 | 1 | 5 | 401 | 85.79 | |||||
34 | 10 | 8 | 16 | 1 | 3 | 184 | 1 | 223 | 82.51 | ||||||||
35 | 5 | 1 | 1 | 1 | 27 | 35 | 77.14 | ||||||||||
41 | 1 | 36 | 1 | 35 | 94.74 | ||||||||||||
42 | 3 | 42 | 45 | 93.33 | |||||||||||||
43 | 1 | 3 | 1 | 25 | 30 | 83.33 | |||||||||||
51 | 30 | 30 | 100.00 | ||||||||||||||
61 | 21 | 21 | 100.00 | ||||||||||||||
总计 | 300 | 580 | 1300 | 30 | 30 | 30 | 3 | 350 | 200 | 30 | 36 | 50 | 30 | 30 | 30 | 3056 | |
用户精度/% | 88.00 | 93.79 | 95.08 | 90.00 | 76.67 | 86.67 | 83.33 | 98.29 | 92.00 | 90.00 | 100.0 | 84.00 | 83.33 | 100.0 | 70.00 | 93.39 |
Fig. 5 Current status of land use/cover in Brazil图5 巴西土地利用/覆盖现状图 |
Tab. 7 Statistics of secondary land use/cover classification area in Brazil表7 巴西二级土地利用/覆盖类型面积统计表 |
地类代码 | 11 | 12 | 21 | 22 | 25 | 31 | 32 | 33 | 34 | 35 | 41 | 42 | 43 | 51 | 56 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
所占面积比例/% | 9.19 | 19.19 | 45.67 | 0.81 | 0 | 0.89 | 0.11 | 12.34 | 6.23 | 1.11 | 1.45 | 1.95 | 0.74 | 0.25 | 0.07 |
Tab. 8 Statistics of primary land use/cover classification area in Brazil表8 巴西一级土地利用/覆盖类型面积统计表 |
地类名称 | 农地 | 农地/自然植被镶嵌 | 林地 | 灌丛和草地 | 水域 | 城市和建成区 | 裸地和冰雪 |
---|---|---|---|---|---|---|---|
地类代码 | 1 | 12 | 2 | 3 | 4 | 5 | 6 |
欧空局产品面积比例/% | 8.49 | 22.73 | 47.28 | 16.87 | 4.50 | 0.06 | 0.07 |
本研究产品面积比例/% | 9.19 | 19.19 | 46.49 | 20.68 | 4.14 | 0.25 | 0.07 |
The authors have declared that no competing interests exist.
[1] |
[
|
[2] |
[
|
[3] |
[
|
[4] |
[
|
[5] |
|
[6] |
[
|
[7] |
[ Qin L J, Zhang Y, Xu H M, et al. Study on the Effect of Land Use Change on the Eco-environment. Scientia Geographica Sinica, 2002,22(4):508-512. ]
|
[8] |
[
|
[9] |
|
[10] |
[
|
[11] |
[
|
[12] |
[
|
[13] |
|
[14] |
[
|
[15] |
[
|
[16] |
IGBP.2005. IGBP Report No.53, IHDP Report No.19: science plan and implementation strategy[R/OL].[2014-01-19].
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
|
[23] |
|
[24] |
|
[25] |
|
[26] |
|
/
〈 | 〉 |