The importance of data type, laser spot density and modelling method for vegetation height mapping in continental China

作者:Liu, CX (Liu, Caixia)[ 1 ] ; Wang, XY (Wang, Xiaoyi)[ 2 ] ; Huang, HB (Huang, Huabing)[ 1 ] ; Gong, P (Gong, Peng)[ 3,4 ] ; Wu, D (Wu, Di)[ 5 ] ; Jiang, JX (Jiang, Jinxiong)[ 6 ]

INTERNATIONAL JOURNAL OF REMOTE SENSING

卷: 37  期: 24  页: 6127-6148

DOI: 10.1080/01431161.2016.1252472

出版年: 2016

摘要

Vegetation height not only has great significance in the field of ecology but also offers a useful contribution to detailed land cover classification. The first vegetation height map was acquired in this study using the ice, cloud, and land elevation satellite /geosciences laser altimeter system (ICESat/GLAS) and other multisource remote sensing data, such as moderate-resolution imaging spectroradiometer (MODIS) tree cover products, leaf area index (LAI) products, Nadir bidirectional reflectance distribution function (BRDF)-adjusted reflectance (NBAR), climatic variables, and topographic indices. We mainly discuss the importance of data type, density of laser spot and modelling method in the generation of this vegetation height map in continental China. It was found that (1) a higher density of laser spot could improve the reliability of modelling in mountainous areas covered by a wide range of forest and shrub land; (2) in terms of the importance of input variables, in the random forest regression modelling, the most important ones are elevation, slope, mean air temperature, temperature variance, precipitation, precipitation variance, and NBAR; (3) when modelling using 50 ecozones covering the whole of continental China, the model showed a good performance with an accuracy of root mean square error (RMSE), correlation coefficient (r), index of agreement (d), and mean absolute error (MAE) at 5.7, 0.7, 0.8, and 3.8 m, respectively. A visual comparison suggests that the spatial pattern of vegetation height is consistent with that of land cover in China. It is very necessary in evaluating the importance of data type, laser spot density, and modelling method in vegetation height mapping in continental China.

通讯作者地址: Huang, HB (通讯作者)

Chinese Acad Sci, Inst Remote Sensing & Digital Earth, State Key Lab Remote Sensing Sci, Beijing 100101, Peoples R China.

通讯作者地址: Gong, P (通讯作者)

Tsinghua Univ, Minist Educ, Key Lab Earth Syst Modeling, Ctr Earth Syst Sci, Beijing 100084, Peoples R China.

地址:

[ 1 ] Chinese Acad Sci, Inst Remote Sensing & Digital Earth, State Key Lab Remote Sensing Sci, Beijing 100101, Peoples R China

[ 2 ] Chinese Acad Sci, Inst Tibetan Plateau Res, Beijing, Peoples R China

[ 3 ] Tsinghua Univ, Minist Educ, Key Lab Earth Syst Modeling, Ctr Earth Syst Sci, Beijing 100084, Peoples R China

[ 4 ] Joint Ctr Global Change Studies, Beijing, Peoples R China

[ 5 ] SBSM, Heilongjiang Geomat Ctr, Harbin, Peoples R China

[ 6 ] State Key Lab Space Ground Integrated Informat Te, Beijing, Peoples R China