Remote Sensing inversion of Classification and Stocking Volume of Tropical Virgin Forest Types Based on Multivariate Data

doi:10.13466/j.cnki.lyzygl.2019.02.006

Abstract

Abstract:

Study on the spatial distribution and inversion of forest ecosystem stocks play a crucial role in carbon stock estimation,biodiversity and global climate change research.However,due to the diversity of forest vegetation types,especially in tropical primary forest areas that are beyond the reach of human,forest survey data is missing,the estimates and inversions of forest stocks still present significant challenges.This study takes the tropical primitive rain forest area of 18.80 million ha in the West Syepik Province of Papua New Guinea as the study area,and uses the high-resolution remote sensing images of RapidEye,QuickBird and Landsat TM combining the field survey data to classify the land cover types in the study area.Based on the forest vegetation parameter information obtained by remote sensing image,the remote sensing inversion model of forest stock quantity is established in cooperation with the ground sample plot.The optimal inversion model is selected to estimate the forest stock volume,and combined with GIS technology to analyze the spatial distribution characteristics of the small class scale.The results show that the land cover types in the study area can be divided into low-altitude plain forests,low-altitude highland forests,low-mountain forests,sparse forests,swamp forests and other types,with a classification accuracy of 79.2%.The multivariate regression model R² of the stock volume remote sensing inversion model is 0.694,which has a good inversion accuracy for the forest stock volume.The distribution of forest stocks in the study area is characterized by a higher central area than the surrounding,northern and central eastern regions,which is significantly higher than the northwest and southeast regions,which corresponds to the distribution of land cover types in the study area.The forest stock inversion model used has important reference value for the estimation of forest resource stocks in tropical forest areas.

Key words: remote sensing image, forest volume, inversion model, spatial distribution, tropical primary forest areas

CLC Number:

S771

CHEN Xinyun, LI Liwei, LIU Chengfang, WANG Liuru, DING Jing. Remote Sensing inversion of Classification and Stocking Volume of Tropical Virgin Forest Types Based on Multivariate Data[J]. FOREST RESOURCES WANAGEMENT, 2019, 0(2): 39-46.

Figures/Tables 9

Fig.1

Fig.2

Tab.1

Tab.2

Tab.3

Tab.4

Tab.5

Fig.3

Fig.4

References 20

[1]	陈仲新, 张时新. 中国生态系统效益的价值[J]. 科学通报, 2000,45(1):17-22.
[2]	王玉辉, 周广胜, 蒋延玲, 等. 基于森林资源清查资料的落叶松林生物量和净生长量估算模式[J]. 植物生态学报, 2001,25(4):420-425.
[3]	方精云, 刘国华, 徐嵩龄. 我国森林植被的生物量和净生产量[J]. 生态学报, 1994,16(5):497-508.
[4]	王斌, 刘某承, 张彪. 基于森林资源清查资料的森林植被净生产量及其动态变化研究[J].林业资源管理, 2009(1):35-43.
[5]	焦燕, 胡海清. 黑龙江省森林植被碳储量及其动态变化[J]. 应用生态学报, 2005,16(12):2248-2252.
[6]	Li Wang, Niu Zheng, Chen Hanyue, et al. Remote estimation of canopy height and aboveground biomass of maize using high-resolution stereo images from a low-cost unmanned aerial vehicle system[J]. Ecological Indicators, 2016,67:637-648.
[7]	Maack J, Kattenborn T, Fassnacht F E, et al. Modeling forest biomass using Very-High-Resolution data—Combining textural,spectral and photogrammetric predictors derived from spaceborne stereo images[J]. European Journal of Remote Sensing, 2015,48(1):245-261.
[8]	方明程. 高分遥感纹理信息对森林蓄积量估测的影响规律研究[D]. 西安:西安科技大学, 2016.
[9]	徐新良, 刘纪远, 庄大方, 等. 基于3S技术的中国东北地区林地时空动态特征及驱动力分析[J]. 地理科学, 2004,24(1):55-60.
[10]	Gemmell F, Varjo J, Strandstrom M, et al. Comparison of measured boreal forest characteristics with estimates from TM data and limited ancillary information using reflectance model inversion[J]. Remote Sensing of Environment, 2002,81(2):365-377.
[11]	Barrett F, Mcroberts R E, Tomppo E, et al. A questionnaire-based review of the operational use of remotely sensed data by national forest inventories[J]. Remote Sensing of Environment, 2016,174:279-289.
[12]	MaKela H, Pekkarinen A. Estimation of forest stand volumes by Landsat TM imagery and stand-level field-inventory data[J]. Forest Ecology and Management, 2004,196:245-255.
[13]	王臣立, 牛铮, 郭治兴, 等. 基于植被指数和神经网络的热带人工林地上蓄积量遥感估测[J]. 生态环境学报, 2009,18(5):1830-1834.
[14]	侯长谋, 杨燕琼, 黄平, 等. 基于RS、GIS的马尾松林分蓄积量判读模型研究[J]. 林业资源管理, 2002 (5):55-58.
[15]	周亦, 张亚亚. 利用eCognition进行高分一号卫星数据土地利用现状解译能力测试[J].测绘通报, 2016(8):77-80.
[16]	焦险峰, 杨邦杰, 裴志远. 基于分层抽样的中国水稻种植面积遥感调查方法研究[J]. 农业工程学报, 2006,22(5):105-110.
[17]	刘三超, 张万昌, 蒋建军, 等. 用TM影像和DEM获取黑河流域地表反射率和反照率[J]. 地理科学, 2003,23(5):585-591.
[18]	赵冰雪, 章勇. 基于Landsat-8 OLI影像的植被信息提取方法研究[J]. 测绘与空间地理信息, 2018,41(1):79-85.
[19]	李亦秋, 冯仲科, 邓欧, 等. 基于3S技术的山东省森林蓄积量估测[J]. 林业科学, 2009,45(9):85-93.
[20]	王新云, 郭艺歌, 何杰. 基于HJ1B和ALOS/PALSAR数据的森林地上生物量遥感估算[J]. 生态学报, 2016,36(13):4109-4121.

一级类型	二级类型
低海拔平原森林	中大树冠森林
	稀疏树冠森林
	小树冠森林
	巴拉斯榄仁树林
低海拔高地森林	大树冠森林
	中树冠森林
	小树冠森林
低山森林	小树冠森林
	小树冠针叶林
稀疏森林
沼泽森林
其它类型

土地覆盖类型	面积/万hm	比例/%	样地数/个
低海拔高地森林	7.52	40.01	33
低海拔平原森林	4.76	25.30	20
沼泽森林	3.16	16.82	14
低山森林	1.16	6.20	6
稀疏森林	1.15	6.09	6
其它类型	1.05	5.58	—
合计	18.80	100.00	79

因子类型	建模因子	来源
波段反射率值	Band1,Band2,Band3,Band4,Band5,Band7	RapidEye,QuickBird,Landsat TM
波段运算	NDVI,EVI,SAVI,MSAVI,NDMI	RapidEye,QuickBird,Landsat TM
主成分	PCA1,PCA2,PCA3
地形数据	Aspect,Slope,Elevation	DEM

类型	总面积		总蓄积		公顷蓄积/(m³/ hm²)
类型	面积/万hm²	比例/%	蓄积/万m³	比例/%	公顷蓄积/(m³/ hm²)
低海拔平原森林	4.76	25.3	1504	35.35	316.1
低海拔高地森林	7.52	40.01	2062	48.49	274.2
低山森林	1.16	6.2	255	5.98	218.5
稀疏森林	1.15	6.09	105	2.46	91.3
沼泽森林	3.16	16.82	328	7.72	103.9
其它类型	1.05	5.58	—	—	—
合计	18.80	100.00	4254	100.00	239.6

建模因子	Pearson相关性	建模因子	Pearson相关性
Band 2	-0.456^**	NDMI	0.074
Band 3	-0.464^**	SAVI	0.388^**
Band 5	-0.083	MSAVI	0.382^**
Band 7	-0.270^*	Elevation	0.035
Slope	0.055	Aspect	0.307^**