Optimized BP Neural Network Model Based on Genetic Algorithm to Estimate The Leaf Area Index of Pinus densata

doi:10.13466/j.cnki.lyzygl.2020.05.018

Abstract

Abstract:

The leaf area index (LAI) is an important index to measure forest productivity.Remote sensing technology provides support for large-scale of LAI estimation.This study takes Pinus densata as the research object in Shangri-La city,uses Sentinel-2 multi-spectral images as the information source,and combines the measured LAI on the ground sample plots,selects vegetation index which is significantly correlated with LAI through correlation analysis,uses BP neural network model and genetic algorithm (GA) to optimize BP neural network to establish LAI of Pinus densata estimation model.Research shows that:1) Sentinel-2 image red-edge band vegetation index has a high correlation with LAI;2) The coefficient of determination (R ²) of the BP neural network model before and after genetic algorithm optimization were 0.289 and 0.508,and the root mean square errors (RMSE) were 0.340 and 0.314.The BP neural network modeling accuracy is higher after the genetic algorithm optimization and the actual measurement the forecast of the change trend of LAI is more accurate.This research result can provide a reference for the study of forest LAI in low latitude and high altitude areas.

Key words: leaf area index, genetic algorithm, BP neural network, Sentinel-2 image, Pinus densata

CLC Number:

S758.4

TAN Dehong, SHU Qingtai, ZHAO Hongying, WANG Keren, YUAN Zijian. Optimized BP Neural Network Model Based on Genetic Algorithm to Estimate The Leaf Area Index of Pinus densata[J]. FOREST RESOURCES WANAGEMENT, 2020, (5): 123-130.

Figures/Tables 11

Tab.1

Tab.2

Tab.3

Vegetation index formula

植被指数	公式	文献号
归一化差值植被指数(NDVI)	(NIR-RED)/(NIR+RED)	21
归一化光谱指数(NDSI)	(R₇₈₃-R₇₀₅/(R₇₈₃+R₇₀₅)	22
比值型光谱指数(RSI)	R₇₈₃/R₇₀₅	22
土壤调节植被指数(SAVI)	(1+L)(R₈₆₅-RED)/(R₈₆₅+RED+L)(L=0.25)	23
垂直植被指数(PVI)	(0.355NIR-0.149RED)²+(0.3355RED-0.852NIR)²	24
红边位置植被指数(S2REP)	705+35×[0.5×(R₇₈₃+R₆₆₅)-R₇₀₅]/(R₇₄₀-R₇₀₅)	25
简单比率(MSR)	(NIR/RED-1)/ $(NIR / RED + 1)$	26

Tab.3

Fig.1

Tab.4

Tab.5

Fig.2

Fig.3

Fig.4

Fig.5

Tab.6

References 28

[1]	Chen Jingming, Black T A. Defining leaf area index for non-flat leaves[J]. Plant,Cell & Environment, 1992,15(4):421-429.
[2]	Running S W, Nemani R R, Peterson D L, et al. Mapping regional forest evapotranspiration and photosynjournal by coupling satellite data with ecosystem simulation[J]. Ecology, 1989,70(4):1090-1101. doi: 10.2307/1941378
[3]	Majasalmi T, Rautiainen M. The potential of Sentinel-2 data for estimating biophysical variables in a boreal forest:a simulation study[J]. Remote Sensing Letters, 2016,7(5):427-436 doi: 10.1080/2150704X.2016.1149251
[4]	Pu Ruiliang, Gong Peng, Biging G S, et al. Extraction of red edge optical parameters from Hyperion data for estimation of forest leaf area index[J]. IEEE Transactions on geoscience and remote sensing, 2003,41(4):916-921. doi: 10.1109/TGRS.2003.813555
[5]	刘洋, 刘荣高, 陈镜明, 等. 叶面积指数遥感反演研究进展与展望[J]. 地球信息科学学报, 2013,15(5):734-743. doi: 10.3724/SP.J.1047.2013.00734
[6]	Darvishzadeh R, Atzberger C, Skidmore A K, et al. Leaf Area Index derivation from hyperspectral vegetation indicesand the red edge position[J]. International Journal of Remote Sensing, 2009,30(23):6199-6218. doi: 10.1080/01431160902842342
[7]	浦瑞良, 宫鹏, 约翰R.米勒. 用CASI遥感数据估计横跨美国俄勒冈州针叶林叶面积指数[J]. 南京林业大学学报:自然科学版, 1993(1):41-48.
[8]	Ercanli I, Gunlu A, Şenyurt M, et al. Artificial neural network models predicting the leaf area index:a case study in pure even-aged Crimean pine forests from Turkey[J]. Forest Ecosystems, 2018,5(1):1-12. doi: 10.1186/s40663-017-0128-5
[9]	朱高龙, 居为民, 范文义, 等. 帽儿山地区森林冠层叶面积指数的地面观测与遥感反演[J]. 应用生态学报, 2010,21(8):2117-2124.
[10]	Hardin P J, Jensen R R. Neural Network Estimation of Urban Leaf Area Index[J]. Giscience & Remote Sensing, 2005,42(3):251-274.
[11]	王强, 舒清态, 罗洪斌, 等. 基于机载LiDAR和光学遥感数据的热带橡胶林叶面积指数反演[J]. 西北林学院学报, 2020,35(4):132-139.
[12]	Yuan Huanhuan, Yang Guijun, Li Changchun, et al. Retrieving soybean leaf area index from unmanned aerial vehicle hyperspectral remote sensing:Analysis of RF,ANN,and SVM regression models[J]. Remote Sensing, 2017,9(4):309. doi: 10.3390/rs9040309
[13]	向洪波. 基于BP神经网络森林叶面积指数估算研究[D]. 重庆:西南大学, 2009.
[14]	Wang Lei, Wang Pengxin, Liang Shunlin, et al. Monitoring maize growth conditions by training a BP neural network with remotely sensed vegetation temperature condition index and leaf area index[J]. Computers and Electronics in Agriculture, 2019: 82-90.
[15]	杨敏, 林杰, 顾哲衍, 等. 基于Landsat8 OLI多光谱影像数据和BP神经网络的叶面积指数反演[J]. 中国水土保持科学, 2015,13(4):86-93.
[16]	Zhang Liangpei, Wu Ke, Zhong Yanfei, et al. A new sub-pixel mapping algorithm based on a BP neural network with an observation model[J]. Neurocomputing, 2008,71(10):2046-2054. doi: 10.1016/j.neucom.2007.08.033
[17]	李雪玲, 董莹莹, 朱溢佞, 等. 基于EnMAP卫星和深度神经网络的LAI遥感反演方法[J]. 红外与毫米波学报, 2020,39(1):111-119.
[18]	马海志. BP神经网络的改进研究及应用[D]. 哈尔滨:东北农业大学, 2015.
[19]	杨帆, 周敏, 金继民, 等. 智能优化算法及人工神经网络在催化裂化模型分析中的应用进展[J]. 石油学报(石油加工), 2020,36(4):878-888.
[20]	王枭轩, 孟庆岩, 张海香, 等. 基于粒子群神经网络模型反演玉米、小麦叶面积指数[J]. 浙江农业学报, 2019,31(7):1170-1176.
[21]	Park J, La S. Assessment of the Ochang Plain NDVI using Improved Resolution Method from MODIS Images[J]. Journal of the Korea Society of Environmental Restoration Technology, 2006,9(6):1-12.
[22]	Li Fei, Mistele B, Hu Yuncai, et al. Comparing hyperspectral index optimization algorithms to estimate aerial N uptake using multi-temporal winter wheat datasets from contrasting climatic and geographic zones in China and Germany[J]. Agricultural and Forest Meteorology, 2013: 44-57.
[23]	Major D J, Baret F, Guyot G, et al. A ratio vegetation index adjusted for soil brightness[J]. International Journal of Remote Sensing, 1990,11(5):727-740. doi: 10.1080/01431169008955053
[24]	Richardsons A J, Wiegand A L. Distinguishing vegetation from soil background information[J]. Photogrammetric Engineering and Remote Sensing, 1977,43(12):1541-1552.
[25]	Frampton W J, Dash J, Watmough G R, et al. Evaluating the capabilities of Sentinel-2 for quantitative estimation of biophysical variables in vegetation[J]. Isprs Journal of Photogrammetry and Remote Sensing, 2013: 83-92.
[26]	Chen Jingming. Evaluation of vegetation indices and a modified simple ratio for boreal applications[J]. Canadian Journal of Remote Sensing, 1996,22(3):229-242. doi: 10.1080/07038992.1996.10855178
[27]	任谢楠. 基于遗传算法的BP神经网络的优化研究及MATLAB仿真[D]. 天津:天津师范大学, 2014.
[28]	都月, 孟晓辰, 祝连庆. 遗传算法和神经网络的重叠光谱解析[J]. 光谱学与光谱分析, 2020,40(7):2066-2072.

波段	波段名称	波长范围/nm	分辨率/m
B1	气溶胶及海岸带波段	422~456	60
B2	蓝波段	439~533	10
B3	绿波段	538~583	10
B4	红波段	646~684	10
B5	植被红边	695~714	20
B6	植被红边	731~749	20
B7	植被红边	769~797	20
B8	近红外波段	773~907	10
B8a	近红外波段	847~881	20
B9	水蒸气波段	932~958	60
B10	卷积云波段	1337~1412	60
B11	短波红外波段	1539~1682	20
B12	短波红外波段	2078~2320	20

植被指数	相关系数	植被指数	相关系数
NDSI(783,705)^*	0.454	MSR(783,705)	0.335
NDSI(783,740)	0.280	NDVI^*	0.504
NDSI(740,705)	0.263	S2REP^*	0.508
MSR	0.398	RSI	0.306
MSR(740,705)	0.263	SAVI^*	0.478
MSR(783,740)	0.280	PVI^*	0.501

LAI取值范围	像元个数	百分比/%
0~2	396719	18.38
2~3	1746277	80.93
3~4	14792	0.69