黑龙江大兴安岭地区森林枯落物含水率遥感反演

发布时间：2018-01-10 11:10

本文关键词：黑龙江大兴安岭地区森林枯落物含水率遥感反演　出处：《东北林业大学》2016年硕士论文　论文类型：学位论文

【摘要】：森林枯落物含水率是影响森林火灾发生的重要因素,新兴发展的遥感技术为森林枯落物含水率预测提供了快捷、方便而又可靠的途径。其中的MODIS (Moderate Resolution Imaging Spectroradiometer)数据由于其周期短、宏观性强、稳定、经济的特点,以逐渐成为自然资源探查、灾害预测预报、国土统计调查、环境保护等研究的重要手段。本研究利用MODIS BRDF(Bi-Directional Reflectance Function)和LAI (Leaf AreaIndex)数据结合基于实测光谱建立的森林枯落物含水率的估算模型对大兴安岭地区森林枯落物含水率进行反演。主要的研究结果如下：(1)利用烘干法实测森林枯落物含水率和SVCHR-1024i全波段地物光谱仪采集其光谱数据,并用统计方法建立原始、一阶导数和去包络线三种不同形式光谱的枯落物含水率的模型,模型的决定系数(R2)、平均相对误差(MRE),均方根误差(RMSE)分别为：0.422、0.17、0.54；0.489、0.33、1.15；0.566、0.306、0.45。通过相关系数分析得出水分敏感波段为468nm～868nm、 998nm～1028nm、1158～1218nm、1318-2248nm、 2378～2488nm (呈现极显著)418nm～468nm、878～988nm、1018nm～1068nm、 1228nm～1248nm、2258nm～2308nm (呈现显著),且在1498nm和1508nm处出现极值分别为0.69和-0.69。(2)借助陈镜明教授的4-scale模型,结合MODIS BRDF和LAI数据产品对大兴安岭地区森林枯落物的反演,结果表明：在建立查找表对多次散射因子(M)和四分量(KT,KZT,KG,KZG)的求解、在M求解中,分别对0。、25。、45。、60。这四个观测角度的模型拟合进行评价和筛选,最后确定45。和60。为最优角度。在选取的MODISBRDF图像5、6波段和多次散射因子(M)建立的模型R2分别为：0.577、0.583;0.95、0.947。在四分量求解中建立的模型R2分别为：0.769、0.867、0.937、0.879;0.822、0.83、0.917、0.82。在之后反演制图中模型的平均绝对误差(MAE)等于61.4%,均方根误差(RMSE)等于696.6%。结果表明：借助MODIS数据反演森林枯落物含水率能够提供准确的连续性好的枯落物含水率数据,对林火预测预报具有一定的意义。
[Abstract]:Forest litter moisture is an important factor affecting forest fire, remote sensing technology developing moisture prediction provides a quick way for forest litter, convenient and reliable. The MODIS (Moderate Resolution Imaging Spectroradiometer) data because of its short cycle, macroeconomic stability, strong economic characteristics, to gradually become the natural resources exploration, disaster forecast, land survey, an important means of environmental protection and so on. This study uses MODIS BRDF (Bi-Directional Reflectance Function) and LAI (Leaf AreaIndex) to establish the estimation model of the measured spectrum of forest litter moisture on a forest in Greater Khingan Range area litter moisture inversion based on data combined. The main results are as follows: (1) the use of drying method to measure forest litter moisture and SVCHR-1024i full band spectrometer collection The spectral data, and use statistical methods to establish the original, first derivative and envelope to three kinds of spectra of the moisture content of litter in the model, the determination coefficient of the model (R2), the average relative error (MRE), root mean square error (RMSE) were: 0.422,0.17,0.54; 0.489,0.33,1.15; 0.566,0.306,0.45. through the analysis of the correlation coefficient the moisture sensitive bands of 468nm ~ 868nm, 998nm ~ 1028nm, 1158 ~ 1218nm, 2378 ~ 2488nm (1318-2248nm, 418nm showed significant) ~ 468nm, 878 ~ 988nm, 1018nm ~ 1068nm, 1228nm ~ 1248nm, 2258nm ~ 2308nm (significant), and the 1498nm and 1508nm value were 0.69 and -0.69. (2) by using the 4-scale model of Professor Chen Jingming, MODIS BRDF and LAI inversion, combined with data on a forest litter in Greater Khingan Range area. The results showed that: in the lookup table is established on multiple scattering factor (M) and four component (KT, KZT, KG KZG), the solution in solving M, respectively for 0., 25., 45., 60. model fitting the four observation angle evaluation and screening, to determine the final 45. and 60. for the optimal angle. In the 5,6 band MODISBRDF image selection and multiple scattering factor (M) model developed by R2 were: 0.577,0.583 R2; model 0.95,0.947. in the four component solution established respectively: 0.769,0.867,0.937,0.879; 0.822,0.83,0.917,0.82. after model inversion in mapping the mean absolute error (MAE) equal to 61.4%, the root mean square error (RMSE) is equal to the result of 696.6%. showed that with the help of MODIS data inversion of forest litter moisture can provide accurate good continuity of litter moisture content data, which has a certain significance for forest fire forecast.

【学位授予单位】：东北林业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S762.1

【参考文献】