基于GF-2号遥感影像的天山云杉林郁闭度估测研究
|
摘要
森林郁闭度作为林业综合评价的一个重要指标,对评价和监测森林生态系统的稳定性具有十分重要的意义。本研究通过提取GF-2号遥感影像的光谱信息、纹理特征因子和地形因子,结合地面样方的实测郁闭度数据,为了筛选出对郁闭度反演影响较大的因子和构建反演精度较高的估测模型,首先对各个因子与郁闭度之间的相关性进行分析,剔除相关性较低的因子;其次对各个纹理特征因子之间进行相关性分析,利用主成分法对各个波段的纹理特征因子进行分析,最终筛选出合理的纹理特征因子与影像的光谱信息、地形因子等特征,并以此作为自变量构建郁闭度估测的逐步回归模型。研究表明:以纹理特征+光谱信息+地形因子为自变量构建的估测模型拟合度为0.823,经精度检验EA%达到89.82%。总体来看,该模型基本上满足了郁闭度反演的需要,为新疆天山云杉森林生态系统的评估和实现精准数字林业提供理论依据和技术支撑。
As a significant index of forest comprehensive evaluation, forest canopy density plays an important role in monitoring and assessing forest ecosystem stability. In this study, spectral information, texture feature factor and topographic factor of GF-2 remote sensing image were extracted, and combined with the measured canopy density data of ground sample, in order to screen out the factors that have a greater impact on canopy density inversion and construct the estimation model with a higher inversion accuracy, the correlation between each factor and canopy density was firstly carried out, eliminating low correlation factor. Secondly, the correlation among texture feature factors was analyzed, the method of principal component factor analysis was carried out on the texture feature factors of each band. Finally, a stepwise regression model for canopy density estimation was constructed by screening out reasonable texture feature factors, spectral information and terrain factors. The experimental results showed that compared with the traditional method only based 0 n spectrum or topographical features, the method combined with texture features greatly improved the estimation accuracy. The optimal R2 value is 0.823, The estimation accuracy increased to 89.82%. Overall, the model basically meets the need of canopy density inversion, provides theoretical basis and technical support for the assessment of the forest ecosystem of Picea schrenkiana in Xinjiang and the realization of accurate digital forestry.
As a significant index of forest comprehensive evaluation, forest canopy density plays an important role in monitoring and assessing forest ecosystem stability. In this study, spectral information, texture feature factor and topographic factor of GF-2 remote sensing image were extracted, and combined with the measured canopy density data of ground sample, in order to screen out the factors that have a greater impact on canopy density inversion and construct the estimation model with a higher inversion accuracy, the correlation between each factor and canopy density was firstly carried out, eliminating low correlation factor. Secondly, the correlation among texture feature factors was analyzed, the method of principal component factor analysis was carried out on the texture feature factors of each band. Finally, a stepwise regression model for canopy density estimation was constructed by screening out reasonable texture feature factors, spectral information and terrain factors. The experimental results showed that compared with the traditional method only based 0 n spectrum or topographical features, the method combined with texture features greatly improved the estimation accuracy. The optimal R2 value is 0.823, The estimation accuracy increased to 89.82%. Overall, the model basically meets the need of canopy density inversion, provides theoretical basis and technical support for the assessment of the forest ecosystem of Picea schrenkiana in Xinjiang and the realization of accurate digital forestry.
引文
[1]谭炳香,李增元,陈尔学,等. Hyperion高光谱数据森林郁闭度定量估测研究[J].北京林业大学学报, 2006, 28(3):95-101.
[2]高云飞,李智广,杨胜天.基于SPOTS影像的郁闭度反演方法[J].水土保持研究, 2012, 19(2):268-270.
[3]伏虎艳,张军,舒清态.基于纹理特征的香格里拉县冷杉林郁闭度遥感估测研究[J].云南地理研究, 2014, 26(3):11-15.
[4]谷金英,冯仲科,葛忠强,等.基于TM影像数据的林分冠层郁闭度反演技术研究[J].山东林业科技, 2014, 44(1):9-12.
[5]吴飏,张登荣,张汉奎,等.结合图像纹理特征的森林郁闭度遥感估测[J].林业科学,2012, 48(2):48-53.
[6]岳彩荣.香格里拉县森林生物量遥感估测研究[D].北京:北京林业大学, 2012.
[7]赵宪文,李崇贵.基于“3S”的森林资源定量估测[M].北京:中国科学技术出版社, 2001, 18-41.
[8]龚建周,夏北成.基于大气辐射校正的广州市植被覆盖度遥感估算[J].应用生态学报, 2007, 18(3):575-580.
[9]陆国富,杜华强,周国模,等.毛竹笋快速生长过程中冠层参数动态及其与光合有效辐射的关系[J].浙江农林大学学报,2012, 29(6):844-850.
[10]郑冬梅,曾伟生,智长贵,等.三峡库区森林郁闭度的遥感定量估测[J].中南林业科技大学学报, 2013, 33(9):1-4.
[11]柏艺.基于3种不同数据源的川西南常绿阔叶林郁闭度估算模型研究[D].雅安:四川农业大学, 2016.
[12]李明诗,谭莹,潘洁,等.结合光谱、纹理及地形特征的森林生物量建模研究[J].遥感信息, 2006(6):6-9.
[13]林伟桦,杨燕琼,侯长谋.用主成分法进行林分及相关因子光谱特征研究[J].华南农业大学学报, 2004, 25(2):22-25.
[14]张雨,林辉,臧卓,等.高光谱遥感影像森林信息提取方法比较[J].中南林业科技大学学报, 2013, 33(1):75-79.
[15]付元元,杨贵军,冯海宽,等.基于高光谱维数约简与植被指数估算冬小麦叶面积指数的比较[J].农业工程学报, 2012,28(23):107-113.
[16]姚春燕,欧阳,付佳,等.玉米营养品质的快速鉴定的模型建立[J].考试周刊, 2013(59):195-196.
[17]DARVISHZADEH R, SKIDMORE A, SCHLERF M, et al. LAI and chlorophyll estimation for a heterogeneous grassland using hyperspectral measurement[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2008(63):409-426.
[18]赵蕊. Hyperion EO-1森林郁闭度反演研究[D].哈尔滨:东北林业大学, 2007.
[19]吴石磊.基于Landsat8 OLI数据的森林郁闭度反演研究[D].北京:北京林业大学,2014.
[20]张瑞英.基于多源遥感数据的森林郁闭度估测方法研究[D].呼和浩特:内蒙古师范大学, 2015.
[2]高云飞,李智广,杨胜天.基于SPOTS影像的郁闭度反演方法[J].水土保持研究, 2012, 19(2):268-270.
[3]伏虎艳,张军,舒清态.基于纹理特征的香格里拉县冷杉林郁闭度遥感估测研究[J].云南地理研究, 2014, 26(3):11-15.
[4]谷金英,冯仲科,葛忠强,等.基于TM影像数据的林分冠层郁闭度反演技术研究[J].山东林业科技, 2014, 44(1):9-12.
[5]吴飏,张登荣,张汉奎,等.结合图像纹理特征的森林郁闭度遥感估测[J].林业科学,2012, 48(2):48-53.
[6]岳彩荣.香格里拉县森林生物量遥感估测研究[D].北京:北京林业大学, 2012.
[7]赵宪文,李崇贵.基于“3S”的森林资源定量估测[M].北京:中国科学技术出版社, 2001, 18-41.
[8]龚建周,夏北成.基于大气辐射校正的广州市植被覆盖度遥感估算[J].应用生态学报, 2007, 18(3):575-580.
[9]陆国富,杜华强,周国模,等.毛竹笋快速生长过程中冠层参数动态及其与光合有效辐射的关系[J].浙江农林大学学报,2012, 29(6):844-850.
[10]郑冬梅,曾伟生,智长贵,等.三峡库区森林郁闭度的遥感定量估测[J].中南林业科技大学学报, 2013, 33(9):1-4.
[11]柏艺.基于3种不同数据源的川西南常绿阔叶林郁闭度估算模型研究[D].雅安:四川农业大学, 2016.
[12]李明诗,谭莹,潘洁,等.结合光谱、纹理及地形特征的森林生物量建模研究[J].遥感信息, 2006(6):6-9.
[13]林伟桦,杨燕琼,侯长谋.用主成分法进行林分及相关因子光谱特征研究[J].华南农业大学学报, 2004, 25(2):22-25.
[14]张雨,林辉,臧卓,等.高光谱遥感影像森林信息提取方法比较[J].中南林业科技大学学报, 2013, 33(1):75-79.
[15]付元元,杨贵军,冯海宽,等.基于高光谱维数约简与植被指数估算冬小麦叶面积指数的比较[J].农业工程学报, 2012,28(23):107-113.
[16]姚春燕,欧阳,付佳,等.玉米营养品质的快速鉴定的模型建立[J].考试周刊, 2013(59):195-196.
[17]DARVISHZADEH R, SKIDMORE A, SCHLERF M, et al. LAI and chlorophyll estimation for a heterogeneous grassland using hyperspectral measurement[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2008(63):409-426.
[18]赵蕊. Hyperion EO-1森林郁闭度反演研究[D].哈尔滨:东北林业大学, 2007.
[19]吴石磊.基于Landsat8 OLI数据的森林郁闭度反演研究[D].北京:北京林业大学,2014.
[20]张瑞英.基于多源遥感数据的森林郁闭度估测方法研究[D].呼和浩特:内蒙古师范大学, 2015.