基于Landsat8 OLI的内蒙古大兴安岭森林健康评价遥感模型研究

Remote Sensing Model of Forest Health Assessment in Greater Khingan Range,Inner Mongolia Based on Landsat8 OLI

为了降低森林健康评价中野外调查人力、物力的大量消耗及对大尺度森林健康评价的需求,对森林健康评价遥感模型的研究具有十分重要的现实意义。本文分别选取内蒙古大兴安岭的针叶林(汗马自然保护区)、针阔混交林(五岔沟林场)及落叶阔叶林(赛罕乌拉自然保护区)为研究对象,分别针对3种森林类型设立30 m×30 m样地(汗马18块,五岔沟90块及赛罕乌拉53块),利用Landsat8 OLI影像数据结合敏感分析法和层次分析法进行了指标筛选、指标权重确定及建立森林健康评价遥感模型;遥感模型检验是以森林野外调查数据计算的健康值为目标值,对遥感模型计算健康值进行相关性检验,遥感模型计算健康值与目标值的相关性显著(P<0.01,R^(2)=0.753 8),针叶林、针阔混交林、落叶阔叶林的遥感模型计算健康值与目标值的相关性分别为(P<0.01,R^(2)=0.530 6)、(P<0.01,R^(2)=0.744 0)、(P<0.01,R^(2)=0.826 3);由于遥感模型计算的健康值与目标值相近,可以判断出本研究所建立的遥感模型适用于内蒙古大兴安岭地区的森林健康评价。

In order to reduce the huge consumption of manpower and material resources in field investigations during the forest health assessment,and the demands for large-scale forest health assessment,the remote sensing model for the forest health evaluation was studied in this paper. The coniferous forests(Hanma National Nature Reserve),coniferous and broad-leaved mixed forests(Wuchagou Forest Farm),and deciduous broad-leaved forests(Saihanwula Nature Reserve)in Greater Khingan Range were selected as the research objects,respectively. 30 m×30 m sample plots were set up in the three kinds of forests(18 blocks in Hanma Nature Reserve,90 blocks in Wuchagou Forest Farm,and 53 blocks in Saihanwula Nature Reserve),respectively. The Landsat8 OLI image data combined with the sensitive analysis method and Delphi-AHP were used to conduct the index screening,index weight determination,and establishing the forest health evaluation remote sensing model. The remote sensing model test took the health values calculated by the forest field investigation data as the target values. The correlation test results of the health values of the remote sensing model showed that the correlation between the health values calculated by the remote sensing model and the target values was significant(P<0. 01,R^(2)=0. 753 8). The correlations between the health values and the target values of the coniferous forests,coniferous and broad-leaved mixed forests,and deciduous broad-leaved forests were(P < 0. 01,R^(2)= 0. 530 6)(P < 0. 01,R^(2)= 0. 744 0)(P < 0. 01,R^(2)= 0. 826 3),respectively. Since the health value calculated by the remote sensing model was close to the target value,it could be concluded that the remote sensing model established in this paper was applicable to the forest health evaluation in the Greater Khingan Range of Inner Mongolia.