CAJ | 학술논문

为了定量评价漓江上游山区复杂地形水源林叶面积指数（LAI）的变化，对阔叶林、针叶林、竹林样地以TRAC仪器测定LAI，利用遥感数据计算归一化植被指数（NDVI）、比值植被指数（SR）、减化比值植被指数（RSR）、土壤调整植被指数（SAVI）、增强植被指数（EVI），并从DEM数据获取高程、坡度、坡向，提出并建立复杂地形最优多植被指数组合估算山区林地LAI的神经网络模型，利用模型对1989-2009年6景TM/ETM遥感图像估算LAI空间分布。结果表明，神经网络解决了LAI与多植被指数的非线性回归方程无法引入地形因素、且方程系数较多较难确定的问题，提高了LAI的估算精度。研究区成熟阔叶林减少代之以大片种植经济幼林，是导致林地LAI变化的原因。1989-2000年，LAI≥6.0的林地面积比例从78.8%逐年急剧下降到44.1%，LAI在1.0～6.0的林地面积比例从20.8%大幅上升到55.4%；2000-2009年，随着幼林的生长、竹林的速生，LAI≥6.0的林地面积比例逐渐上升恢复到74.5%，但仍未恢复到1989年的面积比例，相应LAI在1.0～6.0的林地面积比例逐渐下降到25.1%。研究成果为漓江上游水源林生态评估提供参考。
위료정량평개리강상유산구복잡지형수원림협면적지수（LAI）적변화，대활협림、침협림、죽림양지이TRAC의기측정LAI，이용요감수거계산귀일화식피지수（NDVI）、비치식피지수（SR）、감화비치식피지수（RSR）、토양조정식피지수（SAVI）、증강식피지수（EVI），병종DEM수거획취고정、파도、파향，제출병건립복잡지형최우다식피지수조합고산산구임지LAI적신경망락모형，이용모형대1989-2009년6경TM/ETM요감도상고산LAI공간분포。결과표명，신경망락해결료LAI여다식피지수적비선성회귀방정무법인입지형인소、차방정계수교다교난학정적문제，제고료LAI적고산정도。연구구성숙활협림감소대지이대편충식경제유림，시도치임지LAI변화적원인。1989-2000년，LAI≥6.0적임지면적비례종78.8%축년급극하강도44.1%，LAI재1.0～6.0적임지면적비례종20.8%대폭상승도55.4%；2000-2009년，수착유림적생장、죽림적속생，LAI≥6.0적임지면적비례축점상승회복도74.5%，단잉미회복도1989년적면적비례，상응LAI재1.0～6.0적임지면적비례축점하강도25.1%。연구성과위리강상유수원림생태평고제공삼고。
Leaf area index (LAI) is a crucial vegetation structural parameter that has an influence on a forest ecosystem. In order to monitor the LAI change of the water source forest over the complex mountain areas in the upper reaches of Lijiang River, the ground LAI measurements were made by using the TRAC instrument in broadleaf, coniferous, and bamboo forests during September and October 2009. Then five spectral vegetation indices, NDVI, SR, RSR, SAVI, and EVI, were calculated from TM remote sensing data, and also elevation, slope gradient, and slope aspect were obtained from DEM data. RBF neural network models were established and trained by using the different combination of vegetation index as inputs, and the ground LAI measurements as the output. After the correlation coefficients of linear regression equations and the root mean square errors between estimated LAI and measured LAI were compared, the optimum combination of a multi vegetation index with the highest correlation coefficient and the lowest error was obtained for each of the broadleaf, coniferous, and bamboo species. As the neural network model was extended to complex mountain areas by adding terrain factors to input units, it was used to estimate LAI from six TM/ETM images during 1989 to 2009. Results showed that a neural network could successfully solve the problems that the coefficients of the non-linear regression equation between LAI and multi vegetation index are difficult to calculate and the regression equation can not include terrain factors. The accuracy of LAI estimation from the optimum model added terrain factors was improved as compared to the ground LAI measurements. LAI change in the forests results from the shrinkage of the mature broadleaf forest and the increase of the young economical forests. In the eleven years of 1989-2000, the area percentage of forest with an LAI value more than 6.0 sharply decreased from 78.8% to 44.1%, and the area percentage of forest with an LAI range from 1.0 to 6.0 enormously increased from 20.8% to 55.4%. During 2000-2009, although the area percentage of forest with an LAI value more than 6.0 gradually recovered to 74.5%with the growth of young forest and the fast growing of bamboo forest, it didn’t approach the area percentage in 1989. Meanwhile, the area percentage of forest with an LAI range from 1.0 to 6.0 gradually dropped to 25.1%. The results provide a reference for the ecological assessment of water source forests in the upper reaches of Lijiang river.