Study On Daily Variation Of Forest Reflectance For Tree Height Extraction

Tree height is an important parameter in the field of forest resource management.Obtaining accurate,large-scale,spatially continuous tree height information has significance of forest resource monitoring,management,and carbon cycle research.Lidar methods are accurate but costly and difficult to cover.The method of inversion of forest average tree height based on MODIS BRF data is limited by the coarse resolution and the large observation angle,and the error 1s large.In order to meet the demand for high-resolution tree height inversion,this study proposes a new idea to construct a BRDF data set with a daily changing solar zenith angle and a fixed observation angle,named NBRF,and explore its extraction height.The main content consists of two parts:designing the light direction resetting model disk,making simulated stand lift platforms to form miniature forest observation observations,and exploring the variation of NBRF with tree height.The main results obtained from the study are as follows:(1)The NBRF curve of the stand has a trend of normal distribution at 12 o'clock:the average BRF of the stand in the morning is low,and the BRF gradually increases toward the 12 o'clock and reaches the maximum at noon.As time goes by,BRF gradually decreases again.(2)The NBRF has a tendency of decreasing along with the increase of tree height,that is,NBRF is negatively correlated with tree height,that 1s,at the same time,the average reflectance of forest stands decreases as the tree height increases.(3)The BRF simulated by RAPID model have a good degree of fitting with the observational experiment,and have a good correlation,vwhich can reflect the change law of BRF.The model simulates the BRF of scenes under different zenith angles and stand conditions.The establishment is simple,the key parameters can be adjusted according to needs,the operation speed is fast,the simulation data is accurate,and it has great accuracy and practicality.(4)Using the Multiple stepwise regression statistics,three predictors were selected,and a tree-height estimation model based on BRDF was established.Leave-one-out showed R2 with 0.987 and RBASE=0.184 m,and the fitting effect was goodStudies showed that the varied solar BRF have a good correlation with tree height,and it is hoped that,a more accurate model will be established to invert the average tree height of the real stand in the future.