Study On Vegetation Canopy Nitrogen Inversion Of Burned Land In Greater Khingan Range,Inner Mongolia

Forest fire is an important ecological factor in forest ecosystem,and its occurrence will have a significant impact on forest ecosystem.With the development of remote sensing technology,the monitoring of biochemical parameters in the process of post fire vegetation restoration has become a hot issue.In this study,taking the burned lands in the Greater Khingan Range as the research object,taking remote sensing technology as the means,combined with field investigation and indoor analysis,the nitrogen estimation models of leaves and canopy in the cold temperate forest were constructed to realize the inversion of canopy nitrogen in different time series of burned land in the Greater Khingan Range,to reveal the effect of forest fire on forest nitrogen accumulation in the cold temperate zone.It can provide scientific basis for the restoration of burned land and the study of nitrogen cycle in the Northeast China.The main results were as follows:1.The average nitrogen content of tree leaves in the burned land was lower than that in the control land;The average nitrogen content of tree leaves in the burned area in 1987was higher than that in 2003;The trends of tree leaf spectral curves were basically the same,and the difference was reflected in the"green peak"and"red edge".The reflectances of the burned plots in the green and near-infrared bands were higher than that of the control plots.2.Based on the correlation analysis of spectral reflectance and leaf nitrogen content of trees measured on the ground,a total of 11 variables（after collinearity elimination）were selected to participate in the construction of the leaf nitrogen content model,namely R₄₅₀,R₄₅₃,R₄₅₆,R₄₆₄,R₄₆₇,R₄₇₅,R₄₇₆,R₆₅₂,R₆₆₈,R₆₈₂,R₆₈₉,D_y,SD_y,and SD_r.3.In terms of leaf nitrogen content modelling,the multiple linear regression model with independent variables of R₄₅₀,R₆₅₂,R₆₈₂,and D_y,the BP neural network model with10 hidden layer nodes,and the random forest regression model with R₄₅₀,R₄₆₄,R₆₅₂,and R₆₆₈as input data had higher accuracy.The BP neural network model had the best prediction effect on the nitrogen content of tree leaves by comparisons.4.Based on LAI and leaf scale nitrogen content inversion model,the canopy scale nitrogen content inversion of the burned land was realized.The results showed that the canopy nitrogen contents of the burned land in 1987 and 2003 were close,but it still did not reach the nitrogen content level of the control canopy.The nitrogen content of the burned land canopy in 1987 was significantly higher than that in 2003,which reflected the effect of vegetation restoration time on nitrogen accumulation.