Research On Forest Canopy LAI Remote Sensing Inversion Method

The analysis of terrestrial ecosystem carbon cycle and global climate change has increasing demands on the accuracy of remote sensing LAI product, especially in forest area. So LAI parameters are important in the accuracy of the simulation results about the carbon cycle and it becomes an important scientific problem of vegetation quantitative remote sensing studies about how to effectively improve the accuracy of LAI products in forest area. In the current study of inversion of remote sensing about leaf area index in forest canopy, most of the existing model algorithms ignore the impact of forest background reflections, which will generally result in overestimation about LAI products in the forest areas. Therefore,-it’s important to remove the impact of forest background reflections by taking the inversion forest canopy reflectance into consideration in order to reduce the uncertainty of the inversion results about LAI in forest canopy. In this paper, we chose Five Forest Camp of Xiaoxinganling mountain in the northeast of China and Wuyishan in the southeast of China as the study area. With the use of MODIS BRDF multi-angle remote sensing data products, we calculated pixel components in different proportions based on 4-Scale Remote Sensing Mechanism Model, succeeded in achieving a separation between forest canopy reflectance and undergrowth background spectrum, inversed forest canopy LAI of the study area and did some spatial analysis in LAI of different types of forest canopy.The main conclusions in this paper were as follows:(1) The use of MODIS BRDF model parameters products can effectively inverse forest canopy reflectance. Forest canopy reflectance were lower in the red band, mainly between 0-0.05 in two of the study areas, but it was relatively higher in the near infrared band, mostly between 0.1-0.25,for the reason that shaded canopy components downed the total canopy reflectance values as a whole, making it slightly below normal vegetation spectral.(2) Canopy LAI inversion models were respectively constructed by the use of measured data of canopy LAI and forest canopy vegetation index such as NDVI、SR and MSR, among which the exponential model, the quadratic polynomial model with SR and measured LAI data had the better result and R2 calculated by this model respectively reached 0.66 and 0.55 in Five Forest Camp and Wuyishan.(3) Canopy reflectance in different types of forest had different characteristics. In the red and near-infrared bands, there was no significant difference in the canopy reflectance rate between coniferous and mixed forests, but reflectance rate of broad-leaved forest was relatively significantly higher than that of coniferous and mixed forest. On the distribution of LAI in forest canopy, reflectance rate of broad-leaved forest was highest, followed by mixed and coniferous forest, which also showed the effectiveness of the inversion of LAI in forest canopy.(4) Analysis of the impact of understory background reflections for inversion precision of LAI in forest canopy showed that forest background reflections had a great impact on the inversion of LAI in forest canopy. What’s more, the accuracy of LAI inversion model based on canopy reflectance had been more greatly improved than that without taking understory background spectrum into account, which further confirmed the importance of the canopy reflectance for accurate retrieval of LAI in the forest canopy.