Retrieval Of Forest Canopy Closure In Ji'an, Jiangxi Province Using Remote Sensing Data

Forest is the most important ecological system in the world. Forest canopy closure (FCC) is an important indicator which can represent the status of forest growth and receives attention from many researchers. The method to retrieve FCC and apply this indicator to assess the conditions and status of forest growth is a hot topic of scientific study. With the support by the project "The response of carbon and water cycles to vegetation restoration in Poyan Lake Region (4077501)" from National Science Foundation of China and the project "Study on the simulation of terrestrial carbon sinks and sources in recent 30 years and their projection (2010CB090502)", this study used Landsat TM/ETM+ remote sensing data in 2002-2003 and 2008 and geometric optical model to retrieve FCC in Ji'an City, Jiangxi province. The temporal and spatial patterns of FCC were analyzed. The linkages of FCC with land cover type were explored. The main conclusions and innovation are as followings:(1) Based on field measurements of FCC, potential factors to be used to retrieve FCC were selected from reflectance, vegetation indices, and mathematically transformed variables. These potential factors were further chosen for constructing a BP artificial neutral network model to retrieve FCC from remote sensing data according to the AIC principle. The results show that AIC criterion is applicable for selecting proper variables for retrieving FCC from remote sensing data. The BP artificial neutral network model developed can capture 86% variations of FCC among 35 sampling plots, indicating that the output from this model can be used to validate FCC inversed using 4-scale geometric optical model. The FCC in the study area is relatively high, with the FCC of 47.8% of forests above 0.7. However, FCC exhibits distinguishable spatial pattern. The forests in the westerns and eastern mountainous parts of the study area have higher FCC than forests in the central hilly part.(2) The 4-scale geometric optical model was used to retrieve FCC from 2002 to 2008 in Ji'an city. The spatial and temporal patterns of retrieved FCC in the study area during the period from 2002 to 2008 were analyzed. FCC is lower in areas near the Gan River than in other areas, mainly due to different intensities of human activities. The land cover types are mainly urban, rural areas and farmland in in areas near the Gan River. Forests are sporadically distributed here. With increase in the distance away from the river, the FCC increases due to reduced intensity of human disturbances. The land cover types are mainly broad-leaved forests, coniferous forests and mixed forests. The average of FCC in the entire study area increased from 0.55 in 2002 to 0.60 in 2008, indicating the achievements in vegetation restoration. The changes of FCC among different counties in the study areas are significant. The FCC of evergreen coniferous forests, evergreen broad-leaved forests and needle-broad-leaved mixed forests are usually above 0.6. The FCC of mixed forests and evergreen shrubs is lower, just in the range from 0.1 to 0.2.