Aboveground Forest Biomass Estimation Based On Landsat TM And ALOS PALSAR Data

Forest biomass is an important indicator of forest ecosystem carbon sequestration capacity,and its variation is a comprehensive reflection of the forest ecosystem material cycle and human activities.Estimates of forest biomass has great significance for regional ecosystems and climate change,it has been the focus of global change research.With the popularization and development of remote sensing technology provides the basis for estimates of forest biomass quickly and efficiently carry out large-scale,but the data saturation problem is well recognized and has been regarded as an important factor resulting in poor forest aboveground biomass(AGB)estimation performance However,no research has examined the saturation values for different vegetation types such as coniferous forest and broadleaf forest,due to unavailability of sufficient number of ground truth data.The objective of this study is quantitative to identify data saturation values in Landsat imagery for different vegetation types in the Midwest of Zhejiang Province and to explore approaches to improve forest AGB estimation.The Landsat TM imagery,ALOS PALSAR data,digital elevation model(DEM)data,and field measurements datas were used.Correlation analysis and scatterplot were used to identify the best spectral band for AGB estimation,a spherical model was used to quantitatively determine saturation value of AGB for each vegetation type.Considering the impact of vegetation types and topographical features of slop aspect to AGB,a stratification of vegetation types and/or slope aspects was used to examine the potential to improve AGB estimation performance by developing AGB estimation model for each category.Stepwise regression analysis,which was based on Landsat spectral signatures,backscatter coefficients of PALSAR and theirs textures using Grey-level co-occurrence matrix(GLCM)were used to develop AGB estimation models for different scenarios: non-stratification,stratification based on either vegetation types,slope aspects,or the combination of both vegetation types and slope aspects.The results indicates that:(1)The vegetations of study saturated at 156 Mg/ha,different vegetation types have various AGB saturation values,pine forest(159 Mg/ha)> mixed fores(152 Mg/ha)> Chinese fir(143 Mg/ha)>broadleaf forest(123 Mg/ha)> bamboo forest(75 Mg/ha)> shrub(55 Mg/ha).(2)Compared with non-stratification,the stratification based on vegetation types and slope aspectsprovided smaller root mean squared errors(RMSE)for low(<30Mg/ha)and high(>120Mg/ha)AGB respectively.The AGB estimation models based on stratification of both vegetation types and slope aspects provided the best estimation performance with the smallest RMSE of 24.4 Mg/ha.(3)This research indicates the importance of stratification and integrating multi-source data in mitigating data saturation problem,thus improving AGB estimation performance.However,this research found that the presence of AGB estimating in low biomass(eg <60 Mg / ha)overestimated and high biomass(eg> 120 Mg / ha)underestimate.Further research is needed to reduce the saturation data,improve the biological amount estimation accuracy.