University of Twente Student Theses


Mapping carbon stock in trees outside forest: Comparing a very high resolution optical satellite image (Geo-Eye) and airborne LiDAR data in Chitwan, Nepal

Ngwayi, Itoe Constantine Nfor (2012) Mapping carbon stock in trees outside forest: Comparing a very high resolution optical satellite image (Geo-Eye) and airborne LiDAR data in Chitwan, Nepal.

[img] PDF
Abstract:In this study, object based image analysis (OBIA) was used to compare the segmentation accuracy for trees outside forest (ToF) using a VHR Geo-Eye satellite image and airborne LiDAR data in mapping carbon stock. While CPA that is a proxy for DBH was used for the Geo-Eye, CPA and canopy height model (CHM) was used for the airborne LiDAR data from which a multiple regression analysis was applied. DBH was randomly measured from all the available trees in the field alongside the tree height and the crown diameter. Quantification of AGB was estimated with the use of a mixed allometric equation for tropical forest described in Chave et al., (2005). Regression equations were developed between the field measured parameters, field measured DBH and remote sensing parameters and the result indicated that the coefficient of determination, R2>0.5. The accuracy assessment result of the segmentation process was 79% and 74% for the airborne LiDAR data and Geo-Eye respectively. The result of the relationship between the field measured and that of remotely sensed data was R2=0.74 (RMSE=38.24%) and R2=0.73 (RMSE=41.18%) respectively for the LiDAR and Geo-Eye data sets. The highest quantity of predicted carbon was from the airborne LiDAR data that had an R2 of 0.90 and RMSE of 14.24%. However, the best model is from the multiple regression of the LiDAR data parameters of CHM and CPA with R2=0.69 (RMSE=11.53%). The predicted carbon of the Geo-Eye had R2=0.51 with 42.24% RMSE. The result of OBIA gives airborne LiDAR an edge to Geo-Eye image in estimating CPA due to her 3-D characteristic in determining forest structures. The t-test result rejected the null hypothesis that there was a significant difference between the segmented CPAs of the two data sets at 95% confidence interval. The major problem of this study was the identification of pruned tree crowns on the Geo-Eye image. Further investigation should be carried out with a stratified random sampling of the crown size to uniformly distribute the data set of the study. These findings will be useful to the CFUGs and the Forestry Department of Nepal to propose integration of ToF into their natural resource management scheme for carbon stock estimation. Nevertheless, this study encourages the use of airborne LiDAR data for carbon stock estimation for trees outside forest in a tropical environment like Nepal. Keywords: aboveground carbon stock, OBIA, trees outside forest, Geo-Eye, LiDAR, tree height, crown projection area, Nepal
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page