University of Twente Student Theses


LiDAR and Low vegetation: Extraction of structural characteristics and DTM error

Neelakantan, Neeraja (2010) LiDAR and Low vegetation: Extraction of structural characteristics and DTM error.

[img] PDF
Abstract:LiDAR has proved to an effective tool for mapping terrain and studying vegetation structural characteristics. Past researches have pursued studies related to low vegetation and its disturbing influence in determining the true elevation of the terrain using LiDAR. This study deals with extracting the vegetation structural characteristics of low vegetation and to determine the DTM error of the terrain using a new improved scanner which has a better vertical accuracy (1.5 cm) than the scanners that were used previously by other studies. This is so far have been challenging task because the range of low vegetation is well within the noise of the scanners. The scan angle for each plot is also determined to see if there is any influence of scan angle on the DTM error prediction and the vegetation structural characteristics estimation. The main methods involved in this study are hierarchic robust interpolation used for filtering the terrain points from the non terrain points. Second order spine interpolation has been used as a tool for interpolating two surfaces such as the DTM surface and all the points in order to find the height of the vegetation points above terrain. Vegetation density is found by employing the method of Vegetation Area Index. The scan angle and DTM error do not show any relationship with each other. The error check for the GPS used for field measurements was tested on a plot of asphalt and the error was found to be 0.9 cm. The field data about vegetation height and laser derived height of vegetation points showed good correlation for points above the height of 20cm. From the regression analysis performed between vegetation height and shift, there is a strong correlation seen for vegetation height data ranging from 3 to 7 cm. and shift values between 3 to 15cm. For vegetation height less than 6 cm, a mathematical relationship could be established with corresponding DTM error. Amongst other first order statistical measures that were found, only std deviation and 93rd percentile found to have a strong correlation with vegetation height. Skewness and kurtosis proved poor correlation. For further research, It is recommended to use texture approach for extracting vegetation structural characteristics.
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