STENSON, GRAHAM,STEPHEN (2016) An experimental investigation into ALS uncertainty and its impact on environmental applications. Masters thesis, Durham University.
|PDF - Accepted Version|
This study takes an experimental approach to investigating the reliability and repeatability of an airborne laser scanning (ALS) survey. The ability to characterise an area precisely in 3-D using ALS is essential for multi-temporal analysis where change detection is an important application. The reliability and consistency between two ALS datasets is discussed in the context of uncertainty within a single epoch and in the context of well known point- and grid-based descriptors and metrics. The implications of repeatability, verifiability and reliability are discussed in the context of environmental applications, specifically concerning forestry where high resolution ALS surveys are commonly used for forest mensuration over large areas.
The study used a regular 10-by-10 layout of standard school tables and decreased the separation from 2.5 metres apart to 0.5 metres in order to evaluate the effects of object separation on their detection. Each configuration was scanned twice using the same ALS scanning parameters and the difference between the datasets is investigated and discussed.
The results quantify uncertainty in the ability of ALS to characterise objects, estimate vertical heights and interpret features / objects with certainty. The results show that repeat scanning of the same features under the same conditions result in a laser point cloud with different properties. Objects that are expected to be present in 40 points per metre2 laser point cloud are absent, and the investigation reveals that irregular point spacing and lack of consideration of the ALS footprint size and the interaction with the object of interest are significant factors in the detection and characterisation of features.
The results strongly suggest that characterisation of error is important and relevant to environmental applications that use multi-epoch ALS or data with high resolution / point density for object detection and characterisation
|Item Type:||Thesis (Masters)|
|Award:||Master of Science|
|Keywords:||LiDAR, Uncertainty, Object Detection|
|Faculty and Department:||Faculty of Social Sciences and Health > Geography, Department of|
|Copyright:||Copyright of this thesis is held by the author|
|Deposited On:||26 Sep 2016 11:46|