Introduction

Lidar point clouds and their derived products have become an important source of detailed spatial information for invironmental research. The data contained in a lidar point cloud can be analysed in multiple different ways and lends itself to detailed statistical modelling. Processing point clouds directly can be challenging due to the size of the files. Very large data sets covering regions, rather than sites, are thus best processed on more powerful platforms than a laptop computer. However smaller areas can easily be analysed using open source or free tools.

Obtaining some free Lidar data

The environment agency in the UK has made available a large number of files containing both processed and unprocessed lidar data at this site.

http://environment.data.gov.uk/ds/survey/index.jsp#/survey?grid=SU20

The pre-processed data consists of digital surface models and digital terrain models at a range of resolutions. More on what these files consist of later. They can be used to avoid having to process large amounts of data yourself. However it is important to understand how they have been produced. Raw point clouds are the starting point in any new analysis based on fresh data provided by a lidar flight. We will start off working with a single tile of raw data in order to illustrate the process. The file we will use is taken from a well wooded area in the New Forest.

SU2408_P_8044_20121204_20121205.laz

What does Lidar data consist of?

Raw lidar data is fundamentally quite simple in structure. It consists of a cloud of points with x, y and z coordinates. Asking for information on the file (which has been compressed to laz format from the original las) provides this output.

LASzip compression (version 2.4r1 c2 50000): POINT10 2 GPSTIME11 2 reporting minimum and maximum for all LAS point record entries

There are some components in addition to x,y,z that can be used in processing. An important element is the intensity of the return and the order in which the return arrives at the sensor. First returns represent the top of a forest canopy while last returns have penetrated vegetation and thus usually return from the ground. High intensity returns come from harder surfaces than low intensity returns. The combination of the number of the return and intensity allows the points to be classified by running an algorithm over the data.

  1. = Never classified
  2. = Unassigned
  3. = Ground
  4. = Low Vegetation
  5. = Medium Vegetation
  6. = High Vegetation
  7. = Building
  8. = Low Point
  9. = Reserved

The points in this file have already been classified and we can use this when processing the data.

Visualising point clouds using Plas.io

WebGL is now integrated into most web browsers and there are some nice tools online that can be used to open and view a raw point cloud. For example http://plas.io/

Loading the point cloud produces this visualisation after playing with the intensity slider.