ArcGIS supports LiDAR! It supports both XYZ and LAS files. It can read and display these formats as point clouds or surfaces, in both 3D and 2D. These can be viewed in ArcMap, ArcScene, ArcGIS Pro and as elevation services through ArcGIS for Server. But this is common knowledge right? ArcGIS has been able to do this since 9.3 and we all knew that right?
Well I thought this was the case. Often though it is not. Many people are today just starting to investigate the best tools for LiDAR exploitation not realising the capabilities of the ArcGIS Platform and the wealth of knowledge available to support their investigation. If you want a good place to start take a look at this web page What is LiDAR data?.
But let’s take a step back from the tools right now and ask yourself for what purpose do I want LiDAR data? Do I want it for flood modelling, DEM creation, building extraction or power line detection. Do I want to just differentiate between ground and non-ground or do I need it classified into buildings, trees, ground and water? The reasons for asking these questions is that the answers will dictate the order you place for LiDAR.
Firstly let’s take a look at the classification of LiDAR. Generally speaking you will be asked if you want one of 3 formats, unclassified, ground and nonground or fully classified.
- Very rarely these days do you see unclassified LiDAR however it is the cheapest delivery format but the encumbrance is on you the customer to then separate and classify the returns.
- Ground and non-ground is good if you want to strip away all the above ground features and just perform DEM creation or flood modelling. In the early days of LiDAR this was the most common format and even today you still see a lot of LiDAR being delivered this way.
- Fully classified. LAS is a specification for the delivery of LiDAR data as defined by the international body American Society for Photogrammetry and Remote Sensing (ASPRS). The specification defines how to classify LiDAR data. The latest specification is LAS standard 1.4 full detail can be found at http://www.asprs.org/a/society/committees/standards/LAS_1_4_r13.pdf
Obviously the further you go down the list the more expensive the data becomes. If you have the funds then always go for fully classified as ArcGIS can take advantage of the classification and you can expose the richness in the data.
Second to this is resolution. This is perhaps the least thought of factor when someone orders LiDAR data or consumes LiDAR. I often hear people say they already have LiDAR covering an area and want to perform analysis. When asked the resolution? it is often met with a blank response. I find that resolution of LiDAR is the least understood aspect when someone is starting out. In fact resolution should be paramount in peoples understanding of LiDAR data.
LiDAR resolution is important as it dictates what can be performed with the data. It can be considered in the same way image resolution is considered. The better the resolution the more detail that can be discerned. For example in a Landsat image a building can be identified by its spectral signature but it’s exact shape cannot be defined as the resolution is 30m. However with an aerial photograph where the result on is 10cm the shape of a building is easily identified. The same holds true with LiDAR data.
The most common resolution that people have when they are starting out is 1-2 returns per square meter. For topographic mapping, DEM creation and vegetation analysis this is resolution is great. Large areas of the lands surface can be collected and analysed for relatively inexpensive datasets but if you want to be able to perform object extraction or building recognition then this resolution is not good enough. You need at least 8-10 returns per square meter. With this buildings can be identified and the roof facets extracted. Higher resolutions are also being acquired today. 40-60 returns per square meter. These resolutions are excellent for power line extraction and vegetation encroachment analysis.
So what resolution is best? Well that depends on your purpose. I generally find people start with the lowest resolution data, but as they start to work with it, it doesn’t quite do what they expect. They try to push its capabilities and are often disappointed. If you are unsure I generally go for the medium resolution LiDAR dataset. It provides excellent terrain analysis but also allows you to extend your capabilities to feature extraction at a later time.
So if you are in the market for a LiDAR dataset consider the resolution and consider the classification. If you are unsure of the final intended use, a fully classified, medium resolution dataset is best, if your budget can stretch. It provides the most flexibility for now and future use. Higher resolution datasets are excellent but often are not required unless you are an electrical utility or mining company.