Remote Sensing Lab 5

Goal and Background

LiDAR is an expanding area of remote sensing that has a large potential for job growth. The field of LiDAR will become more prominent as the data becomes more financially feasible. Thus, it is important to understand how to process LiDAR data. The goal of this lab is to gain basic understanding and skills in LiDAR data structure and processing. Specifically, creating surface models, terrain models, and intensity images. In this lab LiDAR point clouds were in LAS file format.

Methods

All of the processes applied to the LiDAR point cloud were done in ArcMap by ESRI.

Creation of a LAS Dataset

First the LAS files provided by Eau Claire County were added to a new LAS Dataset created in ArcCatalog. Next statistics were run on the LAS files in the properties tab of the LAS Dataset (figure 1).

Figure 1. Statistics calculated for the LAS Dataset.

The statistics are based on the information within the headers of each LAS file. Both the statistics for each LAS file and for the overall LAS Dataset are used in quality control checks. One quality control check is comparing the minimum and maximum Z values with the known elevation range of the study area, in this case the City of Eau Claire, to determine whether the values reported in the statistics matched the known elevation range.  After acquiring the minimum and maximum Z values, the horizontal and vertical coordinate systems were defined for the dataset and the range was compared to the z values.

Exploring LiDAR Point Cloud Data

After activating the LAS Dataset tool bar, the dataset was compared to the Eau Claire County shapefile to validate the data is in the correct location (figure 2).

Figure 2. Point cloud area (red) compared to the Eau Claire County shapefile to validate location accuracy.

Next the slope of natural land surface features and cultural, or man-made, features was compared. Filters for the point cloud data were also explored. There are two types of filters: Classification Codes and Returns.  Classification codes refer to the type of object that has reflected the laser pulse. Returns refers to the order of energy that returns to the sensor.

The profile of features, such as a bridge, were displayed using the LAS Dataset Profile View tool and the LAS Dataset 3D View (figure 3, figure 4).

Figure 3. Profile of a bridge on the Chippewa River, WI using the LAS Dataset Profile View Tool.

Figure 4. Profile of a bridge and the surrounding area on the Chippewa River, WI using the LAS Dataset 3D View.

 Generation of DSM, DTM, Hillshades, and Intensity Image

Before any products from this point cloud could be produced, the average nominal pulse spacing (NPS) had to be determined. In this case, the NPS was 1.485. Next, the LAS Dataset to Raster tool was used to create the Digital Surface Model (DSM). The DSM was based on the first return of the LiDAR data. The Binning interpolation method with maximum cell type and natural neighbor void filling was used to accomplish this task. A hillshade for the DSM was then made. The DSM shows the first return data points.

Next a Digital Terrain Model (DTM) was created from the LAS Dataset. The ground filter was applied and the LAS Dataset Raster tool with binning interpolation, minimum cell assignment type, and natural neighbor void filling method created the raster file. Like the DSM a hillshade for the DTM was also made for viewing purposes. In addition, the swipe tool in the effects tool bar was used to compare the hillshade of the first return and the ground.

Since intensity is always capture by the first return, the first return filter was reapplied to the LAS Dataset. The intensity is calculated by the LAS Dataset to Raster Tool used in the DSM and DTM creation. While binning interpolation was used, the cell assignment type was average and the void filling method was natural neighbor.

Results

Figure 5. Z values for the LAS Dataset.

The minimum z value for the dataset is 517.85 feet. The maximum z value is 1845.92 feet. the z range is 1328.07 feet. The minimum and maximum z values in the dataset do not portray a realistic elevation range for the City of Eau Claire because Eau Claire has a very small elevation change, certainly not a difference of 1328.07 feet.

Figure 6. Digital Surface Model (DSM) (top) and accompanying hillshade (bottom) for the City of Eau Claire, WI.

Figure 7. DSM with underlying hillshade raster.

The DSM and hillshade raster in figure 6 are overlaid in figure 7 to add depth to the DSM raster.  Most of the features visible in the first return hillshade are the tops of trees and the tops of buildings. In thickly forested areas, the trees appear as one bumpy surface. There are many small blob-like shapes scattered in the city blocks. In some areas flat areas are present below the trees which are streets. There are also many square or rectangular objects throughout the image that represent buildings but they appear shallow in height. Along the Chippewa River there are several places where a cone feature appears. This is due to the low point density of the LIDAR which creates inaccurate elevation readings.

Figure 8. DTM with underlying hillshade raster.

Many of the features that appeared in the DSM are not present in the terrain model. The landscape is dominated by the outline of the city blocks and the path of the Chippewa River. Flat surfaces appear in Half Moon Park where there was previously a bumpy surface representing the heavy tree cover. The most prominent features in the bare Earth depiction are the old and new flood plain. Although the outline of the city blocks can be seen in both the dark gray and white portions of the picture, the difference between the dark gray and white areas is more prominent. The white, or higher elevation area is the Wissota Terrace, the previous floodplain before down-cutting of the river occurred. The dark gray area, denoting low elevation, is the current floodplain of the Chippewa River.

Figure 9. Intensity image from the LiDAR data viewed in Erdas Imagine.

 The Intensity image shows the first return LiDAR point data. LiDAR data is collected in the near infrared (NIR) at a wavelength of 1.064 micrometers. This image provides a clearer image of objects within the study area and thus can be used as ancillary information in identifying objects in the DSM and DTM. Although this image is captured within the NIR, it is not quite the same  as other NIR sensors such as Landsat TM. LiDAR data is derived from an active remote sensing system that sends out its own energy and records the energy pulses that return. Satellites such as Landsat 5 are passive systems that record reflected electromagnetic energy from the sun. The data is also recorded differently. LiDAR records reflected energy as points while satellite imagery records reflected energy as brightness values in pixels.

After the completion of this lab exercise, basic knowledge and skills were gained in processing LiDAR data.

Sources

Lidar point cloud and tile index from Eau Claire County, 2013.

Price, M. (2016). Mastering ArcGIS (7th ed.). New York, New York: McGraw-Hill.