Introducing... Raster Data

 

Exercise Basics:

Explore the structures of raster data, how attributes are connected to them, and the real world that they represent. 

We’ve put together a series of questions to help you focus on the unique characteristics of raster data and to reinforce some of the critical details you’ve learned in lecture.  

Get Set Up: 

1. Set up your workplace folder
2. Open ArcGIS Pro, start a new project, and save the new project to your folder
3. Set your Folder Connections
4. Add the data from the Introducing Rasters folder from Canvas

 

Part 1

Get Your Data Set Up: 

All of the digital elevation models (DEMs) were downloaded from the UT AGRC.

Stack your data in this order: 

1. Places_of_interest
2. usu_50cm_DEM
3. usu_50cm_DEM_hs
4. usu_50cm_DSM
5. usu_50cm_DSM_hs
6. usu_5m_dem
7. usu_5m_hs

You should notice: There are 3 pairs of data.
Each pair has an elevation model and its corresponding hillshade (*_hs). Hillshades are a visualization of the terrain represented in the elevation model. Elevation rasters and hillshades go together to display the landscape.

Each pair represents the ground in a different way.

• One pair uses a larger pixel with less clarity (usu_5m_dem).
• Two have smaller pixels (0.5 meter) which provides more detail. 
  • usu_50cm_DEM: maps the elevations of the ground (DEM = Digital Elevation Model) and
  • usu_50cm_DSM: maps the elevations of the tops of everything on the landscape when the data was collected (DSM = Digital Surface Model, also called a First Return model). 

You will have to pay particular attention to the names of the datasets in this lab.  

The following is a suggestion for organization. It is not necessary, but note that you will need to keep this data organized.


Group the DEMs with their hillshades in the table of contents (to keep things organized): 

1. Use the shift key to select each DEM and it’s hillshade

a. Click on the USU_5m_DEM
b. Shift click on the USU_5m_hs

2. Right click on either layer > Group

 

3. Name the group “5 meter DEM"

Repeat for the 50 cm DEM and its hillshade. 
Name it “50 cm DEM” (digital elevation model) Repeat for the 50 cm first return DSM and its hillshade.

Name it “50 cm First Return DSM” (digital surface model)

 

Set all 3 elevation rasters (DEMs and DSM) to 50% Transparency:

1. Click on the usu_50cm_dem and set the transparency to 50%

a. appearance tab in the ribbon
b. transparency slider

2. Repeat with the usu_5m_dem
3. Repeat with the usu_50cm_DSM
4. DEMs should be sitting “on top of” their corresponding hillshades
5. Add color to the DEM/DSMs if you’d like. I am assigning "Elevation #3" to my elevation models
          In the symbology window, drop down the 'Color Scheme' options and check the box at the bottom to Show Names
6. Always keep hillshades in black and white with high values = white.

---

That bears repeating:

Experiment

I encourage you to explore the areas represented in each of the rasters (using the Pan tool). Use ‘Zoom to Layer’ to see how the extents differ. 
It is good to know that raster "tiles" have similar File Sizes to make downloading somewhat consistent. 

Larger cell sizes = less detailed information about the surface = larger tiles (more area covered by each raster). 
Smaller cell sizes = more details about the surface = much smaller areas covered by each raster tile.

Use the “Swipe” tool in the Compare section of the Raster Layer tab on the main ribbon to ‘pull back’ upper layers to see the DEM/hillshades beneath.  
(The highlighted layer in the contents pane is the layer that will swipe...)

 

 

Rasters in the Table of Contents: 

The range of values represented by a raster is shown in the Contents pane. 

Check out the raster values (ignore the raster file names, they might differ from your own):

 

These values represent the range of elevations in the area covered by the digital elevation model. 

Larger cell areas average elevations over a larger area. Smaller cells will have better precision on the elevation values stored for each location. 

 

Raster Properties

 

 

Digital Elevation Models (DEMs): 

What data is included in a DEM? In other words, what value is stored for each cell?

Answer: Elevation data.  Typically in meters.


In an elevation raster, very cell has an elevation value for that cell’s location.  

Elevation units are almost ALWAYS meters.  The exception is when someone converts meters to feet.  Units can be found in the layer’s properties > Elevation section: 

 

Explore the Layer properties of the USU 5m DEM to answer the following questions. 

 

 

Hillshade

Look at the range of values for the hillshade layer. 

They are all 0-254.

 

This is an important dataset to understand.
Hillshade is a visualization tool and the values have no analytical significance. 

Hillshades are for seeing, not analyzing.

It's critical to understand that hillshade values in ArcGIS Pro (ranging from 0 to 254) are not true elevation-based or terrain-derived measurements, but rather grayscale values used purely for visualization. These values represent the relative brightness of a surface as illuminated by a hypothetical light source, typically from the northwest (azimuth 315°). A value of 0 corresponds to complete shadow (dark), while 254 represents maximum illumination (bright).

Unlike datasets such as slope, aspect, or terrain roughness, which carry quantitative meaning about the land surface and can be used for analysis or modeling, hillshade values do not contain physical or geomorphometric information. They are influenced by display parameters like sun angle and azimuth, and are contextual to the chosen illumination settings rather than intrinsic properties of the landscape.

Using hillshade values in any form of quantitative analysis — such as statistical summaries, classification, or comparisons — leads to misleading or invalid results, because those values are just visual artifacts, not measurable characteristics.

Hillshade values should never be included on a legend.  

 

 

 

 

Change the basemap to 'Imagery with Labels' so you can compare an image of the ground to the way the DEMs represent it. 
The following set of figures shows the area near First Dam in Logan, Utah. Each of the screenshots have the exact same extent and scale.  

Notice how First Dam is represented in each image. (First dam is located in the south west corner of the map.)
Look for specific features on the aerial image and find them on the two DEMs. Notice that some features can be seen on the elevation models but aren’t visible on the photo basemap.  

 

This second set of images shows the USU quad (northwest corner) and part of the Island neightborhood (bottom half of the image). Again, each image shows the same extent and scale.  

 

Each of the screenshots represent the same piece of land, so why are they so different from each other?  

Two main reasons: 

1. DEMs come in a variety of resolutions. Resolution is the size of the cell (pixel).

The two resolutions utilized in this lab are 0.5 meter and 5 meter. This means that the pixels measure 0.5 m x 0.5 m on the ground and 5 m x 5 m on the ground, respectively.

The two images below represent the same 18 m x 10 m area. You can tell that, even though they are both representing the same location, the resolution of the 1 m DEM allows it to convey the same information in a more precise way.

 

2. Elevation models are also created using different technologies (interpolation, lidar, stereo images…). We can’t go into the details here, but the quality and accuracy of a DEM has a lot to do with how it is created.

 

 

Explore Tool

So, how can we extract information from a raster.  A quick way to read information from the raster is to use the Explore tool in the Navigate section of the Map tab in the main ribbon. 

Zoom in to the west side of Old Main where the flag pole is located. You can use the Places of Interest point file to locate the flagpole.  

• Open the attribute table
• Highlight the flagpole record
• Zoom to selected

 

 

 

Use the Explore tool to find the elevation at this location.

If the pop up window doesn't display cell values from the elevation or surface models:

1. Click on the drop-down tab for the Explore tool

2. Select "Visible Layers"
3. Check the boxes next to the DEMs and DSM in the contents pane to make these layers visible. The hillshade layers do not need to be visible as these layers don't contain usable information. 
4. Zoom in and click the flag pole point (if you are zoomed out and click a neighboring cell you will extract elevation values from cells proximate to the flagpole. If you get the flagpole to appear in the popup window, then you should be in the right place as points are very specific XY locations.)

 

Notice how Zoomed In we are...

 

 

Notice that the popup window has two panels: top and bottom. 

4. The pop-up window shows us the cell values for all the layers in the Contents pane. Read them carefully. The value for this pixel in the USU 50 cm DEM is 1453 meters.  The first return elevation raster reports a higher elevation (~1493 m).  The difference between the two rasters means the flag pole is ~40 meters. Does that seem reasonable for the flag pole in front of old main?

photo credit: Utah State University

Note: if you did not get these same (relatively the same) elevations, you are probably not clicking on the correct pixel.  (Or you are pulling information from the wrong DEM.) The pixels are tiny. You need to be careful and zoom in until you are sure you are pulling the elevation of the cell under the flag pole point. 

 

Now we know that the approximate elevation on the ground for this pixel is 1453 meters and we estimated the height of the flagpole.
You just extracted information about a place. 

GIS is all about the information attached to places, remember?  

Challenge: pan around and find other features on the first return elevation surface and use the Explore tool to find elevations of the top of features (DSM) and the ground (bare earth DEM) to estimate building height, tree canopy height, etc. 

 

Calculate the height of the Flag Pole using the elevation model and the surface model.

 

Do you trust that this is the actual height of the flag pole? 

I do not, for what it's worth. The cell value could have been slightly smoothed via interpolation. If you need the precise height, call your local surveyor. :)

 

 

Raster Attributes

Are you ready to have your mind blown? 

You know how vector attribute tables have a row for every single feature on the map?

Rasters don’t work like that.

Ever heard of a frequency distribution table?  

A frequency distribution table is a chart that organizes data by listing unque values (or categories) and the number of times each one occurs (its frequency).

 

Well, that’s how raster attribute tables are organized. 

 

How many cells share the lowest cell value? How many share the next highest cell value in the raster? And so on… 

Look: 

 

Cell value 1389. That’s the lowest cell value (lowest elevation) for the extent covered by the 5 meter DEM. There is exactly one cell that has that value. But there are 14 cells that have an elevation of 1390 meters.   
Open the attribute table for the 5 meter DEM and scroll through. 

 

Now, open the attribute table for the 0.5 meter DSM (the first return raster). 

What? Grayed out?! Confounding ArcGIS! 

 

This is a floating point raster, meaning the stored elevations contain decimals.

That would mean there would have to be a row for every single unique value between the min and max. And with fractions… that’s too many unique values!   

Therefore, no attribute table for raster datasets containing values that aren’t integers. This is a common hiccup we run into.

OPTIONAL SECTION for your information only:

Solution: convert all the elevation decimal values to integers.  This is done using a raster calculation to truncate the values.

The tool is one you can search for and find in the Geoprocessing tools:

Notice that the tool explains the decimals are truncated.

 

Truncating isn't the best way to preserve whatever vertical accuracy the elevation data has. Wouldn't it be great to have a tool that rounded the values correctly?

Turns out there is a way to force proper rounding.

The idea is to ADD 0.5 to all raster elevation values and THEN truncate. Do the math, it works.
1000.2 + 0.5 = 1000.7 which truncates to 1000. Correct truncation and rounding. 
1000.7 + 0.5 = 1001.2 which truncates to 1001. Correct truncation and rounding. 

Here's a demonstration video:

You are not required to do any of this, but I offer it up as information and a challenge, should you choose to accept it.

 

 

Time to explore and answer some questions.

 

Challenge:
Create an elevation legend so professional it makes people dance in the street. 

Video demonstrating Dynamic Range Adjustments, converting legend to graphic for easy editing, and elevation legend best practices.

Recap:

A raster is a grid of cells.

Each cell contains a value, for example elevation, wind speed, slope, land cover type.

Attributes are the values and a count of the values, similar to a histogram.

Attribute tables can only be viewed for integer rasters.

Elevation rasters (DEMs) can be symbolized with continuous color schemes.
Elevation values can be displayed in a legend (with units of measure).

Hillshade surfaces are essential for illustrating the terrain.
Hillshade values (0-254) signify a shade of illumination determined by elevations and an arbitrary light source location.
These values do not belong in a legend.
Do not confuse hillshade values with values representing slope or terrain roughness.

Resolution is traditionally described in one dimension: 5 meter DEM, 10 meter raster, for example.

 

And that is our introduction to raster data.

See Canvas for submission details.