flying lasers and blue lines

by Joe Kinyon

In late 2013, a plane carrying precision equipment, including a laser, a laser receiver, a GPS receiver, a high-res digital camera and a very accurate clock, took to the air above Sonoma County and “mowed the lawn.”
​
Many people call the aerial imagery they find on current web-based maps “satellite” imagery. However, a large portion of true satellite imagery is obscured by clouds. At that altitude, an image from today’s satellite’s cameras often represents a square meter of earth per pixel. Because a camera in a plane is so close to the earth, it cannot get a clear picture of the county in one shot. To remedy this, the pilots fly back and forth, taking thousands of pictures that are stitched together into a seamless mosaic.
​

“Mowing the lawn” refers to the task of flying a plane at a relatively low altitude along a straight line, then turning around and flying back along an adjacent overlapping swath until the whole area is covered like a mowed lawn. During the flights in 2013, each photo pixel in every part of the digital mosaic represented six square inches of Sonoma County. In urban areas, resolution increased to three square inches of the earth’s surface per pixel.
 
While up above Sonoma, the plane shot lasers — lots of lasers. A minimum of eight pulses of laser light permeated every square meter/yard of Sonoma County’s surface (and because of overlapping swaths, many places saw double that number). While one sensor collected the reflected sunlight as photographs, another sensor collected the laser light pulses reflecting back. With the help of the very accurate clock, each laser pulse was tagged with the time it took to leave the plane and reflect back. Multiply the speed of light with the time a laser pulse takes to travel from the plane to the surface and back to get the total distance of the pulse’s trip.
 
 Subtracting the distance between the plane and the reflected surface from the plane’s location determines the surface’s elevation. The reflected surface can be anything inside one square meter and, with every laser pulse, there can be multiple returning reflections. This can include a tree canopy, the ground, cars, water or buildings below it. Usually, the canopy top’s reflection is the first to return and the ground’s reflection is the last. Vegetation is sometimes so thick that it obscures the ground, but reflections from the multiple pulses will occasionally slip between the leaves. This is the process of airborne LIDAR data collection. Like a bat using echolocation to sense distance to prey, LIDAR systems use light reflections to determine distance. With enough reflections, we can piece together a 3D picture. For every meter of Sonoma County, the airborne LIDAR measured the elevation with enough detail to give us the shape of the undulating land, the height of forests and the shape of all above-ground structures.
​

Image credit: USDA Forest Service Pacific Northwest Research Station Joint Fire Science Program

​
​By taking the last returns (the reflections that took the longest time to leave the plane and return) for every square meter and assembling the derived elevations into a grid of pixels, an image called a “bare earth” Digital Elevation Mode, or DEM, is created.

Below is a sample of a DEM of an area that is three by three meters wide.
Each one-meter square pixel has its elevation stamped on it.
​

​
I like to use color to see number patterns better, so I made the higher spots lighter and lower spots darker. Doing so, I can quickly see that the area goes downhill toward the darker corner in the lower right.
​

Remove the numbers and it might appear clearer.
​

Simulating height with raised blocks reinforces the view.
​

​This example zooms out to include more pixels.
Look closely: Subtle patterns of slope are revealed in the slight variations of similar gray tones.
​

​Keep zooming out to include more of the grid in the DEM and undulating landforms begin to appear.
​

​As you zoom out, you can no longer distinguish the grid of individual elevation pixels.
The topography becomes clearer and one can now see pronounced dark canyons and light ridges in this area between Cazadero and Lake Sonoma.

​Using the same grid, the computer can do some visual tricks.
Instead of coloring the pixels in the DEM lighter for high and darker for low, I simulate higher pixels casting a shadow on lower adjacent pixels and color those shadowed pixels darker.
This process creates a shaded relief image.
Below is the same region from above, shown as if the sun was shining from the northeast corner.
​

The next trick the computer can visualize is to draw lines where water could flow.
Since water flows downhill and toward the steepest slope, we can use the grid of elevation pixels to automate the process.
This is the perfect task for a computer because, in order to do this, the elevation of every pixel must be compared to the pixels around it.
The resulting comparison can be used to calculate the direction downhill.
Using the elevation grid from above, one can easily see that only the pixels with the height of 37 and 22 are lower than the center.
By comparing the difference between pixels, the steepest and lowest point in this grid of nine pixels is the path from the center elevation of 43 to 22, the direction the water will flow and accumulate in these nine pixels representing the surface of Sonoma.
​

By moving the center of this 3x3 comparison window over all the thousands of pixels in a DEM, the computer assigns a direction to every pixel.
Below is a larger extent of the grid, zoomed out to 5x5 pixels wide.
Next to it is a diagram showing the corresponding direction the water would flow from that location.
​

Connecting these directions with blue lines demonstrates how the computer will draw streams.
​

​If we zoom out, we can see more creek lines appear. The angular nature of the grid starts to disappear as the organic meanderings of the water appear through the landscape.
​

If we step back even further, the network of creek and rivers appears. In this next image, the creek line was not drawn until enough land upstream was drained.
​

Adding the creek lines to the shaded relief enhances the visualization.
It’s also easier to see the watershed boundaries for the land draining into a creek.
​

By linking the location of stream gauges to the stream lines, the computer can enhance the cartography by drawing the thickness of the line relative to the average annual flow upstream of that location. 
​
​Less flow equals thin blue lines and more flow equals thicker blue lines, as shown below.
​

​A map is a combination of ingredients, each bit of information chosen and balanced for the way it adds to the final map. This laser-collected elevation information results in enhanced maps that explain other relationships across the landscape.

Below, the elevation, shaded relief and creek lines calculated from LIDAR measurements are layered together to create a stage on which symbols for roads, parcel boundaries, urban boundaries and labels work together to tell a story.
This map is a snippet of our regional display maps that help our staff see where places protected by Sonoma Land Trust relate to other protected lands.
​

​If you would like to learn more about LIDAR and the hydrography of Sonoma County shown above, please see the release of high-res stream data and watershed information by the Sonoma County Vegetation Mapping & LIDAR Program. Sonoma Land Trust is a grateful beneficiary of the scope of this project and the quality of information it creates. Read Mark Tukman’s recent blog post for more technical information and browse the amazing results of this project in online interactive maps here.