A 1.115Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
Things are looking up for both lidar technologists and nontechnical users living along the U.S. coast. A widening circle of coastal developers, planners, resource managers, biologists, and others are using and sharing lidar data to help them make complex decisions. Better still, improved modeling software products continue to hit the market. For instance, the latest version of the Sea Level Affecting Marshes Model (SLAMM) software is more user-friendly and accepts higher-resolution data.
If there’s a downside to lidar’s popularity, it’s the faulty assumption by some that the data represent "the truth, the whole truth, and nothing but the truth." As lidar technologists know, each project comes with its own specifications and limitations. Project partners might be limited by technical factors–for example, maybe their software or hardware is older and cannot process the raw data or the most detailed data available. Limitations are also part and parcel of the project scope, as when coastal partners focus on analyzing tidal marsh, not nearby upland.
What’s more, the process of creating sea level rise scenarios comes with its own limitations and uncertainties. Even top sea level rise analysts provide a range of high-probability to low-probability scenarios, because many variables will play out over the coming decades. Nonetheless, it’s critical to strive for the best information and greatest accuracy, because these scenarios will influence the choices coastal communities make to survive and thrive despite a changing climate.
Elevation errors
A lidar investigation that began with a muddy trek through South Carolina’s saltwater marsh has uncovered important findings on data performance by Keil Schmid and Brian Hadley, two remote-sensing scientists at the National Oceanic and Atmospheric Administration (NOAA) Coastal Services Center. As part of its mission to provide data, tools, and trainings for coastal professionals, the Center houses and distributes downloadable lidar data at http://www.csc.noaa.gov/digitalcoast/data/coastallidar.
"There’s typically no accuracy assessment specifically for coastal marsh," says Schmid. "But like lots of other lidar specialists, we started this project in 2009 thinking that the `scrub/shrub’ land cover category would probably work for marsh areas, in terms of the pulse energy striking and transmitting information in a similar way."
However, as the two researchers cross-checked the lidar findings against information gathered through traditional survey and GPS methods, the numbers did not add up. Sometimes the mismatch was plain to the naked eye. "We’d walk into these wide marsh areas and could see and feel that the land was mostly very flat. But when we looked at the numbers processed through lidar software, the elevation readings were jumping up and down," notes Hadley.
The problem? "Ground returns can be much more difficult to obtain in marsh vegetation than scrub/shrub, because the lidar beam has to penetrate through all that dense leaf material before reaching the ground," notes Hadley. "Possibly only 10 percent of the lidar energy is really hitting the ground, while the rest hits vegetation that’s higher up. Most lidar software is not designed or run to extricate the minute difference between the two areas. So it registers both the ground readings and the vegetation readings as `bare earth,’ since they are very close to one another."
The hazards of distorted elevation readings
When Hadley and Schmid used the scrub/shrub category and protocols in marsh-area ground readings, the results revealed shifts in the elevation measurements by as much as one-quarter meter (roughly 9.8 inches). "That sort of inaccuracy can cause big problems when mapping sea level rise projections, particularly because the elevation errors go just one direction–upward," says Schmid.
Clearly, lidar technologists collecting marsh data are taking a big risk if they simply follow scrub/shrub protocols and stop there. "Future areas might flood that you didn’t expect because these areas are actually lower, and more vulnerable to sea level rise and flooding, than your lidar data showed," emphasizes Schmid. "If your sea level rise projection is one-quarter meter higher than the actual land elevation in an area, it has the potential to throw off the accuracy of your projection for miles, particularly if the area is flat, because rising sea levels will affect flat land more intensely than uplands," he adds.
Vegetation "texture" is misleading: The lidar beam has difficulty penetrating the dense vegetation in coastal marsh. As a result, elevation findings are inaccurately portrayed in this as-received lidar DEM.Credit: Keil Schmid
Customize your digital elevation model
Lidar specialists can improve the accuracy of marsh-area elevation data–and improve their eventual sea level rise scenarios–when they embrace the idea that the base product from the vendor can be modified or sub-sampled based on the local conditions. There are also a variety of methods for generating a DEM by varying the interpolation, gridding routines, or resolution.
Before starting out, says Hadley, make sure you are familiar with the typical features of coastal marsh terrain. Generally, the size of the grid cell should take into account the density of marsh vegetation. "We found that the denser vegetation, which lidar had more difficulty penetrating, required a coarser resolution because the number of ground-point returns are much fewer–here we used a cell of 10 meters by 10 meters. Spartina marsh grass, which is less dense and easier for lidar to penetrate, required a five meter by five meter grid cell," he notes.
The more involved technique of customized point classification (i.e., filtering) might also be a good choice when your DEM extends beyond the marsh or includes upland features. Before beginning, find out whether this marsh area is discontinuous or whether upland areas are common. If the answer in either case is "yes," you might want to add "filtering" to your protocol.
Schmid and Hadley found that using the "minimum bin method" for the DEM garnered better accuracy in sea level rise scenarios when compared with the "as-received" DEM. How does the minimum bin method work?
"Let’s say in collecting lidar for marsh areas that you have approximately 40 points in a typical cell and all points are classified as `ground.’ Take the lowest point in the cell, then assign that elevation to your output grid or raster. By applying the minimum value to each grid cell, you’re likely to improve the data accuracy by reducing the bias toward higherelevation readings," notes Hadley.
Putting these findings into practice, the researchers created two Charleston-area sea level rise scenarios for 2050: one using the "as-received" DEM and the other using a corrected DEM that adds in minimum-bin method and survey findings. After analyzing the two runs, the researchers found that the "as-received" DEM indicated an 11 percent loss of salt marsh by 2050, while the corrected "minimum-bin" DEM indicated a 22 percent loss of salt marsh. In addition, the corrected DEM showed that when tidal flats are lost to water, additional tidal flats spring up near stream channels, leading to a greater overall loss of healthy marsh
"Any coastal lidar user is going to want a less biased, more accurate DEM, whether you’re talking about developers, local planners, coastal managers, or anyone else who uses the data," says Schmid. "For coastal marsh, it may take wading in and getting survey ground control points and altering the DEM generation protocol. When you’re planning for the impacts of sea level rise, it’s better to be safe than sorry."
A detailed and fully referenced technical summary of this coastal lidar investigation can be found at http://www.csc.noaa.gov/digitalcoast/_/pdf/Lidar_marshes_slamm_CSC.pdf.
Kitty Fahey is a writer with I.M. Systems Group at the NOAA Coastal Services Center. The mission of the Center is to support the environmental, social, and economic well-being of the coast by linking people, information, and technology.
Sidebar:
Publication Offers Lay-Friendly Guidance on Marsh-Movement Complexities
The scientific community largely agrees that global sea level is rising, floods are increasing, and coastal marshes are changing as a result. The publication Marshes on the Move is intended for people who use model results for decision-making but do not build models themselves. Coastal resource managers, developers, planners, and others can read Marshes on the Move to understand the ins and outs of marsh modeling: the wetland-migration questions that models can help answer.
How to use models and communicate uncertainty in effective ways
Key dynamics that affect the response of wetlands to sea level rise, plus the data and input parameters needed for modeling
Complicating factors
Good questions to ask about model results
This publication was created by the NOAA Coastal Services Center and The Nature Conservancy and can be downloaded at http://www.csc.noaa.gov/digitalcoast/_/pdf/Marshes_on_the_move.pdf.
A 1.115Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE