Considerations for Coastal Elevation Mapping


High resolution elevation data is critical for many types of spatial analyses. Coastal zone analyses that utilize elevation data include visualizing the future impacts of sea level rise or coastal flooding, estimating shoreline erosion, quantifying vulnerability to coastal hazards, or identifying the sources and pathways of eroded soils entering water bodies. Elevation data is obtained using many different techniques or technologies, and in most cases the raw elevation data is converted into a Digital Elevation Model (DEM) for use in a GIS, CAD, or statistics package. DEMs are commonly grouped into two main types: a bare-earth Digital Terrain Model (DTM) or a top-surface Digital Surface Model (DSM). However I firmly believe that coastal DEMs should also be defined by their intended use: hydraulic DEMs or hydrologic DEMs.

Very often I see DEMs used as if they are one-size-fits-all data. DEMs are not one-size-fits-all. When creating DEMs for use in a model or analysis it is important to always keep the general application in mind. A good rule of thumb (but not always true) is that hydraulic DEMs should be used for applications where water will fill upwards or inwards. Hydrologic DEMs should be used for applications where water will flow downwards or outwards. Examples of applications are sea level rise modeling on a hydraulic DEM and sediment erosion analysis on a hydrologic DEM. So what makes two DEMs so different from one another? Read on for three major factors to consider when using DEMs in spatial modeling.

DATA ACCURACY

Data accuracy is always a major consideration no matter the application. There are two types of elevation data accuracy: relative accuracy and absolute accuracy. While both types of accuracy are important, each type of accuracy can weigh differently on each type of DEM.

Hydraulic Accuracy Considerations:

Absolute accuracy is the major consideration in hydraulic DEMs. Absolute Accuracy refers to the accuracy of an elevation data set in relation to a source of higher accuracy. For elevation data this is typically the National Spatial Reference System. This value should be stated in the DEM’s metadata as an RMSE value. Absolute accuracy is important to hydraulic analysis because the DEM serves as base data that external data will be mapped on (sea level rise rates, tsunami wave heights, etc.). If the base data (the DEM) has high error, that error will propagate into the outputs of the model or analysis. Of course all elevation data has some inaccuracy, which is why NOAA OCM developed a method for mapping inundation uncertainty.

Hydrologic Accuracy Considerations:

Relative accuracy is the major consideration in hydrologic DEMS. Relative accuracy refers to the internal accuracy a sensor can deliver. This value is typically listed on the manufacturer’s specification sheet for a particular sensor. High relative accuracy is important in hydrologic modeling because the relationship of a pixel to all its neighboring pixels is what decides the flow direction, flow amount, and flow accumulation. If absolute error is fairly constant across a data set then the amount of absolute error will not affect the internal, pixel-to-pixel relationships.

BREAKLINES

Breaklines have already been discussed here in the GeoZone in a two part series (Part Un and Part Deux). The takeaway from those two posts is that breaklines are important. Really important! Breaklines are needed for all types of DEMs, however there are different considerations to keep in mind when creating or applying breaklines.

Image of the Enchanted Lakes on Oahu showing how the lack of breaklines causes issues related to bridges and interpolation artifacts that act as artificial dams in the DEM
Notice how not using breaklines causes issues related to bridges (red circles) and interpolation artifacts (orange circles) acting as artificial dams in the DEM.

Hydraulic breakline considerations:

Breaklines will all be set to a common elevation below your initial tide level. When running a sea level rise model the water levels of the ocean, rivers, and lakes should rise in unison. If the ocean level rises it is logical to predict a rise in any hydrologically connected rivers. Lakes should also see a rise due to rising ground water levels.  As a general rule of thumb when creating DEMs for the NOAA Sea Level Rise and Coastal Flooding Impacts Viewer, we usually set our breaklines to flatten all water bodies to -0.5 meters. -0.5 meters is much lower than the mean higher high water level NOAA OCM uses as the starting point for our inundation modeling. Please note that this water flattening results in a hydraulic DEM that is not useful for most other applications.

Breaklines only need to be defined within the areas that could possibly be inundated. There is rarely a practical need to model sea level rise at 40 meters above sea level (at this point you are modeling doomsday). Only the low lying coastal areas near the ocean that could be impacted by sea level rise need to be modeled. As a general rule here at NOAA OCM we typically breakline all water bodies below the 10 meter contour line.

Hydrologic Breakline Considerations:

Breaklines should follow the elevations of the underlying elevation data. Most hydrologic applications simulate water flowing downhill, often to track flow direction and flow accumulation. Most hydrologic models also expect all the simulated water to reach the edge of the DEM.  In this case we need our breaklines to keep the water moving downhill– flattening all of our rivers to a single, flat elevation would result in a pit with no flow direction and massive flow accumulation. Lakes need to be at or near the level of the surrounding terrain, any lakes sunken beneath the earth would also create pits that prevent water from reaching the edges of the DEM.

Breaklines should cover the entire area of interest. Since water could theoretically appear anywhere across our landscape in the form of precipitation, dam/levy failure, etc. the flow dynamics need to be properly captured at all elevations.

SINKS

In GIS terminology ‘sinks’ refer to small depressions in an elevation data set. Sinks can be natural features (wetlands, floodplains, sinkholes, etc.) or man-made features (mining, retention basins, etc.). No matter what the cause of the sink, the result is a DEM pixel with no drainage direction. Water does not escape because no neighboring cells are lower.  Almost all GIS tools have a way to ‘fill sinks’ in DEM surfaces.

Hydraulic Sink Considerations:

Hydraulic DEMs should not have their sinks filled. This is important, because by filling the sinks you are filling areas that could be your future lakes, wetlands, and streams. As the sea rises, and the ground water with it, wetlands, floodplains, etc. will migrate upland with the higher ocean, lake, and river levels. These areas are modeled in the NOAA Sea Level Rise Viewer as light green areas.

Hydrologic Sink Considerations:

Hydrologic DEMs need to have their sinks filled. The reasoning is that a pixel with no drainage direction will ‘trap’ water, which disrupts flow direction and flow accumulation. This is a bad thing for hydrologic modeling, where all water is expected to reach the edge of the DEM.

Image showing two DEMs. One DEM has sinks filled and thus doesn't show flooding that will occur with sea level rise. The other does not have sinks filled and would result in flow accumulation in hydrologic models.
Compare the effects of filling sinks on a sea level rise model of the Kawai Nui Marsh in Kailua, Oahu. Notice how filling sinks causes the low-lying wetlands to show no impacts to sea level rise.

This blog post highlights three of the many different factors to consider when using a DEM for geospatial analysis. Most GIS users strive to create DEMs that closely approximate reality. While approximating reality is important, it is more important to consider the intended application of the DEM. It makes sense to digitally ‘fill sinks’ in a wetland if the goal is to estimate the amount of upland water reaching the near shore environment. Artificially raising the land around a quarry makes sense if that quarry is kept dry by dewatering pumps and you want that reflected in your models. While these operations do not approximate reality, they do help to create more reliable and realistic modeling outputs. Always keep in mind that creating DEMs is equal parts art and science.

Ross Winans

I am a remote sensing specialist with the Geospatial Solutions Program at the NOAA Office for Coastal Management- Pacific Islands. My background is in computer engineering, but I changed direction in college and finished with a geography degree from the University of South Carolina. After graduation I found my way to Honolulu, HI (by way of Charleston, SC), where I landed this gig. I specialize in elevation data, land cover mapping, sea level rise modeling, and data distribution. I also find the occasional opportunity to keep my nerd skills sharp by writing computer code. When I am not nerding out or remotely sensing things I like to play sports, watch sports, and manage fantasy sports teams. I also like to embarrass myself with a guitar and, when not at the office, you're likely to find me at the beach or out on the water.

Leave a Reply. Comments are moderated.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s