Terrain data comes in a variety of display and geometries. Elevation data can come in the form of contours similar to what is seen on a topographic map. Another geometry type would be points in which each point represents an elevation. Of the two datasets mentioned they can each be used to create a triangular irregular network or TIN. Creation of a TIN is important as it acts as a segway between features and the most useful of elevation datasets which would be a raster or grid. When a raster or grid is used and contains elevation data where each pixel value represents an elevation it is called a Digital Terrain Model DTM or Digital Elevation Model. When referring to elevation datasets we will refer to them as DEM in the truest sense, as DTM can be an elevation dataset but also can be a number of different representation of elevation such as slope, aspect, drainage and other terrain attributes (Yures). The DEM’s used in education, research, and industry today are fundamentally the same represent elevation data in a raster format, but are created in drastically different methods and procedures.
The most common DEM is provided from the USGS and was created from 1:24,000 contours at a 30M resolution and covers all the US seamlessly. In certain areas of the US mainly highly urban higher resolutions are available such as 10M (USGS). DEM’s can be created from remotely sensed data from space or aircraft. LiDAR, Light Detection and Ranging are the source for very high resolution DEM creation. LiDAR data is created by flying at low altitude approximately 300-2000 meters, utilizing a swath oflight to and from an onboard sensor which calculates the time distance traveled to return point data with elevation information (NOAA). The light emits a pulse return an elevation this is called the first return and is a measure of the elevations of the canopy, building roof elevations, and other unobstructed surfaces (AeroMap U.S.). The next pulse or second return is a canopy penetrating return; this return can be used by subtracting first return data (canopy) to derive an interpolated bare earth point data. These points can be very dense and take up considerable hard drive space. These points contain X, Y, Z data in which X and Y are ground locational information and Z is the elevation data. The points can be very dense, thus taking up considerable hard drive space. The points are converted to a TIN and then to a DEM that is now in a useable form.
DEM’s are currently the most useable form of elevation datasets because of the capabilities of Raster Math for analysis. The USGS derived datasets are free to the public while LiDAR data can be very expensive even for the wealthiest consumer. What does all this DEM business mean for practical use? In the water resource industry namely surface water modeling the DEM is the single most important dataset used in the analyst of or earth’s river systems. The DEM can derive hydrologic as well as hydraulic datasets used in this type of analyst. DEM’s can delineate stream centerlines, watersheds, longest flowpath of a watershed, and slope all important for the development of hydrology for watersheds. The DEM can also be used to develop hydraulic models, in Raster Math to calculate the extent of the watersheds watersurface elevations, and delineate this data for display. This brings up an interesting situation is the most expensive highest resolution data always the best choice for use in hydrology and hydraulic modeling. This situation will be tested by utilizing ArcHydro tools a watershed delineation and analysis software for ArcGIS. ArcHydro tools will be used to calculate flows for a particular watershed using a side by side comparison approach. The DEM datasets to be compared will be the USGS 30M DEM and LiDAR obtained two foot cell spaced DEM.
Figure 1: An example of one thirty meter (99ft) DEM over lay with 2 foot cell spaced LiDAR data.
Hydrology in the sense of water surface modeling can be considered calculating the behavior of the substance in a natural environment. In a more layman statement hydrology can be considered the calculation of the flow of water usually in CFS (cubic feet per second) at a particular point on a stream. This calculation takes certain parameters into consideration including but not limited to upstream contributing area, upstream centerline length, and slope. Important tools to derive these calculations are hardware, software, data, and people. An important factor in calculating the most accurate hydrology (flows) is the accuracy of the underlying terrain data. Terrain data itself can carry significant cost not only in its acquisition, but in its storage, manipulation, and output. The remainder of this paper will concentrate on hydrology and hydraulic technologies and the underlying terrain data that afford there remarkable ability to model our earth’s water surface features. Namely LiDAR (Light Detection and Ranging) is the emerging king of source elevation data in surface water resource modeling. Other elevation data sources will be discussed and their pros and cons will be unearthed.
For Hydrology ArcHydro tools a free extension from ESRI is used in the development of the perimeters, inputs, for flow calculations. The Preprocessing in ArcHydro tools of the data varies in significantly between the two types of DEM data tested. USGS 30 meter DEM and LiDAR obtained DEM data where tested. Both data sets where clipped to the same extent. The higher resolution LiDAR data is the same extent as the USGS 30 meter DEM but the dense sampling makes the actual computer file size is considerable different. It is this difference that was found to increase computer computation times shown below.
Figure 2: ArcHydro preprocessing steps times are in minutes
When all data is processed the LiDAR data took six times as long to process for the same extent as the USGS 30 meter DEM. The preprocessing is necessary to derive all the inputs for the USGS Regression equation in order to calculate flows. There were slight differences in the watershed area and longest flowpath lengths for the two differing datasets. This difference was insignificant in the final flow values produced with a difference of 10 cfs. The following is the regression equation used in the flow calculation details about the perimeters used are included.
Figure 3: USGS Regression equation note differences in inputs but actual flow value Rural 100 yr has a 10 cfs difference for a 10,800 cfs discharge)
In conclusion the extremely high cost and processing time involved do not warrant utilizing LiDAR data. USGS 30 meter DEM is considered acceptable for developing hydrology for these purposes. It should be noted that further test would need to be conducted in order to substantiate this claim. Other areas of study would be flat coastal terrain verses mountainous terrain, differing regression equations flow calculations, and different resolution of DEM’s.
Hydraulics in the simplest terms can be stated as the behavior of water in a conduit whether a natural riverine system or cylinder. Hydraulics in itself is a complex and time consuming methodology and beyond the scope of this investigation. For hydraulic comparison between the USGS 30 meter DEM and LiDAR DEM already developed hydraulics models were used. The models were run for the 100 year flood event and the delineated floodplains were compared. In floodplain delineation horizontal/vertical accuracy is a must. The following is a comparison of the differing outputs.
Figure 4: Mapped Floodplains
The longer jagged lines represent USGS 30 meter DEM delineation while the tightly spaced lines are LiDAR DEM. The two examples are termed rasterization. This is an undesired trait in floodplain delineation and made minimal by using high resolution LiDAR DEM’s. Although it is undesired in floodplain delineation due to the possibility of loss of life and/or property it might not be necessary for all delineation. If the data is to be used for small scale projects the rasterization might not be noticeable at that scale and lower resolution DEM would be acceptable.
In conclusion the need of LiDAR data is viable in floodplain delineation even with the high cost, expensive hardware/software, and time consumption. It is probable that large government organizations would take on such task and thus warrant the procurement and use of this data. In studies in which the need of highly accurate delineation is not necessary one would consider using data that is of lesser resolution and free in most cases. In carefully planned project these issues would be predetermined and addressed accordingly.
Future research as with hydrology would be flat coastal terrain verses mountainous terrain. The investigation could be on mapped outputs and delineation behaviors in these differing topographic geographies. In previous experience it can be noted that the steeper the terrain, i.e. the closer the contour lines, lessen rasterization provides a more realistic depiction of what is happening on the ground.
Sources:
AeroMap U.S. Company online website Lidar Basics.
NOAA Coastal ServicesCenter. Remote Sensing for Coastal Management LiDAR.
United States Geologic Survey USGS. Online List of Products, National Elevation Dataset NED 1 arc second.
Yures, Gabriel. Online GIS Tutorials: Chapter 3 Digital Terrain Model (DTM)
1