DEM Resolutions

The following are 4 types of Digital Elevation Model resolutions. On this page, I will demonstrate the key differences between resolutions, while providing pros and cons for each.

Resolution Variance and Vertical Accuracy

In this set of maps, you will see a comparison of various resolutions and sampling techniques for elevation data. You will notice the dataset with the highest vertical accuracy is the .5m LIDAR, and lowest is the autocorrelated 5m dataset. While all of the datasets above are focused on the same area, elevation data becomes less and less accurate with larger pixel sizes (compare .5m resolution to 30m resolution).

While it seems obvious that a higher resolution is optimal for all situations, this may not be true in every case. While high resolutions provide higher spatial accuracy, they can often be costly, take up much more disk space, and can be slow to display. For lower resolutions, you may get less spatial accuracy, but they can often be more cost efficient, take up less disk space, and process certain analyses much faster than higher resolution datasets.

Contour and Elevation Variance at Different Resolutions

In this set of maps, I wanted to demonstrate the differences in elevation and contour data for each resolution. You'll notice that .5m resolution contour lines contain far more detail than 30m contour lines, and thus paint a more accurate picture of the true landscape.

At first glance, the Signal Peak elevation datasets appear to be quite similar to each other, and there is not a lot of variance in elevation between resolutions. However, I chose to study an area of steep incline to demonstrate the true difference in resolutions. Accuracy of elevation on the steep incline between the four resolutions varies wildly, and tells us that the higher the resolution, the more accurate our elevation data will be. A 30m resolution will have a much harder time providing true accuracy at steeper inclines because it takes a much wider 'picture', and averages more data together. A .5m resolution takes much smaller 'pictures', and therefore needs to average less data together, providing a clearer and more accurate dataset.

Key Terms

LIDAR

This term stands for Light Detection and Ranging. It is a remote sensing method used to examine the surface of the Earth that uses a pulsed laser to measure variable distances to the Earth. This generates precise three-dimensional information about the shape of the Earth and its surface characteristics.

Autocorrelation

Autocorrelated data is a type of LIDAR data that uses mathematical computations to find common features in two images captured from different perspectives. This allows us to derive a three-dimensional model quickly and affordably. Autocorrelation can provide more horizontal accuracy and higher resolution, although it is not as rigorous as other elevation modeling such as LIDAR. You will notice in the map above that the autocorrelated data has the least amount of vertical accuracy compared to the others.

Bilinear Interpolation

Bilinear interpolation is a resampling or transformation technique used, in this case, for projecting data. In this example, I used bilinear interpolation to project the raster data from the Geographic Coordinate System NAD 1983 to the Projected Coordinate System NAD 1983 UTM Zone 12N, which is the state plane projection for Utah. Bilinear interpolation calculates the value of each pixel within the raster dataset by averaging (weighted for distance) the surrounding 4 pixels. It is suitable for continuous data such as elevation.

Vertical Accuracy

Vertical accuracy is defined as the closeness of a measured or computed value to the true value of a particular quantity. Vertical accuracy is especially important when measuring elevation data.

Page updated

Google Sites

Report abuse