Surveying Levee Damage With UAS



Land surveyors at the New Orleans District are often among the first people responding to a levee slide. Levee slides generally happen as water quickly recedes after a flood. Traditionally levee slides have only been surveyed by laying out cross sections and profiles along the area affected by the slide. These cross sections show the changes in elevation and are used to compute the volume of material needed to fix the damage.
The incorporation of unmanned aerial systems (UAS) as standard equipment for the survey crews has allowed the survey teams to offer the ability to visualize and assess levee damage in ways not possible through conventional survey equipment. UAS allows land surveyors the ability to collect high-resolution aerial imagery (up to 1-inch resolution), and 3D models (elevation point clouds).
During the 2015/2016 winter flood on the Mississippi River, the land surveyors at the New Orleans district responded to 300 feet of levee slide on the Lower Morganza Guide Levee. The survey crews laid out nine traditional cross sections, and operated a fixed-wing UAS to collect aerial imagery and elevation models. A typical flight over a levee slide takes only one to two hours to complete.
Combining Methods
Traditional land surveying is combined with the emerging technologies of unmanned systems to produce the best survey solution to assist engineers in analyzing levee slides. Although the UAS is capable of producing elevations, it is not accurate enough for engineering design. The main limitation for the UAS currently is the inability to penetrate dense vegetation. For instance, on levees the UAS will only derive elevations off the top of the grass and trees. If the grass is a foot tall, the elevations will be off by a foot.
The incorporation of UAS into survey workflows allows engineers to visualize and assess levee slides without the need to go into the field.
Table 1 gives a difference of randomly sampled points that compare the elevation collected by conventional survey methods versus UAS-derived elevations. The root mean square error (RMSE) of the two surveys is 0.33 feet. (RMSE is a common statistic used to compare accuracies of surveys.)
The incorporation of UAS into survey workflows allows engineers to visualize and assess levee slides without the need to go into the field. They can still get a good view of the situation. The photo above, left shows a top-down view of the crack in the levee crown caused by the slide.
The 3D point cloud (similar to LiDAR) allows engineers at the New Orleans district to get a multidimensional view of what the levee slide looks like. The photo above, right is a screenshot from the 3D model. This model can also be utilized to calculate quantities of material that would be required to fix the levee slide. A video was made showing a flythrough of this levee slide.
The addition of UAS into a survey crew is a natural fit. New Orleans survey crews are fully functional survey crews able to access all types of terrains with flat boats, cabin boats, all-terrain-vehicles, etc. Much of the same technology needed to operate the UAS for mapping purposes (survey-grade GPS) is already standard equipment for New Orleans District survey crews. Levee slide response is only one of the many emergency response activities conducted by the survey teams.
Table 1 – Comparing elevations on 25 randomly selected points demonstrated the differences between UAV-collected data and conventional survey methods. Dense vegetation often limited accuracy of UAV observations.
Point # | Shot Code | Survey Elevation | UAS Elevation | Elevation Difference |
---|---|---|---|---|
8 | Slope Shot | 60.75900 | 61.49500 | -0.73600 |
10 | Centerline Levee | 62.56900 | 62.56400 | 0.00500 |
16 | Slope Shot | 46.47400 | 46.37600 | 0.09800 |
22 | Slope Shot | 55.81000 | 55.80300 | 0.00700 |
26 | Centerline Levee | 62.49000 | 62.47031 | 0.01969 |
27 | Flood Side Crown | 62.00300 | 62.09600 | -0.09300 |
30 | Slope Shot | 54.82500 | 55.11500 | -0.29000 |
33 | Flood Side Crown | 45.12300 | 45.26519 | -0.14220 |
45 | Centerline Levee | 61.63800 | 61.57700 | 0.06100 |
51 | Slope Shot | 55.52700 | 55.93400 | -0.40700 |
63 | Slope Shot | 59.32000 | 59.59700 | -0.27700 |
64 | Protected Side Crown | 60.85700 | 60.94172 | -0.08470 |
78 | Slope Shot | 50.03200 | 49.91100 | 0.12100 |
99 | Slope Shot | 52.15600 | 52.11300 | 0.04300 |
104 | Centerline Levee | 61.06700 | 60.89400 | 0.17300 |
109 | Slope Shot | 58.09900 | 58.87900 | -0.78000 |
112 | Slope Shot | 49.02300 | 49.30000 | -0.27700 |
115 | Slope Shot | 47.52200 | 47.80700 | -0.28500 |
122 | Centerline Levee | 61.30300 | 61.25800 | 0.04500 |
132 | Slope Shot | 45.66300 | 45.44700 | 0.21600 |
136 | Slope Shot | 55.11800 | 54.73900 | 0.37900 |
138 | Slope Shot | 60.86100 | 61.26900 | -0.40800 |
140 | Centerline Levee | 62.46400 | 62.40400 | 0.06000 |
156 | Slope Shot | 59.73600 | 60.45100 | -0.71500 |
162 | Flood Side Toe | 43.84200 | 44.23700 | -0.39500 |
Average | Root Mean Square Error (RMSE) | Largest Difference | Smallest Difference |
---|---|---|---|
-0.147 | 0.332 | -0.736 | 0.005 |