Wind Observations on the Morphology and Dynamics of Aeolian Barchanoid Dunes with Unmanned Aircraft

The lower atmosphere has been studied for years, mostly through observations or ground measurements, however, mapping the processes of the lower part of the atmosphere is still sparse. The entire state of Oklahoma derives all atmospheric measurements for forecasting from two weather balloon launches a day at two locations. Until recently, these minimalistic approaches were the only way to gather information and, with them, we have gained a lot of insight into processes in the boundary layer. Given the technological advancement of small unmanned aerial systems (sUAS) we have an opportunity to fill in the data void that sits tantalizingly just above our heads and how it is shaped by the terrestrial expanse under our feet. This study ties atmospheric observations using UAS with photogrammetrically mapped rapid geomorphology to observe and relate the terrain induced effects on lower atmospheric phenomena. Lower atmospheric boundary conditions have been characterized through both kinematic, viz. wind velocity and turbulence, and thermodynamic measurements. In order to achieve this data fusion, empirical data is obtained by UAS mounted anemometers and PTU sensors while photogrammetric terrain modeling is acquired utilizing airborne imagery. The results of both are analyzed and related to compare the coupled effects. Wind induced terrain variations are also compared through multiple terrain models, showing the temporal evolution of the landscape.

This study was performed at Little Sahara is a State park in Oklahoma, a naturally occurring collection of barchan structures and sand dunes that are heavily influenced by the prevailing southern wind. The datasets are months apart, and the changes in the topography demonstrate how closely linked the atmospheric movement and the changes in topography are to each other.

Wind observations are made with multiple direct and indirect methods. The primary systems developed are used to carry commercial off the shelf ultrasonic anemometers, including the Young 81000 3D anemometer, shown in Figure 1, and the more lightweight FT Technologies 205 2D anemometer, shown in Figure 2. The platforms used to carry these sensors are the DJI Matrice 600 and the S900, respectively. A DJI Phantom 4 Pro is used to collect the photogrammetric imagery used to create the high-fidelity terrain model. Survey grade GPS is used to geolocate the models and laser altimeters were integrated into the platforms carrying the anemometers to determine an accurate flight altitude in relation to the dune.

In order to ensure data quality, validation tests are run to characterize platform induced error within the onboard anemometer measurements. These tests are run at different locations including indoors in a quiescent flight test arena and outdoors using both Mesonet towers and additional ground units. Photogrammetric data quality was verified based on survey grade GPS data at ground control point locations within the derived dataset. When examining the preliminary data gathered at Little Sahara, the results are promising. The entire park migrates north at an average rate of 16 inches per year. The setup is ideal for combining atmospheric observations with a time varying terrain to examine the dune migration parameters. An initial point cloud dataset is shown in Figure 3.

The model derived from one of two photogrammetric flights that were performed on the same day at the Little Sahara site are shown in Figure 3a. The digital elevation model (DEM) of the same flight is shown in Figure 3b. Sand is normally a difficult medium to feed into Structure from Motion algorithms and expect usable data on the other side, however the datasets acquired when testing at Little Sahara rendered well.

The datasets from both flights were processed with default Structure from Motion settings and the respective point clouds from both flights were imported into Cloud Compare. The point clouds were registered, a process that performs a best fit function between two 3D data sets and the farthest point removal setting was checked. The comparison was made to determine the fidelity of a homogeneous and notoriously difficult to process surface at a higher altitude. The comparison appeared to be close the noisiest sections were around the foliage. That comparison was then broken into three axes to see where the discrepancies were worst. The distance was lower in the X and Y directions, discrepancies in the Z direction were the highest. Because the sand dunes were difficult to get to, the team flew from a viewing platform around .5 km away. This made it difficult to know how high above the dunes each aircraft was flying, an aspect that will be addressed in the study by implementing the use of an onboard laser altimeter. When tying into topographical and atmospheric data on the scale found in Little Sahara, it was important to know exactly how high the craft is below the crest of the dune.

Preliminary anemometer results are shown in Figure 4. The dune topography is combined with the anemometer data, and the green outline of the dune is an accurate cross-section relative to the profile location. Observations show the attachment and separation patterns of the flow over dune at the base of the leeward side with Re values reaching values on the order of 10⁶. The study provides a follow up to this initial dataset and show further data fusion and map the aeolian based geomorphology on a temporal scale as well as a spatial one.