ValidWind applications: Wind power prospecting, aerosol transport
T. Wilkerson*, A. Marchant, T. Apedaile, D. Scholes, J. Simmons, and B. Bradford
Energy Dynamics Laboratory, 1695 North Research Park Way, North Logan, UT 84341
ABSTRACT
The ValidWind™ system employs an XL200 laser rangefinder to track small, lightweight, helium-filled balloons
(0.33 meters, 0.015 kg). We record their trajectories (range resolution 0.5 meters) and automatically produce local
wind profiles in real time. Tracking range is enhanced beyond 2 km by applying retro-reflector tape to the balloons.
Aerodynamic analysis shows that ValidWind balloon motion is well coupled to the local wind within relaxation
times  1 second, due to drag forces at subcritical Reynolds numbers Re < 2×105. Such balloons are Lagrangian
sensors; i.e., they move with the wind as opposed to being fixed in space. In a field campaign involving many
balloons, slight variations in ground level winds at launch lead to trajectory patterns that we analyze to derive 3D
maps of the vertical and horizontal wind profiles downwind of the launch area. Field campaigns are focused on
likely sites for wind power generation and on facilities from which airborne particulates are emitted. We describe
results of wind measurements in Utah near the cities of Clarkston, Logan, and Ogden. ValidWind is a relatively
inexpensive wind sensor that is easily and rapidly transported and deployed at remote sites. It is an ideal instrument
for wind prospecting to support early decisions required, for example, in siting meteorology towers. ValidWind
provides high-resolution, real time characterization of the average and changing 3D wind fields in which wind
power turbines and other remote sensors must operate.
Key words: wind, wind power, wind profiles, balloons, lidar tracking
1. INTRODUCTION AND BACKGROUND
In response to needs for reliable lower atmosphere wind data for studies of aerosol transport and wind power
resources, our laboratory has initiated an R&D program called VisibleWind, the first phase of which has been to
develop the ValidWind™ sensor. The sensor method consists of optically enhanced lidar tracking of lightweight,
helium-filled balloons. The advantages of this system include low cost, easy deployment, rapid data turn-around,
and operation by a single person, in daytime and nighttime. Balloon trajectories r(t) are automatically recorded and
analyzed for profiles of wind speed, wind direction, and wind shear as functions of time, altitude, and geolocation.
Previous accounts1,2 of ValidWind have described the phases of its development, early results, and the rigorous The
theoretical basis for the authenticity of balloon motion as a wind sensor is treated in a forthcoming journal article3.
Here we describe the development of an automatic balloon tracking system (Section 2), results of recent
measurement campaigns (Section 3), and new data products of ValidWind operations (Section 4). Further
applications to air flow in complex terrain and in inversion-dominated situations are underway.
2. AUTO-TRACKING.
The ValidWind sensor has recently been has been enhanced by incorporation of an automatic target tracking system.
The rangefinder and compass are now mounted to a motorized gimbal (QuickSet Gemineye) along with a digital
camera (Sony, 36x optical zoom). An image recognition system (PerceptiVU PVU-TT-M6) tracks the balloon
within the field of view and controls the gimbal to keep the sensor centered on the balloon. Figure 1 shows the
ValidWind sensor mounted to the autotracking gimbal. The purposes of the autotracker are to eliminate operator
fatigue from tracking balloons, to maximize the tracking accuracy, and thereby to minimize trajectory errors. In this
configuration, ValidWind requires only a single operator for setup, balloon preparation, and tracker operations.
The integration of automatic tracking into the ValidWind architecture is explained by reference to the system block
diagram in Figure 2. Operation begins when the laptop controller commands the rangefinder and compass (across a
wireless link) to begin seeking valid range data. The operator releases a balloon and manually (by joystick) guides
the gimbal to center the camera view on the balloon. The operator sends a joystick trigger command to the image
recognition subsystem to identify the centered image object as the target. The image recognition software
*tdw@sdl.usu.edu; 435-760-2468 (mobile); 435-797-4686 (fax).
camera
compass
Figure 1. ValidWind sensor mounted to its autotracker.
bluetooth
link
laser
rangefinder
motorized
gimbal
camera
monitor
laptop w/
analysis
software
image
recognition
joystick
Figure 2. Architecture of the ValidWind system.
the gimbal to center the camera view on the balloon. The operator sends a joystick trigger command to the image
recognition subsystem to identify the centered image object as the target. The image recognition software begins
autonomously tracking the balloon within the camera field of view. The subsystem continuously feeds the target
offset back to the gimbal as an error signal to keep the camera centered. Because they are optically co-aligned, the
rangefinder tracks the balloon along with the camera. The rangefinder sequentially delivers range and directional
data back to the laptop as long as the tracking remains locked and the balloon is within range. The typical data
collection rate is one trajectory point every 3 seconds. The camera has 32x optical zoom so that the balloon can be
tracked out to an extended range. The operator increases the zoom (manually, using the joystick) as the apparent
balloon size decreases. Nightime use of ValidWind is facilitated by a bright flashlight (40W high-intensity
discharge lamp) mounted on the gimbal next to the rangefinder. The retroreflectors on the balloon create a bright
return that can be tracked to a range of at least 1.5 km. We note that the system’s maximum range can easily be
extended with the use of a higher laser pulse energy.
ValidWind is subject to system limitations under some conditions of use. Structure in the background image can
interfere with image recognition, causing tracking errors if the balloon drifts in front of a background scene (e.g., a
mountain) or high-contrast clouds. High levels of dust or insects (attracted by the flashlight) can cause false
readings from the rangefinder. And image tracking offsets when the balloon has high proper motion increase the
trajectory errors and reduce the rangefinder range. This latter problem is currently being addressed by an upgrade to
the controller software. An integral control term will be added to the error signal sent to the gimbal so that the
tracker will follow the target balloon more closely and will be less susceptible to transient structure in the
background image.
3. RECENT APPLICATIONS OF VALIDWIND
3.1. Canyon winds.
August 19/20, 2009: An overnight study of catabatic canyon winds in Logan, UT was conducted at the mouth of
Logan Canyon on the Logan River. This location has been proposed as a possible wind turbine site for Utah State
University. Figure 3 shows the time dependence of the wind speed at selected altitudes of 75 – 200 meters AGL.
Wind speeds above 15 mph occur during the eleven hour period 22:00 (MST), August 19 – 09:00 (MST), August
20. High winds exceed 30 mph. The optimum altitude range (for maximum wind duration) is 75 – 125 meters. The
3D trajectories demonstrate a very steady direction throughout this period. The nocturnal jet develops from the
ground up, and the optimum height for a wind turbine would be about 100 meters.
Figure 3. Logan Canyon winds, Logan UT, at altitudes of 75 – 200 meters for 13 hours, August 19-20, 2009,
showing optimum wind layer 75 – 175 meters high.
3.2. Ridge winds.
September 19, 2009: Terrain-coupled wind fields showing consistent wind speeds above 15 mph were monitored
during 6 daylight hours above a mountain ridge near Clarkston, UT. Figure 4 shows the set of nine balloon
trajectories collected from 15:50 to 16:41 MDT projected onto a terrain topographic map (brighter shades
correspond to higher altitude). Seven of these balloons were launched at 5-minute intervals during the period 16:14
– 16:41 in a saddle of the ridge and followed a northeast compass course at roughly +52˚. Projecting this set of
trajectories onto that vertical plane, we found the corresponding set of velocity (tangent) vectors shown in Figure 5.
This demonstrates the ability of the ValidWind method to capture cross sections of the wind field. The horizontal
velocity component is relatively steady at 7.8 ± 1.3 m/s, while the vertical component shifts dramatically from + 1.6
m/s to – 3.5 m/s in passing leeward from the ridgeline. Ridgeline behavior is an important consideration in choosing
the locations for not only wind turbines but also the meteorology towers used to test for turbine site suitability.
Figure 4. Terrain map of two groups of ValidWind launches from ridge above Clarkston, UT
followed a northeast compass course at roughly +52˚. Projecting this set of trajectories onto that vertical plane, we
found the corresponding set of velocity (tangent) vectors shown in Figure 5. This demonstrates the ability of the
ValidWind method to capture cross sections of the wind field. The horizontal velocity component is relatively
steady at 7.8 ± 1.3 m/s, while the vertical component shifts dramatically from + 1.6 m/s to – 3.5 m/s in passing
leeward from the ridgeline. Ridgeline behavior is an important consideration in choosing the locations for not only
wind turbines but also the meteorology towers used for qualifying the suitability of turbine sites.
The other two trajectories at the top left of Figure 4 were launched at a point 300 meters windward of the ridge.
Because of the complex air flow they rose more toward the north (+33˚), providing a picture of the rising flow over
higher elevations at a horizontal speed of 7.2 ± 0.8 m/s with an updraft speed of 1 – 2m/s.
Yet another group of eight Clarkston balloon flights on September 19, 2009 between 10:43 and 14:56 were launched
from a higher elevation south of the above site, for comparison with anemometer data being collected from a 60
meter tower. These data sets showed excellent agreement at 60 meters AGL, with winds averaging  8.5 m/s with a
maximum of 11 m/s at 13:30.
Figure 5. Wind velocity vectors: Seven Clarkston trajectories, projected onto plane at + 52˚ and averaged
Horizontal velocity steady at 7.8 ± 1.3 m/s; Vertical component drops from + 1.6 to – 3.5 m/s leeward.
4. NEW DATA PRODUCTS
4.1. Software
The laptop controller for ValidWind includes quick-look software to organize and analyze balloon trajectories in the
field. The quick-look software (Figure 6) is a MatLab application that includes BlueTooth communication with the
rangefinder, data logging, trajectory processing, and graphical display of wind data products.
For a each experimental session, the raw data is recorded as a sequence of 4-vectors. Each vector consists of a timetag, range, altitude angle, and azimuth angle. Between balloon flights, the data file is padded with special values
that enable its segmentation into individual balloon flights. Trajectory analysis and wind characterization may be
requested for one or more balloon flights without interrupting the data session.
The first task of trajectory analysis is elimination of false readings. These are 4-vectors for which there is no valid
range (e.g. the rangefinder was not pointed accurately enough) or the range does not correspond to the balloon (e.g.
interference from the background scene or foreground dust particles). False readings are identified by the
occurrence of invalid range values or by non-physical increments in the range value.
analysis
process
flight 1 data
select flight(s)
pad
eliminate false
readings
flight 2 data
Cartesian
transform
pad
GQLF
trajectory
analysis
data logging
process
3D trajectory
wind distribution
graphical
display
flight 3 data
wind profile
Figure 6. Quick-look software for ValidWind
The valid data points are transformed into local Cartesian coordinates (relative to the tracker location) and then
smoothed by a Gaussian-weighted Quadratic Least-squares Filter (GQLF), a variant2 of locally estimated scatterplot
smoothing (LOESS). The timing of the raw trajectory points is irregular, but GQLF produces a regular time
sequence of trajectory estimates. At each estimate point, GQLF simultaneously estimates vectors for the balloon
location, velocity, and acceleration.
sample rate of 3 readings/second, approximately 8 readings contribute to each trajectory estimate. If the number of
nearby data readings is insufficient at a given estimation time, a blank point is inserted in the smoothed trajectory.
4.2. Weber Canyon, UT
August – September, 2010: A new and more comprehensive study of canyon airflows is being undertaken in Weber
Canyon near Ogden, UT The real-time software allows us to assess the state of the wind field to decide how
frequently to sample. The quick-look software displays the ValidWind data in three graphical formats: 3D
trajectory, wind distribution, and wind profile. The trajectory plot displays the balloon track in a 3D viewer with
full user control. The wind distribution shown in Figure 7 is an intensity plot indicating the extent to which each
wind speed and direction was experienced by the balloon(s). This display provides the same information as a
traditional Wind Rose plot, with a more intuitive interpretation. The wind profile plots for August 4 shown in Figure
8 display the balloon speed and direction (primarily to the west)as a function of altitude (AGL) for a set of 4
sequential balloon flights. These data are generated directly by the GQLF smoothing algorithm and do not require
any secondary differentiation or smoothing. Further ValidWind results will be reported in the near future.
.
Figure 7. A wind probability distribution.
Figure 8. Wind profile plot (August 4, 2010).
REFERENCES
[1] Wilkerson, T., B. Bradford, A. Marchant, C. Wright, T. Apedaile, E. Fowles, A. Howard, T. Naini,
"VisibleWind: A rapid-response system for high-resolution wind profiling," Proc. SPIE Lidar Remote
Sensing for Environmental Monitoring X, Ed. U. N. Singh, 746008 ( 2009).
[2] Wilkerson, T., B. Bradford, A. Marchant, T. Apedaile, C. Wright, "VisibleWind: Wind profile
measurements at low altitude," 74790L, Proc. SPIE Lidar Technologies, Techniques, and Measurements
for Atmospheric Remote Sensing V, Eds. U. N. Singh, G. Pappalardo, 74790L (2009).
[3] Wilkerson, T., A. Marchant, “Wind-field characterization from the trajectories of small balloons”,
submitted to J. Appl. Remote Sensing, (2010).
ACKNOWLEDGEMENTS
Development of ValidWindTM has been supported by funding from the Utah Science, Technology, and Research
Initiative (USTAR). Valuable advice has been provided by Ryan Pierson of Electronic Data Systems and Michael
Wojcik of the Energy Dynamics Laboratory.