Intelligent Sensor Network Study of Dust Devils : yr 1

1
Intelligent Sensor Network Study of Dust Devils
yr 1

• Dr Ralph D Lorenz
• Planetary Exploration Group
• Johns Hopkins University Applied Physics
  Laboratory, Laurel, MD
• Ralph.lorenz_at_jhuapl.edu

NASA Applied Information Systems Research Program
Investigators Meeting Maryland, May 2008
2
Why study dust devils ? Dust Devils occasionally
cause damage to ground structures (also
indirectly fires). Dust devils are implicated
in 100 air accidents in the NTSB database.
Important factor in air quality major conduit
of dust into the air. A major player in Mars
climate and in the operation of assets on the
surface of Mars.
3
Chicken or Egg ? Lofted dust intercepts sunlight,
can heat air causing updraft to accelerate dust
devil. But dust must first be lofted. Dust
devil formation not known ambient vorticity in
wake of obstacles, or kinked convective
roll? Non-dust-devil vortices also exist seen
in snow, leaves etc. Near-surface windfield
likely not axisymmetric.
4
Mars (Phoenix Landing Site) 50m AGL simulation
by Tim Michaels (SwRI) NB Peak gust is 3x
higher than mean windspeed
5

• Observations of dust devils have to date
  typically been of several types
• fixed-site limited-duration observation
  campaigns (e.g. Sinclair, Arizona 1960s)
• short campaigns of pursuit measurements
  vehicle-mounted instrumentation
• data-mining of lander MET and image records.
• parasitic surveys of other datasets (e.g. air
  accident database)
• For comparative in-situ measurements, pursuit
  mode is unsatisfactory generally little context
  information, probable selection effects.
  Statistics generally poor. What about vortices
  that are not flagged by dust - are their
  properties different?
• New approach generate large area-time survey of
  subset of meteorological variables. But needs
  inexpensive sensor stations, and could generate
  very large volumes of data.
• - inexpensive intelligence

6

• Approach
• Inexpensive data acquisition. Commercial
  weather stations often unable to record fast
  enough (1/sec) and are 1K apiece. Our
  approach uses custom datalogging equipment built
  from modern microcontrollers. Even including
  sensors (e.g. pressure) parts cost can fall to
  lt100
• Use of easily-programmable (by the scientist)
  microcontrollers allows implementation of
  on-board autonomy event detection, data
  compression, etc. (Stamp vs BX24 vs PIC vs
  PICAXE...)
• Deployment/recovery/download of array of dozens
  of sensors can be time-consuming. Mitigation
  strategies include use of solar power to avoid
  battery replacement. Use of wireless serial
  communication to download data (lots of Zigbee
  products/applications emerging..)
• Natural extension of 23 is for inter-platform
  communication event detection by one station is
  broadcast, triggers recording by other stations.

7
Tucson, September 2007 Linear array of 20
stations deployed E-W (PICAXE 18X
datalogger) Set to record 1 hour of 1/s 1-byte
data of pressure, temperature, light, microphone
(numerically differenced-summed) Encountered
large dust devil after 30 minutes
8
(No Transcript)
9
Tucson, April 2008 Deployed 2-D (4x6) array of
Crossbow Inc. Wireless Sensor Motes. Datastream
captured by receiver basestation (USB) into
Moteview database application at 3-4 messages per
second.
10
(No Transcript)
11
Findings so far - Dust Devils Good probability
of data acquisition by various techniques
demonstrated (locality) Pressure drop for 1-5s
is simplest and most consistent signature of dust
devil passage. Detection algorithm ('phase
picker' - short-term average minus long term
average gt threshold) easily coded onto
datalogger. Persistent temperature drop after
DD passage is often observed - dust deposition on
platform ? Microphone shows some promise as a
wind proxy - data not yet understood fully.
12
Findings so far - Methods Datation
(synchronization among dataloggers) is essential.
Manual programming of real-time clocks is
laborious and not better than 2-3s. Broadcast
sync marker (20 RF link - could use for
signalling?) Download from PICAXE
microcontrollers at 4800 baud is too slow. (20
stations gt 1hr). Removable media (USB
datastick) will be much faster. Wireless mote
reliability not great. Outages. Wireless motes
hampered by network throughput. Possible
alternative/future commercial solutions may avoid
this limitation. Array still needs PC
base-station. No ESD problems noted.
Communications disrupted by DD (antenna
misconfiguration, or RF noise ?)
13
Future Plans Discussion with atmospheric
modelers Tim Michaels (SwRI), Lori Fenton (SETI)
et al. indicates considerable interest in this
dataset, even when dust devils not present.
(Boundary layer meteorological fields not
typically measured with this temporal or spatial
resolution - those studies possibly favor 1.5D
rather than 2-D array) Large array (60-80) of
stations with USB datastick storage, allowing day
or multiple days of acquisition at 1-2 Hz.
. Stations will record pressure, temperature,
light, dust impact, microphonic wind. Will
record raw stream, and augment with 'summary'
file. Will be made expandable (for wireless
query wind speed/direction, temperature
gradient) Field test of prototype station next
month.
14
http//www.lpl.arizona.edu/rlorenz