Mr. Terry Jameson

1
UAS Data Collection for High-resolution MET
Modeling Ingest
Mr. Terry Jameson Battlefield Environment
Division Army Research Laboratory, WSMR COMM
575-678-3924 terry.c.jameson.civ_at_mail.mil
2
Weather Prediction Models

• Numerical Weather Prediction (NWP) Models
• Predictions of basic Met parameters (winds,
  temperature, pressure, humidity)
• Predictions of derived parameters (turbulence,
  visibility, cloud layers, etc.)
• Predictions at 3-D grid points ( 30 mi. down
  to 8 mi. horizontal spacing)
• Predictions out several hours - up to many days
• Research-grade models (one-hour predictions
  0.6 mi. grid spacing)

• Models require Met data observations input for
  initialization
• Surface weather stations (manned and automated)
  little help for upper
• atmosphere
• Doppler weather radar (intensity and motion
  within storms) good info but
• only when storms are present
• Satellite observations of winds and temps (very
  coarse vertical resolution)
• Vertically-pointing wind profiling radars few
  locations even in U.S.
• Weather balloons (winds, pressure, temperature,
  humidity)
• 70 stations in Lower 48, 700 world-wide
• Twice-daily balloon launches
• Mainstay of NWP model input since its
  inception in late 50s-early 60s

3
But theres a Problem

• In the U.S. all of the above are available,
  but..
• Problem is All of the above leave many gaps
  (time/space),
• especially for high-resolution models
• Problem is In/near the battlefield, only a
  very few weather
• balloon and surface observation stations exist
• Problem is Those few stations can be sporadic
  in their
• observations
• Bottom line
• WE NEED MORE INPUT MET DATA!

4
In-situ Obs from UAVs
Data collected from UAVs - What are we up against?

• Certainly many UAVs have a temperature
  sensor/readout, plus GPS winds

BUT

• Are the data just displayed to the operator?
  cant use in modeling

• Are the data recorded at the ground station?
  probably not

• Are the data recorded on-board somehow?
  probably not

• Are those data date/time/location-stamped?

• What about pressure and humidity? need those
  parameters as well

• How to QC the data? bad data or wrong
  time/place poor performance.

• How to format the data? models are very picky!

5
TAMDAR-What is it?

• TAMDAR (Tropospheric Airborne Met DAta
  Reporting)
• Small meteorological (Met) data
  sensing/transmitting instrument
• AirDat, LLC
• Installed on 150 regional commuter airliners
• Collects Met data for ingest into Numerical
  Weather Prediction (NWP) Models
• TAMDAR-U (TAMDAR-UAV)
• TAMDAR downsized for installation on UAVs
• Stringent restrictions on Size, Weight, and
  Power (SWaP) requirements

6
AirDats Commercial TAMDAR System
Know the Weather
Information used with permission from AirDat, LLC
7
The Team

• NMSU PSL/Technical Analysis Applications Center
  (TAAC)
• The Aerostar-B UAV
• Established COA in southern NM
• Substantial experience in conducting
  instrumentation flight tests

• AirDat, LLC
• The TAMDAR
• Instrumentation facilities (Lakewood, CO)
• Data ground station and NWP modeling facilities
  (Florida)
• Substantial experience in instrumenting
  commercial airline fleets
• Substantial experience in ingesting TAMDAR data
  into models

• ARL
• Long-term history of DOD weather research and
  support
• High-resolution, battlefield-scale NWP model
  development
• Substantial experience in assessing model
  performance

8
TAMDAR-U Sensor (Prototype)
Mounted on Modified Aerostar Nose Cone
Prototype TAMDAR-U
CFD Analysis

• Measures and Reports
• -Ice presence -Relative Humidity
• -Median and peak turbulence -Indicated and True
  Airspeed
• -Static pressure and pressure altitude -Winds
  Aloft (Speed and Dir)
• -Air temperature (Mach corrected) -GPS Position
  and Time
• -Additional sensing possible (CBRN) -Encryption
  Possible

Know the Weather
Information used with permission from AirDat, LLC
9
TAMDAR-U Sensor (Prototype) - SWaP
LRU Dimensions (Volume) Weight Max Power (Estimated)
Probe (External) 2.6x2.5x0.7 3.6 Pitot 2.2 oz (62 g) N/A
Data Acquisition, Processing, and Communications (Internal) 40 in3 12.2 oz (346 g) 8.4W
TOTALS 40 in3 Internal (reductions possible) 14.4 oz (408 g) (reductions possible) 8.4W (reductions possible)
Know the Weather
Information used with permission from AirDat, LLC
10
The Aerostar UAS
11
The Airspace Model Domain
32o 46.00 N 106o 30.00 W
32o 46.00 N 107o 50.00 W
31o 40.00 N 106o 30.00 W
31o 40.00 N 107o 50.00 W
12
Experimental Approach

• Collect TAMDAR-U data within model domain for
  three-hour flight
• Reformat and archive data for later analyses
• Run model in data-ingest mode for 3-hrs,
  simulating ingest during flight
• Continue model run after data ingest cutoff
  generate 6 hr forecast
• Compare output charts with/without TAMDAR-U
  ingest
• Compare against any available observations

13
(No Transcript)
14
Example Test Card
32o 46.00 N 107o 50.00 W
32o 40.00 N 107o 34.00 W Point B
32o 46.00 N 106o 30.00 W
After T/O Normal climb to 10,000 MSL Course
305o True At 10,000 MSL, normal descent to
7,000 MSL At Point B, standard rate turn to
125o True Return to Point A (LRU) At 65 kt IAS
(approx. 75 kt TAS), the R/T to Pt. B will take
approximately 1.15 hr.
305O / 40 nm
125O / 40 nm
LRU A/P 32o 17.21 N 106o 55.19 W Point A
31o 40.00 N 106o 30.00 W
31o 40.00 N 107o 50.00 W
SOUTHERN BORDER ADIZ
15
Example Results
16
What did we find?

• TAMDAR sensor could be adequately
  downsized/configured for UAV ops
• TAMDAR-U data successfully assimilated,
  formatted, ingested given erratic
• flight patterns and altitudes of UAV missions
• From a qualitative standpoint, wind flow
  patterns looked more realistic
• over and near mountain slopes with TAMDAR-U data
  ingest
• Few observations within most of the domain for
  quantitative evaluation
• Weather balloons launched at LRU airport
  compared against vertical profiles
• from the models were inconclusive
• Very benign weather case-study days were not
  conducive to finding clear
• distinctions between models

17
Whats next?

• Collect TAMDAR data within a data-rich model
  domain (commuter fleet)
• Run model ingesting or withholding data as
  before
• Select some bad weather case-study days
  (rainfall, strong winds, etc.)
• Conduct quantitative statistical analyses,
  observation points versus forecasts