Applications of Aircraft Weather Data SENSOR DEVELOPMENT AND CHARACTERISTICS. Ralph Petersen, Univ. of Wisconsin-Madison American Meteorological Society Short Course San Antonio, Texas January 14, 2007 Randy Baker – UPS Airlines Ralph Petersen – University of Wisconsin-Madison, CIMSS Stan Benjamin, Bill Moninger – NOAA ESRL Dave Helms, Kevin Johnston, Rich Mamrosh – NOAA NWS

In the mid-1980s, Stan Benjamin and I had the opportunity to participate in the FAA’s Aviation Weather Forecasting Task Force led by John McCarthy. At that time, flight level wind and temperature forecast errors were costing airlines major losses. e.g., Trans-oceanic flights were often forced to make unscheduled refueling stops in route, requiring overnight lodging for passengers and equipment rescheduling A Brief Historical Perspective

Airlines offered to help. –Several airlines were already downlinking automated temperature and wind data for their own internal use At this time, most major airlines had in-house meteorological staffs – and used the aircraft wind/temperature data to update their own systems flight plans –Resulted in financial advantage to airlines collecting data –Airlines were reluctant to share data with airlines that didn’t invest in down-linking costs. –Relied upon existing digital air-to-ground communincations Aircraft Data Collection has been a Joint Industry/Government effort

Airlines offered to help. –Basic AMDAR Data (Flight Level (Pressure), Temperature and Wind) are, for the most part, copies of observations taken for other purposes –Commercial aircraft need accurate temperature and wind observations to operate most efficiently while in flight Pressure to determine altitude Jet Engine performance is related to the temperature difference between the engine and the atmosphere Flight efficiency depends on minimizing head winds Aircraft Data Collection has been a Joint Industry/Government effort

The benefits of AMDAR data are global and large for forecasts out to 48 hour. Results from ECMWF data denial experiments show benefits at all levels, but most in regions where observations are made. Impact of AMDAR Temp / Wind data depends on number of reports Impact of local detail present in AMDAR Temp / Wind reports is greatest in shorter range forecasts – Satellite data dominates longer ranges ( >48 hrs)

The benefits of AMDAR data are global and large for forecasts out to 48 hour.

Profiles from aircraft ascent/descent improve forecasts further over US

THE Reference for AMDAR data

“On-board” Data Processing Cost of aircraft-to-ground data transmission shared by government and airlines - About 1¢ per report - Data available in real time to research community and government - Airlines consider real-time data to be proprietary for others

Observations possible using “Standard” Aircraft Instrumentation

Available Aircraft Data Systems

Deriving Meteorological Variables from Aircraft Data Systems

Observation Accuracies The observations are dependent upon each other –Pressure is measured by an electronic barometer connected to static pressure ports in the “pitot-static”’ probe Note: All measurement devices mounted on fast jet aircraft must be small to reduce drag and minimize associated fuel costs

Observation Accuracies The observations are dependent upon each other –Pressure is converted to Flight Levels using the ICAO Standard Atmosphere for down-linking as part of the meteorological report Pressure-altitude conversions can produce errors of 2-4 hPa Note: Lower Flight Levels are obtained from a radio-altimeter and are reported in geometric units

Observation Accuracies The observations are dependent upon each other –Temperature observations are made via small tubes protruding from the front sides of the aircraft

Observation Accuracies The observations are dependent upon each other –Temperature observations are made via small tubes protruding from the front sides of the aircraft At high air speeds, the force of the air entering the tubes can increase the pressure in the tube, and therefore bias temperatures through “dynamic” (compressional) heating –Corrections for this dynamic heating are made on-board

Observation Accuracies Observation accuracy can vary from one aircraft type to another and depend on locations of instrumentation on the aircraft –“Determine optimal instrument location is still an art form” Sample of temperature differences between co-located radiosonde and UPS 757 reports

Observation Accuracies The observations are dependent upon each other –Indicated air speed is determined using pressure sensors pointing along the direction of the aircraft and sometimes located at various locations on the aircraft Temperature can affect the pressure/speed relationships

Observation Accuracies The observations are dependent upon each other –Pressure is measured by an electronic barometer connected to static pressure ports in the “pitot-static”’ probe –Temperature observations are made via small tubes protruding from the front sides of the aircraft At high air speeds, the force of the air entering the tubes can increase the pressure in the tube, and therefore bias temperatures –Corrections for this dynamic heating are made on-board –Indicated air speed is determined using pressure sensors pointing along the direction of the aircraft and sometimes located at various locations on the aircraft Temperature can affect the pressure/speed relationships –Moisture can also affect the latter two types of observations

Observation Accuracies – Converting Air Speed to Wind Speed - Indicated air speed is determined using pressure sensors pointing along the direction of the aircraft and located at various locations on the aircraft Temperature can affect the pressure/speed relationships –Larger impacts during maneuvers (direction changes – actually changes in aircraft ‘yaw’) Airplanes normally don’t ‘point’ in the direction they are headed –Usually are ‘pointed’ slightly in the direction of any cross-wind –As long as ‘wind-relative’ direction of airplane doesn’t change, pressure on sensors remains constant –However, as an aircraft changes directions relative to the wind (or the wind changed direction very rapidly), the pressure of the air hitting the sensors changes – producing unreliable results –Data taken during maneuvers are flagged before transmission »Roll angle is used as an indicate that maneuvers may be occurring

A review of aircraft orientation terminology

Effect of Aircraft Maneuvers on Wind Calculations –Data taken during maneuvers are flagged before transmission

Indicated Air Speed must be converted to Wind Speed and Direction (Vector) – assuming no cross-winds –Done by combining Indicated Air Speed with motion of the aircraft (Ground Speed and Direction). –Ground Speed (actually a vector) accuracy depends on the accuracy of aircraft navigation systems. –Ground Speed calculated from difference in aircraft locations over short time intervals –In late 1960’s, inertial navigation systems (complex gyroscopic instruments) were developed to determine precise earth locations without the need for a ‘navigator’ on the aircraft. –Later, these systems were replaced by simpler (less costly and lighter) LORAN and GPS systems –Accuracy of Ground Speed calculation depends on: »Accuracy of navigation systems and »Precision of navigation system and Indicated Air Speed outputs –Wind Report made in terms of Speed and Direction »Precision of output Speed and Direction affects final Wind Vector Observation Accuracies – Converting Air Speed to Wind Speed -

Observation Accuracies – Temperature and Wind Speed - More discussions of observed data quality will follow Comparison of AMDAR and GPS radiosonde winds

AMDAR Data Reporting Precision and Desired Accuracy – Primary Observations Many: ±0.7º Inst. Precision 100’ = 30m Temp error Truncated <±0.3º Nearest <±0.15º

AMDAR Data Reporting Precision and Desired Accuracy – Additional Observations

Measuring Moisture Efforts underway for over a decade –Research instruments not appropriate for “day-to- day”, “real world” application –Initial experiments were made using a “stand- alone” Temperature/Relative Humidity sensor called the Water Vapor Sensing System (WVSS-I) Used humidity sensors “similar” to those used on radiosondes –Test results showed: »Substantial Biases and RMS values that exceeded WMO specification »Systems became contaminated by everyday airport “gunk”, e.g. deicer, dirt on runways, etc.

Measuring Moisture Efforts underway for over a decade –Second-generation Water Vapor Sensing System (WVSS-II) measures Mixing Ratio directly Uses a laser-diode system to measure number of water molecules passing sensor Testing on UPS 757s –Initial tests in Spring 2005 –Second version tested last fall

Measuring Moisture Efforts underway for over a decade –Second-generation Water Vapor Sensing System (WVSS-II) measures Mixing Ratio directly Uses a laser-diode system to measure number of water molecules passing sensor Testing on UPS 757s –Initial tests in Spring 2005 –Second version tested last fall Initial results for ascent data from both tests agree –Some descent reports still questionable

Other Measurements Turbulence –“Eddy Dissipation Rate” derived from high-resolution air speed and vertical accelerometer taken from flight data recorder (“black box”) data feed Binned into approximately 10 data divisions Available primarily from a select number of UAL aircraft for research purposes only Icing –“Ice indicator” indicates presence of ice, not the process of icing – Icing detectors under development

Other Data Sources - TAMDAR Efforts underway for nearly a decade –Objective – To fill in data between major airports using commuter airlines A private venture –Built upon concept supported –by FAA and NASA Originally intended for slower-flying aircraft Does not need airlines to subscribe (pay for) ARINC digital communications services Consists of: –Self contained observing system »Temp, Wind, Pressure, Icing, Turbulence, … –Satellite-based air-to-ground communications »Requires new comms systems All data are proprietary

Other Data Sources - TAMDAR

Improvements in NWP over past 10 years AMDAR data have made a difference SAA pilot said recently that flight times from South Africa to Australia are now typically within 1 minute of predictions

Other observations made from Aircraft can be interesting too!! ©Ralph Petersen 2006