On the impact and benefits of AMDAR observations in operational forecasting: Part I: a review of the impact of automated aircraft wind and temperature reports.
Low-cost automated aircraft wind and temperature observations now constitute the third most important dataset for short-range global NWP and, in areas of ample reports, have become the single most important dataset for use in regional NWP applications.
Data obtained from aircraft have played important roles in meteorological research and operations for over a century (Wendisch and Brenguier 2013). Although manual meteorological observations have been collected from commercial aircraft since the 1930s, these pilot reports (PIREPs) often provided minimal value, owing both to transcription and telecommunication errors and to inaccuracies in aircraft location reports and related wind determinations. The first automated aircraft data acquisition approach was implemented during the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE; Julian and Steinberg 1975), when meteorological observations were stored using onboard recording systems. Automated collection and transmission of aircraft observations began during the First GARP Global Experiment (FGGE) in 1978/79. Although the initial use of these limited aircraft reports (AIREPs) was restricted primarily to research applications, the number, quality, and operational use of these data have expanded substantially in the years since. Critical to this success was the introduction of accurate Long Range Navigation (Loran; U.S. Coast Guard 1962) and Inertial Navigation Systems (INS; Woodman 2007) instruments that provided reliable aircraft position and Earth-relative wind data as routine by-products.
Regular real-time transmission of automated weather observations began using dedicated equipment mounted on commercial aircraft as part of the Aircraft to Satellite Data Relay (ASDAR) program (Fleming et al. 1979; Sparkman et al. 1981) that remained available after FGGE. More than 20 ASDAR-equipped aircraft provided reports during the next 20 years from eight air carriers around the globe. Although the program provided beneficial information in data-sparse regions, it required that extra equipment be installed and maintained on participating aircraft. The project, however, demonstrated that high-quality wind and temperature observations could be obtained from commercial aircraft, especially near the high kinetic energy regions around the jet stream that can influence predictions of cyclogenesis and storm evolution.
With the development of modern aircraft equipped with flight computers and improved navigation systems, it became apparent that these types of observations could be made much more efficient and affordable by installing ASDAR-like software to collect data from sensors already present on commercial aircraft. The system in the United States was originally known as the Meteorological Data Collection and Reporting System (MDCRS; Martin et al. 1993). Reports were sent to ground stations using systems available through several telecommunication service providers, including Aeronautical Radio Incorporated (ARINC) and Societe Internationale de Telecommunications Aeronautiques (SITA). The messages included location, temperature, and wind reports that were used by individual airlines to monitor aircraft performance and to improve flight planning and systems efficiency. As the LORAN and INS systems used on earlier wide-body aircraft were replaced by global positioning system (GPS) location finders on a much broader fleet of aircraft and internal aircraft communications systems were upgraded, the availability of automated aircraft reports increased further, without loss of quality. These observations are now part of the broader World Meteorological Organization (WMO) Aircraft Meteorological Data Relay (AMDAR) program (WMO 2003, 2014a). (1)
In the mid-1980s, U.S. airlines approached the Federal Aviation Administration (FAA) with a request to improve the quality of the flight-level wind forecasts used in their flight planning systems. At that time, errors in short-range upper-level wind forecasts from numerical weather prediction (N WP) models over the Northern Hemisphere (NH) ranged between 9 and 10 m [s.sup.-1] [root-mean-squared vector (RMSV) wind error]. This uncertainty had large negative impacts on airline operating costs, including the need to carry excessively large amounts of extra fuel when tailwinds were larger than expected or to make unplanned refueling stops on long international routes when headwinds were higher than forecast. Further, the airlines needed better quality and vertical- and horizontal-resolution forecast wind profiles near airports to optimize fuel use, especially during aircraft descent.
To address these and other aviation weather forecasting issues, the FAA established an "Aviation Weather Forecasting Task Force" (NCAR 1986). A principal objective of this effort was to improve flight-level wind and temperature forecasts provided by NWP models through the World Area Forecast Centers (WAFCs) at the Met Office and the U.S. National Weather Service (NWS). See WAFC Washington (2011) for details.
A major outcome of the task force was agreements by five U.S. airlines to allow the WAFCs to acquire and use the previously proprietary weather data portions of automated aircraft reports in their real-time operations and NWP systems. In return, the WAFCs made significant efforts to enhance their NWP systems to make better use of these new data. For example, the vertical resolution of the U.S. global NWP system was enhanced in the area near the tropopause and the horizontal resolution of the model and output fields were increased. Development was also begun on a new domestic NWP system designed specifically around the new observations and intended to provide improved very-short-range forecasts of wind and temperatures for use in flight planning and air traffic management systems across the United States (Benjamin 1989; Benjamin et al. 1991).
Improvement in global-scale wind forecasts in the two decades after 1984 (Fig. 1) can in part be attributed to use of these new automated aircraft data in improved data assimilation (DA) systems. Between 1984 and 2004, the average NH wind errors decreased by approximately 40%, from 10 to 6 m [s.sup.-1]. More importantly, aircraft were less likely to encounter unexpected areas of excessive head- or tailwinds, as noted in the nearly 45% reduction in wind errors near the jet stream (defined as areas with greater than 40 m [s.sup.-1] wind speeds) from above 13 m [s.sup.-1] to less than 8 m [s.sup.-1], thereby enhancing airline efficiency and reducing fuel consumption. Although is it difficult to determine the exact contributions of enhancements in observations and DA systems, it should be noted that between 1986 and 2001, a time when few changes were made to the National Centers for Environmental Prediction (NCEP) global DA system, wind forecast errors declined as aircraft observations increased.
Since the late 1990s, the AMDAR program has expanded to include more than 3500 aircraft from 39 airlines globally. Ten different national and regional AMDAR programs currently provide more than 680,000 wind and temperature reports daily. The observation intervals range between 5-7 and 1-3 min when aircraft are at cruise levels and more frequently during aircraft takeoff and landing, as frequently as every 6-20 s (see appendix for details). High-quality moisture observations are now also available from more than 100 aircraft, primarily over the United States.
AMDAR data are routinely available over large portions of the globe, with the highest density of observations over the heavily traveled areas of North America (NA) and Europe (Fig. 2). Oceanic observations are generally limited to upper-tropospheric reports along intercontinental routes. Elsewhere, reports are taken both at cruise levels and during ascent and descent. Although flight-level reports are most prevalent, the number of aircraft profiles is increasing steadily, even in some data-sparse regions where AMDAR reports are playing increasingly important roles. The quality of the wind and temperature data has been determined by various authors to be very high (e.g., Moninger et al. 2003; Benjamin et al. 1999) and is monitored operationally by NCEP as part of WMO data protocols. (2)
This paper reviews the impact of AMDAR observations on operational NWP models at weather forecast offices at both regional and global scales.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
Although much of the discussion of regional NWP impact builds upon a literature review of the use of abundant automated aircraft observations available over the United States during the past two decades, discussion of recent global impacts relies on materials from the Fifth WMO Workshop on the Impact of Various Observing Systems on Numerical Weather Prediction conducted as part of the World Weather Watch in Sedona, Arizona, in May 2012 (WMO 2012) and related meetings. An evaluation of the increasingly available AMDAR moisture observations is presented in Petersen et al. (2017, hereafter Part II).
ASSESSMENTS OF AMDAR TEMPERATURE AND WIND DATA QUALITY. Integral to any operational NWP improvement effort must be a parallel program to monitor the datasets and develop appropriate quality control (QC) procedures. A number of schemes have been developed at global NWP centers to assure that erroneous AMDAR data are excluded from operational DA systems (e.g., Ballish and Kumar 2008). Outputs, including lists of aircraft producing questionable reports, are frequently updated and transmitted to other NWP centers and participating airlines so that errors in suspect aircraft instrumentation and communication systems can be corrected. [See Jacobs et al. (2014) for more details regarding candidate data correction procedures.]
A major problem in melding the variety of different data used in modern DA systems is the need to recognize and remove biases contained in each dataset (Dee and Uppala 2009). If this is not done, the biased data reports can reduce the impacts of other datasets. For example, if one large dataset has a cold or warm bias compared to all other data sources, those data can have the detrimental effect of cooling (or warming) the entire analysis, even though the dataset may contain valuable information about the spatial and temporal temperature variations. The situation can be made even worse if information from a reliable data source available only once or twice per day is countered by biased reports available many times daily. As noted at the recent NOAA Aircraft Data Workshop [see WMO (2014b)], however, the availability of multiple reports in proximity is also valuable for cross validation and QC.
Although numerous studies have shown that AMDAR data have very small random errors, Ballish and Kumar (2008) identified that individual AMDAR temperature reports can have systematic warm biases of as much as 1[degrees]C when compared with analysis background fields and can fluctuate by altitude, phase of flight (i.e., ascent, cruise, and descent), and aircraft type. In general, the magnitudes of the biases increase with altitude, ranging from values of a few tenths of a degree near the surface to as much as 1[degrees]C at flight levels. In part, they may be produced by combinations of hysteresis effects in the onboard sensors (especially during ascent and to a lesser degree descent), dynamic heating within the sensor at high speeds (at cruise levels), and truncations by the onboard data processing systems (which vary between aircraft). The biases also can fluctuate between seasons (with slightly larger values noted in summer) and aircraft type (e.g., three different Boeing 737 models had biases that ranged from 0.8[degrees] to 1.2[degrees]C). A scheme designed by Isaksen et al. (2012) to correct these biases and implemented at the European Centre for Medium-Range Weather Forecasts (ECMWF) has improved upper-level temperature analyses and agreement with stratospheric satellite measurements, especially near and above the tropopause. Other NWP centers are planning to implement similar bias correction schemes in the near future, some of which include relationships between aircraft ascent and descent rates and vertical variations in biases (Zhu et al. 2015).
REVIEW OF THE IMPACT OF AMDAR OBSERVATIONS ON SHORT-RANGE REGIONAL FORECASTS. Although data impact tests generally focus initially on the global and longer-time scales and then later on regional scales, the availability of large volumes of automated aircraft observations taken throughout the day (called asynoptic observations) over the United States provided the opportunity to develop a succession of short-range DA and NWP models tailored specifically to use these unique datasets. The first of these was the Rapid Update Cycle (RUC), developed by the National Oceanic and Atmospheric Administration (NOAA) at the Earth Systems Research Laboratory (ESRL) and run hourly at NCEP, provides the focus of the following discussion. The evolution of the model and the importance of various asynoptic and synoptic datasets to the quality of short-range RUC forecasts are described in a series of papers by Benjamin et al. (1991, 2004). Results of short-range impact tests using other analysis and forecast systems can be found in Cardinali et al. (2003, 2004), Gao et al. (2012), Huang et al. (2013), Laroche and Sarrazin (2010), and Lupu et al. (2011).
In an early attempt to quantify the impact of aircraft observations, Benjamin et al. (1991) compared results including MDCRS data in an 80-km-resolution developmental version of the RUC against equivalent twice-daily forecasts from the NWS's then-operational Nested Grid Model (NGM) that relied primarily on rawinsonde observations (raobs) for upper-air information. Aircraft temperature and wind reports at this time were typically collected only at regular 7-10-min intervals over the United States, mostly at cruise levels near the jet stream. The continual insertion of aircraft data had a positive impact on short-range forecasts throughout the day. By 2100 UTC, the inclusion of 9 h of MCDRS reports available since the previous raob reports reduced errors in forecasts of 250-hPa winds over 40 m [s.sup.-1] by approximately 10% when compared to 12-h NGM forecasts valid at the same time. This in turn lowered flight duration estimate errors along several major U.S. flight routes by 30%, thereby offering airlines a means of refining fuel-use estimates.
In 2000, Schwartz et al. examined the effect of model resolution and physics on the utility of MDCRS observations. For these tests, a 60-km version of the RUC was compared to a 40-km version of the RUC-2, which included more sophisticated surface physics, improved orography, and higher vertical resolution. The RUC-2 performed better at all forecast ranges, with the largest impacts noted at the shortest forecast ranges. Improvements above 300 hPa and below 700 hPa were attributed primarily to increased vertical model resolution and improved boundary layer parameterizations, while the reduced improvements at midlevels were likely related to the paucity of ascent and descent reports available at the time.
[FIGURE 3 OMITTED]
Recognizing the need to obtain profiles during aircraft ascent and descent, the NWS and FA A worked with participating U.S. airlines in the late 1990s to increase observing frequencies to approximately every 10 hPa in the lowest 100 hPa during takeoff and every 50 hPa until an aircraft finished its climb out (typically near 400 hPa), with a higher vertical reporting rate used again during descent. (See appendix for details.) To quantify the impact of these higher-frequency wind and temperature profiles, Petersen (2004b) conducted a limited series of tests using the then-operational 20-km-resolution RUC-2. For 3 weeks in June 2002, operational runs of the RUC-2 using all AMDAR data were compared to experimental runs in which all aircraft wind and temperature reports below 350 hPa were removed. Unlike some of the global data-denial tests to be discussed later, no thinning or averaging of the aircraft data were done.
Comparisons of analyses and 3-, 6-, 9-, or 12-h forecasts from both systems to raobs over the contiguous United States (CONUS) at 0000 and 1200 UTC are shown in Fig. 3. The asynoptic ascent and descent temperature and wind aircraft profiles had positive impacts on wind analyses at both the 0000 and 1200 UTC synoptic times. Although improvements were noted at all levels, impacts below 300 hPa were 3-4 times greater than at upper levels where abundant cruise-level data were retained in both tests. Improvements at these synoptic analysis times, when raobs typically dominated analyses over the United States, were the combined result of both the additional aircraft ascent and descent reports available at these times and enhancements in the analysis background fields resulting from the inclusion of those reports during the previous 12 h. The latter effect is especially apparent above 350 hPa, where the amount of data available to both tests was unchanged.
The impacts of analysis improvements on 12-h forecasts were also positive across all variables. Wind forecast improvements at and above 300 hPa were comparable to the initial analysis differences. Below 300 hPa, the impacts of the ascent and descent data on the forecasts were smaller than in the analyses, but still averaged about 4%, more than double the improvement found at upper levels where cruise-level observations dominated. Inclusion of aircraft ascent and descent data in the midtroposphere also improved most 3-9-h forecasts by 5%-9%, with the largest error reductions in the shortest forecast ranges when AMDAR data dominated both the initial condition and analysis background fields. These improvements were equivalent to doubling the model resolution, which would have required a tenfold increase in computing resources (Benjamin et al. 2002). The smaller changes in the 9-h forecasts from 0300 and 1500 UTC above 400 hPa and slight degradation near 250 hPa may be the residual effect of the assimilation system adjusting to biased cruise-level aircraft temperature reports in the periods immediately after raob data were used.
The cumulative impact of AMDAR observations received throughout the day was determined by comparing forecasts made with and without ascent and descent data from successive hourly RUC analysis updates between 0000 and 0900 UTC and 1200 and 2100 UTC with 12-h forecasts from 0000 and 1200 UTC. Improvements using all AMDAR data collected during the updating interval ranged from 0.2 to 1.2 m [s.sup.-1] across all levels (Fig. 4), equivalent to improvements from about 5% at lower levels to 20% aloft. The larger error reduction aloft is likely a combination of the inherent presence of larger errors (and therefore larger margin for improvement) near the jet stream and improvements in the underlying thermal structures supporting the vertical wind structures provided by the ascent and descent temperature data. Forecasts using only upper-level aircraft reports showed slightly smaller improvements aloft and degradations in the middle troposphere, illustrating both the quality of the full-tropospheric aircraft data and the importance of using them frequently in N WP data assimilation systems.
[FIGURE 4 OMITTED]
Figure 5 summarizes the overall impact of automated aircraft wind and temperature profiles on regional analyses and very-short-range forecasts valid at 0000 and 1200 UTC. AMDAR reports have positive impacts at all times and for all parameters, including indirect improvements in moisture forecasts. Improvements were greatest for the 3- and 6-h ranges when the largest cumulative amount of asynoptic data had been assimilated. The especially large temperature improvements reflected not only the availability of thermal data in the aircraft reports, but also the use of more accurate wind and temperature fields in advection calculations employed in determining analysis background fields. Because no new humidity data were included in these tests, the enhancements in the humidity fields must similarly be attributed to improved advection computations. The slightly reduced impact of the aircraft data in the 0000 and 1200 UTC analyses (relative to the 3-h forecasts) was again likely due to conflicts and redundancies between the biased AMDAR reports and raob profiles at these times.
More recent tests have documented the increasing influence of aircraft reports in short-range forecasts over the United States where automated aircraft data abound. "Data denial" tests were conducted using 13-km versions of both the last operational version of the isentropic-hybrid coordinate RUC and its successor, the sigma-coordinate Rapid Refresh (RAP) Model, which includes advanced DA systems and enhanced NWP physics (Benjamin et al. 2010, 2012, 2014). Results from the "control run" analysis and forecasts using all data sources were compared with reruns in which different sources of data were removed individually. Differences in forecast errors between the control run and various tests used as a measure of the impact of each data source (Fig. 6). It should be noted that for these tests, aircraft reports included not only AMDAR reports available on the WMO Global Telecommunications System (GTS), but other sources. Most notably, 10%-15% of the observations were from Tropospheric Airborne Meteorological Data Reporting [TAMDAR; see Daniels et al. (2006), Moninger et al. (2007), and Gao et al. (2012)] equipped aircraft, generally made below 500 hPa and near airports not usually served by AMDAR aircraft.
Overall, aircraft data dominated as the single most important dataset over the CONUS for both the analyses and forecasts out to at least 12 h. This was especially apparent in shorter-range forecasts and in the cold season when baroclinic processes dominate the midlatitudes (not shown). The next most important datasets were raob and surface observations.
[FIGURE 5 OMITTED]
[FIGURE 6 OMITTED]
Only for 12-h forecasts during winter did raob reports show comparable impact to AMDAR data. Automated aircraft observations continued to show their largest impacts at shorter forecast ranges and during the daytime and early evening hours when observations are most abundant. These results provide evidence that collaborations between U.S. airlines and the NWS to promote improved access to and use of AMDAR wind and temperature reports has benefited both the data providers and other users of products issued by NWP centers, as well as forecasters using these data.
REVIEW OF THE IMPACT OF AMDAR OBSERVATIONS ON GLOBAL FORECASTS.
As with regional N WP systems, the impact of aircraft observations in global N WP results from a combination of enhancements in the global AMDAR coverage and improvements in analyses. Global DA systems must integrate the in situ aircraft data with a huge number of different types of observations, the vast majority of which are satellite based and provide critical information over the oceans globally. Within this mix of data, automated aircraft reports are unique in that they are the only nonsurface, globally distributed asynoptic dataset that directly measures both temperature (mass), wind (momentum), and, in some cases, moisture (Part II). Raobs provide these same types of observations, but only at synoptic times and are generally not available over oceans. All other data systems need to infer one of these variables from the others based upon a variety of different assumptions and constraints within the DA systems. In addition, because AMDAR observations tend to be concentrated near the jet stream level, they can be particularly useful in defining the sources of kinetic energy that drive many weather systems, especially in data-sparse regions. It is worth noting that in situ AMDAR measurements not only provide independent observations in the same places observed by satellite data to improve their value but also supply additional horizontal and vertical detail.
Since the volume of automated aircraft reports began increasing in the late 1980s, a number of major advances have also occurred in operational DA systems, led to a large degree by the need to make better use of the increasing number of higher-resolution satellite observations that are available (Kalnay, 2003). In general, these advances followed a progression from using techniques that allowed the satellite data to be treated in their native form (i.e., as radiances, rather than converting them to the NWP parameters of temperature, wind, and moisture) to implementing procedures that use more data throughout longer assimilation periods. Had additional resources been put toward improving the availability and use of aircraft observations in DA systems in the past, it is likely that the impacts of satellite systems would have been improved even further.
To ascertain the relative importance of the different datasets, a variety of gobal observing systems experiments (OSEs) have been conducted during the past two decades. The following brief synopsis of earlier impacts of AMDAR data will focus on ECMWF results, followed by a review of more recent findings from a broader set of global NWP centers.
One of the longer global studies by Kelly et al. (2004) used the ECMWF global analysis and forecast system [T-159 (~120 km) four-dimensional variational data assimilation (4D-Var) analysis and T-511 (~40 km) forecast model] with data from August to September 2002 and December 2002 to January 2003. Fifteen airlines provided AMDAR data internationally during the period, less than half the number of current participants, and some data thinning was done [see Andersson et al. (2005) for details]. Although the importance of satellite data collected throughout the day from multiple spacecraft on medium-range 500-hPa forecasts increased notably in the improved version of 4D-Var implemented in 2004 (Kelly and Thepaut 2007), AMDAR temperature and wind observations had greater impact at shorter forecast ranges and in the upper troposphere, as discussed next.
Using data provided by Kelly et al. (2004), Petersen (2004a) performed a more detailed analysis of the specific impact of AMDAR observations in the ECMWF global forecast system on shorter-range global (12-48 h) temperature and wind predictions for the NH and the more-aircraft-data-rich NA area (Fig. 7). In contrast to the regional tests described earlier that used both en route and vertical profile reports over the United States, most oceanic AMDAR reports used in these tests only provided cruise-level data.
When evaluated over the full NH, AMDAR data had positive impact on all forecasts out to at least 48 h. The impacts were largest in the first day, with average 12-h temperature and wind errors reduced by 14% and 8%, respectively. The improvements were most pronounced near the primary flight levels between 200 and 300 hPa and extending downward to 500 hPa. Above the levels where AMDAR reports are available, improvements in both parameters were smaller (3%-6% and 2%-5% at 100 hPa). In general, the reports had greater relative influence on temperature forecasts than wind forecasts (especially in the first forecast day).
Over NA, where more AMDAR data were available both at flight levels and during aircraft ascents and descents, the impacts were larger than over the entire NH at all levels below 100 hPa and extended over a longer range of forecast times. During the first day, 12-h temperature forecasts between 200 and 300 hPa improved by nearly 23% and wind forecast errors were reduced by 12%--14%. Improvements in 200-300-hPa temperature and wind forecasts continued through 2 days. Relative improvements were again larger for temperature than for winds.
The impacts of AMDAR data on wind forecasts over other portions of the globe are depicted in Fig. 8. Although AMDAR reports available then were concentrated in the NH, the data benefited most regions of the globe, including the tropics and the Southern Hemisphere (SH). The largest forecast improvements occurred at cruise levels in the tropics and in the regions that had the highest data concentrations at the time of this study, particularly over the United States, Europe, and oceanic flight routes in the NH, as well as in the area around Indonesia. Similar improvements should be expected in other areas as data availability increases in the future.
On average, forecasts for the NH (Fig. 8, right) improved by more than 0.25 m [s.sup.-1] between 200 and 300 hPa. From the North Pacific into Europe (Fig. 8, top), average improvements ranged from 0.3 to nearly 0.5 m [s.sup.-1] for 12-h forecasts. In individual events, improvements were notably larger, often exceeding 5 m [s.sup.-1]. For longer forecast periods, impacts decreased by 25%-75% at different levels and locations. Similar improvements and trends were also noted in temperature forecasts for these same levels for each region (not shown). Enhancements in global forecasts such as these not only allow aviation users to adjust flight plans to reduce fuel use and optimize efficiency (NCAR 1986) but also benefit medium-range NWP systems by improving background fields used in their data assimilation procedures.
[FIGURE 7 OMITTED]
RECENT TESTS OF THE IMPACT OF AMDAR OBSERVATIONS. Since 1996, the WMO has sponsored a series of quadrennial scientific meetings to understand and assess the impact of various observing systems on NWP skill. The following summarizes results from the fifth such meeting held in Sedona, Arizona, in May 2012 (WMO 2012). Participants included representatives of all major global and regional DA and NWP centers as well as experts in various observing systems and forecast applications.
During the past several years, more advanced techniques have been developed to increase understanding and assess the relative impact of various observing systems using sophisticated DA systems. Although the number of satellite observations available to DA systems has increased dramatically, typically fewer than 5% of the available radiance data are used (Derber and Collard 2011). By contrast, only about 5% of AMDAR-equipped aircraft are excluded through rejection lists at any one time (Pauley et al. 2014). Automated aircraft observations now represent the largest nonsatellite data source, an indicator of their quality and increasing importance, and supply nearly 35% of the impact of all "conventional" observations globally and 60% over the United States (Isaksen 2014).
Figure 9 shows the average impact of various observations obtained by compositing independent results from five participating DA centers using similar data combinations, including the Met Office, NCEP, ECMWF, Meteo-France, and the National Aeronautics and Space Administration (NASA) Global Modeling and Assimilation Office (GMAO). It should be noted that variations in results among centers are expected owing to a number of factors, such as the use of different data selection criteria and DA techniques, including varying ability to maximize benefits from hyperspectral infrared (IR) satellite observations.
[FIGURE 8 OMITTED]
AMDAR reports rank third in importance among all observations for improving 24-h global weather forecasts, behind only microwave satellite observations [primarily Advanced Microwave Sounding Unit A (AMSU-A) observations, which afford more continuous global coverage than AMDAR reports] and raob reports (which furnish important information above the highest AMDAR reporting levels) when using evaluation criteria similar to those described by Cardinali (2009). Further, AMDAR data exhibit the most consistent impact across the five centers, as indicated by their small variability compared to other major datasets. These results attest not only to the quality of the data, but also to the importance of using reports made both near airports in ascent and descent and around the major source of atmospheric energy near the jet stream, the ease and economy with which the temperature and wind information can be incorporated into DA systems, and the importance of multiple reports along flight routes for use in cross validating individual observations. Other tests, some using different combinations of data sources, show even greater aircraft data impact regionally and globally, in some instances, providing the greatest impact of any dataset (e.g., Kang et al. 2014; Alexander et al. 2014; Lupu et al. 2012; Ota et al. 2013).
Several additional studies have addressed other important aspects about the role played by AMDAR data within the total Global Observing System (GOS). For example, Andersson and Radnoti (2012) saw only small degradation in ECMWF forecast skill when the number of raobs was reduced selectively but the number and distribution of AMDAR reports was held constant. Cress et al. (2012) also noted that AMDAR data retained impact longer than raobs in some regional forecast systems. Other studies indicate that if AMDAR reporting variables are assimilated separately, wind observations had a somewhat greater impact than temperature data but that the greatest impact occurred when both parameters were assimilated simultaneously [see WMO (2012) for more details].
[FIGURE 9 OMITTED]
For individual events, the impact of AMDAR data can be much larger. For example, Hoover et al. (2014) studied the impact of AMDAR data on forecasts of Hurricane Sandy using the Navy Global Environmental Model and Naval Research Laboratory Variational DA System. In these tests (Fig. 10), AMDAR data had a greater impact than any other data source on improving forecasts of the location and timing of landfall of this major storm, both at 24 and 48 h. In this case, the impact was nearly double that of raobs. These enhancements can be especially important for airlines to minimize unnecessary disruptions to flight operations while maximizing the safety of both passengers and their aircraft.
ECONOMIC ASPECTS OF AMDAR OBSERVATIONS. Eyre and Reid (2014) introduced an approach to obtain needed information about the cost effectiveness of the AMDAR observations relative to all other data sources used in global N WP systems. They first showed the impact of each of the individual data systems available for real-time application in 2013 using the Met Office global analysis and forecast model assimilation and forecast system. Results indicate that the largest global forecast improvements are due to microwave and hyperspectral infrared (H-IR) observations from multiple (four or more) polar-orbiting satellites, as shown by the gray bars on the left side of Fig. 11. Other analyses have also shown that increasing the amount of satellite data increases their impacts at least linearly. The remaining members of the five highest-impact datasets have similar magnitudes and include rawinsonde reports and two aviation-related observing systems: surface synoptic observations plus aviation routine weather reports (SYNOP+METAR) surface observations and AMDAR+AIREP aircraft reports.
When estimates of the costs of each observing system are included to approximate which observations have the largest impact per unit cost, the rankings change. (2a) As shown in the right side of Fig. 11, by far the two most cost-effective observing systems are AMDAR+AIREP and drifting buoys, the latter of which, however, has approximately 5 times less impact than automated aircraft reports. Confidence in the relative rankings of these two systems was quite high, despite the uncertainties in the precise costs estimates used in the study. For reference, AMDAR+AIREP observations have an annual cost globally on the order of $6-$8 million [U.S. dollars; about one-four-hundredth of the annual cost of the GOS. Of the five highest-impact systems, all others appear to be less cost effective than AMDAR observations and contribute substantially larger portions of the total GOS costs that can be attributable to NWP use, with raobs appearing to be the least economically effective.
[FIGURE 10 OMITTED]
Based on results shown earlier in this paper and findings from multiple previous studies that varied the amounts of satellite data available to NWP systems, it is anticipated that increasing the spatial coverage of AMDAR reports will further enhance their impact and continue to improve the quality of operational NWP products used by the general public and aviation community, including terminal and en route forecasts. Even in the unlikely scenario that there would be no change in impact, a doubling of the current number of AMDAR reports would still leave the observations nearly 3 times more cost effective than any other observing system. Although an amalgam of many different datasets is essential for sustaining improvements in NWP into the future, AMDAR data are the most cost effective of any global dataset and one that can be expanded quickly and easily, making them particularly attractive for use in areas where raob availability may be in jeopardy. In particular, expansion of automated aircraft reporting systems may offer a highly cost-effective means of improving local aviation services, as well as short- and medium-range weather prediction, for regions such as Africa and South America, as well as outlying areas, such as the Pacific Islands and data-sparse, high-latitude areas such as Alaska and Russia.
[FIGURE 11 OMITTED]
SUMMARY AND RECOMMENDATION.
Tests conducted by numerous NWP centers for over 25 years have demonstrated that high-quality and high-frequency AMDAR temperature and wind observations increase the skill of forecasts at both regional and global scales and for both short- and medium-range forecasts. Results show that aircraft data taken at cruise levels and during ascent and descent provide important information for improving forecasts, both in terms of long-term average performance and for individual events. Although global, "all weather" satellite microwave observations have the largest average influence on medium-range global forecasting systems (especially in the SH), AMDAR observations have become recognized as a critical component of these systems around the world. Aircraft observations rank third in importance globally (especially in the NH) and contribute between 10% and 15% to 24-h forecast skill improvement, with impacts extending to 48 h and beyond. In areas with denser data coverage aloft and abundant ascent and descent reports, they have become the single most important dataset for use in shorter-range, regional NWP applications, especially when moisture data are also included (Part II).
A unique feature of AMDAR reports is that they provide both temperature and wind data at the same locations and in profiles made during ascent and descent, thereby furnishing explicit two-dimensional information on baroclinic adjustments needed in DA systems. Because the data are available continuously along flight routes, the observations also provide information about gradients of wind and temperature near high-energy jet stream regions. The availability of multiple reports along flight routes is also important for cross validation and QC (WMO 2014b).
Results presented here attest to 1) the quality of the data and the value of bias correction, 2) the importance of the reports made both as profiles during ascent and descent and at cruise level near the major reservoir of energy in the atmosphere, and 3) the ease of use of the temperature and wind information in DA systems. Additional experiments are needed to understand more fully how temperature and wind gradient information that can be derived from cruise-level AMDAR reports may contribute to the enhanced importance of AMDAR data relative to other, more costly datasets.
The long-term, consistently positive impacts of automated aircraft reports on regional and global operational NWP presented here provide ample evidence that the collaborative effort between airlines and national meteorological services to improve access to and use of automated wind and temperature reports has benefited both the data providers and other users of products issued by NWP centers and forecasters using these data. Future addition of automated turbulence and icing observations could further increase safety-related services important for aircraft and airspace operations.
AMDAR growth has occurred chiefly over developed countries, whereas, for developing and least-developed countries, the progress in implementation has lagged and is now well behind. Because improvements attributable to AMDAR observations have been concentrated in areas of highest data availability, greater improvements can be expected in other more data-sparse regions as the spatial and temporal coverage of AMDAR reports increases globally. As a means of fostering further AMDAR expansion, cooperative means (including possible cost-sharing opportunities) should be developed both to continue expanding the AMDAR observing network into areas not currently covered adequately and to increase the number of aircraft providing data, especially over data-sparse regions of the globe. This should include establishing dedicated efforts at regional and global NWP centers to continue evaluation of the impact and cost effectiveness of all observing systems components, with the goal of promoting rapid expansion of those systems that have both high value and low cost.
ACKNOWLEDGMENTS. This paper was commissioned and funded by the WMO in support of an effort to expand the global collection of automated aircraft observations. Particular thanks go to Dean Lockett and Frank Grooters for their encouragement and support. I thank the reviewers for their recommendations, which have been particularly useful in strengthening the paper. Finally, the roles of Kenneth Macleod, Charles Sprinkle, and Jeff Stickland in developing the AMDAR program through WMO deserve recognition.
APPENDIX: DESCRIPTION OF AMDAR DATA REPORTING PROCESSES. Early automated aircraft meteorological observations used a number of formats, as well as different reporting frequencies, spatial density, reporting precisions, or phase of flight needed to assess potential wind and temperature errors. As part of the WMO AMDAR effort, standard reporting practices were adopted by the participating airlines and air-to-ground communication centers, as shown schematically in Fig. A1.
Observations of aircraft location, time, phase of flight, and altitude, along with the primary temperature and wind observations (as well as moisture, turbulence, and other optional variables if available) are made at specific intervals throughout an aircraft's flight. These data are collected onboard the aircraft using specially developed software, stored for a short interval, and then transmitted to the ground via radio or satellite as a small addition to normal messages containing information about engine performance, fuel use, etc., using existing digital air-to-ground communications systems. The meteorological portions are then encoded into WMO standard formats and forwarded to operational meteorological services via the WMO Global Telecommunication System (GTS).
A consolidated set of standards for providing AMDAR observations has been recently updated by the WMO (2003, 2014a). Meteorological variables are reported at different intervals depending on aircraft phase of flight, as shown in Fig. A2. Increased reporting during takeoff and landing captures thermal structures and shear zones across narrow inversions and fronts. Nominally, reports are obtained or calculated from the highest-frequency (typically 1 s-1) instantaneous measurements available made closest to the observation time, using minimal smoothing, and subject to established data validation requirement procedures. On occasion, reporting data precisions can exceed the accuracy of the aircraft onboard measurements, resulting in additional errors.
[FIGURE A1 OMITTED]
[FIGURE A2 OMITTED]
Alexander, C., and Coauthors, 2014: The High-Resolution Rapid Refresh (HRRR): A maturation of frequently updating convection-allowing numerical weather prediction. Extended Abstracts, World Weather Open Science Conf, Montreal, QC, Canada, CIRES/ESRL/NCEP, 45 pp. [Available online at http://ruc.noaa.gov/ppt_pres/WWOSC2014_Alexander_Final.pdf.]
Andersson, E., and G. Radnoti, 2012: Impact of satellite and conventional observations on the performance of the ECMWF data assimilation system by OSEs. Proc. Fifth Workshop on the Impact of Various Observing Systems on Numerical Weather Prediction, Sedona, AZ, WMO, la (4). [Available online at www.wmo.int/pages/prog/www/OSY/Reports/NWP-5_Sedona2012.html.]
--, C. Cardinali, B. Truscott, and T. Hovberg, 2005: High frequency AMDAR data--A European aircraft data collection trial and impact assessment. ECMWF Tech. Memo. 457, 15 pp.
Ballish, B. A., and V. K. Kumar, 2008: Systematic differences in aircraft and radiosonde temperatures--Implications for NWP and climate studies. Bull. Amer. Meteor. Soc., 89, 1689-1707, doi: 10.1175/2008BAMS2332.1.
Benjamin, S. G., 1989: An isentropic mesoa-alpha scale analysis system and its sensitivity to aircraft and surface observations. Mon. Wea. Rev., 117, 1586-1605, doi: 10.1175/1520-0493(1989)117<1586: AIMSAS>2.0.CO;2.
--, K. A. Brewster, R. Briimmer, B. F. Jewett, T. W. Schlatter, T. L. Smith, and P. A. Stamus, 1991: An isentropic three-hourly data assimilation system using ACARS aircraft observations. Mon. Wea. Rev., 119,888-906, doi: 10.1175/1520-0493(1991)119<0888: AITHDA>2.0.CO;2.
--, B. E. Schwartz, and R. E. Cole, 1999: Accuracy of ACARS wind and temperature observations determined by collocation. Wea. Forecasting, 14,1032-1038, doi:10.1175/1520-0434(1999)014<1032:AOAWAT >2.0.CO;2.
--, and Coauthors, 2002: RUC20--The 20-km version of the Rapid Update Cycle. NOA A Tech. Memo. OAR FSL-28, 30 pp.
--, and Coauthors, 2004: An hourly assimilation forecast cycle: The RUC. Mon. Wea. Rev., 132, 495-518, doi: 10.1175/1520-0493(2004)132<0495: AHACTR>2,O.CO;2.
---, B. D. Jamison, W. R. Moninger, S. R. Sahm, B. E. Schwartz, and T. W. Schlatter, 2010: Relative short-range forecast impact from aircraft, profiler, radiosonde, VAD, GPS-PW, METAR, and mesonet observations via the RUC hourly assimilation cycle. Mon. Wea. Rev., 138, 1319-1343, doi:10.1175/2009MWR3097.1.
--, H. Lin, S. Weygandt, S. R. Sahm, and W. Moninger, 2012: Impact of upper-air and near-surface observations on short-range forecasts from NOAA hourly assimilation cycles (RUC and Rapid Refresh). Proc. Fifth Workshop on the Impact of Various Observing Systems on Numerical Weather Prediction, Sedona, AZ, WMO, 2a (1). [Available online at www.wmo.int/pages/prog/www/OSY/Reports/NWP-5_Sedona2012.html.]
--, E. James, W. Moninger, B. Jamison, S. Weygandt, and C. Alexander, 2014: Impact of aircraft observations on short-range aviation weather forecasts from NOAA regional models. Proc. Aircraft Data Workshop, Annapolis, MD, NOAA, 46 pp. [Available online at www.arinc.com/resources/customers/meetings/NOAA/NOAA_Aircraft_Data_Workshop_2014/.]
Cardinali, C., 2009: Monitoring the observation impact on the short-range forecast. Quart. J. Roy. Meteor. Soc., 135, 239-250, doi:10.1002/qj.366.
--, L. Isaksen, and E. Andersson, 2003: Use and impact of automated aircraft data in a global 4DVAR data assimilation system. Mon. Wea. Rev., 131, 1865-1877, doi:10.1175//2569.1.
--, L. Rukhovets, and J. Tenenbaum, 2004: Jet stream analysis and forecast errors using GADS aircraft observations in the DAO, ECMWF, and NCEP models. Mon. Wea. Rev., 132, 764-779, doi:10.1175/1520-0493(2004)132<0764:JSAAFE>2 ,0.CO;2.
Cress, A., H. Anlauf, H. W. Bitzer, A. Rhodin, C. Schraff, K. Helmert, and K. Stephan, 2012: Global impact studies at the German Weather Service and regional impact studies at the German Weather Service (DWD). Proc. Fifth Workshop on the Impact of Various Observing Systems on Numerical Weather Prediction, Sedona, AZ, WMO, la (5). [Available online at www.wmo.int/pages/prog/www/OSY /Reports/NWP-5_Sedona2012.html.]
Daniels, T. S., W. R. Moninger, and R. D. Mamrosh, 2006: Tropospheric Airborne Meteorological Data Reporting (TAMDAR) overview. 10th Symp. on Integrated Observing and Assimilation Systems for Atmosphere, Oceans, and Land Surface, Atlanta, GA, Amer. Meteor. Soc., 9.1. [Available online at https://ams.confex.com/ams/Annual2006/techprogram/paper_104773.htm.]
Dee, D., and S. Uppala, 2009: Variational bias correction of satellite radiance data in the ERA-Interim reanalysis. Quart. J. Roy. Meteor. Soc., 135, 1830-1841, doi:10.1002/qj.493.
Derber, J. C., and A. Collard, 2011: Current status and future of satellite data assimilation. Proc. 11th Annual Seminar on Data Assimilation for Atmosphere and Ocean, Reading, United Kingdom, EMCWF, 81-93.
Eyre, J., and R. Reid, 2014: Cost-benefit studies of observing systems. Met Office Forecasting Research Tech. Rep. 593, 11 pp.
Fleming, R. J., T. M. Kaneshige, and W. E. McGovern, 1979: The Global Weather Experiment 1. The observational phase through the first special observing period. Bull. Amer. Meteor. Soc., 60, 649-661, doi:10.1175/1520-0477(1979)060<0649:TGWETO>2.0.CO;2.
Gao, F., X. Zhang, N. A. Jacobs, X.-Y. Huang, X. Zhang, and R R Childs, 2012: Estimation of TAMDAR observational error and assimilation experiments. Wea. Forecasting, 27, 856-877, doi:10.1175/WAF-D-11-00120.1.
Hoover, B. T., C. S. Velden, and R. H. Langland, 2014: Adjoint-based analysis of observational impact on tropical cyclone forecasts. 31st Conf. on Hurricanes and Tropical Meteorology, Orlando, FL, Amer. Meteor. Soc., 7A.3. [Available online at http://aurora.aos.wisc.edu/~hoover/7A.3_abstract.pdf.]
Huang, X.-Y., F. Gao, N. A. Jacobs, and H. Wang, 2013: Assimilation of wind speed and direction observations: A new formulation and results from idealized experiments. Tellus, 65A, 19936, doi:10.3402/tellusa.v65i0.19936.
Isaksen, L., 2014: Data assimilation overview--Why aircraft observations are valuable for NWP. Proc. Aircraft Data Workshop, Annapolis, MD, NOAA, 19 pp. [Available online at www.arinc.com/resources/customers/meetings/NOAA/NOAA_Aircraft_Data_Workshop_2014/.]
--, D. Vasiljevic, R. Dee, and S. Healy, 2012: Bias correction of aircraft data implemented in November 2011. ECMWF Newsletter, No. 131, ECMWF, Reading, United Kingdom, 6. [Available online at http://old.ecmwf.int/publications/newsletters/pdf/131.pdf.]
Jacobs, N. A., D. J. Mulally, and A. K. Anderson, 2014: Correction of flux valve-based heading for improvement of aircraft wind observations. J. Atmos. Oceanic Technol., 31, 1733-1747, doi:10.1175/JTECH-D-13-00175.1.
Julian, P. R., and R. Steinberg, 1975: Commercial aircraft as a source of automated meteorological data for GATE and DST. Bull. Amer. Meteor. Soc., 56, 243-251. Kalnay, E., 2003: Atmospheric Modeling, Data Assimilation and Predictability. Cambridge University Press, 364 pp. Kang, J.-S., B.-J. Jung, Y. Jo, and Y. Oto, 2014: Estimation of forecast sensitivity to observations within KIAPS. Sixth EnKF Workshop, Buffalo, NY, KIAPS, 22 pp. [Available online at http://hfip.psu.edu/fuz4/EnKF2014/Day3/Kang_EnKF2014_EFSO_diff_loc.pdf.]
Kelly, G., and J.-N. Thepaut, 2007: Evaluation of the impact of the space component of the Global Observing System through observing system experiments. Proc. Seventh Annual ECMWF Seminar, Reading, United Kingdom, ECMWF, 327-348.
--, T. McNally, J.-N. Thepaut, and M. Szyndel, 2004: OSEs on all main data types in the ECMWF operation system. Proc. Third Workshop on the Impact of Various Observing Systems on Numerical Weather Prediction, Alpbach, Austria, WMO, 63-94. [Available online at www.wmo.int/pages/prog/www/GOS/Alpbach2004/Proceedings.pdf.]
Laroche, S., and R. Sarrazin, 2010: Impact study with observations assimilated over North America and the North Pacific Ocean on the MSC global forecast system. Part II: Sensitivity experiments. Atmos.-Ocean, 48, 26-38, doi:10.3137/A01007.2010.
Lockett, D., 2015: Aircraft-based observations contributing to WMO Global Observing Systems. Aircraft Data Workshop, Annapolis, MD, NOAA, 29 pp. [Available online at www.arinc.com/resources/customers/meetings/ NOAA/NOAA_Aircraft_Data_Workshop_2014/presentations.html.]
Lupu, C., P. Gauthier, and S. Laroche, 2011: Evaluation of the impact of observations on analyses in 3D- and 4D-Var based on information content. Mon. Wea. Rev., 139, 726-737, doi:10.1175/2010MWR3404.1.
--, --, and --, 2012: Assessment of the impact of observations on analyses derived from observing system experiments. Mon. Wea. Rev., 140, 245-257, doi:10.1175/MWR-D-10-05010.1.
Martin, R. C., M. M. Wolfson, and R. G. Hallowell, 1993: MDCRS: Aircraft observations collection and uses. Preprints, Fifth Conf. on Aviation Weather, Vienna, VA, Amer. Meteor. Soc., 317-321.
Moninger, W. R., R. D. Mamrosh, and P. M. Pauley, 2003: Automated meteorological reports from commercial aircraft. Bull. Amer. Meteor. Soc., 84, 203-216, doi:10.1175/BAMS-84-2-203.
--, S. G. Benjamin, B. D. Jamison, T. L. Smith, E. J. Szoke, and T. W. Schlatter, 2007: TAMDAR and its impact on Rapid Update Cycle (RUC) forecasts. 22nd Conf. on Weather Analysis and Forecasting and 18th Conf. on Numerical Weather Prediction, Park City, UT, Amer. Meteor. Soc., 4A.2. [Available online at https://ams.confex.com/ams/pdfpapers/124821.pdf.]
NCAR, 1986: Final report of the Aviation Weather Forecasting Task Force. FAA/NWS Final Rep. DTFA01-84-Z-2051, Research Applications Program, National Center for Atmospheric Research, 87 pp.
Ota, Y., J. C. Derber, E. Kalnay and T. Miyoshi, 2013: Ensemble-based observation impact estimates using the NCEP GFS. Tellus, 65A, 20038, doi:10.3402/tellusa.v65i0.20038.
Pauley, P. M., and Coauthors, 2014: Quality control for aircraft data in numerical weather prediction systems. Proc. Aircraft Data Workshop, Annapolis, MD, NOAA, 38 pp. [Available online at www.arinc.com/resources/customers/meetings/ NOAA/NOAA_Aircraft_Data_Workshop_2014/presentations.html.]
Petersen, R. A., 2004a: Do automated meteorological data reports from commercial aircraft improve forecasts? ICAO J., 59, 4-7. [Available online at www.icao.int/publications/journalsreports/2004/5904.djvu.]
--, 2004b: Impact of AMDARdata on numerical prediction models. Proc. Third Workshop on the Impact of Various Observing Systems on Numerical Weather Prediction, Alpbach, Austria, WMO, 298-307. [Available online at www.wmo.int/pages/prog/www/GOS/Alpbach2004/Proceedings.pdf.]
--, L. Counce, R. Mamrosh and R. Baker, 2017: On the impact and future benefits of AMDAR observations in operational forecasting. Part II: Water vapor observations. Bull Amer. Meteor. Soc., in press.
Schwartz, B. E., S. G. Benjamin, S. M. Green, and M. R. Jardin, 2000: Accuracy of RUC-1 and RUC-2 wind and aircraft trajectory forecasts by comparison with ACARS observations. Wea. Forecasting, 15, 313-326, doi:10.1175/1520-0434(2000)015<0313:AORARW>2.0.CO;2.
Sparkman, J. K., J. Giraytys, and G. J. Smidt, 1981: ASDAR: A FGGE real-time data collection system. Bull. Amer. Meteor. Soc., 62, 394-400.
U.S. Coast Guard, 1962: The LORAN-C system of navigation. U.S. Coast Guard Rep., 135 pp.
WAFC Washington, 2011: The World Area Forecast System (WAFS) Internet File Service (WIFS) users guide, version 4.1. U.S. National Weather Service, 32 pp.
Wendisch, M., and J.-L. Brenguier, Eds., 2013: Airborne Measurements for Environmental Research: Methods and Instruments. Wiley-VCH, 641 pp.
WMO, 2003: Aircraft Meteorological Data Relay (AMDAR) reference manual. World Meteorological Organization, 84 pp.
--, 2012: Final report of the Fifth WMO Workshop on the Impact of Various Observing Systems on Numerical Weather Prediction. WMO Tech. Rep. 2012-1, 25 pp.
--, 2014a: WIGOS WMO Integrated Global Observing System: Requirements for the implementation and operation of an AMDAR programme. WMO Tech. Rep. 2014-2, 31 pp.
--, 2014b: Aircraft Observations Workshop held April 2014 in Annapolis, USA. AMDAR Observing System Newsletter, No. 8, World Meteorological Organization, Geneva, Switzerland. [Available online at https://sites.google.com/a/wmo.int/amdar-newsand-events/.]
Woodman, O. J., 2007: An introduction to inertial navigation. University of Cambridge Computer Laboratory Tech. Rep. 696, 37 pp.
Zhu, Y., J. Derber, R. J. Purser, J. Whiting, and B. R. Ballish, 2015: Variational correction of aircraft temperature bias in the NCEP's GSI analysis system. Mon. Wea. Rev., 143, 3774-3803, doi:10.1175/MWR-D-14-00235.1.
(1) AMDAR observations are obtained from the WMO global AMDAR observing system, which is comprised of those aircraft-based observing systems that derive meteorological data from an aircraft platform according to WMO standards and specifications and make it available on the WMO Global Telecommunications System (GTS). The WMO AMDAR observing system comprises the national and regional member AMDAR systems, which are implemented and operated in collaboration with AMDAR partner commercial airlines.
(2) Recent statistics are available at http://amdar.noaa.gov.
(2a) It must be noted that, because of large uncertainties in the costs for many of the observing systems, readers should be careful not to draw quantitative conclusions about all systems.
AFFILIATIONS: Petersen *--Space Science and Engineering Center, Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin-Madison, Madison, Wisconsin
* Performed under contract to the World Meteorological Organization
CORRESPONDING AUTHOR: Ralph Alvin Petersen, Cooperative Institute for Meteorological Satellite Studies (CIMSS), University of Wisconsin-Madison, Space Science and Engineering Center (SSEC), 1225 West Dayton Street, Madison, WI 53706
The abstract for this article can be found in this issue, following the table of contents.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Aircraft Meteorological Data Relay|
|Author:||Petersen, Ralph Alvin|
|Publication:||Bulletin of the American Meteorological Society|
|Date:||Apr 1, 2016|
|Previous Article:||Impact of aviation on climate: FAA's Aviation Climate Change Research Initiative (ACCRI) Phase II: current and future climate impacts of aviation...|
|Next Article:||The USArray transportable array as a platform for weather observation and research: the data, design, and numerous meteorological applications...|