1111metron 冬天機場容量模型tim myers

1、AIAA 2012-553512th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference and 14th AIAA/ISSM 17 - 19 September 2012, Indianapolis, IndianaWinter Weather Airport Capacity ModelTim Myers1, Mark Andrews2, and Jimmy Krozel, Ph.D. 3Metron Aviation, Inc. Dulles, VA, 20166The impact of win

2、ter weather conditions on airport capacity involves a complex relationship among several factors including precipitation rate and type, temperature, deicing capabilities, snow removal equipment, and airport procedures. This paper describes an airport capacity model which accounts for winter weather

3、conditions. The model is based on statistical relationships observed among weather conditions, historical departure counts, and derived weather-related metrics across three consecutive winter seasons. A set of linear coefficients is defined specific to each airport for relating reported winter weath

4、er conditions with observed deviations in airport departure rates. These coefficients are applied to translate winter weather forecasts into airport capacity estimates accounting for empirical distributions in weather forecast uncertainty. Model predictions are compared against observed airport thro

5、ughput during an actual winter weather event.I. Introductionwide variety of aviation weather impact models have been recently developed; surveys are provided in K10, K11. Some of these models address the en route and terminal domains, while others address airport capacity.AThe majority of existing a

6、irport capacity models address the effects of ceiling, visibility, wind, and terminal-areaconvective weather on airport operations KCM11, KKS11. While these weather conditions are responsible for about two-thirds of all weather-related air traffic delays, around a quarter of weather-related air traf

7、fic delays are due to winter weather including snow, freezing precipitation, and icing KKS11. Moderate to heavy snow has been found to impose the greatest impact on departure and arrival rates of any weather phenomenon more so than thunderstorms and low ceilings KKS11. An overall advancement in airp

8、ort capacity modeling and forecasting techniques could be achieved through the integration of airport capacity impacts resulting from winter weather conditions.Two fundamental challenges were addressed in developing a Winter Weather Airport Capacity Model (WWACM): Airport capacity cannot be directly

9、 measuredSpecifically, reductions in capacity due to winter weather conditions had to be inferred relative to some notion of a “baseline” capacity.No data sources exist providing hourly observations of snow on the runwayThe amount of snow, or more accurately the water equivalent of snow, on the runw

10、ay surface had to be estimated based on assumptions related to snow melt off and snow removal operations at each airport.Interviews with airport operations Subject Matter Experts (SMEs) revealed that the effects of winter weatherdiffer by airport based on several complex factors including airport la

11、yout, runway length, snow removal anddeicing capabilities, and the efficiency with which these capabilities are utilized. It was also recommended that a single winter-weather-to-capacity relationship would not be applicable to all airports. This is because individual airports might respond different

12、ly under similar weather conditions.The option of developing a theoretical capacity model was considered. The design for such a model would have centered on factors such as braking distance and deicing procedures as inputs to simulating the maximum possible arrival and departure rates for each possi

13、ble runway configuration under all winter weather conditions. The difficulty in developing such a theoretical model lies in configuring the model to be applicable to each individual airport. As indicated by SME inputs, each airport utilizes snow removal and deicing capabilities in different ways. Ac

14、curately modeling these usage characteristics would have been overly complex for this initial modeling exercise.1 Senior Analyst, Concept Engineering Group, 45300 Catalina Ct, Suite 101, AIAA Senior Member2 Principal Weather Subject Matter Expert, Advanced Research and Engineering Dept., 45300 Catal

15、ina Ct, Suite 1013 Sr. Engineer, Advanced Research and Engineering Dept., 45300 Catalina Ct, Suite 101, AIAA Associate FellowCopyright 2012 by the American Institute of Aeronautics and Astronautics, Inc. The U.S. Government has a royalty-free license to exercise all rights under the copyright claime

16、d hereinDownloaded by Nanjing University of Aero & Astro on April 1, 2014 | | DOI: 10.2514/6.2012-5535The remaining sections of this paper define the methodology and results of the WWACM. Conclusions and references are provided at the end of the paper.II. MethodologyWWACM was deve

17、loped based on the empirical relationship between historical winter weather conditions and the observed airport capacity. Airport capacity was inferred from observed reductions in the hourly departure rate relative to a baseline departure rate accounting for time of day and day of week. This Relativ

18、e Departure Rate (RDR) is described further in Section III.B. Historical winter weather conditions were obtained from archived Meteorological Aerodrome Reports (METARs). The severity of winter weather conditions was based on estimates of the water content of snow on the runway, accounting for temper

19、ature-based snow melt off rates that are presented in Section III.D. To account for varying snow removal capabilities among airports, nine snow removal rates were modeled to determine the best correlation between observed capacity and winter weather conditions (see Section III.E). Correlations were

20、analyzed between actual departure rates and a combination of observed and derived metrics across three consecutive winter seasons: 20082009, 20092010, and 20102011. The winter season was defined by periods outside of daylight savings.SME interviews suggested that winter weather impacts, specifically

21、 the time requirements associated with deicing procedures, result in airport departure rates being more severely constrained by winter weather than arrival rates. WWACM was therefore initially designed to correlate reductions in observed departure rates with reported winter weather conditions at eac

22、h airport. Follow-on analysis is planned to better understand the impact of winter weather on airport arrival rates and to develop a corresponding capacity model for arrivals.Weather-impacted airport capacity cannot be measured directly due to interactions with other factors (e.g., congested overhea

23、d streams, congested taxiways) that may haveTable 1: Airports included in the WWACM study.contributed to observed departure rates. Therefore,wequantify reductions in airport capacity due to weather by comparing the observed departure rate during a weather event at a specific time of day and day of w

24、eek with what is normalfor that same time. This empirically-based methodology was applied to individual airports across the three aforementioned winter seasons to quantify how each airport responds to varying severities of winter weather. Table 1 lists the 32 airports selected for the study based on

25、 the high frequency of winter weather conditions occurring at those airports.The output of the WWACM correlation study is a set of airport-specific linear coefficients which relate airport capacity with winter weather conditions. These airport-specific coefficients are then applied in Section VI to

26、predict airport capacity accounting for winter weather forecasts.III.ParametersNext, we describe the key parameters used in correlating airport capacity subject with winter weather conditions.A. Baseline Departure RateFigure 1 illustrates the process by which a departure rate baseline was defined fr

27、om hourly departure counts (blue dots). The hourly counts themselves exhibit large variances from hour to hour. Attributing variations in the departure rate to winter weather conditions, as opposed to random fluctuations, would have been difficult in the absence of some form of trending. A Gaussian

28、filter with standard deviation of three hours was applied to the hourly departure counts in order to define a Trendline Departure Rate (TDR) in Figure 1. The TDR follows a daily pattern which also cycles through weekly peaks during weekdays and reduced peaks during weekends.Downloaded by Nanjing Uni

29、versity of Aero & Astro on April 1, 2014 | | DOI: 10.2514/6.2012-5535IdentifierIdentifierIdentifierIdentifierBOSDTWMCIPITBUFEWRMDWPVDBWIGYYMHTRFDCLEHPNMKESDFCVGINDMSPSLCDAYISPOMASTLDCAJFKORDSWFDENLGAPHLTEBObserved hourly departure countTrendline Departure Rate (TDR) using sigma =

30、3 hoursBaseline Departure Rate (BDR) = Average of TDR by time of day and day of weekDepartures Per HourMidnight ZuluDeparture rate is lower than normal for this hour on a WednesdayFigure 1: Baseline departure rate.The Baseline Departure Rate (BDR) was defined by computing the average TDR specific to

31、 each hour of day and day of week. There were approximately 15 weeks in each winter season. For example, the BDR at 11Z on Thursdays was computed as the average TDR occurring at 11Z across all 15 Thursdays throughout the winter season. This baseline is shown in light blue in Figure 1 and provides a

32、reference point for the typical departure rate specific to that hour of day and day of week.B. Relative Departure RateA Relative Departure Rate (RDR) was computed as the ratio of the TDR and BDR as shown in Figure 2. The RDR has a value of 1.00 whenever the hourly trendline matches the baseline, whi

33、ch indicates departure rates at normal levels. Periods during which the RDR drops below 1.00 may indicate reduced airport capacity. Figure 2 shows that during the later half of Wednesday, January 12, 2011, for example, the observed RDR dropped from 1.00 to almost zero. The RDR gradually returned to

34、1.00 the following day, indicating a return to normal levels.Departures Per HourRelative Departure RateDeparture rateisclose to normalDeparture rate dropsDeparture rate below normal levels returns tonormal Figure 2: Relative departure rate.The general approach for modeling winter weather impacts inv

35、olved correlating reductions in RDR with winter weather conditions. A refinement made during the modeling was to only consider times of day when peak demand normally occurred. This additional filter was implemented by including RDR data only during periods when the BDR exceeded the median BDR for ea

36、ch particular airport. For the case of Boston Logan International Airport (BOS) shown in Figure 3, the median baseline departure rate was 19 flights per hour for the 20102011 winter season. RDR values were later correlated with winter weather conditions limited to times when the BDR exceeded 19 flig

37、hts per hour as shown by red dots in the lower plot of Figure 3.Downloaded by Nanjing University of Aero & Astro on April 1, 2014 | | DOI: 10.2514/6.2012-5535Baseline Departure Rate (BDR) Median Baseline Departure Rate (BDR) = 19 flights per hourDeparturesPer Hour Relative Departu

38、re RateRDR for times when BDR median BDRFigure 3: Relative departure rate limited to peak times.C. Estimated Precipitation Rate of SnowSnow density, or the water equivalent of snow, varies greatly based on cloud temperature and physical processes during snow crystal formation. SMEs indicated that it

39、 is the water equivalent of snow as opposed to the actual snow depth that drives the impact on airport operations. For example, dry fluffy snow is easier to remove from runways than dense, wet snow of the same depth. The dense snow has the higher water content. Therefore, rather than attempting to c

40、orrelate airport capacity with snow depth, it was more appropriate to model the water content of snow as the independent variable.METAR data were obtained from an online archive WU11 to provide information about observed winter conditions including snow, ice, freezing rain, and temperature at each o

41、f the modeled airports. METAR observations are typically generated hourly, with more frequent updates occurring during periods of significant weather activity. METAR reports provide two fields which can indicate the water equivalent of snowfall:Precipitation RateThis numerical field is designed to p

42、rovide the amount of precipitation that has occurred since the previous hourly METAR report. Precipitation rates are reported in inches of water.ConditionsThis text-based field provides a categorical description of reported conditions. Possible values related to winter weather conditions include “Li

43、ght Snow,” “Snow,” and “Heavy Snow.”A METARs Precipitation Rate and Conditions do not always agree.Figure 4 shows a METAR report from BOS during New Years Eve 2008. In this example, METAR conditionsread “Heavy Snow” whereas the Precipitation Rate ranges from 0 to 0.02 inches of water content per hou

44、r.Local Time10:34 AM10:54 AM11:10 AM11:54 AM12:03 PM12:11 PM12:27 PM12:54 PM1:54 PM2:30 PM2:51 PM2:54 PM3:38 PM3:54 PM4:04 PM4:40 PM4:54 PMTemp232324.82321.92119.419.419.919.419.919.421.221Vis.0.50.8122Precip. Conditions0 Light Snow0 Light Snow0 Light

45、SnowFigure 4: Contrast between METAR Conditions and Precipitation.To address thisapparent discrepancy, the entire correlation analysis was conducted twice. The first passinvolved the use of Precipitation Rate as an indicator of winter weather conditions. A second pass was made usingDownloaded by Nan

46、jing University of Aero & Astro on April 1, 2014 | | DOI: 10.2514/6.2012-5535Precipitation between 0 and 0.02 does not match Conditions of “Heavy Snow”Using Conditions provided superior linear correlations0.010Snow Heavy Snow0.02Heavy Snow0.010.010.010.020.020.010.010.010.010.010H

47、eavy SnowHeavy Snow Snow Heavy Snow Heavy Snow SnowSnow Snow Snow Snow SnowConditions to indicate the rate of snowfall. In the end, using the METAR Conditions field yielded a stronger correlation between winter weather conditions and capacity degradations inferred by deviations in the RDR. Precipita

48、tion rates were estimated for each category of Conditions simply as the average of all reported Precipitation values associated with each category of Conditions.D. Estimated Snow Melt-Off RateThe amount of snow remaining on the ground accounting for melt-off is an important input to WWACM. Archived

49、snow depths are available on a daily basis N11. However, correlating snow depth with airport capacity required a higher temporal fidelity of snow depth reporting. Snow depths were estimated at the time of each METAR report, roughly once per hour, based on an estimated accumulation derived from METAR

50、 Conditions and a simple snow melt-off model based on METAR-reported temperature.The snow depth model used in the current research assumed a melt-off rate R of 0.01 inches of water per hour per degree Fahrenheit (F) above freezing:R = 0.01 max(T 32, 0)(1)For example, a snow pack in an ambient temper

51、ature of T = 42F would melt at a rate of 0.1 inches of water per hour.4 Note that the units for the melt-off rate R are inches of water as opposed to inches of snow. The actual rate at which snow melts would depend on the snows density. Snow having a density of 10% relative to water, for example, wo

52、uld recede at rate of one inch per hour at 42F based on this simpl e model.For all of its simplicity, the modeled snow melt-off rate compared reasonably well with reported snow depths as shown in Figure 5. The upper plot shows the hourly estimated water content of snow accounting for precipitation a

53、nd melt-off. The lower plot gives the daily reported snow depths. There appears to be a good match between occurrences of complete melt-off and peak accumulation between the estimated and reported snow depths.Hourly estimated snow on ground based on precipitation rates and temperatureSnow Depth (est

54、imated) Peak accumulationDaily reported snow on groundSnow Depth (reported)Melt offMelt off2/22/32/42/52/62/72/82/9 2/10 2/11 2/12 2/13 2/14 2/15 2/16 2/17 2/18 2/19 2/20Figure 5: Comparison of estimated and reported snow depths (February 2010).E. Estimated Snow Removal RateIn addition to natural sn

55、ow melt-off, each airport may employ snow removal equipment such as snow plows, blowers, and brushes to accelerate the resumption of normal operations following winter weather events. The Snow Removal Rate (SRR) at each airport, measured in inches of water per hour, is a key factor in determining ai

56、rport capacity.The Water content of Snow on Runway (WSR) is an estimate of the amount of water contained in snow on the airport movement surfaces accounting for the combined effects of precipitation, melt-off and SRR. WSR defines the independent variable in the correlation between winter weather and

57、 airport capacity in this study.Snow removal rates vary by airport. Therefore, rather than assuming a single SSR for all airports, multiple correlations were computed to identify the best snow removal rate coefficient for each airport. Figure 6 shows the effect of various SSRs on WSR alongside departure rates, relative departure rate, and temperature on the same timeline. The WSR estimates were calculated based on the aforementioned snow melt-off rate (Eq. (1) combined with nine SRR values ranging from 0.00 to 0.04 inches of water per hour in 0.005 increments. WSR