• Appleman, H. S., 1953: The formation of exhaust condensation trails by jet aircraft. Bull. Amer. Meteor. Soc., 34 , 1420.

  • Bromwich, D. H., J. J. Cassano, T. Klein, G. Heinemann, K. M. Hines, K. Steffen, and J. E. Box, 2001: Mesoscale modeling of katabatic winds over Greenland with the Polar MM5. Mon. Wea. Rev., 129 , 22902309.

    • Search Google Scholar
    • Export Citation
  • Busen, R., and U. Schumann, 1995: Visible contrail formation from fuels with different sulfur content. Geophys. Res. Lett., 22 , 13571360.

    • Search Google Scholar
    • Export Citation
  • Cassano, J. J., J. E. Box, D. H. Bromwich, L. Li, and K. Steffen, 2001: Evaluation of Polar MM5 simulations of Greenland’s atmospheric circulation. J. Geophys. Res., 106 , 3386733890.

    • Search Google Scholar
    • Export Citation
  • Downie, C. S., and B. A. Silverman, 1957: Jet aircraft condensation trails. Handbook of Geophysics for Air Force Designers, Geophysics Research Directorate, Air Force Cambridge Research Laboratories, Air Research and Development Command, United States Air Forces, 19-1–19-9.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1993: A nonhydrostatic version of the Penn State/NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev., 121 , 14931513.

    • Search Google Scholar
    • Export Citation
  • Elliott, W. P., and D. J. Gaffen, 1991: On the utility of radiosonde humidity archives for climate studies. Bull. Amer. Meteor. Soc., 72 , 15071520.

    • Search Google Scholar
    • Export Citation
  • Goff, J. A., and S. Gratch, 1946: Low pressure properties of water from −160° to 212°F. Trans. Amer. Soc. Heat. Vent. Eng., 52 , 95122.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., J. Dudhia, and D. R. Stauffer, 1994: A description of the fifth-generation Penn State/NCAR mesoscale model (MM5). NCAR Tech. Note, NCAR/TN-398+STR, National Center for Atmospheric Research, Boulder, CO, 122 pp.

  • Hack, J. J., B. A. Boville, B. P. Briegleb, J. T. Kiehl, P. J. Rasch, and D. L. Williamson, 1993: Description of the NCAR Community Climate Model (CCM2). NCAR Tech. Note, NCAR/TN-382+STR, National Center for Atmospheric Research, Boulder, CO, 138 pp.

  • Jensen, E. J., and Coauthors, 1998: Environmental conditions required for contrail formation and persistence. J. Geophys. Res., 103 , D4,. 39293936.

    • Search Google Scholar
    • Export Citation
  • Meerkötter, R., U. Schumann, D. R. Doelling, P. Minnis, T. Nakajima, and Y. Tsushima, 1999: Radiative forcing by contrails. Ann. Geophys., 17 , 10701084.

    • Search Google Scholar
    • Export Citation
  • Miloshevich, L. M., H. Voemel, A. Paukkunen, A. J. Heymsfield, and S. J. Oltmans, 2001: Characterization and correction of relative humidity measurements from Vaisala RS80-A radiosondes at cold temperatures. J. Atmos. Oceanic Technol., 18 , 135156.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., U. Schumann, D. R. Doelling, K. M. Gierens, and D. W. Fahey, 1999: Global distribution of contrail radiative forcing. Geophys. Res. Lett., 26 , 18531856.

    • Search Google Scholar
    • Export Citation
  • Peters, J. L., 1993: New techniques for contrail forecasting. Air Weather Service Tech. Rep. AWS/TR-93/001, 26 pp. [Available from HQ Air Weather Service, Scott AFB, IL 62225.].

  • Pilie, R. J., and J. E. Jiusto, 1958: A laboratory study of contrails. J. Meteor., 15 , 149163.

  • Pratt, R. W., 1985: Review of radiosonde humidity and temperature errors. J. Atmos. Oceanic Technol., 2 , 404407.

  • Reisner, J., R. M. Rasmussen, and R. T. Bruintjes, 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 mesoscale model. Quart. J. Roy. Meteor. Soc., 124B , 10711107.

    • Search Google Scholar
    • Export Citation
  • Rind, D., P. Lonergan, and K. Shah, 2000: Modeled impact of cirrus cloud increases along aircraft flight paths. J. Geophys. Res., 105 , D15,. 1992719940.

    • Search Google Scholar
    • Export Citation
  • Sausen, R., K. Gierens, M. Ponater, and U. Schumann, 1998: A diagnostic study of the global distribution of contrails. Part I: Present day climate. Theor. Appl. Climatol., 61 , 127141.

    • Search Google Scholar
    • Export Citation
  • Schmidt, E., 1941: Die Entstehung von Eisnebel aus den Auspuffgasen von Flugmotoren. Schr. Dtsch. Akad. Luftfahrtforsch., 44 , 115.

  • Schrader, M. L., 1997: Calculations of aircraft contrail formation critical temperatures. J. Appl. Meteor., 36 , 17251729.

  • Schumann, U., 1996: On conditions for contrail formation from aircraft exhausts. Meteor. Z., 5 , 423.

  • Shull, J. D., 1998: A validation study of the Air Force Weather Agency (AFWA) JETRAX contrail forecast algorithm. M.S. thesis, Dept. of Meteorology, Air Force Institute of Technology, 120 pp. [Available from Airforce Institute of Technology, Wright Patterson AFB, OH 45433.].

  • Stuefer, M., and G. Wendler, 2004: Contrail studies and forecasts in the subarctic atmosphere above Fairbanks, Alaska. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., CD-ROM, P8.13.

  • Tilley, J. S., A. H. Lynch, W. L. Chapman, W. Wu, and G. Weller, 1999: On the coupling of land surface packages within a regional climate model of the Arctic. Proceedings of a Workshop on Arctic Regional Climate Models, Tech. Doc. 981, H. Cattle, Ed., World Meteorological Organization, C12.1–12.5.

    • Search Google Scholar
    • Export Citation
  • Wendler, G., and M. Stuefer, 2002: Improved contrail forecasting techniques for the subarctic setting of Fairbanks, Alaska. Geophysical Institute of the University of Alaska Rep. UAG R-329, 35 pp.

  • Zhang, J., and X. Zhang, 2004: Modeling study of Arctic storm with the coupled MM5-Sea Ice-Ocean model. Preprints, Fifth WRF/14th MM5 User’s Workshop, Boulder, CO, NCAR. [Available online at http://www.mmm.ucar.edu/mm5/workshop/workshop-papers_ws04.html.].

  • View in gallery

    All-sky digital camera image displaying a contrail formed by a Boeing 747-400 (Northwest Airlines flight 69 from Detroit to Osaka/Japan, 31 Jul 2002).

  • View in gallery

    Temperatures vs air pressure shown at the altitudes of aircraft observations. Different symbols refer to the persistence of contrail observations.

  • View in gallery

    Errors in critical temperatures derived due to a dry bias of relative humidity of 10% for different ambient relative humidity values. Critical temperatures were calculated with p = 250 hPa and CF = 0.036 g kg−1 K−1.

  • View in gallery

    Mean altitudes of contrail layers from reference sounding data vs the forecasted mean altitudes. The (left) 0- and (right) 72-h forecast comparisons are shown.

  • View in gallery

    Mean absolute deviation of mean altitudes of contrail layers derived from MM5 in reference to altitudes derived from sounding measurements. Absolute deviations are shown for Arctic and AFWA MM5 for 0–72-h forecasts.

  • View in gallery

    Same as in Fig. 5, but for mean absolute deviation of contrail layer thickness.

  • View in gallery

    Example of contrail layer thickness derived from (left) reference sounding and (right) AFWA-MM5 forecasts.

  • View in gallery

    The MO indicating the agreement of modeled layers and layers from radiosonde measurements. Arctic (solid lines) and AFWA (dashed lines) MM5 overlap parameters are indicated for the different model runs.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 170 79 1
PDF Downloads 140 63 1

MM5 Contrail Forecasting in Alaska

View More View Less
  • 1 Geophysical Institute, University of Alaska Fairbanks, Fairbanks, Alaska
Full access

Abstract

The fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) is being used for forecasting the atmospheric layers of aircraft condensation trail (contrail) formation. Contrail forecasts are based on a conventional algorithm describing the adiabatic mixing of aircraft exhaust with environmental air. Algorithm input data are MM5-forecasted temperature and humidity values at defined pressure or sigma levels, and an aircraft-relevant contrail factor that is derived statistically from a contrail observation database.

For comparison purposes a mean overlap (MO), which is a parameter quantifying the overlap between forecasted contrail layers and contrail layers derived from radiosonde measurements, is introduced. Mean overlap values are used to test for the altitude and thickness of forecasted contrail layers. Contrail layers from Arctic MM5 and Air Force Weather Agency (AFWA) MM5 models agree well with contrail layers derived from corresponding radiosonde measurements for certain forecast periods; a steady decrease of the MO shows a decrease of contrail forecast accuracy with the increasing forecast period. Mean overlaps around 82% indicate reasonable results for the 48-h forecasts. Verification of MM5 with actual contrail observations shows a slightly better performance of Arctic MM5. A possible dry bias might occur in humidity measurements at low temperature levels due to temperature-dependence errors of the humidity sensor polymer, which might also affect forecasts of humidity of the upper troposphere or lower stratosphere. Despite this fact, this contrail verification study shows hit rates higher than 82% within forecast periods up to 36 h using Arctic MM5.

Corresponding author address: Dr. Martin Stuefer, Geophysical Institute, University of Alaska Fairbanks, 903 Koyukuk Dr., P.O. Box 757320, Fairbanks, AK 99775-7320. Email: Stuefer@gi.alaska.edu

Abstract

The fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) is being used for forecasting the atmospheric layers of aircraft condensation trail (contrail) formation. Contrail forecasts are based on a conventional algorithm describing the adiabatic mixing of aircraft exhaust with environmental air. Algorithm input data are MM5-forecasted temperature and humidity values at defined pressure or sigma levels, and an aircraft-relevant contrail factor that is derived statistically from a contrail observation database.

For comparison purposes a mean overlap (MO), which is a parameter quantifying the overlap between forecasted contrail layers and contrail layers derived from radiosonde measurements, is introduced. Mean overlap values are used to test for the altitude and thickness of forecasted contrail layers. Contrail layers from Arctic MM5 and Air Force Weather Agency (AFWA) MM5 models agree well with contrail layers derived from corresponding radiosonde measurements for certain forecast periods; a steady decrease of the MO shows a decrease of contrail forecast accuracy with the increasing forecast period. Mean overlaps around 82% indicate reasonable results for the 48-h forecasts. Verification of MM5 with actual contrail observations shows a slightly better performance of Arctic MM5. A possible dry bias might occur in humidity measurements at low temperature levels due to temperature-dependence errors of the humidity sensor polymer, which might also affect forecasts of humidity of the upper troposphere or lower stratosphere. Despite this fact, this contrail verification study shows hit rates higher than 82% within forecast periods up to 36 h using Arctic MM5.

Corresponding author address: Dr. Martin Stuefer, Geophysical Institute, University of Alaska Fairbanks, 903 Koyukuk Dr., P.O. Box 757320, Fairbanks, AK 99775-7320. Email: Stuefer@gi.alaska.edu

1. Introduction

Aircraft contrails are mostly formed in the upper troposphere and in some cases may remain for several hours. Persistent contrails that have a long narrow cloudlike appearance are easily identified shortly after formation. A distinction between contrails and naturally formed cirrus clouds may be difficult after a while because of wind drift and related spreading. Sausen et al. (1998) reported a comparatively significant contribution of contrails to the total cirrus cloud coverage. Persistent contrails influence the radiation balance of the atmosphere and hence have a possible effect on surface climate (Meerkötter et al. 1999; Minnis et al. 1999). Aircraft contrails are also of importance for the military, which in times of expensive stealth development is interested in nonformation of contrails for preservation of an aircraft’s “low observability.”

The intent of this paper is to verify contrail forecasts that can be useful especially to military pilots for flight planning purposes. The accuracy of the model forecasts is discussed with operational contrail observations and with contrail predictions derived from radiosonde measurements directly.

Dichotomous contrail forecasting applications were developed by specifying critical temperatures that serve as threshold to separate if a contrail will be formed or not. Schmidt (1941) and Appleman (1953) described originally the thermodynamics of an air parcel that is influenced by the entrainment of moist and warm exhaust gases. Schumann (1996), Schrader (1997), and Jensen et al. (1998) published more recent reviews and explanations of the physics involved in contrail formation processes. We calculated the contrail layers by comparing critical temperatures for contrail formation with forecasted temperatures. In the following section we give a summary of the physics and the derivation of critical temperatures for contrail formation.

2. Critical temperature theory

Appleman (1953), Schrader (1997), Schumann (1996), and others published similar equations showing the contrail formation theory. Efforts are ongoing for implementation of the derived algorithm in operational contrail forecast models; the U.S Air Force Weather Agency (AFWA) uses the algorithm in their contrail prediction technique called JETRAX. JETRAX has been developed for military air operation support; it incorporates a fixed humidity parameterization scheme for atmospheric pressure levels below 300 hPa. For illustration of the iterative process needed in generating solution we give a short review of equations for the explicit calculation of critical temperatures for the formation of contrails.

Assuming an isobaric mixing process, the temperature increase, dT, of the affected ambient air due to the combustion of one mass unit of fuel is calculated according to
i1520-0493-133-12-3517-e1
The parameter Q denotes the liberated heat by the combustion of one mass unit of fuel, and k is the ratio of exhaust gas to the mass of fuel (k ≈12 kg kg−1). The temperature difference is large at the beginning of the entrainment of exhaust gases; Jensen et al. (1998) showed typical lapse rates of temperature differences with increasing exhaust gas dilution derived from model studies. The mass ratio N accounts for the amount of entrained environmental air to the exhaust gas; thus the product k × N characterizes the mass of environmental air that is affected by the combustion of one mass unit of fuel. The value of N depends strongly on the distance of the considered mixing parcel behind the aircraft, the combustion efficiency of the engines, and the density and stability of the atmosphere controlling the spreading of the exhaust gases. The mass ratio ranges from 0 immediately behind the aircraft engine to infinity. The temperature increase dT further depends on the specific heat of air (cp = 1004 J kg−1 K−1). Estimating the emission index for water vapor as the amount of water vapor produced by the combustion of 1 mass unit of fuel, kH2O (≈1.4 kg kg−1), the increase of mixing ratio drf (g kg−1) in the considered air parcel is derived:
i1520-0493-133-12-3517-e2
The combination of Eqs. (1) and (2) results in a relation independent of the state of mixing (N):
i1520-0493-133-12-3517-e3
The ratio drf/dT is called the contrail factor (CF), which is a characteristic of aircraft engine combustion and thus is not constant during the different states of a flight. Busen and Schumann (1995) discuss the dependence of combustion to fuel, aircraft engine, and flight parameters. Contrail factors were estimated for certain aircraft and engine types; these contrail factors might be representative for an airplane during cruise and level flight.

Appleman derived an original value of 0.0336 g kg−1 K−1. Schrader (1997) compiled values of typical contrail factors ranging from 0.0300 to 0.039 g kg−1 K−1. Busen and Schumann (1995) calculated a minimum contrail factor of 0.028 g kg−1 K−1. The maximum contrail factor was published with 0.049 g kg−1 K−1 (Peters 1993).

Jensen et al. (1998) report that for ambient air temperatures between the critical temperatures for liquid water saturation and ice saturation no visible contrails were found during the Subsonic Aircraft: Contrail and Cloud Effects Special Study (SUCCESS). For visible contrails to form, supersaturation with respect to water was observed and a phase change from water droplets to ice crystals might occur immediately. We use the saturation of water for critical temperature calculations.

The relation of the saturation-mixing ratio to temperature change (drs/dT) is compared with the contrail factor to derive a threshold temperature for contrails formation, which is further denoted as the critical temperature for a saturated environmental atmosphere, Tcrit,100. With p as the air pressure, and eS as the saturation vapor pressure, the saturation-mixing ratio is calculated. For eSp, a good approximation is
i1520-0493-133-12-3517-e4
Assuming isobaric processes, Eq. (4) yields the change of the saturation-mixing ratio with temperature as function of the pressure p and the air temperature T:
i1520-0493-133-12-3517-e5
Downie and Silverman (1957) and Pilie and Jiusto (1958) used the approach of Goff and Gratch (1946) in order to specify des/dT (T). Goff and Gratch introduced a numerical approximation of the Clausius–Clapeyron equation. The threshold temperature Tcrit,100 is defined for
i1520-0493-133-12-3517-e6
The critical temperatures Tcrit,h for nonsaturated conditions (relative humidity h < 100%) are derived from Eq. (6) according to
i1520-0493-133-12-3517-e7

Equations (6) and (7) are solved iteratively in order to obtain the critical temperatures Tcrit,100 (p) and Tcrit,h (p, h) for a previously estimated CF. The determination of Tcrit,h requires accurate humidity measurements. An example programming code is given by Wendler and Stuefer (2002).

3. Source data

a. Observation

We used contrail observations conducted at the Geophysical Institute of the University of Alaska Fairbanks (Wendler and Stuefer 2002). Observers carried out visual contrail identification overhead Fairbanks. In addition a digital camera photographed the sky every minute through a fish-eye lens pointed to the zenith (Fig. 1). The aircraft flight information, including airplane type, altitude, origin, destination, and speed was available directly to the observers in near–real time using an online Federal Aviation Administration (FAA) data display. In this way, aircrafts approaching Fairbanks were identified on the display and an observer was alerted to check for the presence of a contrail. Observations were carried out continuously from June 2001 to March 2004 during daylight hours. The contrail database was compared with atmospheric measurements from radiosondes, launched by the National Weather Service (NWS) twice daily (0000 and 1200 UTC) from Fairbanks Airport. From our observation database we selected 377 observations, which were within a time range of 2 h before or after the respective sounding. The chosen observations were divided into three groups; 55 negative contrail events (no contrails), 168 transient (threshold) contrails with lifetimes of a few seconds to 1 min, and 154 contrails lasting longer than 1 min. Due to drifting and spreading effects, contrails sometimes could not be identified and distinguished from surrounding natural cirrus clouds, therefore the lifetimes in our database represent a lower boundary value. The longest clearly identified contrail lasted 6 h.

Most frequently observed aircraft types were Boeing 747-200 and Boeing 747-400, together composing 65% of all observations. Cruising speeds were between 400 (741) and 550 kt (1019 km h−1). The cruising altitudes varied from 7223 (23 700 ft) to 11277 m (37 000 ft).

b. Atmospheric sounding

Atmospheric parameters were obtained from the radiosonde ascents. Linear interpolation between two respective data points was used for obtaining temperature and dewpoint temperature at flight level and the pressure was interpolated with the barometric height equation. Figure 2 shows the aircraft ambient temperatures and pressures for the three contrail observation classes. No-contrail cases were observed at a temperature range from −56° to −36°C, threshold contrails occurred from −65° to −39°C, and slightly colder temperatures from −67° to −44°C were measured for contrails persisting longer than 1 min. Typical pressure values for all observations were between 330 and 200 hPa.

Inaccuracies in critical temperature calculations might occur as the humidity reported by radiosondes at low temperatures is subject to sources of error (Pratt 1985; Elliott and Gaffen 1991). Miloshevich et al. (2001) investigated a strong bias toward dry humidity values measured with Vaisala RS80-A radiosondes at cold temperatures. Correction factors for the relative humidity measurements of 1.3 at temperature of −35°C increasing to 2.4° at −70°C were suggested for this particular radiosonde instrument. In Fairbanks, the NWS used Vaisala RS80-57H radiosondes for upper-air measurements. As possible correction factors depend strongly on sensor type (Miloshevich et al. 2001) and as no simultaneous hygrometer measurements to Vaisala RS80-57H measurements were available for comparison purpose, we used the radiosonde humidity measurements without correction. Schrader (1997) reported a little sensitivity of critical temperatures to ambient relative humidity values between 0% and 70%. Critical temperature changes due to a 10% error in relative humidity are shown in Fig. 3; the changes are similar for different pressure levels and contrail factors. Errors in critical temperatures increase with increasing relative humidity; more than 1°C lower critical temperatures results from 10% humidity errors at ambient relative humidity values above 80%.

c. Forecast models

For contrail layer forecasts and forecast verification we used two different versions of the nonhydrostatic, fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5; Dudhia 1993; Grell et al. 1994). A coupled modeling system, referred to as the Arctic MM5 was developed by the Mesoscale Modeling and Applications Group at the Geophysical Institute of the University of Alaska, Fairbanks (Tilley et al. 1999; Zhang and Zhang 2004). The Arctic MM5 treats radiative processes and microphysical cloud and precipitation physics according to the Polar MM5 (Bromwich et al. 2001; Cassano et al. 2001). Hack et al. (1993) described the longwave and shortwave radiation scheme, which is also used in the NCAR Community Climate Model, version 2 (CCM2).

Experimental Arctic MM5 runs were executed 3 times daily by nudging meteorological fields with analysis of actual observations. The initialization times were at 0600, 1200, and 1800 UTC, the forecast range was from 0 to 48 h. Most forecast products were saved for model verification purpose; forecasts from mid-December 2002 to the end of February 2004 were available from the Arctic MM5 for contrail layer verification. Due to the designated atmospheric sounding times at 0000 and 1200 UTC, near-real-time (0 h) forecast verification with radiosonde data was restricted to Arctic MM5 initialized at 1200 UTC. The different model initialization times and forecast ranges had to be considered for verification of MM5 forecasts with atmospheric soundings and actual observations of contrails.

Starting in June 2003, we received, in addition to Arctic MM5, AFWA MM5 initialized at 0000 and 1200 UTC. (For AFWA MM5 characteristics we refer the reader to http://meted.ucar.edu/nwp/pcu2/afintro.htm.) AFWA MM5 produced 0–72-h forecasts with time intervals of 3 h. The forecasts covering the Alaska theater (region) were obtained after actual model runs via the Applied Physics Laboratory of Johns Hopkins University. Due to the amount of data transfer (maximum ∼253 MB, compressed *.tar format), gaps in forecast data acquisition occurred.

Both the AFWA MM5 and the Arctic MM5 use a 45-km horizontal grid. The cloud and precipitation physics are equivalent in both models and follow the Reisner scheme (Reisner et al. 1998). The vertical resolution of AFWA MM5 consists of 24 levels, forecasts are performed every 50 hPa for pressure levels above 850 hPa. Arctic MM5 is characterized by a higher vertical resolution with 41 terrain-following σ levels (Zhang and Zhang 2004). Contrail layers calculated from MM5 forecasts for the nearest model grid point to Fairbanks were verified with radiosonde-generated contrail layers, which we refer to as reference layers in the following section.

4. Verification

Critical temperatures were calculated with atmospheric data derived from Fairbanks radiosonde ascents and from MM5 forecasts from Fairbanks. Contrail formation was predicted for those layers, where critical temperatures exceeded the measured or forecasted temperatures.

a. Verification of contrail observations with predicted contrail occurrence based on radiosonde measurements

To verify the algorithm with radiosonde data we used various contrail factors ranging from 0.025 to 0.05 g kg−1 K−1. The results of dichotomous contrail forecasts for selected contrail factors are shown in Table 1. The contingency of forecasted hits (x), misses (y, not forecasted but observed), the number of cases when a contrail was forecasted but not observed (z, false alarm) and the correct negative forecasts (w) were calculated for no contrails, threshold contrails (0 < lifetime 1 min) and contrails persisting longer than 1 min. In addition the probability of detection [POD = x/(x + y)], the probability of detection for no-contrail events [(PODnil = w/(w + z)], the false alarm rate [FAR = z/(x + z)], the hit rate [HR = (x + w)/(x + y + w + z)] and a mean probability of detection [PODm = (POD + PODnil)/2] are given in Table 1. A steady increase of the hit rate (HR) combined with low false alarm rates (FARs) were obtained due to the larger number of contrail incidence compared to no-contrails. The probability of detection showed major differences between threshold contrails (PODth) and contrails lasting longer than 1 min (POD>I). The mean value PODm is considered appropriate for the derivation of a threshold between contrails and no contrails. The radiosonde prediction provided the best results (PODm = 90%) with a contrail factor CF = 0.036 g kg−1 K−1. This contrail factor represents an average value for different aircraft types with different combustion characteristics for cruising commercial flights over Fairbanks. Contrail layers were defined as the range with critical temperatures warmer than ambient temperatures. PODnil values decrease with increasing contrail factors due to a thickening of the forecasted contrail layers.

b. Verification of contrail layers from radiosonde measurements with contrail layers derived from MM5

For the verification of MM5 we derived all contrail layers from critical temperature calculations using a contrail factor of 0.036 g kg−1 K−1. Temperatures and critical temperatures were linearly interpolated between those successive reference (or model) levels where the sign of the difference between temperature and critical temperature changed. The altitudes where the critical temperatures equaled the measured (forecasted) temperatures correspond to the boundaries of contrail layers. A total of 870 radiosonde profiles were available for the derivation of reference contrail layers for the verification period from December 2002 to February 2004. Most reference profiles resulted in one contrail layer; in 6% of the cases we found two layers, and another 6% of the profiles showed no contrail probability with critical temperatures continuously below the measured temperature. Mainly because of different vertical resolution of the models, the number of multiple contrail layers in forecasted profiles is less than in reference profiles. For simplicity we omitted cases with multiple-layer occurrences. Table 2 shows the number of contrail layers that were derived from MM5 forecast outputs and were compared with corresponding reference layers.

The mean altitudes of reference contrail layers during winter months were typically between 8000 and 9000 MSL; a minimum value of 5940 m was derived for the period of verification. The altitudes rose in summer frequently above 10 000 MSL. The comparison of mean altitudes from the different models and forecast times with corresponding reference data resulted mostly in increasing deviation with increasing forecast time. Figure 4 shows example plots of mean layer altitudes for 0- and 72-h forecasts derived from AFWA MM5 data. Significant spreading of the data and a reduced correlation coefficient was found for the 72-h forecast period. Mean values of the absolute deviation of forecasted layer altitudes from reference altitudes are shown in Fig. 5 for all forecasts. The altitudes derived from the 0-h forecasts for all models are on the average about 200 m above or below the contrail layer altitudes derived from reference sounding. For 24-h forecasts we obtained differences between 270 and 395 m; the difference increased to about 500 m for the 72-h forecasts. Steadily increasing altitude differences with increasing forecast time also resulted for the 0–48-h Arctic MM5 forecasts. AFWA MM5 showed for both initialization times partly irregular performance characteristics, however general increasing altitude deviations indicated a significant decrease of contrail layer forecast accuracy.

To estimate effects of contrail forecast errors, layer thickness data have to be considered besides altitudes. Thickness depends strongly on the contrail factor; for a contrail factor of 0.036 g kg−1 K−1 the contrail layer thickness derived from Fairbanks radiosonde measurements is 1700 m on the average during the summer months of June, July, and August. For winter months from December to February we found more than twice as thick layers with an average value of 3700 m. Model verification differences might result due to pronounced seasonal thickness differences. The thickness differences were almost coincident using 0600 and 1800 UTC initialization Arctic MM5 data. Figure 6 shows slight differences for the 1200 UTC Arctic model, which is most likely a result of the limited number of data available (Table 2). For near-real-time 0-h model forecasts, the modeled thickness was on the average about 400 m thinner or thicker than the reference thickness. For 48-h forecasts the thickness errors increased to 690 m (Arctic MM5) and 800 m (AFWA MM5). Strong coincident thickness differences were obtained from AFWA MM5 for forecast periods from 48 to 72 h. Errors of more than 30% of the reference thickness could be expected on the average for 72-h forecasts.

A mean overlap (MO) was calculated for verification of the altitude and the thickness of contrail layers. The MO has been introduced as a forecast skill parameter, which relates overlapping contrail layers derived from MM5 forecasts and from atmospheric sounding measurements; values of MO may range from 0% to 100%. Defining dr as the thickness of the reference layer from the sounding profile, df as the thickness of the forecasted layer, and db as the section of the overlapping contrail layer from both profiles, we calculated the mean overlap in percent as MO = [(db/dr) + (db/dr)]/ (2 × 100) (Fig. 7). For a perfect forecast showing coinciding contrail layers the value of MO would be 100%. A small MO may result from a vertical shift in the model and reference layers even at perfectly forecasted thickness. Figure 8 shows the resulting decrease of the MO with an increasing forecast hour. Arctic MM5 verification results coincided well with AFWA MM5; all 0-h model runs showed mean overlaps of about 90%. A steady decrease of 4% (24 h)−1 on the average of all forecast datasets was observed. The MO values between 82% and 83% were found for 48-h forecasts.

c. Contrail observations verified with forecasted contrail occurrence based on MM5

The algorithm for the derivation of critical temperatures as threshold for contrail formation [Eq. (7)] was tested with actual observations of contrails overhead Fairbanks and MM5 forecasts. Only daytime traffic within a time range of 0000 UTC ±2 h was considered for verification; therefore, initialization times with forecasts for 0000 UTC were used. The number of observation to forecast pairs varied with available model and initialization times within the period from mid-December 2002 to the end of February 2004, thus results have to be compared with some reservations (Table 3). Nevertheless, the overall observation counts between 95 and 167 cases for Arctic MM5 resulted in high mean probability of detection (≥87%) and hit rates (≥78%) for all forecasts (6–42 h). No-contrail events were forecasted correctly except for the 18-h forecast, when three observed no-contrail cases were forecasted as contrails due to a small margin between critical temperatures and forecasted temperatures. Errors in forecasts of observed contrails were mostly caused by threshold contrails with short lifetimes; a ratio y(th)/y higher than 86% was observed.

The performance of the AFWA MM5 was slightly less than that of the Arctic MM5 (Table 3). Similar to Arctic MM5, most forecast errors originated from threshold contrails; almost no errors were obtained for no-contrail cases. Mean probability of detection values of less than 80% were found after a forecast period of only 36 h. We related the differences in model performance mostly to the higher vertical resolution of the Arctic MM5. Contrails with short lifetimes were often characterized by little differences TcritT (Jensen et al. 1998); small temperature differences occurred mostly in marginal zones of contrail layers or in thin layers, which might be inaccurate, or missed, by the AFWA vertical resolution of 50 hPa.

5. Summary and discussion

Both Arctic MM5 and AFWA MM5 performed similar in forecasting the altitude of contrail layers; contrails were observed in Fairbanks typically at altitudes higher than 8000 MSL. Slightly better results were obtained for layer thickness forecasts with Arctic MM5 than with AFWA MM5. This difference in performance might result from the different vertical resolution of the available forecast data, as the differences in the physics of AFWA and Arctic MM5 mainly affect processes in the boundary layer and lower troposphere. Different model performance in the considered altitudes of contrail formation may not result from model initial and boundary conditions, which are similar for Arctic and AFWA MM5 models; differences may occur at lower altitudes over the Arctic Ocean, where Arctic MM5 uses a different sea ice parameterization scheme.

A mean overlap (MO) was defined as an overall check of the forecast contrail layer. For short-term forecasts MOs of between 85% and 90% imply good agreement of forecasted contrail layer altitude and thickness with reference data. The MO was consistent for all model runs, a steady decrease of performance with increasing forecast period was found. The MO for 48-h forecasts is above 80%. Due to a decrease in the MO, forecast errors are expected to occur mostly in marginal regions of contrail layers with small (TcritT) <values. These cases were observed mostly as threshold contrails with lifetimes of less than 1 min.

The forecast verification of visual aircraft observations from the ground confirmed that main errors originate from threshold contrails; for some forecast model runs all cases of observed contrails, which were wrongly classified as no contrails, were contrails with lifetimes ≤1 min. Though transient (fast dissolving) contrails can be considered to have little potential to affect directly the radiation budget at the surface, they are of great interest for military when producing contrail forecasts. Transient contrails may have importance for climate studies due to the added water vapor in the atmosphere and an increased potential for heterogeneous nucleation, which facilitates the formation of persistent contrails by succeeding aircrafts (Rind et al. 2000).

The AFWA JETRAX contrail forecast model uses parameterized constant humidity values depending on the altitude in reference to the tropopause (Shull 1998). Humidity parameterization avoids inaccurate contrail forecasts due to a possible poor quality of humidity measurements and forecasts. Shull (1998) showed in his validation study of JETRAX hit rates of 84.4%, a mean value of 18-, 24-, and 30-h AFWA MM5 forecasts; his findings were based on 397 aircraft observations. The JETRAX hit rates are comparable to our derived hit rates using Arctic MM5 and using forecasted humidity values. Our comparison of contrail observations with AFWA MM5 source data showed less agreement; reasons for the lack of forecast quality might be subject to the considerable number of threshold contrails in our observation database, and the reduced number of cases, and hence less significant statistics. The smaller vertical resolution of AFWA MM5 reduces the probability of detection of threshold contrails. A small vertical resolution may also hamper the detection of accurate tropopause altitudes reducing JETRAX performance.

The obtained quality of contrail forecasts in Fairbanks provides confidence of using MM5 especially for short-term forecasts of contrail layers. Contrail forecasts could be incorporated for flight planning purposes. Even for 60-h forecasts MOs of forecasted contrail layers of 80% suggest good model performance at least for the detection of those atmospheric layers where persistent contrails are expected to form. Arctic MM5 performed slightly better when compared to AFWA MM5 due to an enhanced vertical resolution and, hence, a more accurate detection of contrail layers.

A graphical interpretation of the discussed MM5 contrail layer forecasts for Fairbanks and the whole Alaska region has been developed (Stuefer and Wendler 2004); we refer the reader to our Web site (online at http://contrail.gi.alaska.edu/).

Acknowledgments

The University Partnering for Operational Support (UPOS) supported this research in collaboration with the Applied Physics Laboratory, Johns Hopkins University, by a grant from DoD. We thank B. Moore, M. Shulski, B. Hartmann, and M. Robb, who participated in observation and analysis of contrails. Special thanks go to the anonymous reviewers of this paper. Jeff Tilley and the University of Alaska Fairbanks, Mesoscale Modeling and Applications Group provided Arctic MM5 source data.

REFERENCES

  • Appleman, H. S., 1953: The formation of exhaust condensation trails by jet aircraft. Bull. Amer. Meteor. Soc., 34 , 1420.

  • Bromwich, D. H., J. J. Cassano, T. Klein, G. Heinemann, K. M. Hines, K. Steffen, and J. E. Box, 2001: Mesoscale modeling of katabatic winds over Greenland with the Polar MM5. Mon. Wea. Rev., 129 , 22902309.

    • Search Google Scholar
    • Export Citation
  • Busen, R., and U. Schumann, 1995: Visible contrail formation from fuels with different sulfur content. Geophys. Res. Lett., 22 , 13571360.

    • Search Google Scholar
    • Export Citation
  • Cassano, J. J., J. E. Box, D. H. Bromwich, L. Li, and K. Steffen, 2001: Evaluation of Polar MM5 simulations of Greenland’s atmospheric circulation. J. Geophys. Res., 106 , 3386733890.

    • Search Google Scholar
    • Export Citation
  • Downie, C. S., and B. A. Silverman, 1957: Jet aircraft condensation trails. Handbook of Geophysics for Air Force Designers, Geophysics Research Directorate, Air Force Cambridge Research Laboratories, Air Research and Development Command, United States Air Forces, 19-1–19-9.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1993: A nonhydrostatic version of the Penn State/NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev., 121 , 14931513.

    • Search Google Scholar
    • Export Citation
  • Elliott, W. P., and D. J. Gaffen, 1991: On the utility of radiosonde humidity archives for climate studies. Bull. Amer. Meteor. Soc., 72 , 15071520.

    • Search Google Scholar
    • Export Citation
  • Goff, J. A., and S. Gratch, 1946: Low pressure properties of water from −160° to 212°F. Trans. Amer. Soc. Heat. Vent. Eng., 52 , 95122.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., J. Dudhia, and D. R. Stauffer, 1994: A description of the fifth-generation Penn State/NCAR mesoscale model (MM5). NCAR Tech. Note, NCAR/TN-398+STR, National Center for Atmospheric Research, Boulder, CO, 122 pp.

  • Hack, J. J., B. A. Boville, B. P. Briegleb, J. T. Kiehl, P. J. Rasch, and D. L. Williamson, 1993: Description of the NCAR Community Climate Model (CCM2). NCAR Tech. Note, NCAR/TN-382+STR, National Center for Atmospheric Research, Boulder, CO, 138 pp.

  • Jensen, E. J., and Coauthors, 1998: Environmental conditions required for contrail formation and persistence. J. Geophys. Res., 103 , D4,. 39293936.

    • Search Google Scholar
    • Export Citation
  • Meerkötter, R., U. Schumann, D. R. Doelling, P. Minnis, T. Nakajima, and Y. Tsushima, 1999: Radiative forcing by contrails. Ann. Geophys., 17 , 10701084.

    • Search Google Scholar
    • Export Citation
  • Miloshevich, L. M., H. Voemel, A. Paukkunen, A. J. Heymsfield, and S. J. Oltmans, 2001: Characterization and correction of relative humidity measurements from Vaisala RS80-A radiosondes at cold temperatures. J. Atmos. Oceanic Technol., 18 , 135156.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., U. Schumann, D. R. Doelling, K. M. Gierens, and D. W. Fahey, 1999: Global distribution of contrail radiative forcing. Geophys. Res. Lett., 26 , 18531856.

    • Search Google Scholar
    • Export Citation
  • Peters, J. L., 1993: New techniques for contrail forecasting. Air Weather Service Tech. Rep. AWS/TR-93/001, 26 pp. [Available from HQ Air Weather Service, Scott AFB, IL 62225.].

  • Pilie, R. J., and J. E. Jiusto, 1958: A laboratory study of contrails. J. Meteor., 15 , 149163.

  • Pratt, R. W., 1985: Review of radiosonde humidity and temperature errors. J. Atmos. Oceanic Technol., 2 , 404407.

  • Reisner, J., R. M. Rasmussen, and R. T. Bruintjes, 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 mesoscale model. Quart. J. Roy. Meteor. Soc., 124B , 10711107.

    • Search Google Scholar
    • Export Citation
  • Rind, D., P. Lonergan, and K. Shah, 2000: Modeled impact of cirrus cloud increases along aircraft flight paths. J. Geophys. Res., 105 , D15,. 1992719940.

    • Search Google Scholar
    • Export Citation
  • Sausen, R., K. Gierens, M. Ponater, and U. Schumann, 1998: A diagnostic study of the global distribution of contrails. Part I: Present day climate. Theor. Appl. Climatol., 61 , 127141.

    • Search Google Scholar
    • Export Citation
  • Schmidt, E., 1941: Die Entstehung von Eisnebel aus den Auspuffgasen von Flugmotoren. Schr. Dtsch. Akad. Luftfahrtforsch., 44 , 115.

  • Schrader, M. L., 1997: Calculations of aircraft contrail formation critical temperatures. J. Appl. Meteor., 36 , 17251729.

  • Schumann, U., 1996: On conditions for contrail formation from aircraft exhausts. Meteor. Z., 5 , 423.

  • Shull, J. D., 1998: A validation study of the Air Force Weather Agency (AFWA) JETRAX contrail forecast algorithm. M.S. thesis, Dept. of Meteorology, Air Force Institute of Technology, 120 pp. [Available from Airforce Institute of Technology, Wright Patterson AFB, OH 45433.].

  • Stuefer, M., and G. Wendler, 2004: Contrail studies and forecasts in the subarctic atmosphere above Fairbanks, Alaska. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., CD-ROM, P8.13.

  • Tilley, J. S., A. H. Lynch, W. L. Chapman, W. Wu, and G. Weller, 1999: On the coupling of land surface packages within a regional climate model of the Arctic. Proceedings of a Workshop on Arctic Regional Climate Models, Tech. Doc. 981, H. Cattle, Ed., World Meteorological Organization, C12.1–12.5.

    • Search Google Scholar
    • Export Citation
  • Wendler, G., and M. Stuefer, 2002: Improved contrail forecasting techniques for the subarctic setting of Fairbanks, Alaska. Geophysical Institute of the University of Alaska Rep. UAG R-329, 35 pp.

  • Zhang, J., and X. Zhang, 2004: Modeling study of Arctic storm with the coupled MM5-Sea Ice-Ocean model. Preprints, Fifth WRF/14th MM5 User’s Workshop, Boulder, CO, NCAR. [Available online at http://www.mmm.ucar.edu/mm5/workshop/workshop-papers_ws04.html.].

Fig. 1.
Fig. 1.

All-sky digital camera image displaying a contrail formed by a Boeing 747-400 (Northwest Airlines flight 69 from Detroit to Osaka/Japan, 31 Jul 2002).

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Fig. 2.
Fig. 2.

Temperatures vs air pressure shown at the altitudes of aircraft observations. Different symbols refer to the persistence of contrail observations.

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Fig. 3.
Fig. 3.

Errors in critical temperatures derived due to a dry bias of relative humidity of 10% for different ambient relative humidity values. Critical temperatures were calculated with p = 250 hPa and CF = 0.036 g kg−1 K−1.

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Fig. 4.
Fig. 4.

Mean altitudes of contrail layers from reference sounding data vs the forecasted mean altitudes. The (left) 0- and (right) 72-h forecast comparisons are shown.

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Fig. 5.
Fig. 5.

Mean absolute deviation of mean altitudes of contrail layers derived from MM5 in reference to altitudes derived from sounding measurements. Absolute deviations are shown for Arctic and AFWA MM5 for 0–72-h forecasts.

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Fig. 6.
Fig. 6.

Same as in Fig. 5, but for mean absolute deviation of contrail layer thickness.

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Fig. 7.
Fig. 7.

Example of contrail layer thickness derived from (left) reference sounding and (right) AFWA-MM5 forecasts.

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Fig. 8.
Fig. 8.

The MO indicating the agreement of modeled layers and layers from radiosonde measurements. Arctic (solid lines) and AFWA (dashed lines) MM5 overlap parameters are indicated for the different model runs.

Citation: Monthly Weather Review 133, 12; 10.1175/MWR3048.1

Table 1.

Contingency table with the number of forecasted and observed contrails. Here, x = the number of observed and forecasted contrails; y = the false observed, but not forecasted; w = not observed and not forecasted; z = false not observed, but forecasted. The total probability of detection is given for all observed contrails (POD), for threshold contrails [POD(th), 0 < lifetime ≤ 1 min], and for contrails persisting longer than 1 min. PODnil = POD for no-contrail events; FAR (defined in text); HR = the number of correct contrail/no-contrail forecasts related to the total number of observation; PODm: mean value of POD and PODnil.

Table 1.
Table 2.

Numbers of contrail layers available for comparison with layers calculated from atmospheric sounding at Fairbanks. The 0000, 0600, 1200, and 1800 UTC columns correspond to the different model initialization times.

Table 2.
Table 3.

Contingency table with the number of MM5-forecasted and observed contrails. The model with initialization time and respective forecast hour (Fc) is given. Numbers x(th) and y(th) refer to observed threshold cases with lifetimes less or equal 1 min. Abbreviations are the same as in Table 1.

Table 3.
Save