geophysical institute
Transcription
geophysical institute
GEOPHYSICAL INSTITUTE UNIVERSITY OF ALASKA, FAIRBANKS Improved Contrail Forecasting Techniques for the Subarctic Setting of Fairbanks, Alaska Gerd Wendler and Martin Stuefer UAG R-329 Aug 2002 1 Improved Contrail Forecasting Techniques for the subarctic setting of Fairbanks, Alaska by G. Wendler and M. Stuefer, with help from B. Moore, J. Broussard, C. Cole, J. Curtis, S. Nakanishi, M. Robb, H. Stone ABSTRACT .......................................................................................................................................................... 3 INTRODUCTION .................................................................................................................................................. 4 OBSERVATIONAL METHODS .............................................................................................................................. 5 Digital photography......................................................................................................................................... 5 FAA flight data ................................................................................................................................................ 7 Radiosonde data from Fairbanks ................................................................................................................... 8 CONTRAIL DATABASE AND OVER-FLIGHT CHARACTERISTICS .......................................................................... 9 ATMOSPHERIC CONDITIONS AND STATISTICS.................................................................................................. 14 VALIDATION OF EXISTING CONTRAIL FORECASTING ALGORITHMS WITH UAF DATASET ............................. 19 Appleman Algorithm ..................................................................................................................................... 19 Hanson Algorithm ......................................................................................................................................... 21 Schumann Algorithm .................................................................................................................................... 22 Model verification.......................................................................................................................................... 25 DISCUSSION ..................................................................................................................................................... 27 FUTURE WORK................................................................................................................................................. 32 REFERENCES .................................................................................................................................................... 33 APPENDIX A..................................................................................................................................................... 35 THIS RESEARCH WAS SUPPORTED BY THE UNIVERSITY PARTNERING FOR OPERATIONAL SUPPORT (UPOS) IN COLLABORATION WITH THE APPLIED PHYSICS LABORATORY, JOHNS HOPKINS UNIVERSITY, BY A GRANT FROM DOD. 2 Abstract Jet contrails can be frequently observed in the subarctic setting of Fairbanks, Alaska, much like in the contiguous United States. Since March of 2000, continuous digital imagery of the sky was obtained, supported by FAA flight data and radiosonde ascents at the Fairbanks International Airport. There were a total of 2504 over-flights (March 2000July 2002) at Fairbanks, but for a great number of these, contrail observations were not possible due to clouds and/or darkness. For 590 cases, the formation of contrails could be confirmed; their life span varied widely from a few seconds to several hours. In general, cold temperatures and high relative humidity at flight level favored the formation of contrails. These conditions are frequently found in the upper troposphere close to the tropopause. Using our substantial database, different existing algorithms were tested and, in part, improved in order to predict contrail formation and lifetime. The best results were obtained with an algorithm described by Schumann (1996) and an aircraft specific contrail factor of 0.036 g/kgK. For contrails within 4 hours of the radiosonde ascents, a combined hit rate for correctly forecasting the occurrence and non-occurrence of contrails of 92% was obtained. 3 Introduction Aircraft contrails have been investigated since the advancement of high-flying aircrafts. Large differences in contrail occurrence can be observed; some of them disappear after a few seconds, others last for longer time periods and might drift with the winds aloft, while some might spread over parts of the sky or even the whole sky. Schmidt (1941) and Appleman (1953) have performed pioneer work in describing the thermodynamics involved in contrail formation. The water vapor saturation pressure of air is a logarithmic function of the temperature; therefore cold air masses are especially susceptible for the formation of contrails (e.g. Schrader 1997). At very low temperatures, contrails can form in relatively dry air; however, they will only persist if the ambient air temperature is moist and supersaturated with respect to ice. Further, the stability of the air and the wind shear are also of importance. Aircraft characteristics such as aircraft speed, engine type, fuel type and consumption along with the sulfur content of the fuel play a major role in the characteristics of contrails (Busen and Schumann 1995). For significant horizontal spreading across the sky, a strong wind shear aloft combined with super saturation with respect to ice is essential. Observations of contrails have been carried out by ground-based observers, all sky cameras and lidar instruments (Sassen et al 2001), and also from satellites (Carleton and Lamb 1986). Travis et al. (1997) have done modeling efforts, and a comprehensive study (IPCC 1999) discusses possible influences of increasing air traffic on the structure of the atmosphere and the climate of the earth. Contrail formation facilitates the detection of an aircraft, which is important for military purposes. Further, contrails have a strong influence on radiation budget and as such, are of importance for climate change (Seinfeld 1998). In the short wave region of the solar spectrum, cirrus contrails tend to scatter the incoming radiation (mostly forward) (e.g. Gayet et al. 1996). In the infrared region they contribute to the back radiation of the atmosphere. Kuhn (1970) carried out an early investigation on the infrared radiation, while Smith et al. (1998) performed more recent studies. Combined, these processes can contribute either to cooling or warming, depending on the time of the day, season and latitude. However, integrated over the whole Earth and the annual cycle, they contribute to warming (Seinfeld 1998). In the present study, we examine the atmospheric conditions favorable to the formation/non-formation of contrails for the subarctic setting of Fairbanks, which lies below the much traveled flight corridor from Europe to Asia. We have collected the most comprehensive data set on contrail formation for the subarctic, and one of the largest worldwide, on which our analyses is based. 4 Observational Methods Digital photography Continuous all sky digital dome camera imagery has been taken since 1 March 2000. This camera is directed at the zenith and equipped with a fish-eye lens. It is situated below a transparent plexiglas dome on the roof of the Geophysical Institute of the University of Alaska in Fairbanks. The temporal spacing between successive images is 2.5 minutes. The characteristics of the typical dome camera image and the camera position are comprised in Table 1. Table 1: All sky digital imagery characteristics Size Resolution Colors File Format File Length Exposure Time Scanner Latitude/ Longitude Altitude 1024 x 1536 72 x 72 ppi 16.7 Mill. Colors (32 Bit) JPEG >150 KB (185 KB typical during day) 1/250 sec KODAK DC265 ZOOM DIGITAL 64° 51' 22" N / 147° 50' 58" W 225 m a.s.l. In Fig. 1 a typical example is presented for 13:21h LT on 31 July 2002. This contrail lasted more than 10 minutes. In addition to single image analysis, we produced daily digital movies from all the images, so that the occurrence, evolution, and duration of contrails can be evaluated. Direct, visual observations of aircraft over-flights occasionally reveal the formation of contrails, which last only a few seconds. These cases will not necessarily be captured by the operational all-sky camera as the time intervals of obtaining images is 2.5 minutes. Hence, the flight pass in relation to the station is important. If the flight takes place at the edge of our visibility range, it is unlikely that we will obtain an image. However, if the flight is more toward zenith, the time the plane is visible will likely be longer than the time interval between two subsequent images. In addition to the all-sky camera, we installed an Axis Network 2110 camera on the 3rd floor of the International Arctic Research Center building adjacent to the Geophysical Institute. The camera faces south towards the Alaska Range, and supplies additional information for this field of view. 5 Figure 1: All sky digital dome camera image on 31 July 2002 showing a long lasting contrail. 6 FAA flight data Using Flyte Trax 2001 Version 2.0 software we are able to track all commercial flights overhead Fairbanks in real time. Besides aircraft position, the flight identification number, aircraft type, departure and destination airport, flight altitude, ground-speed and estimated time of arrival at the destination are obtained. For contrails analysis, all daytime aircraft passages within 50 miles of Fairbanks are recorded. Figure 2 shows an example print- screen of the flight tracking system. Figure 2: Example of flights in the Fairbanks area at 23:54 h GMT, 27 March 2002. The circles are centered at Fairbanks with a radius spacing of 10 miles (16 km). Flight KLM 9196 from Anchorage to Amsterdam, flying of an altitude of 9450 m produced a contrail lasting only few seconds. We have recently changed our FlyteCom service, which will summarize all daily flights in the Fairbanks area (daily e-mailed), reducing the time needed for data processing, and hence leaving more time for data analysis. 7 Radiosonde data from Fairbanks Atmospheric soundings are carried out at Fairbanks Airport twice a day at 00:00 h and 12:00 h GMT nominal time. Sounding data are available online via http://weather.uwyo.edu/upperair/sounding.html. The main station characteristics are given in Table 2. Table 2: Radiosonde data, Fairbanks Station identifier Station number Latitude Longitude Station elevation PAFA 70261 64.81 N -147.86 E 138 m a.s.l Figure 3: Atmospheric sounding carried out from the Fairbanks International Airport at 00:00 h GMT, 27 March 2002. 8 Figure 3 shows the Fairbanks temperature, dew-point temperature and wind profiles up to an altitude of 30,350 m for 27 March 2002. A number of atmospheric indices characterizing the atmospheric stability are also presented. Contrail database and over-flight characteristics Systematic observations of contrails started on 1 March 2000; each observation includes the following parameters: 1. Date and time of the position of closest approach of an aircraft to the University of Alaska, Fairbanks. The GMT and local time are included. 2. Aircraft specific data at the time of passing Fairbanks: • Aircraft identification (ID) • Origin and destination of flight • Route direction • Aircraft type • Altitude • Speed • Duration to destination airport • Closest approach to the University of Alaska Fairbanks (estimated from flight visualization program). 3. Contrail data • Contrail formed (yes/no) • Number of digital dome camera frames showing the contrail • Duration of contrail visibility derived from dome camera pictures • Duration derived from direct (by eye) observation (short/middle/long lasting) • Duration derived from direct observation in minutes • Total duration representing the larger value of direct observation or domecamera duration in minutes • The degree of determination in case of a 'no contrail' observation. If the aircraft can be spotted directly and there exists no contrail (rare case), the degree of determination is 100%. A sky already covered by clouds, 'nocontrail' observation based only on digital dome camera images (also no aircraft visible) reduces the degree of determination. • Comments about the dissolving process, the drifting of contrails, the observing person and additional information such as the present Fairbanks weather situation (visibility) or hand-held pictures taken of the aircraft. From 1 March 2000 to 31 July 2002, 2504 over-flights have occurred. The most frequent direction of the flights was towards the West (33.5 %) with destination airports in Japan. A number of 29.9 % of the aircrafts originated in Anchorage going north to Europe via the polar route; such a case was depicted in Figure 2. Nearly 1/5 of all flights came from Europe going south to Anchorage, while all other directions were less frequent. See Figure 4 for more details. 9 Figure 4: Directional frequency distribution of aircraft over-flights at Fairbanks, Alaska, between March 2000 and June 2002. The most frequently observed airplane types are Boeing 747-200 and Boeing 747-400, together comprising 65 % of all observations. Typical cruising speeds are between 400 knots (741 km/h) and 550 knots (1019 km/h); 96% of all flights fall in this speed range. In Figure 5 a frequency distribution of the airplane speeds is presented. The cruising altitudes vary from flight-level 210 (6,400 m) to flight-level 450 (13,716m). Nineteen percents of all flights of our database are between flight-levels 280 (8,534m) and 300 (9,144m), and 40.9 % of the flights are between flight-levels 340 (10,363m) and 360 (10,973m). The most frequent altitudes are flight levels 290 (8,840m) and 360 (10,973 m). The observed aircraft types overhead Fairbanks are shown in Table 3. 10 Table 3: Aircraft types and frequencies Aircraft Type Boeing 737-200 Boeing 737-Q Boeing 747-200 Boeing 747-300 Boeing 747-400 Boeing 767-300 Boeing 777-200 MD11 Airbus 340 Airbus 343 others unknown # 119 97 848 119 643 14 230 30 7 11 61 281 Percentage 4.8% 3.9% 34.5% 4.8% 26.1% 0.6% 9.3% 1.2% 0.3% 0.4% 2.5% 11.4% 70% 60% 50% 40% 30% 20% 10% 0% 551600 501550 451500 401450 351400 301350 251300 201250 Groundspeed (knots) Figure 5: Histogram presentation of aircraft cruising speeds at Fairbanks, Alaska, March 2000-July 2002 11 The closest horizontal distance between the aircrafts and our point of observation at UAF is estimated from the 'FlyteComm' visualization software (http://www.flytecomm.com/products/flytetrax.html) in order to ensure correct identification of contrail observations. The technical specifications of the digital dome camera and the fish-eye lens allow a direct observation of contrails within a radial spacing of about 16 km depending on the weather situation and the sun-angle. Sometimes contrails drift with prevailing high wind speeds into the camera field-of-view, therefore we record over-flight data within a range of 50 km to UAF. Prior to 1 January 2002 the contrails analysis is mostly based on dome camera imagery. As the camera picture frequency is 2.5 minutes, the separation of 'no contrail' cases from short contrails, which lasted only a few seconds, is not always possible. Therefore in 2002 we have increased direct observations 'by eye'. Figure 6: Histogram presentation of observed contrail/no-contrail cases with different lifetimes, Fairbanks, Alaska, March 2000-July 2002. A non-linear scale is applied to the x-axis. For comparison with atmospheric sounding data, the contrail database has been restricted to cases with accurate observations of the contrail duration (lifetime) and to observations within 4 hours of the sounding ascent. According to the duration we classified 3 different contrail types: the short duration contrails with a lifetime less or equal to 1 minute, the 12 medium duration contrails lasting between 1 and 10 minutes and the long duration contrails lasting longer than 10 minutes. This reference database contains a number of 20 'no contrails', which refer to cases when an aircraft has been spotted while not forming a contrail; these cases require 'by eye' observations and excellent visibility. A number of 223 contrails were confirmed, ranging from short tails, which dissolve within a few seconds, to long lasting ones, which could be seen in some cases for several hours. Our database includes 78 short-lasting contrails, 47 medium duration contrails (<10 minutes) and 98 contrails lasting longer than 10 minutes (see Figure 6). The maximum observed duration has been 6 hours, which occurred on 23 May 2002 when a Boeing 777-200 overflying Fairbanks from East to West formed a contrail at 10,360 m altitude. Table 4: Altitude distribution of the different contrail classes relative to 10,000m level and to the height of the tropopause. at or below 10000 m above 10000 m at or below tropopause above tropopause no contrail 20.0% 80.0% 25.0% 75.0% short duration 41.0% 59.0% 42.3% 57.7% medium duration 61.7% 38.3% 59.6% 40.4% Altitude above/below Tropopause (m) 4000 long duration 53.1% 46.9% 68.4% 31.6% no contrail short duration medium duration long duration 3000 2000 1000 0 0% 10% 20% 30% 40% 50% 60% -1000 -2000 -3000 -4000 Figure 7: Altitude distribution of different contrail classes in reference to the tropopause. Positive values refer to contrails (no-contrails) above the tropopause. 13 Table 5: Number of contrail cases in different altitude ranges above/at/below the tropopause Altitude Range (m) >3500 2501 - 3500 1501 - 2500 501 - 1500 -499 - 500 -1499 - -500 -2499 - -1500 -3499 - -2500 <= -3500 sum no contrail 2 3 5 5 0 2 2 1 0 20 short medium long duration duration duration 0 0 0 0 0 0 4 2 0 29 7 5 25 19 59 15 15 29 5 4 4 0 0 0 0 0 1 78 47 98 The flight levels show statistically significant different frequencies for the 'no-contrails' compared to contrail cases. At flight-level 290 (8,840m), the preferred altitude of the lower traffic corridor overhead Fairbanks, a frequency peak is obtained for the contrail cases. 'No contrails' have been observed frequently at altitudes above 10,000m (Table 4, Table 5). Medium and long lasting contrails show an almost symmetric frequency distribution around the tropopause (Figure 7). Short lasting contrails are most frequently observed above the tropopause, whereas no-contrails showed 2 maxima; the dominant frequency maximum is in the lower stratosphere, while a secondary maximum was found in the upper troposphere. At the height of the tropopause, no 'no-contrails' were observed. The tropopause altitudes were calculated for all atmospheric soundings available between March 2000 and July 2002. Figure 8 indicates daily and seasonal variations; we obtained average tropopause levels overhead Fairbanks for summer of 10,650m (June-August) and for winter of 9,470m (December-February). Atmospheric conditions and statistics Atmospheric characteristics favorable for contrails formation are investigated for the contrails reference database from June 2000 to July 2002. Statistical analysis has been carried out for temperature, humidity, wind data and different indices including stability and wind-shear data. Histograms of temperature, relative humidity, mixing ratio and mixing ratio deficit were calculated (Figures 9-12). The mixing ratio deficit was defined as the difference between saturation mixing ratio and the measured mixing ratio. The temperatures at aircraft altitudes derived from atmospheric soundings ranged from -35ºC to -70ºC (Figure 9). Almost all contrail observations occurred at altitudes with 14 temperatures below -45ºC; one long lasting contrail was observed at a temperature of -35ºC indicating an error in the temperature measurement of this specific radiosonde ascent. Eighty-five percent of the no-contrails occurred at temperatures above -52.5ºC. This temperature threshold of -52.5ºC for the separation between contrails and nocontrails yields to a “success rate” of 83%. The humidity measurements at aircraft altitudes show significant differences especially between no-contrails and long lasting contrails (see Figure 10). Eighty-five percent of ‘no contrails’ were observed with relative humidity values of less than 25%. In contrast 97% of the 'long-lasting contrails’ occurred during situations with relative humidity values higher than 25%. Nevertheless, a success rate for the separation between no-contrails and all contrails of only 75% was derived, as 64 % of the short lasting contrails were observed at a relative humidity less than 25%. Figure 11 shows mixing ratio values; the mixing ratio is in contrast to the relative humidity an absolute measure of the water content in air (g water vapor/kg air). Values are small for no-contrail cases, and increase with the increasing lifetime of the contrails. For example, the most frequent value for long lasting contrails is about four times as high (0.04g/kg) as for no contrail occurrences (about 0.01g/kg). The mixing ratio deficit is statistically significant different between the contrail and nocontrail cases (Figure 12). Mixing ratio deficits of less than 0.15 g/kg were derived for 98.7% of all contrail cases, whereas 70% of the no-contrails showed deficits above 0.15 g/kg. A mixing ratio deficit of 0.10% as threshold between contrail and no-contrail cases yields a hit rate of 88.5%. The figure indicates similar mixing ratio deficit frequencies for the middle and long lasting contrails. Atmospheric indices were calculated using the 250 hPa and 500 hPa sounding levels. The indices contain temperature, dew point, wind direction and wind speed data. We compared the indices with our contrail observations; the statistical analysis revealed no significant differences between contrail classes. This indicates that for the formation of contrails, stability criteria of the atmosphere are of secondary importance. However, we are still investigating their effect on spreading of contrails. 15 Figure 8: Altitudes of the tropopause derived from atmospheric soundings at Fairbanks between March 2000 and July 2002. The red line shows the 7-point running average. 16 70% no contrail 60% medium duration short duration long duration Frequency 50% 40% 30% 20% 10% 0% -70 -65 -60 -55 -50 -45 -40 -35 Temperature (ºC) Figure 9: Temperatures at flight level for different contrail lifetimes and “no-contrail” cases at Fairbanks, Alaska, March 2000 – July 2002. 60% no contrail short duration 50% medium duration Frequency long duration 40% 30% 20% 10% 0% 0 10 20 30 40 50 60 70 80 Humidity (%) Figure 10: Relative humidity at flight level for different contrail lifetimes and “nocontrail” cases at Fairbanks, Alaska, March 2000 – July 2002 17 50% no contrail 45% short duration medium duration 40% long duration Frequency 35% 30% 25% 20% 15% 10% 5% 0% 0 0.02 0.04 0.06 0.08 0.1 Mixing Ratio (g/kg) Figure 11: The mixing ratio at flight level for different contrail lifetimes and “nocontrail” cases at Fairbanks, Alaska, March 2000 – July 2002 long duration medium duration short duration no contrail 100% 90% 80% Frequency 70% 60% 50% 40% 30% 20% 10% 0% 0.24 0.20 0.16 0.12 0.08 0.04 0.00 Mixing ratio deficit (g/kg) Figure 12: The saturation mixing ratio deficit for different contrail lifetimes and “nocontrail” cases at Fairbanks, Alaska, March 2000 – July 2002 18 Validation of existing contrail forecasting algorithms with UAF dataset Schumann (1996) has given an overview of the history of aircraft condensation contrails research and the different studies, which have been carried out in order to understand the factors responsible for the contrail formation. The objectives of previous research efforts have not been uniform. For example, the military is interested to avoid flight levels at which contrails form, as it facilitates the detection of aircraft. Hence, scientists working for the Air Force, have developed contrail forecast algorithms, which are in operational use (Shull 1998). Furthermore, due to the steadily increasing aircraft traffic, contrails are of interest for scientists investigating atmospheric radiation transfer processes, the chemical state of the atmosphere, and their potential for climatic change (Grassl 1990). Appleman Algorithm Pioneer work to describe the thermodynamics of the contrails formation process has been carried out by Schmidt (1941) and Appleman (1953). The Schmidt/Appleman algorithm provides a threshold decision for contrails, and most of the succeeding algorithms are based on their investigations. Theoretical calculations were used for describing the thermodynamics and the possible phase changes, which might occur when mixing the aircraft engine exhaust with the ambient air. The input parameters for the contrail forecast model are the ambient air pressure, temperature and relative humidity. By considering the entrainment of the heated and moist exhaust gas to the ambient air, a 'critical' temperature is calculated as the threshold temperature to determine if saturation occurs. A contrail is composed of ice crystals and to visibly detect a contrail, a minimum ice crystal content of 0.004 gm-3 is assumed. The contrail formation requires saturation with respect to water, a phase change from water droplets to ice crystals will occur immediately. The temperature increase, DT , of the affected ambient air due to the combustion of one mass unit of fuel is calculated according to: DT = Q k ⋅ N ⋅ cp (1) † where Q denotes the liberated heat by the combustion of one mass unit of fuel 10000cal † = 41.84 ⋅ MJ ), and k is the ratio of exhaust gas to the mass of fuel ( k =12 (Q ª g kg kg/kg). The mass†ratio of entrained environmental air to exhaust gas is denoted by N; thus k ⋅ N characterizes the mass of environmental air, which is affected by the † ratio N depends strongly on the distance combustion of one mass unit of fuel. The mass † 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 value of N ranges from 0 immediately behind the exhaust to infinity. The temperature increase DT further depends on the specific heat of air ( c p ); the specific heat of dry air considering constant pressure is used for c p (=0.24 cal/gºC =1004 J/kgK). Estimating the emission index for water vapor as amount of water vapor produced by the † † † 19 † combustion of 1 kg of fuel, k H 2O (ª1.4 kg/kg), the increase of mixing ratio Drf (g/kg) is derived: k ⋅1000 (2) Drf = H 2O k⋅N † † The combination of equations (1) and (2) leads to: Drf k H 2O†⋅ c p ⋅1000 = DT Q (3) The increase of the mixing ratio per degree temperature increase due to the entrained Drf exhaust is† independent of N. The ratio is called the contrail factor (CF), it is DT characteristic of aircraft engine combustion; an original value of 0.0336 g/kgK has been derived. Appleman compared the mixing ratio deficit (difference between saturation mixing ratio res and actual atmospheric mixing ratio re ) with the contrail factor to obtain † the ambient temperature T and relative humidity fw thresholds for certain pressure levels p and mixing states N, which are necessary for saturation. Using equation 2, the amount of water vapor Dr (g/kg) necessary for saturation with respect to water in an air-parcel † † which is mixed with aircraft exhaust is given by: Dr = Drf - (res - re ) = k H 2O ⋅1000 f - res ⋅ (1- w ) †k ⋅ N 100 (4) As the affected air is undergoing heating due to the exhaust, saturation requires moisture excess (to avoid a lowering of f w below 100%). This moisture excess due to the heating was formulated as the difference of the saturation mixing ratios: † Drh = rs (T + DT) - rs (T) (5) † as DT(N) using equation 1. Thus for the The temperature difference DT is calculated maintenance of saturation equation 4 becomes: Dr = Dr -†(r - r ) - Dr (6) f es e h † † Appleman (1953) calculated critical temperatures ( Tcrit ) based on equation 6 for certain pressure levels and relative humidity values for Tcrit = T(Dr = 0) , and for a range of N values from † 58 to 7000. Using the stipulation that a faint visible and a distinct visible contrail require solid water contents of 0.004 and 0.010 g/m3 respectively, critical † temperatures were also calculated for saturation with respect to ice taking into account † the excess moisture content for the production of visible contrails. Because the saturation vapor pressure is higher over water than over ice at an identical temperature, the comparison of the critical temperatures for saturation with respect to water and for saturation with respect to ice showed that the critical temperatures for water trails must be colder. Thus, taking into account the excess moisture content, which is necessary for the formation of a visible contrail, saturation with respect to water satisfies the requirement for the formation of contrails. For given environmental pressure and relative humidity values, critical temperatures were calculated for various mixing states between the 20 environmental air and the exhaust (N). These critical temperatures show a maximum at a certain mixing state, which can be used as a general threshold for contrail-formation. Appleman showed with his algorithm that no contrails are likely to form at temperatures above -29ºC. He further compared his results with the U.S. Standard Atmosphere, and found a favorable pressure range for the formation of contrails to be between 240 and 205 hPa. Hanson Algorithm Hanson and Hanson (1995) published an algorithm, which is based on Appleman's concept. The development of different aircrafts and different engines required changes in the estimation of the characteristic values defining the fuel combustion and thus in the contrail factor derived originally by Appleman (1953). They also calculated critical temperatures for the formation of contrails. Visible contrails are likely to be formed if the air is at or below a critical temperature, formulated as a function of pressure, humidity, and an aircraft specific contrail factor. The formation of contrails requires enough moisture, supplied by the exhaust and is mixed with the ambient air, in order to maintain Dr saturation with respect to water. Instead of a linear ratio , the curvature in a vapor DT pressure versus temperature curve for saturated air is considered. In order to specify the temperature dependence of the saturation vapor pressure, Hanson has used the approach of Goff and Gratch (1946) instead of the Clausius-Clapeyron equation. Goff & Gratch † obtained more accurate pressure values, especially for low temperatures. The Goff & Gratch formulation is included in the programming script in Appendix A. With the pressure for dry air, p d, the pressure for humid air, p (p= pd +e), and the gas constants for dry air, Rd, and for vapor, R v, the mixing ratio (in g/kg) is converted into terms of vapor pressure by: r= Rd ⋅ e Rd ⋅ e ⋅1000 = ⋅1000 Rv ⋅ pd Rv ⋅ ( p - e) (7) For p>>e a good approximation is: r@ Rd ⋅ e † 622 ⋅ e ⋅1000 = Rv ⋅ p p (8) The amount of moisture released during the mixing between exhaust and environment, drf which was = CF ), is † estimated by Appleman (equation 3) as contrail factor ( dT converted into terms of vapor pressure change using relation (8): de p ⋅ CF @ dT 622 (9)† † 21 An isobaric mixing process is assumed. Hanson used contrail factors defined by Peters (1993); the values were for the non-bypass engine type 0.030 g(kgK)-1, the low-bypass engine 0.034 g(kgK)-1, and for the high-bypass engine 0.039 g(kgK)-1. By replacing the vapor pressure, e, with the saturation vapor pressure, es, in equation 9, Hanson calculates a tangent to the saturation vapor pressure curve (formulated by Goff & Gratch) in order to get a stipulation for critical temperatures. The critical temperature, Tcrit, for contrail formation is derived, where the slope of the saturation vapor pressure versus temperature de p ⋅ CF curve coincides with the mixing line from the exhaust ( s = ). This formulation is dT 622 valid for initially saturated conditions. Relative humidity values less than 100% (f<100%) are taken into account by the product of the saturation vapor slope with a value 100 : † f 100 des p ⋅ CF ⋅ = f dT 622 † (10) In accordance with observations, decreasing critical temperatures are resulting for decreasing relative humidity values; nevertheless a physical explanation of the direct † implementation of the relative humidity, f, as a factor in equation 10 would fail. Schumann Algorithm In analogy to former studies, Schumann's algorithm (1996) is based on the calculation of a critical slope in order to derive a threshold temperature for contrail formation. Besides the effects of the phase changes due to mixing of exhaust fumes with ambient air, Schumann also considers the transformation of combustion heat into kinetic energy of the aircraft's wake motion. Appleman described the contrail factor, CF, as the ratio of the change in the mixing ratio, r, to the temperature change due to the mixing of exhaust gases to the ambient air. According to equation 9, the contrail factor can be considered as ratio of changes of the partial pressure of the vapor, e, to temperature changes: CF = Drf 622De 622G @ = DT pDT p (11) Considering the propulsion efficiency, h, of an aircraft's engine, Schumann defined the De parameter†G (= ) as: DT G= k H 2O c p p 0.622Q(1- h) (12) † The combination of equations 11 and 12 leads to: † 22 CF = k H 2O c p 1000 (13) Q(1- h) This equation differs from equation 3, derived by Appleman (1953), only by the factor ( 1- h ) in the denominator, which accounts for the amount of work performed for the † aircraft drag. The propulsion efficiency, h, is defined as: Fv (14) † Qm f The parameter F denotes the aircraft thrust, v denotes the true air speed, and m f , the rate of fuel flow. Typical values of propulsion efficiencies h for modern bypass turbofan engines are † between 0.3 and 0.4 (Schumann, 1996). He obtained critical temperatures Tcrit for contrail formation for a saturated ambient air (relative humidity f=100%) in analogy † to Hanson (1995) by calculating the temperatures, where the slope of the saturation vapor de pressure equals the slope G in a vapor pressure e versus temperature T curve ( s = G ). dT The non-saturated conditions (f<100%) are derived according to: h= Tcrit, f = Tcrit,100 - (es T crit,100 -eT crit , f ) G = Tcrit,100 - (es T crit,100 - f es T crit , f G ) † (15) The index 100 in equation 14 accounts for the previously calculated saturated threshold condition, and the index f considers the values for the actual relative humidity in the † environmental air. Schumann calculated critical temperatures in equation 15 with Newton’s iterations. Besides calculating the critical temperatures for saturation over water, ice saturation was also considered. Due to the larger contrail factors, the threshold temperatures derived by Schumann are significantly higher than Appleman's temperatures. A similar method to calculate critical temperatures for contrail formation and a brief explanation of the basic physics involved in the mixing of aircraft exhaust gases with the ambient air was given by Schrader (1997). He pointed out errors in the physics of the algorithms by Hanson (1995) and Peters (1993). Schrader's derivation of critical temperatures as threshold for contrail formation coincides for initially saturated air with the Hanson method. For a relative humidity less than 100%, critical temperatures are derived as intersections of the mixing line with the respective vapor pressure curves (see Figure 13). Schrader's solution is in general equivalent to Schumann's method; equation 15 is solved for the vapor pressure, e, instead of the temperature, T. Relations 11 and 15, and the same indices as before yield to: eT crit , f = es T crit ,100 - (Tcrit,100 - Tcrit, f )( pCF ) 622 (16) † 23 With f = eT crit , f es T Schrader showed a possibility to calculate Tcrit, f iteratively: crit , f es T crit ,100 pCF - (Tcrit,100 - Tcrit, f )( ) = f es T crit, f 622 † † (17) Critical temperatures were calculated for the contrail factors also used by Hanson (1995) and Appleman (1953, CF=0.0336 g(kgK)-1). † Figure 13: Water vapor partial pressure as function of temperature (Mollier-Schmidt diagram) for relative humidity values of 100 % (saturation – dark blue) and 60 % and 30%, respectively (light blue). The red straight line represents the threshold line for isobaric mixing of exhaust with environmental air (p=200 hPa and CF=0.036 g(kgK)-1. Summarizing, it can be stated that the algorithms are based on the following assumptions and have the following similarities: • A critical temperature can be calculated as a threshold, whether a contrail will be formed or not. • Condensation starts to form water droplets instead ice crystals. Below the freezing point, the water is super-cooled. • Contrails originate from emitted water vapor and subsequent condensation on preexisting nuclei in the environment or on nuclei from the exhaust. 24 • • • • During the cooling of the exhaust due to the mixing with the ambient air, the decrease of the absolute humidity in the exhaust is directly proportional to the decrease of the temperature. Hanson, Schumann and Schrader used ‘Mollier– Schmidt’ diagrams (water vapor partial pressure versus temperature). For saturated air their algorithms yield the same results. Combustion specific contrail factors, CF, define the mixing between aircraft plume and ambient air. The critical temperatures are functions of the contrail factors, the atmospheric air pressure, and the relative humidity. There are, however, some differences in the algorithms: • • • • Appleman used a linear extrapolation from saturated to the non-saturated condition in the ‘Mollier- Schmidt’ diagram for describing the mixing of the exhaust plume with the ambient air. Hanson made an error in the physics. He calculated the slope of the vapor pressure versus temperature during the mixing process proportional to the relative humidity. The contrail factors used by the different authors vary significantly due to different engine types. Schrader compiled published contrail factors ranging from values of 0.0295 g(kgK)-1 (Pilié and Jiusto 1958) to 0.049 g(kgK)-1 (Peters 1993). Busen and Schumann (1995) and Schumann (1996) discussed contrail factors; they considered aircraft aerodynamics and flight-parameters besides combustion specific parameters. Their formulation of contrail factors yielded minimum values of 0.028 g(kgK)-1. Model verification The critical temperatures were calculated using the different algorithms and were compared to our database (see Figure 6). The vapor pressure-temperature relationship after Mollier-Schmidt and the water vapor pressure, which were derived from the radiosonde measurements, are presented in Figure 14. Vapor pressures close to a saturated state characterized the long lasting contrails. In contrast, 65% of our no-contrails showed vapor pressures below 0.005 hPa. In order to obtain appropriate contrail factors for the separation of contrail from no-contrail cases, we applied a wide range of contrail factors from 0.02 g(kgK)-1 to 0.05 g(kgK)-1 to each algorithm. Using the different algorithm and contrail factors, we could use our data set to test the quality of the prediction. A successful algorithm is characterized by a large number of forecasted and observed contrails as well as predicted no-contrails, which were observed. In Figure 15 the results are presented for various contrail factors. Applemann’s and Schumann’s algorithms coincided and gave a hit rate of 92% for contrail factors of 0.035 and 0.036 g(kgK)-1 , corresponding to mean values between low-bypass and high-bypass engine types. 25 Figure 14: Mollier-Schmidt diagram with radio-sounding measurements corresponding to the different observation classes. The lines indicate relative humidity of 20% and 60% and the saturation vapor pressure over water. 95% Appleman Schumann Hanson Hitrate (%) 90% 85% 80% 75% 70% 65% 0.020 0.025 0.030 0.035 0.040 0.045 0.050 -1 -1 Contrail Factor (gkg K ) Figure 15: Validation of different contrail forecasting algorithms using different contrail factors (CF) 26 Hanson's method resulted in a less perfect agreement, 84% and 85%, respectively were found using contrail factors of 0.039 g(kgK)-1 (high bypass) and 0.047/ 0.048 g(kgK)-1. The high and not realistic contrail factor of 0.047 g(kgK)-1 indicates the errors in the physics for non-saturated air in the Hanson algorithm. A maximum hit rate of 90% was obtained without the consideration of the relative humidity and comparing only the critical temperatures for saturation over water (Tcrit,100) with the measured temperatures. Figure 16 shows a scatter plot of critical temperatures versus measured temperatures derived from the Schumann algorithm with a contrail factor of 0.036 g(kgK)-1. The range of the critical temperatures found is from 218.5 K to 225.7 K. Ninety percent of the nocontrail temperatures correspond to correct algorithm results ( Tmeasured > Tc ). The overall numbers of the agreement between prediction and observation for all contrails and nocontrails are given in Table 6. Fifteen observed contrails are not correctly classified as contrails by the algorithm. The majority of these algorithm errors occur due to fast dissolving, short contrails; Figure 19 shows an example-image of a short † lasting contrail. The critical temperatures of 11 short contrails are below the corresponding temperatures at the aircraft altitudes; these cases would produce incorrect no-contrail predictions, although contrails occurred (Figure 17). Nevertheless a high number of 93.3% of all contrail cases is classified correctly by the Schumann algorithm (CF=0.036 g(kgK)-1). Almost no errors occur for the medium (1 case) and for long lasting contrails (3 cases). Table 6: Contingency table including the total algorithm agreement of observed and predicted contrails and no-contrail cases. Contrail Prediction Contrail Observation yes no yes 208 2 no 15 18 Discussion Radiosonde data allow the prediction of the layers in the atmosphere at which contrails are most likely to form. An example is given in Figure 18. The contrail layers are mostly situated close to the tropopause, which is defined by a temperature minimum and is often characterized by relatively high humidity values. In the lower stratosphere the air is already substantially drier, and hence the likelihood of formation of contrails is reduced. The height as well as the thickness of the layer varies seasonally. During the winter months the contrail layers overhead Fairbanks are more than twice as thick than during summer; the layer base altitudes are significantly lower in winter (compare Table 7). 27 Table 7: Seasonal characteristics of contrail layers. The mean values refer to the 0:00h GMT radiosonde ascents at Fairbanks Airport from January 2000 to July 2002. Dec-Jan-Feb Mar-Apr-May Jun-Jul-Aug Sep-Oct-Nov Altitude of layer Thickness below Thickness above base tropopause tropopause (m a.s.l.) (m) (m) 7606 1861 987 8095 1427 811 9718 962 299 8141 1540 835 Total layer thickness (m) 2849 2238 1261 2374 U.S. Standard Atmospheres for latitudes of 60º North are available for winter (January) and summer (July), (U.S. Standard Atmosphere 1966). The comparison of these standards with critical temperatures for different relative humidity values reveal an atmospheric pressure layer between 180 hPa und 350 hPa, where contrails are most likely to form in winter even at very dry conditions. Most of our observed over-flights are situated within this pressure layer, which corresponds to altitudes between 7,500m and 12,500m. In contrast to the winter situation the likelihood of contrail formation seems to decrease significantly in summer. The U.S. Standard Atmosphere is significantly warmer in summer than in winter at most pressure levels; also below 300 hPa the seasonal temperature differences are pronounced. These warm temperatures in summer would require relative humidity values above 78% in order to form contrails (see Appleman type chart: Figure 20). From atmospheric sounding measurements at Fairbanks we have observed only minor seasonal temperature differences above the 300 hPa level. In 2001, for instance, the mean temperature differences between summer (June-July-August) and winter (January-February-December) were 1.3ºC and 0.4ºC at pressure levels of 250 hPa and 200 hPa respectively. These cold summer-temperatures observed at high levels favor contrail formation also during the warm season at certain heights. 28 Temperature at Aircraft Altitude (K) 230 no contrail short duration medium duration long duration - CF=0.036 g/(kgK) 225 220 215 210 210 215 220 225 230 Critical Temperature (K) Figure 16: Scatter plot of critical temperatures for the observation classes for CF=0.036 g(kgK)-1. 100% wrong prediction agreement 80% 60% 40% 20% long duration medium duration short duration no contrail 0% Figure 17: Hit rate of contrail prediction (blue columns) and wrong prediction percentages (red columns) using the Schumann algorithm with a contrail factor CF=0.036 g(kgK)-1. 29 Figure 18: Temperature and dew-point sounding of 22 June 2000. The critical temperature derived from the Schumann (CF=0.036 g/kgK) algorithm is included. Contrails are likely to be formed at levels, where Tcrit > T (yellow marked level from 8500 m to 10250 m). 30 Figure 19: Example of fast dissolving contrail (~ 4 sec) produced on 22 March 2002 by a Boeing 747-200 jet in 10,700 m (Japan Airline Flight 6422 from Frankfurt to Anchorage) overhead the International Arctic Research Center (UAF). 31 Figure 20: Critical Temperatures for specific relative humidity values and various atmospheric pressure, and the U.S. Standard Atmosphere 1966 in 60ºN for winter (blue) and summer (red) respectively. The critical temperatures were calculated according to Schumann with a contrail factor of 0.036 g(kgK)-1. The ordinate is in logarithmic scale. Future Work • We have been successful in predicting most of the observed contrails. For further prediction-improvement the significance of radiosonde data will be analyzed in more detail. Frequently a time lag between aircraft passages and radiosonde ascents exists. Algorithm errors may occur especially in cases of fast changing meteorological conditions at flight level (frontal passages). Furthermore the quality of radiosonde data will be checked by comparison with re-analysis data from meteorological models. 32 • The evaluation of the characteristics of engine combustion from different aircraft types will be possible with an extended database. An appropriate contrail factor for each aircraft engine might reduce errors in the prediction of contrails. • At present, we are verifying the occurrence/ non-occurrence of contrails with the radiosonde ascents from the Fairbanks International Airport. However, the goal is the prediction, for which a contrail forecast model will be needed, most likely MM5. Forecast charts with favorable altitude levels for the formation of contrails will be produced. This should lead to closer cooperation with the US Air Force Weather Agency. • So far our emphasis has been on the occurrence/ non-occurrence of contrails. We distinguished 3 different types, short, medium and long lasting ones. Spreading of contrails, even observed, was not systematically analyzed. Parameters, which were of minor importance for the formation of contrails, such as the wind speed aloft, wind shear, and atmospheric stability of flight level seem to be of primary importance for spreading and dissolving of contrails. We plan to investigate this in greater detail, including the radiative effects of contrails on the atmosphere and surface. References Appleman, H. S. 1953: The Formation of Exhaust Condensation Trails by Jet Aircraft, Bulletin American Meteorological Society, 34, p 14-20. Busen, R. and Schumann, U., 1995: Visible contrail formation from fuels with different sulfur content. Geophysic. Res. Letters, 22, p 1357-1360. Carleton, A. and P. Lamb 1986: Jet contrails and cirrus clouds: a feasibility study employing high resolution satellite imagery. BAMS, 67, 301-309 Gayet, J., G. Febvre, G. Brogniez, H. Chepfer, W. Renger and P. Wendling 1996. Microphysical and optical properties of cirrus and contrails. J. Atmos. Sci. 53, 126-138 Goff, J. A. and Gratch, S., 1946: Low pressure properties of water from -160 to 212 F. Trans. Amer. Soc. Heat. Vent. Eng., 52, 95 Grassl, H., 1990. The climate at maximum entropy production by meridional and atmospheric heat fluxes. Quarterly Journal of the Royal Meteorological Society 107: 153-166. Hanson, H. M. and Hanson, D. M., 1995: A Reexamination of the Formation of Exhaust Condensation Trails by Jet Aircrafts, Journal of Applied Meteorology, p 24002405. 33 Iribarne, J. V. and Godson, W. L., 1981: Atmosperic Thermodynamics, 2nd ed., D. Reidel Publishing, 128 pp. IPCC 1999: Aviation and the global atmosphere. A special report of IPCC working groups I and III, Cambridge University Press, 373pp Kuhn, P. 1970: Airborne observations of contrail effects on the thermal radiation budget. J. Atmos. Sci. 27, 937-942 Nakanishi, S., Curtis, J. and Wendler, G., 2001: The influence of increased jet airline traffic on the amount of high level cloudiness in Alaska. Theor. Appl. Climatol. 68, 197-205 Peters, J. L. 1993: New techniques for contrail forecasting. AWS/TR-93/001, 26 pp. Pilié, R. J., Jiusto, J. E., 1958: A laboratory study of contrails. J. Meteor, 15, p 149-154. Sassen, K., J.M. Comstock, Z. Wang, and G.G. Mace 2001: Cloud and Aerosol Research Capabilities at FARS: The Facility for Atmospheric Remote Sensing. Bull. of the Americ. Meteorol. Soc., Vol. 82, Nr. 6, 1119-1138 Schmidt, E. 1941: Die Entstehung von Eisnebel aus den Auspuffgasen von Flugmotoren. Schriften der Deutschen Akademie der Luftfahrtforschung, Verlag R. Oldenbourg, Muenchen und Berlin, Heft 44, p 1-15. Schrader, M.L. 1997: Calculations of aircraft contrail formation critical temperatures. J. Appl. Meteorol. ,36, 1725-1729 Schumann, U. 1996: On conditions for contrail formation from aircraft exhausts, Meteorol. Zeitschrift, 5, p 4-23. Seinfeld, J. 1998: Clouds, contrails and climate. Nature, 391, 837-838 Shull, J., D., 1998: A validation study of the air force weather agency (AFWA) jetrax contrail forecast algorithm. Thesis, 118 p. Smith,W., S. Ackerman, H. Rivercomb, H. Huang, D. DeSlover, W. Feltz, L. Gumley and A. Collard 1998: Infrared spectral absorption of nearly invisible cirrus clouds. Geophys. Res. Lett. 25, 1137-1140 Travis, D.J., A.M. Carlton and S.A. Changnon 1997: An empirical model to predict widespread occurrences of contrails J. Appl. Meteorol. 36, 1211-1220 US Standard Atmosphere, Supplements, 1966. ESSA, NASA, US-Airforce, Washington D.C. 289 p. 34 Appendix A Scripting The Perl script calculates critical temperatures with the mixing cloud algorithm described by Schumann (1996) and Schrader (1997). Air temperature, humidity and pressure at flight level and an aircraft specific contrail factor have to be entered as input data. #!/usr/bin/perl -w # # # # # # # # # # Perl script for the calculation of jet contrails formation. Author: Stuefer August 2002 Input: Air Pressure, Relative Humidity, Temperature, Contrail Factor Critical temperatures for contrail formation decision are calculated formation (saturation stipulation). The saturation vapor pressure for low temperatures is derived using a Goff Gratch Formulation. (Goff, J.A. and S. Gratch, 1946, 'Low pressure properties of water from -160 to 212 F.', Trans. Americ. Soc. Heat. Vent. Eng., 52, 95). #============================================================================== # Initialisation of parameters #============================================================================== $ks = 0; $ka = 0; $esat100 = 0; $esatf = 0; $h9 = 1; # # # # # # # # derivative of saturation vapor pressure versus temperature, $ks = 10**($h1-$h2-$h3-$h4) * (log(10)) * ($h5-$h6+$h7+$h8); conversion of vapor pressure per kelvin from aircraft specific contrail factors. saturation vapor pressure at Tc100 saturation vapor pressure at Tc help variable #============================================================================== # Input of the atmospheric parameters and the contrail factor #============================================================================== print chomp print chomp print chomp print chomp 'Enter the air temperature at flight level in deg C: '; ($Ta = <STDIN>); 'Enter the atmospheric pressure at flight level in hPa: '; ($p = <STDIN>); 'Enter the relative humidity at flight level in %: '; ($rf = <STDIN>); 'Enter the Contrail Factor in g/kgK: '; ($CF = <STDIN>); #============================================================================== # Calculation Routine #============================================================================== $Tc100 = 253.16; $Tc = 253.16; # # # # Initial critical temperature in saturated state used as limit value for contrail formation algorithm initial critical temperature at relative humidity f used as limit value for contrail formation algorithm $ka = ($p*$CF/622); 35 Goff_Gratch_slope_comparison(); #=> Tc100 calculated $esat100 = 10**(23.832241-5.02808*(log($Tc100)/log(10))-(1.3816*10**(7))*10**(11.334-0.0303998*$Tc100)+8.1328*10**(-3)*10**(3.491491302.884/$Tc100)-2949.076/$Tc100); $Tc = $Tc100; while (0 <= $h9) { $Tc -= 0.01; $esatf = 10**(23.832241-5.02808*(log($Tc)/log(10))-(1.3816*10**(7))*10**(11.334-0.0303998*$Tc)+8.1328*10**(-3)*10**(3.491491302.884/$Tc)-2949.076/$Tc); $h9 = $esat100-($Tc100-$Tc)*$ka-($rf/100)*$esatf; } $Tc=$Tc-273.16; $Tc100=$Tc100-273.16; printf ("CF = %.4f g/kgK, p = %.0f hPa, Tc100 = %.4f deg C, Tc(f=%.0f percent) = %.4f deg C\n", $CF, $p, $Tc100, $rf, $Tc); if ($Ta < $Tc) { print "Warning: Contrails are expected to form!\n"; } else { print "No contrails are expected to form.\n"; } #============================================================================== # sub program sub Goff_Gratch_slope_comparison { #============================================================================== # Goff Gratch Formulation for the calculation of the temperature derivative of # the saturation vapor pressure #============================================================================== $h1 $h2 $h3 $h4 $h5 $h6 $h7 $h8 $s1 = = = = = = = = = ''; ''; ''; ''; ''; ''; ''; ''; 0; # # # # # # # # # help variable help variable help variable help variable help variable help variable help variable help variable slope difference 1, => help variable #============================================================================== # Calculation of the critical temperature representing a limit for contrail # formation #============================================================================== while (0 >= $s1) { $Tc100 -= 0.01; $h1 = 23.832241+8.1328*10**(-3)*10**(3.49149-1302.8844/$Tc100); $h2 = 1.3816*10**(-7)*10**(11.334-0.0303998*$Tc100); $h3 = 2949.076/$Tc100 ; $h4 = 5.02808*(log($Tc100)/log(10)); $h5 = 2949.076/$Tc100**2; $h6 = 5.02808/$Tc100/log(10); $h7 = 1.3816*3.03998*10**(-9)*10**(11.3340.0303998*$Tc100)*log(10); $h8 = 8.1328*log(10)*10**(-3)*10**(3.491491302.8844/$Tc100)*1302.8844/$Tc100**2; $ks = 10**($h1-$h2-$h3-$h4) * (log(10)) * ($h5-$h6+$h7+$h8); $s1 = $ka-$ks }} 36