Analysis of community departure noise exposure variation using airport noise monitor networks and operational ADS-B data

Ara Mahseredjian 1

Massachusetts Institute of Technology 77 Massachusetts Avenue, Cambridge MA 02139

Jacqueline Huynh 2

University of California Irvine 4200 Engineering Gateway, Irvine CA 92697

R. John Hansman 3

Massachusetts Institute of Technology 77 Massachusetts Avenue, Cambridge MA 02139

ABSTRACT Causes of variation in airport noise monitor network measurements due to departures remain an important source of uncertainty in the development of departure noise abatement procedures. Variation is observed to be between 10 - 15 dB at individual noise monitors for Airbus A320 and Boeing 737NG aircraft flying the same departure trajectories. In order to understand this variation, aggregate departure noise and flight procedures were examined so that factors that correlate with measured noise could be isolated. This paper identifies these factors. Operational flights at Seattle-Tacoma International Airport conducted in March and August of 2019 were examined using a framework that includes ADS-B data, a force balance kinematics model to estimate aircraft performance, and noise monitor recordings from the Port of Seattle Aircraft Noise Monitoring System. Variation in measured departure noise at six monitors within the network was examined as a function of observed data, including aircraft type, aircraft trajectory, airline, wind, temperature, and relative humidity; and inferred variables, including aircraft configuration, weight, and thrust. Altitude is shown to have the strongest e ff ect on community noise exposure. Airline-specific departure procedures are shown to impact noise measurements. Procedures with higher thrust and higher initial climb gradients were observed to have lower measured noise. Ambient environmental conditions, including wind, temperature, and relative humidity, were found to impact noise variation.

1 aramahs@mit.edu

2 huynhlj@uci.edu

3 rjhans@mit.edu

a slaty. inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS O ¥, ? GLASGOW

1. INTRODUCTION

A data-driven exploration of factors that contribute to the variation in departure noise monitor measurements seen at Seattle-Tacoma International Airport (SEA) is presented. The causes of variation in airport noise monitor network measurements of departing aircraft remain a source of uncertainty which must be understood in order to improve existing noise models [1] and develop new departure noise abatement procedures. Variation in departure noise is of specific interest because it is found to be up to 15 dB when aircraft type, departure procedure, and month are held constant. Understanding the potential contributors to this variation is therefore the aim of this work.

The variables which may potentially contribute to this variation include those that arise as a result of flight procedures, including operator-specific practices relating to thrust, airspeed, and configuration management on departure; those that arise as a result of the environment, including ambient wind, relative humidity, and temperature; and those that are specific to individual flights, including departure weight. The Port of Seattle Noise Monitoring System was chosen for this study because of the extensive placement of monitors ranging from those at the airport boundary to those further from the airport than monitors at other airports. Noise monitor recording data for Airbus A320 and Boeing 737NG departures taken in March and August of 2019 was used for this study. March and August were chosen to determine whether variation in relative humidity between the two months would impact monitor recordings. Operational ADS-B data from SEA for the same time periods was taken from the OpenSky Network [2] and was matched with flyovers triggering noise monitor recordings. Weather data, including wind, relative humidity, and temperature, was taken from the NOAA Rapid Refresh numerical weather model [3] and was treated both as a raw variable potentially impacting monitor recordings and as a variable in modeling flight performance. The purpose of this study is to measure how noise measurements may correlate with raw data, but variables not included in surveillance data, including weight, thrust, and configuration, are modeled by necessity.

The impact of both raw and modeled variables on sound exposure level (SEL) recordings at various monitors throughout the SEA noise monitoring network is demonstrated. Trends that consistently appear at monitors at distances varying between 3-12 nautical miles from SEA are illustrated using data from the network. Data suggesting that specific airline operational techniques impact community noise exposure is shown, and areas for future work and model refinement are identified. While data that may explain some of the variation in the recordings is presented, they may not be the only contributors. Future model refinement or analysis using flight data recorder (FDR) data for multiple flights may help address this ambiguity. The causes of variation in departure noise may be used to improve noise models and contribute to the development of future departure noise abatement procedures.

2. DEPARTURE NOISE DATA EXPLORATION AND FLIGHT PERFORMANCE MODELING METHODOLOGY USING OPERATIONAL FLIGHTS AND GROUND NOISE MEASUREMENTS

2.1. Identification of Variables with Potential Departure Noise Variation Impact The variables investigated are organized into three categories: Observed aircraft data, environmental data, and aircraft performance parameters. Observed aircraft data includes the data that characterizes the aircraft type, position, and velocity. Environmental data characterizes the ambient wind, temperature, and relative humidity. Performance parameters include takeo ff weight; landing gear, slat and flap configuration; flight path angle; and takeo ff thrust. Noise data, measured in Sound Exposure Level (SEL) at discrete monitor locations and correlated with specific flights, was provided

by the Port of Seattle. The noise, aircraft, and environmental data were observed, while performance parameters were modeled using observed data. The variables used in this study are listed in Table 1.

Table 1: Variables with Potential Noise Variation Impact

Noise Data Aircraft Data Environmental Data Aircraft Performance

SEL at Monitor Locations Aircraft Type Relative Humidity Takeo ff Weight

Aircraft Operator Northward Wind Aircraft Configuration

Altitude Eastward Wind Takeo ff Thrust

Lateral Position Temperature

Groundspeed

Flight Path Angle

2.2. Data Sources and Seattle Noise Monitoring Network

Noise Data

Noise data from the Port of Seattle Noise Monitoring System was used to obtain flyover noise measured in SEL. Each flyover was correlated with a specific flight number by the Port of Seattle. The noise monitoring system is shown in Figure 1. The south monitors measure noise from aircraft departing to the south. The north monitors measure noise from aircraft departing to the north. The six monitors chosen for this study are highlighted green. These six were chosen because they track departures at close, medium, and far distances from the airport, and because each recorded su ffi cient data for both Airbus A320 and Boeing 737NGs. Close, medium-distance, and far monitors are examined because the variation at each monitor depends on its proximity to the airport.

Figure 1: Port of Seattle noise monitor network. Monitors analyzed shown in green

The lateral tracks of all departures to the south and north are shown in Figure 2 (a) and (b), respectively. Aircraft depart from all three runways.

(a) South departures

(b) North departures

Figure 2: Lateral tracks of Seattle departures

Flights were filtered so that only aircraft that flew within a 0.25 nautical mile lateral track distance of the monitor being analyzed were considered. This filtering was done as a means of holding flyover distance approximately constant.

Aircraft Data

$307) Bo \ Bothell Poulsbo @) Redmond yerdale Bremerton Issaquah

ADS-B Data from the OpenSky Network [2] was used to obtain aircraft data including aircraft type, aircraft operator, altitude, lateral position, and groundspeed. Flight path angle was estimated using the change in altitude and lateral position at two successive ADS-B data points. There may be random fluctuations in the ADS-B data. ADS-B data was correlated with the flights that generated noise monitor recordings. Aircraft operator was used to determine whether any airline-specific operational practices impacted measured noise.

Environmental Data

Environmental data including wind, temperature, and relative humidity was obtained from the NOAA Rapid Refresh (RAP) numerical weather model [3], a grid-based model updated hourly. Weather data taken at the time closest to the aircraft flyover was used. Temperature, northward wind, and eastward wind were averaged between the surface and the aircraft altitude at the point of closest approach to the monitor. Relative humidity was averaged between the surface and 1000 ft above ground level.

Aircraft Performance Data

The Base of Aircraft Data, 4 (BADA4), a database of performance parameters from commercial aircraft manufacturers [4], was used to obtain drag data for the A320 and B737NG. Drag data was used to calculate thrust for each aircraft type.

2.3. Departure Flight Performance Modeling Framework Operational ADS-B and weather data were used to model flight performance. Aircraft position, groundspeed, and altitude were included in ADS-B data, while ambient wind, temperature and relative

Issaquah § Auhiurn

humidity were included in the weather data. Wind data was used to convert aircraft groundspeed to true airspeed. True airspeed was converted to indicated airspeed using atmospheric pressure and density estimates based on the temperature data. Aircraft weight was modeled as a function of true airspeed and altitude 10 nautical miles from the runway, which correlated with FDR data as explained in [5]. This method was chosen because it allowed departure weight to be modeled using only surveillance and weather data. Once weight was estimated, thrust was modeled. The thrust modeling required assumptions about aircraft landing gear, slat, and flap configuration to be established. Landing gear retraction was assumed to occur 0.25 nautical miles after lifto ff . Flaps and slats were assumed to be extended from the takeo ff roll up until 10 knots below the maximum flap extension speed. Airbus A320s were assumed to take o ff with a slats and flaps extended to CONF2, and Boeing 737NGs were assumed to take o ff with slats and flaps extended to Flaps 5. Flap and slat retraction thus occurred at 190 KIAS and 200 KIAS for the A320 and 737NG, respectively. Once configuration assumptions were defined, thrust was calculated. A force-balance kinematics model was used to estimate thrust using performance characteristics including the drag as a function of configuration setting. These drag characteristics were obtained from BADA4. The aircraft was treated as a point mass for simplicity in the performance modeling framework. Further details of the aircraft performance framework are detailed in [1]. The flight performance modeling framework is summarized in Figure 3.

Figure 3: Flight Profile Modeling Framework

3. ANALYSIS OF VARIATION IN AIRCRAFT NOISE MEASUREMENTS USING SEATTLE ADS-B AND NOISE MONITOR MEASUREMENT DATA

3.1. Boeing 737NG Noise Trends at South Monitors The noise impact of each variable for the B737NG at the south close monitor is given in Figure 4. Altitude, thrust per engine, true airspeed, and flight path angle were taken at the point of closest approach to the monitor. Environmental data was averaged as described in Section 2.2. Results at the close, mid, and far monitors are generally consistent. Plots are color-coded by airline so that the noise impact of airline-specific operating procedures can be seen. The correlation coe ffi cient between SEL and each variable, as well as the slope of the linear regression between SEL and each variable are included.

Indicated Airspeed —> Assumed Flap/Slat Schedule Operational ADS-B Data —_ Takeoff Weight —> {Flight Profile | "89 Altitude Weather Velocity Data True Airspeed —> | Configuration Calculated Thrust

Correlation: -0.162 Linear Regression Slope: -2.451 #2 * : jt RE ae GE 9 0% All ALA ALS ® ALT] ALS ALS ALE. 75 0.0 25 5.0 78 Flight Path Angle [deg] 10.0

(a) Altitude impact, south close monitor

(b) Weight impact, south close monitor

(c) Thrust impact, south close monitor

Correlation: -0.124 Linear Regression Slope: -0.305 #8 All ALA ALS ® ALT] ALS ALS ALE. 4 oO 5 10 15 20 Temperature [°C]

Correlation: -0.060 Linear Regression Slope: -0.378 ii All ALA ALS ® ALT] ALS ALS ALE. Northward Wind [kts] “10 0 10 «202-30

(d) True airspeed impact, south close monitor

(e) Flight path angle impact, south close monitor

(f) Temperature impact, south close monitor

Correlation: 0.098 Linear Regression Slope: 0.630 22. Tes All ALA ALS ® ALT] ALS ALS ALE. Eastward Wind [kts] “10 0 10 «20230

Correlation: 0.116 Linear Regression Slope: 0.110 i All ALA ALS ® ALT] ALS ALS ALE. 20 40 60 80 100 Relative Humidity [%]

(g) Northward wind impact, south close monitor. Northward wind positive to the north

(h) Eastward wind impact, south close monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, south close monitor

Correlation: -0.397_ Linear Regression Slope: -2.572 ly All ALA ALS ® ALT] ALS ALS ALE. 2000 3000 4000 Altitude [ft]

Figure 4: Trends for the Boeing 737NG at the south close monitor

As shown in Figure 4 (a), there is strong correlation between noise and altitude at the point of closest approach to the monitor, with lower noise at higher altitude. This trend is consistent with spherical spreading and attenuation losses.

In Figure 4 (b), a strong correlation between noise and takeo ff weight is shown. This trend is not expected given that heavier aircraft typically climb more slowly and with more thrust than light aircraft. This trend is only observed at the close monitor and could be impacted by airline-specific operational practices, such as policies regarding de-rated thrust, in the early phases of the climb.

Correlation: -0.081 Linear Regression Slope: -0.040 8, All ALA ALS ® ALT] ALS ALS ALE. 110 120 130 140 150 Takeoff Weight [1000 Ib]

Figure 4 (c) shows strong correlation between noise and thrust per engine. This trend is also unexpected. This indicates that the reduction in noise that arises with the increased altitude is greater than the increase in thrust associated with higher takeo ff thrust.

Correlation: -0.235 Linear Regression Slope: -0.412 8, fon All ALA ALS ® ALT] ALS ALS ALE. 75. 5.0 75 10.0 12.5 15.0 Thrust per Engine [1000 Ib} 175

Correlation: 0.198 400 Linea Regression Slope: 0.421 ad All Ald ALS © ALT] ALS ALB ALS. 75 160 180 200 220 (240 True Airspeed [KTAS]

As shown in Figure 4 (d), noise increases with true airspeed. This trend is consistent with more aggressive climb rates at lower airspeeds, or with increased airframe noise at higher airspeed.

Noise is shown to increase with flight path angle in Figure 4 (e). This trend is expected given that higher flight path angles correlate with higher altitude.

In Figure 4 (f), noise is shown to decrease with temperature. This trend is not expected since increased temperature is known to reduce noise attenuation. This trend may be a result of airline- specific thrust corrections based on temperature.

Noise is shown to decrease with northward wind in Figure 4 (g). For departures to the south, positive northward wind is a headwind. The trend is consistent with improved climb performance with headwinds.

Figure 4 (h) shows that noise increases with eastward wind. As shown in Figure 1, the south close monitor is east of the airport, so wind blowing towards the east may cause noise from airplanes to advect towards the monitor. The impact of advection on measured noise depends on the relative locations of the noise monitor and the aircraft.

As shown in Figure 4 (i), noise increases with relative humidity. This trend is consistent with the findings in [6], which demonstrates lower noise attenuation for increased relative humidity values above 20 percent.

Significant di ff erences between noise recordings produced by di ff erent airlines are seen in the monitor data. Airline 6 (depicted as the orange triangles) had the lowest average noise measurements in the data observed. Airline 7 (shown as the black stars) had the highest average noise measurements in the data observed. Both Airline 6 and 7 depart with similar takeo ff weight but with di ff erent operational procedures. Airline 6 appears to use an initial climb procedure with high thrust, high climb angle, and low airspeed, whereas Airline 7 appears to operate with lower takeo ff thrust, resulting in lower climb gradients and lower altitudes over the noise monitors.

The trends at the south close monitor are also seen at the south mid and south far monitors, as shown in Figure 5 for the south mid monitor and Figure 6 for the south far monitor. However, at these monitors, noise increases with departure weight as expected. This may be because thrust de-rating occurs during the early phases of climb, so the impact of de-rates seen at the south close monitor are no longer observed at the south mid and far monitors.

Noise measurements for the B737NG at the north monitors are consistent with the results at the south monitors, with the exception that noise measurements increase with the northward wind for northbound departures. This is likely because tailwinds decrease climb performance. Results for northbound B737NG departures are given in Appendix A.

a = a a o 2 £ 5 8 3 id 5 § = @ 8 a 8 2 Correlation: -0.503 Linear Regression Slope: -2.818 jg8 5 All” AL4— ALS ALT) ALS AL ALG 3000 4000 Altitude [ft]

Correlation: 0.102 Linear Regression Slope: 0.062 85, ° 8 eg & 2 8 All” AL4— ALS ALT) ALS AL ALG Monitor Recording SEL [4B] 110 120 130 140 150 Takeoff Weight [1000 Ib]

(a) Altitude impact, south mid monitor

(b) Weight impact, south mid monitor

(c) Thrust impact, south mid monitor

Correlation: -0.154 Linear Regression Slope: -0.283 =” = a8 ® 280 a4 8 8 75 4 270 All AL€ ALS ALT § Als ALB ALB = 65 75 100 «128150175 Thrust per Engine [1000 Ib}

Correlation: 0.102 Linear Regression Slope: 0.191 ;2, ° 8 eg & 2 8 All” AL4— ALS ALT) ALS AL ALG Monitor Recording SEL [4B] 180 200 220 (240 260 True Airspeed [KTAS]

(d) True airspeed impact, south mid monitor

(e) Flight path angle impact, south mid monitor

(f) Temperature impact, south mid monitor

Correlation: -0.205 Linear Regression Slope: -2.437 #2 zo = aes o 280 g 8 875 id 5 270 ALt > AL4 ALS ® ALT| § Als Al ALB = 65 0.0 50 10.0 15.0 Flight Path Angle [deg]

Correlation: -0.043 Linear Regression Slope: -0.120 ° 8 eg & 2 All” AL4— ALS ALT) ALS AL ALG a = a a o 2 £ 2 5 8 3 id 5 § = 15 -10 5 O 5 10 15 20 Temperature [°C]

Correlation: -0.178 Linear Regression Slope: -1.167 ;g8- ° 8 eg & 2 8 All” AL4— ALS ALT) ALS AL ALG Monitor Recording SEL [4B] -20 10 oO 10 20 30 Northward Wind [kts]

(g) Northward wind impact, south mid monitor. Northward wind positive to the north

(h) Eastward wind impact, south mid monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, south mid monitor

a = a a o 2 £ g 5 8 3 id 5 § = @ 8 a 8 2 Correlation: 0.008 Linear Regression Slope: 0.053 58 All” AL4— ALS ALT) ALS AL ALG 10 oO 10 20 Eastward Wind [kts]

Figure 5: Trends for the Boeing 737NG at the south mid monitor

Correlation: 0.180 Linear Regression Slope: 0.183 zo = wes o 280 g 8 875 id 5 270 Alt ALA ALS g AL3 © ALB ALG oO 20 40 60 80 100 Relative Humidity [%]

Correlation: -0.483 Linear Regression Slope: -2.186 23> ° 8 eg & 2 8 Alt” AL4 ALE) ALS AL8_* ALT, Monitor Recording SEL [4B] 3000 4000 5000 6000 Altitude [ft]

(a) Altitude impact, south far monitor

(b) Weight impact, south far monitor

(c) Thrust impact, south far monitor

Correlation: 0.216 Linear Regression Slope: 0.156 eg & Alt” AL4 ALE) ALS ALT) Monitor Recording SEL [4B] 110 120 130 140 Takeoff Weight [1000 Ib]

Correlation: -0.253 Linear Regression Slope: -0.539 zo = wes o 280 g 8 875 id 270 al ALA ALS § As ALB ® ALT, 2. us 8 50 75 100 126 160 175 Thrust per Engine [1000 Ib}

Correlation: 0.109 Linear Regression Slope: 0.338 ;2, ° 8 eg & 2 8 Monitor Recording SEL [4B] 200 220 (240 260 280 True Airspeed [KTAS]

(d) True airspeed impact, south far monitor

(e) Flight path angle impact, south far monitor

(f) Temperature impact, south far monitor

Correlation: 0.006 Linear Regression Slope: 0.087 #2 ° 8 eg & 2 8 Alt” AL4 ALE) ALS ALS ALT, Monitor Recording SEL [4B] g oh & 25 5.0 75 10.0 12.5, Flight Path Angle [deg]

Correlation: -0.035 Linear Regression Slope: -0.112 2% ° 8 eg & 2 8 Als ALG) ALS ALT) Monitor Recording SEL [4B] 20-15 -10 -5 0 5 10 15 20 Temperature [°C]

(g) Northward wind impact, south far monitor. Northward wind positive to the north

(h) Eastward wind impact, south far monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, south far monitor

Correlation: -0.135 Linear Regression Slope: -0.849 a ° 8 eg & 2 8 Alt” AL4 ALE) ALS ALS ALT, Monitor Recording SEL [4B] “20-10 oO 10 20 30 40 Northward Wind [kts]

Figure 6: Trends for the Boeing 737NG at the south far monitor

3.2. Airbus A320 Noise Trends at South Monitors The noise impact of each variable for the A320 at the south close monitor is given in Figure 7. Altitude, thrust per engine, true airspeed, and flight path angle were taken at the point of closest approach to the monitor. Environmental data was averaged as described in Section 2.2. Results at the close, mid, and far monitors are generally consistent. Results for the A320 are generally consistent with results for the B737NG. Plots are color-coded by airline so that the noise impact of airline- specific operating procedures can be seen. The correlation coe ffi cient between SEL and each variable, as well as the slope of the linear regression between SEL and each variable are included.

Correlation: -0.039 Linear Regression Slope: -0.306 ;g8 ° 8 eg & 2 8 Alt” AL4 ALE) ALS ALT) Monitor Recording SEL [4B] “20-10 oO 10 20 30 40 Eastward Wind [kts]

Correlation: 0.089 Linear Regression Slope: 0.119 jf; ° 8 eg & 2 8 Alt” AL4 ALE) ALS ALS ALT, Monitor Recording SEL [4B] oO 20 40 60 80 100 Relative Humidity [%]

Correlation: -0.099 Linear Regression Slope: -0.599 22> 80 ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 75. 7000 2000 3000 4000 Altitude [ft]

(a) Altitude impact, south close monitor

(b) Weight impact, south close monitor

(c) Thrust impact, south close monitor

Correlation: 0.235 Linear Regression Slope: 0.093 85, 400 eal p Tow ad ALT ALS ALTO. ALIQ| ALS © ALA ALT AL, 75, 120 130 140 150 Takeoff Weight [1000 Ib]

Correlation: 0.063 Linear Regression Slope: 0.096 85, 400 eal ps Tow ad ALT ALS ALTO. ALIQ| ALS 9 ALA ALT ALT 75 75 10.0 12.5 15.0 175, 20.0 Thrust per Engine [1000 Ib}

Correlation: 0.028 Linear Regression Slope: 0.051 ;2%, 100 a ad ALT ALS ALTO. ALIQ| ALS 9 ALA ALT ALT 75 140 160 180 200 220 (240 True Airspeed [KTAS]

(d) True airspeed impact, south close monitor

(e) Flight path angle impact, south close monitor

(f) Temperature impact, south close monitor

Correlation: 0.168 Linear Regression Slope: 2.770 % ALL ALS ALTO. ALI2) ALO > ALA © ALI © AL7 25 5.0 75 10.0 Flight Path Angle [deg]

(g) Northward wind impact, south close monitor. Northward wind positive to the north

(h) Eastward wind impact, south close monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, south close monitor

Figure 7: Trends for the Airbus A320 at the south close monitor

As shown in Figure 7 (a), there is strong correlation between noise and altitude at the point of closest approach to the monitor, with lower noise at higher altitude. This trend is consistent with spherical spreading and attenuation losses.

Correlation: -0.153 Linear Regression Slope: -0.374 an ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 oO 5 10 15 20 25 Temperature [°C]

Figure 7 (b) shows that noise increases with takeo ff weight. This trend is expected given that heavier aircraft are expected to climb more slowly and with more thrust than light aircraft.

Noise is shown to increase with thrust per engine in Figure 7 (c). This trend is expected because noise is known to increase with thrust when all other potential factors are held constant.

As shown in Figure 7 (d), noise does not vary significantly with true airspeed. This trend is not expected. Aircraft with more aggressive climb rates fly at lower airspeeds.

Correlation: -0.071 Linear Regression Slope: -0.420 ;g8 Be. ae ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 10 oO 10 20 30 Northward Wind [kts]

Figure 7 (e) shows that noise increases with flight path angle. This trend not is expected because higher flight path angles correlate with higher altitude.

Noise decreases with temperature, as shown in Figure 7 (f). This trend is not expected since increased temperature is known to reduce noise attenuation. This trend may be a result of airline- specific thrust corrections based on temperature.

Correlation: 0.119 Linear Regression Slope: 0.749 22. Tes ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 10 oO 10 20 30 Eastward Wind [kts]

Noise decreases with northward wind, as shown in Figure 7 (h). For departures to the south, positive northward wind is a headwind. Thus, the negative correlation is expected because headwinds improve climb performance.

Figure 7 (h) shows that noise increases with eastward wind. As shown in Figure 1, the south close monitor is east of the airport, so wind blowing towards the east may cause noise from airplanes to

Correlation: 0.158 Linear Regression Slope: 0.141 4 ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 20 40 60 80 100 Relative Humidity [%]

advect towards the monitor. The impact of advection on measured noise depends on the relative locations of the noise monitor and the aircraft.

As shown in Figure 7 (i), noise increases with relative humidity. This trend is consistent with the findings in [6], which demonstrates lower noise attenuation for increased relative humidity values above 20 percent.

Correlation: -0.492 Linear Regression Slope: -2.539 85 Fc zo = wes o 280 8 875 id 5 270 ALT ALS ALIO. = ALI2. 2 AS) ALA ALT» ALT 65 1000 2000 3000 4000 5000 Altitude [ft]

Airlines are not as clearly segregated for Airbus A320s as they were for Boeing 737NGs. However, the aircraft with the highest observed noise measurements are operated by Airline 1 (shown as a purple circle), Airline 4 (depicted as a red diamond) and Airline 9 (drawn as a pink triangle). Airline 1, Airline 4, and Airline 9 operate A320s powered by both CFM and IAE engines, so an engine-specific noise impact is unlikely.

The trends at the south close monitor are also evident at the south mid and south far monitors, as shown in Figure 8 for the south mid monitor and Figure 9 for the south far monitor. However, noise is shown to decrease with thrust and flight path angle at these monitors. This trend is consistent with the finding that climbing to higher altitude lowers community noise exposure.

Correlation: 0.228 Linear Regression Slope: 0.101 85, ior, zo = wes o 280 8 875 id 5 270 ALT ALS ALIO. = ALI2. 2 ALS) ALA ALT» ALT 65 120 130 140 150 Takeoff Weight [1000 Ib]

Noise measurements for the A320 at the north monitors are consistent with the results at the south monitors, with the exception that noise measurements increase with the northward wind for northbound departures. This is likely because tailwinds decrease climb performance, increasing noise exposure on the ground. Results for northbound A320 departures are given in Appendix B.

Correlation: -0.036 Linear Regression Slope: -0.050 2, Toon zo = wes o 280 8 875 id 5 270 ALT ALS ALIO. = ALI2. g Ag © ALA ALT» ALT 65 75 10.0 125 15.0 17.5 20.0 Thrust per Engine [1000 Ib}

(a) Altitude impact, south mid monitor

(b) Weight impact, south mid monitor

(c) Thrust impact, south mid monitor

Correlation: 0.138 Linear Regression Slope: 0.17 ° 8 eg & 2 8 ALL ALS ALIO. ALI ALO © AL4 © ALIT © ALT Monitor Recording SEL [4B] g & 140 160 180 200 220 240 260 True Airspeed [KTAS]

(d) True airspeed impact, south mid monitor

(e) Flight path angle impact, south mid monitor

(f) Temperature impact, south mid monitor

Correlation: -0.164 Linear Regression Slope: -2.167 zo = wes o 280 8 875 id 5 270 ALT ALS ALIO. = ALI2. § Als) ALA ALT» ALT = 65 00 25 50 75 100 125 Flight Path Angle [deg]

Correlation: -0.128 Linear Regression Slope: -0.335 8 ° 8 eg & 2 8 ALS ALIO. ~ ALI2/ ALA © ALI» ALT Monitor Recording SEL [4B] 10 5 0 5 10 15 20 Temperature [°C]

Correlation: -0.208 Linear Regression Slope: -1.210 ;g8- ° 8 eg & 2 8 ALL ALS ALIO. ALI ALQ © AL4 © ALIT © ALT Monitor Recording SEL [4B] -20 10 oO 10 20 30 Northward Wind [kts]

Correlation: 0.091 Linear Regression Slope: 0.587 ji ° 8 eg & 2 8 ALL ALS ALIO. ALI ALQ © AL4 © ALIT © ALT Monitor Recording SEL [4B] -20 10 oO 10 20 30 Eastward Wind [kts]

(g) Northward wind impact, south mid monitor. Northward wind positive to the north

(h) Eastward wind impact, south mid monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, south mid monitor

Correlation: 0.253 Linear Regression Slope: 0.238 if ° 8 eg & 2 8 ALL ALS ALIO. ALI ALQ © AL4 © ALIT © ALT Monitor Recording SEL [4B] oO 20 40 60 80 100 Relative Humidity [%]

Figure 8: Trends for the Airbus A320 at the south mid monitor

Correlation: -0.403 Linear Regression Slope: -1.887 jg8 5 a0 zo = wes o 280 g 8 875 id 270 § ‘AL = AL7] = 3000 4000 5000 6000 Altitude [ft]

Correlation: 0.215 Linear Regression Slope: 0.143 lf, ° 8 eg & 2 8 ALT= ALT) Monitor Recording SEL [4B] 120 130 140 150 Takeoff Weight [1000 Ib]

(a) Altitude impact, south far monitor

(b) Weight impact, south far monitor

(c) Thrust impact, south far monitor

Correlation: -0.091 Linear Regression Slope: -0.134 i, ° 8 eg & 2 8 ALT= ALT) Monitor Recording SEL [4B] oh S 10.0 15.0 20.0 Thrust per Engine [1000 Ib]

Correlation: -0.080 Linear Regression Slope: -0.241 #8 ° 8 eg & 2 8 AL Monitor Recording SEL [4B] 220 (240 260 280 True Airspeed [KTAS]

(d) True airspeed impact, south far monitor

(e) Flight path angle impact, south far monitor

(f) Temperature impact, south far monitor

Correlation: -0.116 Linear Regression Slope: -1.661 ° 8 eg & 2 8 ALT= ALT) Monitor Recording SEL [4B] oh & 25 5.0 75 10.0 12.5, Flight Path Angle [deg]

Correlation: -0.109 Linear Regression Slope: -0.303 22 ° 8 eg & 2 8 ALT= ALT) Monitor Recording SEL [4B] 20-15 -10 -5 0 5 10 15 20 Temperature [°C]

Correlation: -0.267, Linear Regression Slope: -1.671 ;g8 ° 8 eg & 2 8 ALT= ALT) Monitor Recording SEL [4B] -20 10 oO 10 20 30 Northward Wind [kts]

(g) Northward wind impact, south far monitor. Northward wind positive to the north

(h) Eastward wind impact, south far monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, south far monitor

Correlation: -0.092 Linear Regression Slope: -0.729 iB ° 8 eg & 2 8 ALT= ALT) Monitor Recording SEL [4B] -20 10 oO 10 20 30 Eastward Wind [kts]

Figure 9: Trends for the Airbus A320 at the south far monitor

Correlation: 0.200 Linear Regression Slope: 0.280 jf ° 8 eg & 2 8 AL Monitor Recording SEL [4B] oO 20 40 60 80 100 Relative Humidity [%]

4. CONCLUSION

Variation in departure noise can be attributed to operator-specific climb procedures, aircraft weight, and environmental factors. Altitude is shown to have the strongest e ff ect on community noise exposure. Airline-specific procedures with higher thrust and higher initial climb gradients were observed to have lower noise exposure. This finding may help inform the development of new noise abatement departure procedures. Future validation studies may examine the impact of specific airline standard operating procedures on aircraft noise. Data from flight data recorders can also be used to obtain precise configuration and weight data.

Environmental factors including ambient wind and relative humidity are shown to have impacts on climb performance (headwind), advection of noise (crosswind), and attenuation of noise (relative humidity).

ACKNOWLEDGEMENTS

This work was sponsored by the Federal Aviation Administration (FAA) under ASCENT Center of Excellence Project 44. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the United States Government. The authors would like to acknowledge the support of Chris Dorbian, Joseph DiPardo, and Bill He of the FAA O ffi ce of Environment and Energy as well as Thomas Fagerstrom and Stan Shepherd the Port of Seattle.

REFERENCES

[1] Ara Mahseredjian, Jacqueline Thomas, and R. John Hansman. Advanced Procedure Noise Model Validation using Airport Noise Monitor Networks. In 50th International Congress and Exposition on Noise Control Engineering , Virtual Event, 2021. [2] M. Schäfer, M. Strohmeier, V. Lenders, I. Martinovic, and M. Wilhelm. Bringing up opensky: A large-scale ads-b sensor network for research. Proceedings of the 13th IEEE / ACM International Symposium on Information Processing in Sensor Networks (IPSN) , pages 83–94, April 2014. [3] Stanley G. Benjamin, Stephen S. Weygandt, John M. Brown, Ming Hu, Curtis R. Alexander, Tatiana G. Smirnova, Joseph B. Olson, Eric P. James, David C. Dowell, Georg A. Grell, Haidao Lin, Steven E. Peckham, Tracy Lorraine Smith, William R. Moninger, Jaymes S. Kenyon, and Geo ff rey S. Manikin. A North American Hourly Assimilation and Model Forecast Cycle: The Rapid Refresh. Monthly Weather Review , 144(4):1669–1694, 2016. [4] A. Nuic. User Manual for the Base of Aircraft Data (BADA) Revision 3.12. Technical Report 12 / 11 / 22-58, Eurocontrol, 2013. [5] Sandro Salgueiro, Jacqueline L Huynh, and R. John Hansman. Aircraft Takeo ff and Landing Weight Estimation from Surveillance Data. In AIAA Science and Technology Forum , San Diego, CA, 2022. [6] Cyril M. Harris. Absorption of sound in air versus humidity and temperature. The Journal of the Acoustical Society of America , 40(1), 1966.

APPENDIX A: BOEING 737NG NOISE TRENDS AT NORTH MONITORS

North Close Monitor

@ & gs 8 e & 2 8 a ‘Aut AL2 Monitor Recording SEL [4B] 70, Correlation: -0.372 ALS ALS Als ALG 1000 2000 3000 Altitude [ft] Linear Regression Slope: -2.510 ;g8 5 TOO » au7| 4000

Correlation: 0.091 Linear Regression Slope: 0.031 85, @ & gs 8 e & 2 8 a Al ALS ALS ALT) AL2 © AL ALG Monitor Recording SEL [4B] ~ 3 100 110 120 130 140 150 Takeoff Weight [1000 Ib]

(a) Altitude impact, north close monitor

(b) Weight impact, north close monitor

(c) Thrust impact, north close monitor

Correlation: -0.085 Linear Regression Slope: -0.010 85, Ect we = 90 ® Pes a4 8 8 80 4 275 All ALSCALS ALT § Al2 9 ALA ALB = 70 25 75 125 178 228 278 Thrust per Engine [1000 Ib}

Correlation: 0.278 Linear Regression Slope: 0.580 ;2, @ & gs 8 e & 2 8 a Al ALS ALS ALT) AL2 © AL4 ALG a = a a o 2 £ 5 8 3 id 5 § = ~ 3 180 200 220 (240 True Airspeed [KTAS]

(d) True airspeed impact, north close monitor

(e) Flight path angle impact, north close monitor

(f) Temperature impact, north close monitor

Correlation: -0.067, Linear Regression Slope: -0.289 @ & gs 8 e & 2 8 a Al ALS ALS ALT) AL2 9 AL4 ALG Monitor Recording SEL [4B] ~ of & 5.0 10.0 15.0 20.0 25.0 30.0 Flight Path Angle [deg]

Correlation: 0.112 Linear Regression Slope: 0.240 2% @ & gs 8 e & Au AL ALI 76 AQ ALIO. © ALT ALS ALIT—_ALI3) Monitor Recording SEL [4B] 100-5 oO 5 10 15 20 25 Temperature [°C]

(g) Northward wind impact, north close monitor. Northward wind positive to the north

(h) Eastward wind impact, north close monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, north close monitor

Figure 10: Trends for the Boeing 737NG at the north close monitor

a = a a o 2 £ g 5 8 3 id 5 § = 8 8 8. & e Correlation: 0.038 Linear Regression Slope: 0.271 42 Al ALS ALS ALT) AL2 © AL4 ALG 10 oO 10 20 Northward Wind [kts]

Correlation: 0.120 Linear Regression Slope: 0.755 2 @ & gs 8 8 Al ALS ALS ALT) AL2 © AL4 ALG a = a a o 2 £ g 5 8 3 id 5 § = 10 oO 10 20 Eastward Wind [kts]

Correlation: 0.223 Linear Regression Slope: 0.236 iy @ & gs 8 e & Al ALS ALS ALT) AL2 © AL ALG Monitor Recording SEL [4B] 0 20 40 60 80 100 Relative Humidity [%]

North Mid Monitor

Correlation: -0.149 Linear Regression Slope: -0.952 22 All ALS ALS ® ALT] AL2 9 ALA ALE. 4000 5000 += 6000S 7000 Altitude [ft]

Correlation: -0.041 Linear Regression Slope: -0.030 8; All ALS ALS ® ALT] AL2 9 ALA ALE. 110 120 130 140 Takeoff Weight [1000 Ib]

(a) Altitude impact, north mid monitor

(b) Weight impact, north mid monitor

(c) Thrust impact, north mid monitor

Correlation: 0.029 Linear Regression Slope: 0.025 lf, All ALS ALS ® ALT] AL2 9 AL4 ALE. 75 12.5 175 22.5 275 Thrust per Engine [1000 Ib}

Correlation: -0.031 Linear Regression Slope: -0.082 28 All ALS ALS ® ALT] AL2 9 AL4 ALE. 200 220 240 260 280 True Airspeed [KTAS]

Correlation: 0.090 Linear Regression Slope: 0.684 All ALS ALS ® ALT] AL2 9 ALA ALE. 5.0 10.0 15.0 20.0 25.0 30.0 Flight Path Angle [deg]

(d) True airspeed impact, north mid monitor

(e) Flight path angle impact, north mid monitor

(f) Temperature impact, north mid monitor

Correlation: -0.300 Linear Regression Slope: -0.960 #8 AS ALS. AL2 9 AL4 ALE. * AL] 10 5 0 5 10 15 20 25 Temperature [°C]

Correlation: 0.229 Linear Regression Slope: 2.004 22. Tes All ALS ALS ® ALT] AL2 9 AL4 ALE. 10 oO 10 20 30 Northward Wind [kts]

(g) Northward wind impact, north mid monitor. Northward wind positive to the north

(h) Eastward wind impact, north mid monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, north mid monitor

Correlation: -0.180 Linear Regression Slope: -1.536 ;22 Toes All ALS ALS ® ALT] AL2 9 AL4 ALE. 10 oO 10 20 30 Eastward Wind [kts]

Figure 11: Trends for the Boeing 737NG at the north mid monitor

North Far Monitor

Correlation: 0.006 Linear Regression Slope: 0.012 if All ALS ALS ® ALT] AL2 9 AL4 ALE. 20 40 60 80 100 Relative Humidity [%]

Correlation: -0.253 Linear Regression Slope: -1.199 ly 80 Alt) ALa AL ALS ALS ALT, 50. 5000 6000 7000 8000 9000 10000 Altitude [ft]

(a) Altitude impact, north far monitor

(b) Weight impact, north far monitor

(c) Thrust impact, north far monitor

Correlation: 0.163 Linear Regression Slope: 0.150 85, alt AL@ ALB! ALS ALS ALT, 110 120 130 140 Takeoff Weight [1000 Ib]

Correlation: 0.058 Linear Regression Slope: 0.128 85, Tow aye A a Ag + att 75 10.0 Thrust per Engine [1000 Ib} 125

Correlation: -0.069 Linear Regression Slope: -0.316 #8 80 Alt) ALa ALG! ALS ALS ALT) 50 240 260 280 300 True Airspeed [KTAS]

(d) True airspeed impact, north far monitor

(e) Flight path angle impact, north far monitor

(f) Temperature impact, north far monitor

Correlation: 0.033 Linear Regression Slope: 0.747 Alt ALA ALG! ALS ALS ALT! 25 5.0 75 10.0 Flight Path Angle [deg]

(g) Northward wind impact, north far monitor. Northward wind positive to the north

(h) Eastward wind impact, north far monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, north far monitor

Figure 12: Trends for the Boeing 737NG at the north far monitor

Correlation: 0.089 Linear Regression Slope: 0.322 an 80 Alt) ALa ALG! ALS ALS ALT, 50, -25 -20 -15 -10 5 O 5 10 15 Temperature [°C]

Correlation: 0.138 Linear Regression Slope: 0.963 22. Tes Ala ALG! ALS ALS ALT! 10 oO 10 20 30 Northward Wind [kts]

Correlation: 0.090 Linear Regression Slope: 0.610 28. Tes Alt ALA ALG! ALS ALS ALT! Eastward Wind [kts] “10 0 10 «20230

Correlation: 0.135 Linear Regression Slope: 0.307 i Alt AL4 ALG! ALS ALS ALT! 20 40 60 80 100 Relative Humidity [%]

APPENDIX B: AIRBUS A320 NOISE TRENDS AT NORTH MONITORS

North Close Monitor

Correlation: -0.239 Linear Regression Slope: -1.773 ily @ & gs 8 e & 2 8 Alt > AL» ALI2/ Monitor Recording SEL [4B] 75 ALQ = ALIO » ALT ALS ALITY ALI3) 70, 1000 2000 3000 4000 Altitude [ft]

Correlation: 0.402 Linear Regression Slope: 0.104 8, @ & Alt > AL» ALI2/ ALQ = ALIO » ALT ALS ALITY ALI3) Monitor Recording SEL [4B] 120 140 160 Takeoff Weight [1000 Ib]

(a) Altitude impact, north close monitor

(b) Weight impact, north close monitor

(c) Thrust impact, north close monitor

Correlation: -0.104 Linear Regression Slope: -0.012 3; Tar Pad = m9 o Pes 8 go 5 At ALA ALT). 275 ALQ— ALIO. © ALT § ALS ALIT—_ ALS = 70 25 75 125 175 225 275 Thrust per Engine [1000 Ib}

Correlation: 0.227 Linear Regression Slope: 0444 j4%- ina Pad = m9 o Pes 8 § 80 id 5 At ALA ALI 275 ALQ— ALIO. © ALT 2 AL ALIT ALIS 70 150 170 190 210 True Airspeed [KTAS]

(d) True airspeed impact, north close monitor

(e) Flight path angle impact, north close monitor

(f) Temperature impact, north close monitor

Correlation: -0.108 Linear Regression Slope: -0.456 Pad = m9 o Pes 8 § 80 id 5 At ALA ALI 275 ALQ— ALIO. © ALT § ALS ALIT ALS = 70 00 50 100 150 200 250 300 Flight Path Angle [deg]

Correlation: -0.188 Linear Regression Slope: -0.700 22 2 2 5 8 id 37° Al” ALA ALI) ees alg ALIO. =» ALT g ALS ALIT ALIS 5 10 Temperature [°C] 15 20 25

(g) Northward wind impact, north close monitor. Northward wind positive to the north

(h) Eastward wind impact, north close monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, north close monitor

Figure 13: Trends for the Airbus A320 at the north close monitor

Correlation: 0.023 Linear Regression Slope: 0.183 22. finn Pad = m9 o Pes 8 go 5 At ALA ALI 275 ALQ— ALIO. © ALT § ALS ALIT—_ ALS = 70. -20 “10 0 10 20 Northward Wind [kts]

Correlation: 0.166 Linear Regression Slope: 1.154 ;@& finn Pad = m9 o Pes 8 go 5 At ALA ALI 275 ALQ— ALIO. © ALT § ALS ALIT—_ ALS = 70. -20 “10 0 10 20 Eastward Wind [kts]

Correlation: 0.305 Linear Regression Slope: 0.350 i we = 90 ® Paes.-- a4 8 ¢ 8 80 4 5 ALT” ALA LIQ) 275 ALQ— ALIO. © ALT § ALS ALIT—_ ALS = 70 0 2 40 6 80 100 Relative Humidity [%]

North Mid Monitor

Correlation: -0.007, Linear Regression Slope: -0.051 ie Alt > AL» ALI2/ ALQ = ALIO » ALT ALS ALIT ALIS 5.0 10.0 Flight Path Angle [deg] 15.0 20.0 25.0 30.0

Correlation: -0.188 Linear Regression Slope: -0.700 22 2 2 5 8 id 37° Al” ALA ALI) ees alg ALIO. =» ALT g ALS ALIT ALIS 5 10 Temperature [°C] 15 20 25

(a) Altitude impact, north mid monitor

(b) Weight impact, north mid monitor

(c) Thrust impact, north mid monitor

Correlation: 0.214 Linear Regression Slope: 2.136 ao Tes, Alt > AL» ALI2/ ALQ = ALIO » ALT ALS ALITY ALI3) -20 10 oO 10 20 Northward Wind [kts]

Correlation: -0.109 Linear Regression Slope: -0.974 ;g8- Alt > AL ALI2) ALQ = ALIO » ALT ALS ALITY ALI3) -20 10 oO 10 20 Eastward Wind [kts]

Correlation: -0.023 Linear Regression Slope: -0.047 jf Alt > AL» ALI2/ ALQ = ALIO » ALT ALS ALITY —_AL13) 0 20 40 60 80 100 Relative Humidity [%]

(d) True airspeed impact, north mid monitor

(e) Flight path angle impact, north mid monitor

(f) Temperature impact, north mid monitor

Correlation: -0.377, Linear Regression Slope: -1.977 yy 6 ALT ALS ALTO. ALIQ| ALS 9 ALA ALT ALT 50. 5000 6000 7000 8000 9000 10000 Altitude [ft]

Correlation: 0.362 Linear Regression Slope: 0.300 lf, ee: ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 50, 120 125 130 135 140 145 Takeoff Weight [1000 Ib]

(g) Northward wind impact, north mid monitor. Northward wind positive to the north

(h) Eastward wind impact, north mid monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, north mid monitor

Correlation: 0.098 Linear Regression Slope: 0.180 85, Tow ‘AIO. = ALI2| ALIt © ALT 5.0 75 10.0 12.5 15.0 Thrust per Engine [1000 Ib}

Figure 14: Trends for the Airbus A320 at the north mid monitor

North Far Monitor

Correlation: -0.129 Linear Regression Slope: -0.950 22 70 All AUd ALR Sos AL9ALIO. = ALT 5 AL3 ALI ALIS 60 3000 4000 5000 «6000 Altitude [ft]

Correlation: 0.052 Linear Regression Slope: 0.029 i, 2 e 8 « 37 Ali ALe > ALR 65 ALQ ALIO. © ALT 8 ALS ALIY ALIS = 6 0 110 120 130 140 150 160 Takeoff Weight [1000 Ib]

(a) Altitude impact, north far monitor

(b) Weight impact, north far monitor

(c) Thrust impact, north far monitor

Correlation: 0.019 Linear Regression Slope: 0.014 85, Tow 70 ALY ALA ALI = 65 ALQ— ALIO. © ALT 2 AL ALIT ALIS 60 75 12.5 175 22.5 275 Thrust per Engine [1000 Ib]

Correlation: -0.158 Linear Regression Slope: -0.311 28 70 ALY ALA ALI = 65 ALQ— ALIO. © ALT 2 AL — ALIT ALIS 60 180 200 220 240 260 280 True Airspeed [KTAS]

Correlation: -0.010 Linear Regression Slope: -0.053 ;¢8 400 serena pet Ee 6 ALT ALS ALTO. ALIQ| AQ © ALA ALT ALT 50, 230 240 250 260 270 280 290 300 True Airspeed [KTAS]

(d) True airspeed impact, north far monitor

(e) Flight path angle impact, north far monitor

(f) Temperature impact, north far monitor

Correlation: 0.028 Linear Regression Slope: 0.609 ALL ALS ALTO. ALI2) ALO > ALA © ALI © AL7 25 5.0 75 10.0 Flight Path Angle [deg]

(g) Northward wind impact, north far monitor. Northward wind positive to the north

(h) Eastward wind impact, north far monitor. Eastward wind positive to the east

(i) Relative humidity wind impact, north far monitor

Figure 15: Trends for the Airbus A320 at the north far monitor

Correlation: 0.177, Linear Regression Slope: 0.697 an 6 ALT ALS ALTO. ALIQ| ALS © ALA ALT ALT 50, -25 -20 -15 -10 5 O 5 10 15 Temperature [°C]

Correlation: 0.130 Linear Regression Slope: 1.024 ;@8 Tes ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 20, -10 oO 10 20 30 Northward Wind [kts]

Correlation: 0.196 Linear Regression Slope: 1.432 22. Tes ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 20, -10 oO 10 20 30 Eastward Wind [kts]

Correlation: 0.248 Linear Regression Slope: 0.589 4 ALL ALS ALTO. ALI2) ALO > AL4 © ALI © AL7 20 40 60 80 100 Relative Humidity [%]