Analysis and preprocessing of Czech Mode-S observations · ATR 42 Boeing 787 ATR 72 Boeing 767 Boeing 747 Airbus A340 Fokker 100 Canadair CRJ Airbus A380 Boeing 757 Bombardier Dash

Analysis and preprocessing of CzechMode-S observations

Benedikt Strajnar, Slovenian Environment Agency (ARSO)

in cooperation with Alena Trojáková, CHMI

Stay report, CHMI Prague, 6-24 July 2015

1 IntroductionModern surveillance techniques allows air traffic controls (ATC) to collect a large num-ber of flight parameters from aircraft in real time. Among the available communicationtechniques, Mode-S (selective mode communication between ATC and airplane) is re-alized with a ground requesting radar (the so-called secondary radar), transponders onboard the aircraft, and Mode-S ground sensors to collect the replies. These systems areof interest for meteorology because it is possible to extract meteorological information.This can be done either by using mandatory Mode-S data, the Enhanced Surveillance(EHS) parameters which are available in any Mode-S system (de Haan, 2011) or by di-rectly requesting wind and temperature messages contained in a non-mandatory Mode-Sdata register Meteorological Routine Air Report (MRAR) (Hrastovec and Solina, 2013;Strajnar, 2012).

Both Mode-S EHS and MRAR methods have their advantages and disadvantages. Agreat advantage of Mode-S EHS observations is that EHS parameters are available inany Mode-S equipped ATC system and from all Mode-S equipped aircraft. However,the computation is indirect, especially for temperature, which can only be computedthrough speed of sound equation from Mach number and airspeed (de Haan, 2011). Thewind is computed from a vector difference between air speed (including heading) andground speed. The magnetic heading in particular has been shown to include aircrafttype specific biases. A correction procedure was proposed by (de Haan, 2011) and thenrefined in (de Haan, 2013). On the other hand, Mode-S MRAR includes direct wind andtemperature observations and whose quality was shown to be the same as the qualityof Automated Meteorological Data Relay (AMDAR) on the AMDAR-equipped aircraft(Strajnar, 2012). However, because this data are not mandatory in a standard Mode-Ssystem, the ATC radars must be configured to interrogate an additional MRAR registercontaining this data (where this is possible at all with respect to the density of airtraffic). The support for MRAR also depends on the type of the transponder on boardthe aircraft. In the Slovenian case (Strajnar, 2012), MRAR was available from 5% of allMode-S reports.

Since few years, Mode-S observations (both EHS and MRAR) have been collectedby Air Navigation Services of Czech Republic (ANS-CZ). The data was recently madeavailable for evaluation also to Czech Hydrometeorological Institute (CHMI). This workis devoted to a basic analysis and preprocessing of these Mode-S observations. We firstdescribe the data and provide examples of flight profiles. As a next step, raw Mode-SMRAR and Mode-S EHS (as computed for the moment at ANS CZ) are compared toAMDAR in a colocation study. A further evaluation is provided by a comparison toALADIN/CZ numerical weather prediction model. This comparison enables an evalua-tion for each Mode-S observation (also non-AMDAR equipped aircraft). Description ofscripts, software and data can be found last in the Appendix A.

2 Data description

2.1 Mode-S data set

The Czech Mode-S data includes 3 Czech Mode-S radars Praha/Ruzyně, Písek (Jince)and Buchtův kopec (Sněžné). Those provide Mode-S EHS every 4 s and Mode-S MRARevery 10 seconds. Mode-S EHS is also available form 2 additional radars near Auersberg

1

Figure 1: (left) Mode-S MRAR and (right) Mode-S EHS coverage in 24 hours (from JanKráčmar and Jiří Frei, ANS CZ). Note the much higher density and slightly larger extentof Mode-S EHS.

(D) and Bratislava (SK). The horizontal coverage is shown in Fig. 1. There are around3.5 million Mode-S EHS and around 140 thousand Mode-S MRAR observations per day.The time period used for analysis is 12 June - 9 July 2015.

2.2 ICAO aircraft addresses

Within Mode-S, ICAO address is provided as an unique aircraft identification. It is as-signed by flight authorities in each country. Consequently, no centralized public databaseis publicly available. It is possible to relate ICAO address with aircraft information byonline resources (e.g. various live flight tracking or plane spotter websites). Anotherpossibility is to retrieve aircraft information from flight plans available at ATC. We useda combination of both to identify as many ICAO addresses as possible (more than 8000of around 9000 different aircraft). Figure 2 shows the distribution of Mode-S EHS byaircraft type (without detailed subtypes). It can be noted that most frequent type isBoeing 737 followed by Airbus 320,319 and 321. The most frequent aircraft reportingMode-S MRAR is Canadair Regional Jet (CRJ).

2.3 Flight profiles

The easiest way to observe properties of Mode-S data sets is by plotting profiles of mea-sured variables during the flight. Figure 3 shows an ascent and a descent of a CRJ aircraftwhich provides both Mode-S EHS and Mode-S MRAR observations. For wind, a goodagreement between Mode-S EHS and MRAR can be seen, although there is a small sys-tematic difference (smaller wind speeds in EHS observations). MRAR wind time seriesseems to be slightly more smooth, but oscillations of about 1-2 m/s can be observed bothtypes. In the case of temperature (right part of Fig. 3 ), EHS is much more noisy andoscillates around MRAR values.

Another example of temperature profile demonstrates sensitivity of EHS temperatureson ground speed (which is an input in EHS calculations, Fig. 4). Figure shows MRAR

2

Pilatus PC

Cessna 510 Citation Mustang

Avro RJ100

McDonnell Douglas MD

Fokker 70

Airbus A318

Airbus A300B4

ATR 42

Boeing 787

ATR 72

Boeing 767

Boeing 747

Airbus A340

Fokker 100

Canadair CRJ

Airbus A380

Boeing 757

Bombardier Dash 8 Q400

Embraer ERJ

Airbus A330

Boeing 777

Airbus A321

Airbus A319

Airbus A320

Boeing 737

Number of Mode−S EHS observations per day

Number [million]

0.0 0.2 0.4 0.6 0.8 1.0

Figure 2: Number of Mode-S EHS observations from different aircraft types (only the 25most frequent are shown). The aircraft which also provide MRAR are plotted in red.

temperature profile in blue and EHS temperatures in green. The oscillations of EHStemperature in this case are large (5-10 K). There are also some unphysical values whichare related to errors in ground speed values (red line). The ground speed is constantlydecreasing over time but there are a few zero values in the time series, causing unrealistictemperature calculation. A filtering would be needed to remove such ground speedsfrom the data set. A time smoothing is also suggested to suppress oscillations in EHStemperature (as already done by de Haan (2011)). EHS temperature calculation mayespecially be problematic at low ground and air speeds.

3

Flight profile

4000

5000

6000

7000

8000

9000

1000

0Al

titud

e(m

)

180

190

200

210

220

Airc

raft

spee

d (m

/s)

05

1015

Win

d (m

/s)

2030

4050

6070

80Ai

rcra

ft he

adin

g (°

)Time

Altitude

Mode-S EHSAircraft SpeedAircraft heading

Mode-S MRAR

Flight profile

4000

5000

6000

7000

8000

9000

10000

Altitude(m

)

0.0

0.2

0.4

0.6

0.8

1.0

Aircra

ft s

peed (

m/s

)

−45

−40

−35

−30

−25

−20

−15

Tem

pera

ture

[°C

]

20

30

40

50

60

70

80

Aircra

ft h

eadin

g (

°)

Time

Altitude

Mode-S EHS

Mode-S MRAR

Aircraft heading

Figure 3: Evolution of (left) wind and (right) temperature (and other parameters) duringascent and descent. Mode-S EHS is shown in green, Mode-S MRAR in blue.

Landing profile

1000

2000

3000

4000

Altitu

de(m

)

020

4060

8010

012

014

0Air

craft s

peed

(m/s)

−50

510

1520

25Te

mpera

ture (

degre

es)

050

100

150

200

250

300

350

Aircra

ft hea

ding (

°)

10:00 12:00 14:00 16:00 18:00 20:00 22:00Time (minute)

Altitude

Mode-S EHS

Aircraft SpeedAircraft headingMode-S MRAR

Figure 4: Evolution of EHS and MRAR temperatures and other reported parametersduring descent. Aircraft is ATR 42, Czech Airlines.

3 Colocation of AMDAR, Mode-S EHS and Mode-SMRAR

To validate Mode-S observations other meteorological observations which are close inspace and time are needed. An ideal observational reference for Mode-S is AMDAR

4

Figure 5: Vertical and temporal distribution of colocated points.

which is available on several types of aircraft. Although AMDAR and Mode-S obser-vations originate from the same sensors, they differ due to the preprocessing or simplythe reporting frequency and temporal averaging. Such a comparison can not highlightpossible systematic instrumental errors affecting both observation sources.

To find Mode-S EHS/MRAR and AMDAR observation pairs (colocated observations)we allowed a time mismatch of 30 s, maximal height difference of 50 m and 10 kmhorizontal separation. Such a tolerance is essential because AMDAR reports are notavailable so frequently as Mode-S. Due to preprocessing (e.g. smoothing or averaging)the exact time and space of the AMDAR measurement does not fully agree with flightdata from Mode-S. If more possible pairs are found within these criteria the closest inspace is selected. Figure 6 shows time and space distribution of colocated observationpairs. There were around 73200 EHS - AMDAR pairs and around 4850 MRAR - AMDARpairs, slightly depending on the variable.

3.1 General difference statistics

A general comparison between AMDAR and Mode-S is shown in Fig. 7. The differencebetween Mode-S and AMDAR are mostly normally distributed. The spread of differencesagainst EHS is much larger than the spread against MRAR, both for wind and temper-atures. The wind difference distribution even shows a two peak distribution. To verifythis result, the statistics are shown against the subset of MRAR observations (around

5

Distribution by height

Height [m]

Fre

quen

cy

0 2000 6000 10000

040

0080

0012

000

Distribution by height difference

Height difference [m]

Fre

quen

cy

−40 −20 0 20 40

050

0010

000

1500

0

Distribution by horizontal distance

Distance [km]

Fre

quen

cy

0 5 10 15

050

0015

000

Distribution by time separation

Time difference [s]

Fre

quen

cy

−30 −20 −10 0 10 20 30

020

0060

0010

000

Figure 6: Distribution of Mode-S EHS/MRAR - AMDAR matches by (a) height, and (b)vertical, (c) horizontal and (d) time separation.

6

Mode−S EHS − AMDAR

Temperature [K] raw data, N = 73211 , mean= 0.34 , sd= 1.29 K

maxrange= −300.4 −> 141.8

Num

ber

−6 −4 −2 0 2 4 6

020

0060

0010

000

Mode−S MRAR − AMDAR

Temperature [K] raw data, N = 4851 , mean= −0.09 , sd= 0.32 K

maxrange= −3.3 −> 4.7

Num

ber

−6 −4 −2 0 2 4 6

050

010

0015

0020

00

Mode S EHS − AMDAR

Wind speed [m/s] raw data, N = 71296 , mean= 0.35 , sd= 2.13 m/s

maxrange= −24.2 −> 485.4

Fre

quen

cy

−6 −4 −2 0 2 4 6

010

0030

0050

0070

00

Mode S MRAR − AMDAR

Wind speed [m/s] raw data, N = 4829 , mean= 0 , sd= 0.7 m/s

maxrange= −5.7 −> 5.1

Fre

quen

cy

−6 −4 −2 0 2 4 6

050

010

0015

0020

00


Wind direction [Deg] raw data, N = 72500 , mean= 0.34 , sd= 18.11 Deg

maxrange= −180 −> 179.9

Fre

quen

cy

−150 −100 −50 0 50 100 150

050

0010

000

1500

0



maxrange= −161.8 −> 100.6

Fre

quen

cy

−150 −100 −50 0 50 100 150

050

010

0015

0020

0025

00

Figure 7: Histograms of (left) Mode-S EHS and (right) Mode-S MRAR - AMDAR differ-ences. Shown are (top) temperature , (middle) wind speed and (bottom) wind direction.To compute histograms, only the displayed value range is shown and extreme values areindicated below the plots.

4850 cases, 8). Here, the calculation is made for the same aircraft reporting both kinds ofobservations. While most of the distributions do not change, wind speed for EHS seemsto become somewhat biased. This may be linked to the type of the aircraft (in this casemostly CRJ as explained later) and the errors in either ground or air speed or magneticheading.

3.2 Type dependent statistics

The colocated observations stem from different aircraft types. Figure 9 presents thedistribution of collocated data set between different aircraft types. It can be observedthat colocated Mode-S EHS observations come mostly from Airbus A319, A320 and A321,and that Mode-S MRAR comes from CRJ aircraft. It is important to keep in mind thatthe colocation result will be limited only to these aircraft types.

In Figs. 10,11,12 the differences are shown separately for aircraft types. Because the

7

Mode−S EHS − AMDAR

Temperature [K] raw data, N = 4780 , mean= 0.52 , sd= 1.36 K

maxrange= −12.5 −> 141.8

Num

ber

−6 −4 −2 0 2 4 6

020

040

060

0

Mode−S MRAR − AMDAR

Temperature [K] raw data, N = 4851 , mean= −0.09 , sd= 0.32 K

maxrange= −3.3 −> 4.7

Num

ber

−6 −4 −2 0 2 4 60

500

1000

1500

2000


Wind speed [m/s] raw data, N = 4783 , mean= 0.71 , sd= 1.67 m/s

maxrange= −6.9 −> 10.8

Fre

quen

cy

−6 −4 −2 0 2 4 6

010

020

030

040

050

060

0


Wind speed [m/s] raw data, N = 4829 , mean= 0 , sd= 0.7 m/s

maxrange= −5.7 −> 5.1

Fre

quen

cy

−6 −4 −2 0 2 4 6

050

010

0015

0020

00



maxrange= −179.9 −> 172.1

Fre

quen

cy

−150 −100 −50 0 50 100 150

020

040

060

080

012

00



maxrange= −161.8 −> 100.6

Fre

quen

cy

−150 −100 −50 0 50 100 150

050

010

0015

0020

0025

00

Figure 8: Same as Fig. 7 except that Mode-S EHS comparison is only displayed for thesubset where MRAR is also available.

8

Airbus A300B4Airbus A310Airbus A318Airbus A319Airbus A320Airbus A321Airbus A330Airbus A340Airbus A380

ATR 42ATR 72

Avro RJ100Beech C90GTi King Air

Beech C90GTX King AirBoeing 717Boeing 737Boeing 747Boeing 757Boeing 767Boeing 777Boeing 787

Bombardier Challenger 300Bombardier CRJ

Bombardier Dash 8 Q400Bombardier Global 6000

Canadair CRJCessna 510 Citation MustangCessna 525B CitationJet CJ3

Cessna 525 CitationJet CJ1Cessna 560XLS Citation Excel

Dassault Falcon 2000LXDassault Falcon 900EX

Embraer ERJEmbraer Legacy 600

Fokker 100Fokker 70

Gulfstream 550Gulfstream G550

Hawker 800XPiMcDonnell Douglas MD

Piper PAPiper PA44Saab 2000

Number of colocated Mode−S EHS by type

Number of obs. (T or wind)

0 5000 10000 15000 20000

Beech C90GTX King Air

Bombardier CRJ

Canadair Challenger 604

Canadair CRJ

Dassault Falcon 900EX

Hawker Beechcraft 400XP

Saab 2000

Number of colocated Mode−S MRAR by type

Number of obs. (T or wind)

0 1000 2000 3000 4000

Figure 9: Number of (left) Mode-S EHS and (right) Mode-S MRAR by aircraft type.

MRAR colocated set is dominated only by CRJ, these plots are shown only for Mode-SEHS. The interpretation should be limited to the above mentioned Airbus types andCRJ because the other types are rarely colocated with AMDAR over the region. A largetemperature error standard deviations of around 10 K can be seen for A319-A321 whiletheir biases are reasonable but positive. A large warm bias appears for A318. In the caseof wind speed, the errors of A319 seem to be larger that for the other Airbuses and theyalso include significant biases.

9


ATR 42ATR 72





Canadair CRJCessna 510 Citation MustangCessna 525A CitationJet CJ2Cessna 525B CitationJet CJ3


Cessna 560XLS Citation Excel +Dassault Falcon 2000LXDassault Falcon 900EX


Eurocopter EC135 P2Fokker 100

Fokker 70Gulfstream 550

Gulfstream G550Hawker 800XPi

McDonnell Douglas MDPiper PA

Piper PA44Saab 2000

Mode−S EHS vs. AMDAR

Temperature standard deviation [K]

0 20 40 60 80 100 120 140


ATR 42ATR 72





Canadair CRJCessna 510 Citation MustangCessna 525A CitationJet CJ2Cessna 525B CitationJet CJ3


Cessna 560XLS Citation Excel +Dassault Falcon 2000LXDassault Falcon 900EX


Eurocopter EC135 P2Fokker 100

Fokker 70Gulfstream 550

Gulfstream G550Hawker 800XPi

McDonnell Douglas MDPiper PA

Piper PA44Saab 2000


Temperature mean [K]

−100 −50 0

Figure 10: (left) Standard deviation and (right) mean temperature difference betweenMode-S EHS and AMDAR by aircraft type.


ATR 42ATR 72






Cessna 560XLS Citation ExcelDassault Falcon 2000LXDassault Falcon 900EX


Fokker 100Fokker 70





Wind speed standard deviation [m/s]

0 5 10 15 20


ATR 42ATR 72








Fokker 100Fokker 70





Wind speed mean [m/s]

−40 −20 0 20 40

Figure 11: (left) Standard deviation and (right) mean wind speed difference betweenMode-S MRAR and AMDAR by aircraft type.

10


ATR 42ATR 72








Fokker 100Fokker 70





Wind direction standard deviation [deg]

0 10 20 30 40 50 60


ATR 42ATR 72








Fokker 100Fokker 70





Wind direction mean [deg]

−50 0 50

Figure 12: (left) Standard deviation and (right) mean wind direction difference betweenMode-S MRAR and AMDAR by aircraft type.

Table 1: Thresholds used to generate white list of aircraft reporting Mode-S MRAR.variable number of observations mean std.

temperature (K) 3000 < 1 K < 2 Kwind speed (m/s) 3000 < 1 m/s < 5 m/s

wind direction (deg) 3000 < 10◦ < 100◦

4 Validation against NWPThe comparison against AMDAR offers a good observations reference to Mode-S data butis limited to AMDAR-equipped aircraft. An evaluation of data from all aircraft is possibleonly by comparison against NWP model. This allows sufficiently large samples for eachsingle aircraft and therefore a reliable error estimates. However, such a comparison islimited by forecast and model errors. In this study we used ALADIN/CZ forecasts ofvarious lengths, depending on the hour of day. As ALADIN/CZ performs an analysisevery 6 hours, we used forecasts of lengths 6-11 hours. Analysis was avoided becauseAMDAR observations are also used, which may in some cases prevent fair comparison.We assume that the method is robust with respect to slight decrease of the quality offorecasts with range. To find the model counterpart of each Mode-S observation weused screening configuration of the ALADIN model (e002) and saved obs-minus-guessdepartures stored in the observational database (ODB).

The analysis was first performed for Mode-S MRAR. Partly following Strajnar et al.(2015), criteria were defined to regard measurements from an aircraft as of a good quality(Table 1). Because wind speed and direction are measured separately by measuring airspeed and heading, criteria with respect to those wind variables were preferred over u andv wind components. To achieve this, the wind departures were first recalculated fromobservations and u and v component differences which are output of screening.

11

Figure 13 shows difference statistics computed from aircraft on the white list, dis-tributed by type. For example, the turboprop ATR aircraft (often flying in the Czechairspace and reporting MRAR) are not of sufficient quality and thus not displayed. Thefinal white list includes 127 aircraft addresses for temperatures and 116 addresses forwind.

The same procedure can also be applied to Mode-S EHS. The aggregation of thestatistics takes longer because the data set is large (> 30 GB of data). For tempera-ture, 2036 out of 7278 aircraft pass the criteria defined in Table 1. For wind, the whitelists was not yet computed because the model departures are so far only defined for uv components. Figure 14 shows the number of observations per observation type forwind and temperature and Fig. 15 displays the distributions of Mode-S EHS - ALADINdifferences computed separately for each aircraft (wind is shown only as u components)within thresholds defined in Table 1. This suggests that for a large part of aircraft thequality of Mode-S EHS is also potentially acceptable. To create a white list of Mode-SEHS observations to be used in data assimilation experiments, the Mode-S - ALADINdifferences should be recomputed in the form of wind speed and direction, like in the caseof Mode-S MRAR.

12

AMDBeech 390 Premier

Bombardier BD100 Challenger 300Bombardier Challenger 300

Bombardier CRJCanadair Challenger 604Canadair Challenger 850

Canadair CL601 ChallengerCanadair CL604 ChallengerCanadair CL605 Challenger

Canadair CRJCanadair CRJ700

Cessna 525 CitationJet CJ1+Cessna 560 Citation Encore+Cessna 560XL Citation ExcelCessna 560XL Citation XLS+

Cessna 560XLS Citation ExcelCessna 560XLS Citation Excel +

Dassault Aviation Falcon 2000Dassault Falcon 2000

Dassault Falcon 900Dassault Falcon 900EX

Gulfstream G200 GalaxyHawker 800XPHawker 900XP

Learjet 35ALearjet 60

Raytheon 390 PremierSaab 2000

Selected Mode−S MRAR by type

Std. of Mode−S − ALADIN temperature difference [K]

0.0 0.5 1.0 1.5 2.0 2.5 3.0

AMDBeech 390 Premier

Bombardier BD100 Challenger 300Bombardier Challenger 300

Bombardier CRJCanadair Challenger 604Canadair Challenger 850



Cessna 525 CitationJet CJ1+Cessna 560 Citation Encore+Cessna 560XL Citation ExcelCessna 560XL Citation XLS+

Cessna 560XLS Citation ExcelCessna 560XLS Citation Excel +

Dassault Aviation Falcon 2000Dassault Falcon 2000

Dassault Falcon 900Dassault Falcon 900EX

Gulfstream G200 GalaxyHawker 800XPHawker 900XP

Learjet 35ALearjet 60

Raytheon 390 PremierSaab 2000


Mean Mode−S − ALADIN temperature difference [K]

−1.0 −0.5 0.0 0.5 1.0

AMDBeech 200GT Super King Air

Beech 300 Super King Air 350Beech B200GT Super King Air

Beech C90GTi King AirBeech C90GTi KIng Air

Beech C90GTX King AirBombardier BD100 Challenger 300


Canadair Challenger 604Canadair Challenger 850



Cessna 525A Citation CJ2+Cessna 525A CitationJet CJ2+

Cessna 525B CitationJet CJ3Cessna 525 CitationJet CJ1+Cessna 560 Citation Encore+Cessna 560XL Citation ExcelCessna 560XL Citation XLS

Cessna 560XL Citation XLS+Cessna 560XLS Citation Excel

Cessna 560XLS Citation Excel +Dassault Aviation Falcon 2000

Dassault Falcon 2000Dassault Falcon 900

Dassault Falcon 900EXGulfstream G100

Gulfstream G200 GalaxyHawker 750

Hawker 800XPHawker 800XPiHawker 900XP

Learjet 60Piper PA

Saab 2000


Std. of Mode−S − ALADIN wind speed difference [m/s]

0 1 2 3 4 5

















Learjet 60Piper PA

Saab 2000


Mean Mode−S − ALADIN wind speed difference [m/s]

−2 −1 0 1 2

















Learjet 60Piper PA

Saab 2000


Std. of Mode−S − ALADIN wind direction difference [deg]

0 20 40 60 80 100

















Learjet 60Piper PA

Saab 2000


Mean Mode−S − ALADIN wind direction difference [deg]

−10 −5 0 5 10

Figure 13: (left) Standard deviation and (right) mean difference between Mode-S MRARand ALADIN forecast by aircraft type for (top) temperature, (middle) wind speed and(bottom) wind direction.

13

AirbusAirbus A300B4Airbus A300C4

Airbus A310Airbus A318Airbus A319Airbus A320Airbus A321Airbus A330Airbus A340Airbus A350Airbus A380

Airbus KC2 Voyager (A330AirbusA320Boeing 717Boeing 737

Boeing 737 Boeing 747Boeing 757Boeing 767Boeing 777Boeing 787

Bombardier Challenger 300Canadair CL605 Challenger


Embraer EMBB500 Phenom 100Embraer ERJ

Fokker 100McDonnell Douglas MD

Selected Mode−S EHS by type

Number of temperature obs. [million]

0 5 10 15

AirbusAirbus A300B4Airbus A300C4

Airbus A310Airbus A318Airbus A319Airbus A320Airbus A321Airbus A330Airbus A340Airbus A350Airbus A380

Airbus KC2 Voyager (A330ATR 42ATR 72

Beech 300 Super King Air 350Boeing 737Boeing 747Boeing 757Boeing 767Boeing 777Boeing 787



Canadair CL604 ChallengerCanadair CL605 Challenger


Cessna 172S Skyhawk SPCessna 208B Grand CaravanCessna 510 Citation Mustang

Cessna 525A Citation CJ2+Cessna 560XL Citation XLS+

Cessna 680 Citation SovereignCessna T182T Turbo Skylane

Cessna T206H Turbo StationairCirrrus SR22Cirrus SR22

Dassault Falcon 2000LXDassault Falcon 2000S

Diamond DAEmbraer ERJ

Fokker 100Fokker 70

Gulfstream G550Learjet 60

Let LMooney M20TN

Raytheon 390 PremierTecnam P2006T

Selected Mode−S EHS by type

Number of wind obs. [million]

0 5 10 15

Figure 14: (left) Number of Mode-S EHS (left) temperature and (right) wind observationsby aircraft type.

5 Conclusions and outlookIn this preliminary evaluation of Mode-S data available over the Czech airspace andsurroundings, we demonstrated the observation quality with respect to AMDAR obser-vations and ALADIN NWP model, separately for each reporting aircraft or aircraft type.The difference in availability and quality between Mode-S MRAR and Mode-S EHS wasalso demonstrated. While Mode-S MRAR observations are of overall good quality evenwithout preprocessing, they can be used for meteorological applications including data as-similation just after the basic data selection based on statistics of differences with respectto ALADIN model.

In the case of Mode-S EHS a larger variability in the data and their errors was ob-served, and these could be further improved by smoothing the data or applying correctionsto temperature wind speed and wind direction, as done at KNMI (de Haan, 2013). Itappears that Mode-S EHS generally include a slight warm bias. The possible future stepsinclude:

• checking for unrealistic air speeds resulting in unphysical temperatures in somecases (see Fig. 4)

• temporal smoothing of variables which appear in EHS temperature calculation(Mach number, air speed) and recomputing EHS temperatures, investigating theeffect on warm bias

• computation of magnetic heading correction separately for each aircraft and re-computation of EHS wind observations, optional temporal smoothing (currentlynot performed at KNMI)

• repeated colocation with AMDAR and evaluation of impact of such corrections

14

Mode−S EHS − ALADIN distribution by aircraft

Std. temperature difference [K]

Fre

quen

cy

1.0 1.2 1.4 1.6 1.8 2.0

010

020

030

040

050

0


Mean temperature difference [K]

Fre

quen

cy

−1.0 −0.5 0.0 0.5 1.0

010

020

030

040

0


Std. u wind component difference [m/s]

Fre

quen

cy

2.0 2.5 3.0 3.5 4.0 4.5 5.0

010

020

030

040

0


Std. u wind component difference [m/s]

Fre

quen

cy

2.0 2.5 3.0 3.5 4.0 4.5 5.0

010

020

030

040

0

Figure 15: Distributions of (left) standard deviation and (right) mean Mode-S EHS -ALADIN differences computed separately for each aircraft. Shown are (top) temperatureand (bottom) u wind component.

15

• repeated comparison against ALADIN model, recalculation of wind speed and di-rection difference from wind components and data selection based on thresholdsused in Table 1 to generate a white list for EHS observations.

The data assimilation experiments should be carried out first with Mode-S MRARand then Mode-S EHS. Recent studies suggested that improvements in the first few hoursof forecasts can be expected (de Haan and Stoffelen, 2012; Strajnar et al., 2015). As thequality of EHS temperatures were shown to be less than the quality of wind and if thisdoes not improve significantly with further correction it may be reasonable to assimilateEHS winds only on top of all Mode-S MRAR observations.

AcknowledgmentWe thank ANS CZ for providing Mode-S observations used in this work. Jan Kračmarand Jiří Frei helped with description of data, discussion and reconstructing the aircraftdetails from flight plans. We are also grateful for organizing the visit to ANS CR. Manythank to colleagues from CHMI for support during the stay.

6 Referencesde Haan, S. (2011). High-resolution wind and temperature observations from aircrafttracked by Mode-S air traffic control radar. J. Geophys. Res., 116.

de Haan, S. (2013). An improved correction method for high quality wind and tempera-ture observations derived from mode-s ehs. Technical report.

de Haan, S. and Stoffelen, A. (2012). Assimilation of high-resolution Mode-S wind andtemperature observations in a regional NWP model for nowcasting applications. Wea.Forecasting, 27:918–937.

Hrastovec, M. and Solina, F. (2013). Obtaining meteorological data from aircraft withMode-S radars. Aerospace and Electronic Systems Magazine, 28(12).

Strajnar, B. (2012). Validation of Mode-S Meteorological Routine Air Report aircraftobservations. J. Geophys. Res., 117.

Strajnar, B., Žagar, N., and Berre, L. (2015). Impact of new Mode-S MRAR observationsin a mesoscale model. J. Geophys. Res., 120(9).

A Scripts and data setsThe following scripts and data sets were prepared on server yaga, user mma233. Thedata sets are archived on directory /̃work/data/modes (Table A). Scripts are locatedunder /̃scripts/R and /̃python and possible their sub directories (Table 3).

16

Table 2: List of data sets.data set descriptioncolocated_AMDAR csv files containing colocated AMDAR, EHS and

MRAR observational datascreened_mrar_csv csv files containing screened MRAR observationsscreened_ehs_csv csv files containing screened EHS observationsmrar_merged_1h csv files with merged MRAR observations,

ALADIN departures and aircraft detailsehs_merged_1h csv files with merged EHS observations,

ALADIN departures and aircraft detailsehs_merged_byICAO csv files with merged EHS observations,

separate file for each aircraft address

Table 3: List of R and python scripts.

script purposeprofiles.R plot flight profiles

compute_colocations.R compute colocated AMDAR and Mode-S points,merge both information and outprint

analyse_amdar_colocation.R compute and plot statistics from colocated dataanalyse_amdar_colocation_by_type.R compute and plot type-dependent

statistics from colocated datamerge_ModeS_ehs_NWP.R prepare merged data sets for EHSmerge_ModeS_mrar_NWP.R prepare merged data sets for MRAR

create_whitelists_and_stats_mrar.R prepare statistics and create white lists for MRARcreate_whitelists_and_stats_ehs.R prepare statistics and create white lists for EHS

(not fully completed)create_stats_ehs_by_address.R create files with NWP difference statistics

separately for each aircraft addresssuperobs.R create smoothed super observations for MRAR

(not used in this work)split_ehs_by_ICAOaddr.py split EHS data by aircraft registrationcreate_screened_csv_ehs.py creation of screened EHScreate_screened_csv_mrar.py create of screened MRAR

ecma2modescsv.py combine variables into one row for asingle observation, called by the two above scripts

17

Documents

Analysis and preprocessing of Czech Mode-S observations · ATR 42 Boeing 787 ATR 72 Boeing 767 Boeing 747 Airbus A340 Fokker 100 Canadair CRJ Airbus A380 Boeing 757 Bombardier Dash