EMADDC: high quality, quickly available and high volume wind and temperature observations from aircraft using the Mode-S EHS infrastructure

de Haan, Siebren; de Jong, Paul; Koutek, Michal; Sondij, Jan

doi:https://doi.org/10.5194/amt-2024-110

Preprints

https://doi.org/10.5194/amt-2024-110

Preprints

28 Jun 2024

| 28 Jun 2024

Status: this preprint is currently under review for the journal AMT.

EMADDC: high quality, quickly available and high volume wind and temperature observations from aircraft using the Mode-S EHS infrastructure

Siebren de Haan, Paul de Jong, Michal Koutek, and Jan Sondij

Abstract.

Temperature and wind observations from aircraft are regarded of major importance for aviation meteorology and numerical weather prediction (NWP). The European Meteorological Aircraft Derived Data Center (EMADDC) system processes aircraft surveillance data received from air traffic control (ATC) and converts it into observations of wind and temperature. Only so-called Mode-S Enhanced Surveillance data can be used, because this data contains the air vector and ground vector of the aircraft, from which a wind vector can be inferred. To acquire high quality observations, the data is processed in three steps: pre-processing, processing, and post-processing. The pre-processing is needed to obtain high-quality information and to calculate several correction values such as temperature corrections and heading corrections. Processing converts the aircraft data into meteorological information and finally post-processing guarantees that only high-quality information is made available. The quality of the observations produced is verified by comparing these observation to other upper air wind and temperature observations from radiosondes and comparing them with NWP data. This paper presents the EMADDC system, operational since 2019.

Received: 20 Jun 2024 – Discussion started: 28 Jun 2024

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Siebren de Haan, Paul de Jong, Michal Koutek, and Jan Sondij

Status: final response (author comments only)

CC1:
'Comment on amt-2024-110', Edmund Stone, 29 Jul 2024

The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2024-110/amt-2024-110-CC1-supplement.pdf

Citation: https://doi.org/10.5194/amt-2024-110-CC1
- AC4: 'Reply on CC1', Siebren de Haan, 30 Sep 2024
  
  The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2024-110/amt-2024-110-AC4-supplement.pdf
  
  Citation: https://doi.org/10.5194/amt-2024-110-AC4
RC1:
'Comment on amt-2024-110', Anonymous Referee #1, 31 Jul 2024

The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2024-110/amt-2024-110-RC1-supplement.pdf

Citation: https://doi.org/10.5194/amt-2024-110-RC1
- AC3: 'Reply on RC1', Siebren de Haan, 30 Sep 2024
  
  The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2024-110/amt-2024-110-AC3-supplement.pdf
  
  Citation: https://doi.org/10.5194/amt-2024-110-AC3
CC2:
'Comment on amt-2024-110', Ralph Petersen, 01 Aug 2024

See detailed comments in attachment. I will be happy to discuss my comments with the authors via email.

Citation: https://doi.org/10.5194/amt-2024-110-CC2
- AC5: 'Reply on CC2', Siebren de Haan, 30 Sep 2024
  
  The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2024-110/amt-2024-110-AC5-supplement.pdf
  
  Citation: https://doi.org/10.5194/amt-2024-110-AC5
RC2:
'Comment on amt-2024-110', Anonymous Referee #2, 01 Aug 2024
Given the importance that observations derived from MODE-S EHS data are taking on, both for operational meteorology and for research, and the amount of algorithm development and data processing techniques that have been necessary to achieve the ability to reliably produce so much useful data, the publication of such an article is fully justified.
The article follows a logical order, provides a synthesis between work already described, which is referenced and placed in the context of the present work, and more recent aspects, such as aircraft-dependent heading correction (§7.1). It ends with insightful characterisation of the data produced versus NWP and radiosondes.
Nevertheless:
the choice of journal remains, in my views, open to question. For example, could Earth System Science Data (ESSD) be a more appropriate choice?

In its current form, the article needs to be corrected or reworked before it can become a solid reference for all future work using these data, which it ultimately deserves.

In several places, the article lacks precision of description, and relies on the reader's implicit understanding. This needs to be corrected in a definitive publication. In particular:
- in §8 Processing Infrastructure, the text let the reader think that the duplicate removal process applies to calculated observations, and not to input (Mode-S and ADS-B) messages. This would means that each data supply channel is treated individually, and that the duplicate removal process is applied at the end, when all the groups of observations produced are merged. Is this really the case? As the article states that “A processing job starts by gathering all data available in the time window of interest.” (line 212), it seems that there is room for de-duplication of input data (for example, same MODE-S message received by 2 receivers). Is this carried out, or not ? This should be clarified.
- are the whitelistings described in §5.3 “Output control” and whitelisting at the end of §8 “Processing Infrastructure” (around line 230) the same? If so, it could be better described in one place, and simply referred to in the other one (“observations are within three times standard deviation of the measurement with NWP model equivalents” does not make much sense to me).
- Also : some processing techniques depend on assumptions (for example, magnetic declination tables or the form of corrections for static pressure, Mach number or airspeed). Overall, the final results on the quality of the measurements produced validate the work carried out and the assumptions made, but for this article to give the reader a full understanding of the measurement and processing techniques, quantified indicators should be given for the various stages. For example: are there any aircraft for which minimisation of the cost function for magnetic declination (eq. 18) does not converge? Is so, do the authors have any clues about these aircraft (particularly old or recent, or else)? What is the typical amplitude of the true air speed correction mentioned in §7.3? In addition to the ongoing research mentioned to develop a more physical method, did the authors try to check that this correction was indeed uncorrelated with a spatial characteristic, or some bias in the model, or else (for example, simply by drawing maps of typical values, or scatter plots, ...) ? What is the typical percentage of aircraft that are whitelisted? Is it evolving over time ? Is it possible to learn anything from the list of aircraft that are rejected ? Are they simply aircraft that transmit incorrect data, or are they particular types of aircraft for which other assumptions and calculation methods could be used? This could be an avenue for development if these aircraft fly where others do not.

Here are some more specific remarks, along the text :
Line 15 : "For many years, aircraft observations form the backbone of the global observing system". The wording "form the backbone" appears a bit overstated. Associated references support the value and importance of aircraft observation, but they do not assess with certainty that it is central and structuring. Could be rephrased as "aircraft observations are an essential component of the global observation system"

Line 17: De Haan, 2013 is an article which deals with Assimilation of GNSS ZTD and radar radial velocity for the benefit of very-short-range regional weather forecast. It recalls the importance of aircraft-based observation, but only marginally demonstrate it. I don’t think that its brings much here.

Line 33: Since there is a causal relationship, the wording ‘However, as (or since) some airlines have continued to fly’ seems more appropriate to me than ‘However, whilst some airlines have continued to fly’ (but I’m not a native English speaker).

Line 58-59 : You might want to quote a technical article on these decoding techniques, to give the interested reader the information they need to find out more about these difficulties and the techniques for overcoming them. For example : J. Sun, H. Vû, J. Ellerbroek and J. M. Hoekstra, "pyModeS: Decoding Mode-S Surveillance Data for Open Air Transportation Research," in IEEE Transactions on Intelligent Transportation Systems, vol. 21, no. 7, pp. 2777-2786, July 2020, doi: 10.1109/TITS.2019.2914770.

Line 66: missing reference at the end on the sentence “[…] transmitted frequently and could be used in data assimilation (?).” . For example, Bruce Ingleby mentioned such a technique in is poster at the 2023 International Symposium on Data Assimilation (ISDA-2023), titled "ECMWF use of Mode-S winds and changes to aircraft thinning."

Line 80: Since reference is made to a personal communication, details of the calculation should be given in the present text.

Line 83 : “the timestamp is supplied by the receiver and not by the aircraft” is not the root cause of the difference between receiver and radar data, since "The timestampis created at the moment of arrival of the information”, as stated in §5.1. The advantage of the radars probably rather comes from the synchronisation between the positioning, carried out when the echo from the aircraft is received, and the reception of the Mode-S message, practically simultaneously

Line 95, eq.1 : it could be worth noting that the numerical constant used here are valid for dry air, and later, for example in in §7.3, consider the possibility of controlling the applied correction in areas known to be particularly humid (boundary layer in the Mediterranean or the Canaries)

Line 104 : given the prior presentation of the difference between receiver and radar data, and even if I agree with the need to choose “The (most relevant) parameters”, I would expect table 1 to contain two different lines for positioning : (latitude-longitude) from ADS-B at 0.5-2 seconds period, and (range-azimuth) from radar at 5s - 20s period. Or EMADCC never uses radar positioning and mixes radar and ADS-B receiver to assign a position to an observation ? Also, in table 1, the headings of the ‘frequency’ and ‘reported accuracy’ columns are reversed.

Line 112: “check the input for obvious errors,” could be completed by “or measurements in conditions where calculation is not possible”.

Line 146, eq.6 : Even if their meanings can be guessed, I believe that "V" and "d" have not been formally defined before, and it would be more rigorous to do so. Please also recall quickly the hypothesis behind this formula (this is a 2D formula, not valid at large values of roll and pitch angles, which justifies the criteria roll <2.5% in table 2, and this formula assumes that the airspeed is aligned with the axis of the aircraft (the heading), and therefore that sideslip is zero, which is mostly true for airliners, but not necessarily during the aircraft's rapid manoeuvring phases)

Line 207 : after the sentence beginning with “For receiver data…”, the reader expects another one describing what is done for radar/tracker data. Or does this sentence apply to both receiver and radar data ?

Line 241 and further ; Although I know that this is common practice, the choice of the word 'error' to designate the difference between an observation and a model analysis is, to say the least, debatable. There are cases where the RMS time series of these deviations have changed significantly, without any change in the observation system, but when the model version was updated. Especially since the article later shows that "the comparison between radiosonde and Mode-S EHS show to have a standard deviation lower than that of the comparison is model and Mode-S EHS or radiosonde", which suggests that part of the variance in the MODE-S/model difference is due to a discrepancy between the model and the reality.

Figure 4 legend : an “i” is missing in “Observatons”
Line 262 : “858” is probably a typo for “850”

Line 271 : I don’t fully understand the sentence “Note that, although the data is corrected using ECMWF forecast, the data is independent because a forecast lead time of minimal 9 hours is used”. Is the forecast lead time of 9 hours used for the computation of the magnetic declination table, or the True airspeed correction mentioned in §7.3? Wouldn’t this sentence be better placed closer to the correction description ? Does it imply that the impact of Mode-S assimilation in the model forecast does not extend beyond 9 hours?
Citation: https://doi.org/10.5194/amt-2024-110-RC2
- AC1: 'Reply on RC2', Siebren de Haan, 30 Sep 2024
  
  The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2024-110/amt-2024-110-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/amt-2024-110-AC1
RC3:
'Comment on amt-2024-110', Anonymous Referee #3, 01 Aug 2024

EMADDC: high quality, quickly available and high volume wind

and temperature observations from aircraft using the Mode-S EHS infrastructure

General
The EMADDC processing of Mode-S messages into meteorological reports is a major and

very useful undertaking. I am a user of the EMADDC reports.
In places the jargon (eg ASTERIX CAT48 format) is perhaps too prominent.

I have made suggestions about this and minor improvements to the English.

The text is sprinkled with more commas than I would use.
The statistics all use flight level as the vertical coordinate - the equivalent pressure

levels should also be given (at least once).
I would like to see some discussion of the wider context, both the impact of Mode-S on NWP and

the future of Mode-S and what might replace it.

Piecing information together from messages designed for another purpose is not how one would

design a meteorological observing system. Also, from my perspective, reports every 4-seconds are

overkill. I would hope that, in the longer term it would be replaced by a a better designed

aircraft reporting system that provides high resolution data in a single report without the need

for heading corrections, rederivation of temperature from Mach number etc. I would like to see some

discussion of any moves in that direction, timescale etc and whether it needs a directive from the

EU to ensure that such a system becomes widely used over Europe and perhaps elsewhere.

Detailed
Title: I would suggest 'timely' in place of 'quickly available' and possibly

moving 'high volume' before 'high quality' - it is the number of reports that

really sets Mode-S apart from other aircraft data sources.

What does 'infrastructure' mean here? 'using Mode-S EHS messages' might be better.

1 'Temperature and wind observations from aircraft are regarded of major importance'

I suggest 'Wind and temperature ...' - the winds are more important.
3 'converts it' - 'converts them' ('data' is a plural noun)
4 'this data' - 'these data'
5 'To acquire' - 'To produce'? 'the data is' - 'the data are'
13 'for example its height, and velocity' 'its' should be 'their' or can be omitted.
15 'aircraft observations form the backbone of the global observing system'

'the backbone' is a bit too stong - 'an important part'?
19 '01/2020 2020', just '2020'
22 'The last decade' - 'Over the last decade'
25,26 'intended heading, airspeed etc.' - I think 'intended' should be deleted

(they are reporting actual heading and airspeed)
28 'the most of Mode-S' - 'most Mode-S'
32,33 'observations performed by dedicated aircraft ... (AMDAR)' - 'observations from

AMDAR aircraft' (the acronym was introduced earlier).
35 'ECMWF-IFS' explain acronym (perhaps just ECMWF, need not mention IFS?)
49,50 suggestion: 'not mandatory; fewer than 5% of aircraft respond to such interrogation requests (Strajnar, 2012) and few countries actively interrogate this register.'
51 2.1 Mode-S EHS Interrogation

I think parts of this section could be rewritten more concisely.
59 'shall be applied' - 'are applied'
75-78 'Data can be of ASTERIX CAT48 format, which is mono-radar data ...'

I struggle a bit with the jargon and whether it is useful for me and other users to know.

It might be better to put the jargon in brackets, perhaps (assuming that I have understood correctly):

'Data can be from a single radar (in ASTERIRIX CAT48 format) or multiple radars (in CAT62 format,

tyically sampled at 4 second intervals; the Mach number is at lower resolution giving derived

temperature of lower quality).'
78,79 'For this ... MUAC to develop a solution.'

Perhaps delete the first sentence and replace the second with

'EMADDC is working with EUROCONTROL MUAC to develop a solution that provides temperature data with

consistently good quality.' Also 'MUAC' explain acronym
107 'information of' - 'information on'
112 'Measurements fulfilling one of these checks are discarded' 'failing' better than 'fulfilling'
115 'Output control is necessary to obtain good quality observations.'

Please provide details (or possibly a reference) of the quality control applied.

Also some indication of the proportion of 'bad' data remaining (1% or 0.1% or whatever),

all observing systems have some gross errors. I have recently become aware of some spikes -

wind speeds much higher than in the forecast - what might be causing these?
Table 1. I think that the 'frequency' and 'reported accuracy' headings should be swapped, and

'reported precision' might be better. Do you know if values are rounded or truncated when they

are reported?
135 'pressure, at low altitude, is less accurate.' Why?
136 'an improved pressure value that' insert 'is calculated' before 'that'
198 '(minimal 15 days)' - '(at least 15 days)'
231 'is outputted' - 'is output' or 'is written'
236 '9.1 Model comparison'

There should be some mention of quality control to remove 'bad' observations (radiosondes as well

as Mode-S).
Table 4. 'flight level' - 'number' heading is missing 'bias' and 'std.dev' misplaced
248 '9.2 Comparison with Radiosondes observations' - 'Radiosonde' (delete final s)
249 'Radiosondes are regarded as the anchor observation for meteorology' delete 'the'

(For satellite soundings GNSS-RO are now more important anchor observations than radiosondes.)
250 'with some sites launching also at 06 UTC and 18 UTC' replace 'some' with 'a few'
250,251 'Due to budget optimization, the number of launches per day was decreased to one or two.'

Delete? or replace with 'Due to budget restrictions some radiosondes are only launched once a day.'

WMO GBON regards two launches per day as standard and most, but not all, European radiosondes follow

this pattern.
251 'Aircraft observations are regarded as replacement to collect upper air observations' -

'Aircraft observations are regarded as supplemental upper air observations'
251-251 'Aircraft and observations will never be collocated in both space and time, ....

avoids the balloon.' Perhaps just 'will never be exactly collocated' and delete the rest of the

senteence.
255 'of 50 km' delete 'of'
255 'The table below' - 'Table 5'
257 'show to have' - 'has'
259 'the mean difference between aircraft and balloon is small.'

Both Mode-S and radiosondes have slightly stronger mean wind speeds than 'NWP', I assume that

this is because the NWP fields are on a ~9 km grid, whereas the observations are closer to point

measurements and have a contribution to the kinetic energy from scales unresolved by the model.
Table 6. Column headings missing. Caption too brief.
264,265 'derived from Mode-S EHS aircraft observations'

'reports' or 'messages' better than 'observations' here?
269 'this heading correction is unique for each aircraft individually', delete 'individually'
271,272 'although the data is corrected using ECMWF forecast, the data is independent because a

forecast lead time of minimal 9 hours is used' ('minimal' - 'at least')

This is only partially true, if the forecast model used has a bias then this will be reflected

with reduced magnitude as a bias in the 'corrected' observations, Eyre (2016).

Because aircraft heading is not related to forecast values it seems unlikely that the heading

correction will cause this type of problem. The temperature and airspeed corrections might be

susceptible to problems from model biases. This should be mentioned.
Eyre, J.R. (2016), Observation bias correction schemes in data assimilation systems:

a theoretical study of some of their properties. Q.J.R. Meteorol. Soc., 142: 2284-2291.

https://doi.org/10.1002/qj.2819
283,284 'The change in declination is ... close to zero for low latitude regions (middle panel).'

It is confusing having deep red for very small values on this panel - would be better just to use

blue scales (white for near zero).
Figure A1. define 'WMM' or omit. Add note that the contour intervals are different for the three plots.
Appendix B.

'number of observation' - 'number of observations'
296 'in casu'?
302 '1207, E.: Commission' - 'European Commission'?
334 'Painting, J. D.: WMO AMDAR Reference Manual, WMO-No.958, WMO, Geneva, 2003.'

WMO regards this manual as superseded (although it is still available), see

https://community.wmo.int/en/activity-areas/aircraft-based-observations/resources/manuals-and-guides

Should the reference be changed? If not is WMO wrong in regarding it as superseded?

Citation: https://doi.org/10.5194/amt-2024-110-RC3
- AC2: 'Reply on RC3', Siebren de Haan, 30 Sep 2024
  
  The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2024-110/amt-2024-110-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/amt-2024-110-AC2

Siebren de Haan, Paul de Jong, Michal Koutek, and Jan Sondij

Viewed

Total article views: 1,005 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
401	138	466	1,005	15	11

HTML: 401
PDF: 138
XML: 466
Total: 1,005
BibTeX: 15
EndNote: 11

Views and downloads (calculated since 28 Jun 2024)

Month	HTML	PDF	XML	Total
Jun 2024	45	25	4	74
Jul 2024	145	37	8	190
Aug 2024	111	33	111	255
Sep 2024	41	22	3	66
Oct 2024	35	12	129	176
Nov 2024	16	6	187	209
Dec 2024	8	3	24	35

Cumulative views and downloads (calculated since 28 Jun 2024)

Month	HTML	PDF	XML	Total
Jun 2024	45	25	4	74
Jul 2024	145	37	8	190
Aug 2024	111	33	111	255
Sep 2024	41	22	3	66
Oct 2024	35	12	129	176
Nov 2024	16	6	187	209
Dec 2024	8	3	24	35

Viewed (geographical distribution)

Total article views: 1,009 (including HTML, PDF, and XML) Thereof 1,009 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 14 Dec 2024

Short summary

This manscript describes the operational method of extracting meteorological information from all airborne aircraft in the European airspace every 20 seconds to 1 minute. The methodology is described and the quality of the wind and temperature observations are evaluated against radiosonde observations.


Total:	0
HTML:	0
PDF:	0
XML:	0