OpenSky Report 2024: Analysis of Global Flight Contrail Formation and Mitigation Potential

Abstract

With the increasing coverage of OpenSky, flight data gathered from the network of receivers has become a primary source of open data for aviation research. This year marks ten years of OpenSky. In this report, we employ a year's worth of OpenSky data to analyze the formation of contrails and study the potential mitigation with altitude diversions. Persistent contrails, often formed under humid atmospheric conditions, significantly trap outgoing terrestrial radiation. More insights into contrails are essential for studying aviation climate impact. We estimated the potential formation of persistent contrails based on the numerical weather assimilation data. An efficient approach is employed to fuse meteorology data in our analysis, which allows the fast evaluation of these contrail conditions at a very large scale. We designed a simple yet effective algorithm for flight with persistent contrails to study the shortest altitude diversion that would have prevented the contrail formations. We have estimated that between 5% and 9% of total flight distances each month could have formed persistent contrails. Furthermore, altitude diversions could have mitigated 70% of these contrails.

Full text

OpenSky Report 2024: Analysis of Global Flight
Contrail Formation and Mitigation Potential
Junzi Sun∗§ , Xavier Olive∗†, Esther Roosenbrand§, Céline Parzani†, Martin Strohmeier∗‡
∗OpenSky Network
Switzerland
lastname@opensky-network.org
§Faculty of Aerospace Engineering
Delft University of Technology
Delft, The Netherlands
†ONERA DTIS
Université de Toulouse
Toulouse, France
‡armasuisse W+T
Switzerland
Abstract—With the increasing coverage of OpenSky, flight data
gathered from the network of receivers has become a primary
source of open data for aviation research. This year marks ten
years of OpenSky. In this report, we employ a year’s worth of
OpenSky data to analyze the formation of contrails and study the
potential mitigation with altitude diversions. Persistent contrails,
often formed under humid atmospheric conditions, significantly
trap outgoing terrestrial radiation. More insights into contrails are
essential for studying aviation climate impact. We estimated the
potential formation of persistent contrails based on the numerical
weather assimilation data. An efficient approach is employed
to fuse meteorology data in our analysis, which allows the fast
evaluation of these contrail conditions at a very large scale. We
designed a simple yet effective algorithm for flight with persistent
contrails to study the shortest altitude diversion that would have
prevented the contrail formations. We have estimated that between
5% and 9% of total flight distances each month could have formed
persistent contrails. Furthermore, altitude diversions could have
mitigated 70% of these contrails.
Index Terms—Sustainability, contrails, meteorology, OpenAP,
optimization, OpenSky, open data
I. INTRODUCTION
Founded in 2013, The OpenSky Network [1] operates as
a crowdsourced network of ADS-B receivers, consistently
gathering surveillance data from equipped aircraft and offering
it for scientific purposes. Over the years, its coverage has
continuously expanded, with currently over 6,000 sensors reg-
istered worldwide. This platform has played a significant role
in facilitating research across various domains, including radio
frequency, signal security, and many topics related to air traffic
management, including safety, efficiency, and sustainability of
air transport.
A. OpenSky supported aviation sustainability research
With facilitated access to crowdsourced data, researchers
have been explicitly improving the transparency of data-driven
studies, especially for aviation sustainability-related topics,
including aircraft emissions [2], [3], operation efficiencies [4],
[5], [6], and air transport’s climate impact [7]. Several recent
research studies have also combined OpenSky data and remote
sensing data to analyze contrail formations [8], [9], showing
potential adoption for open-source data in contrail analysis.
In another recent study [10], some authors of this paper
combined OpenSky data with global radiosonde measurements
to study the local formation of contrails at these locations
and propose simple altitude diversions that could reduce a
significant amount of persistent contrail formation. However,
the study is limited to the areas where radiosonde stations are
available. Moreover, only two hours of flights at midnight and
noon were analyzed.
B. Brief background of contrails and impacts
The overall climate impact of aviation results not only from
carbon dioxide (CO2) emissions but also for a significant part
from non-CO2effects. These include contrails and contrails
cirrus formation from emitted water vapor, which represent
one of the most significant radiative forcing contributors from
the aviation sector [11], [12]. Contrails usually appear at high
altitudes with very low ambient temperatures. They can also
persist up to a day in the region of the atmosphere where the
relative humidity with respect to ice is greater than 100% [13].
Persistent contrail clouds can either reflect sunlight back into
space during the day, leading to a cooling effect, or trap large
amounts of heat that would otherwise leave the atmosphere,
leading to a warming effect. This warming effect dominates
according to previous research [12], which suggests a net
positive global radiative forcing caused by contrails. However,
the significant uncertainties still cause debate regarding the
magnitude of the impact.
Since contrail’s effects on the climate depend on several
parameters, such as altitude and ice-supersaturated regions, one
way to reduce them could be to construct flight trajectories that
avoid areas that favor contrails being formed. Thus, one key
point is determining which flight routes will create contrails.
Recent research has focused on these contrail inventories.
Air traffic data obtained through NATS (the UK air navigation
service provider) and ECMWF’s ERA5 reanalysis data were
used to quantify contrail formation in the North Atlantic from
2016 to 2019 in [14]. In [15], CARATS provides flight track
data over the Japanese airspace and uses ERA5 data. Data from
a commercial airline is used in [16] to assess the feasibility of
contrail avoidance based on ECMWF HRES forecast data over
several weeks in 2023 and 2024.
C. Contributions of this paper
Building upon this aforementioned work, we analyze the
formation of contrails in 2022 with all the available data from
the OpenSky Network over the North American, Europe, and
the Atlantic region. Approximately 5 TB of raw flight data for
the whole year globally are extracted and combined with the
entire year of weather data from ERA5 from the European
Centre for Medium-Range Weather Forecasts (ECMWF), using

fastmeteo, a tool that we recently developped [17]. When
necessary, altitude diversion strategies are employed to prevent
persistent contrail formation. This extensive analysis will pro-
vide concrete conclusions regarding the formation of contrails.
Overall, by exploring the global flight and meteorology data
and studying specifically for one region, we aim to bring better
insights into contrail formation and mitigation strategies. This
report addresses the following research questions:
•What are the statistics on contrail formation on a global
scale, subject to the data availability of OpenSky?
•To what extent can altitude diversions prevent the forma-
tion of persistent contrails based on openly available data?
The remainder of this paper is structured as follows. Sec-
tion II presents the necessary background information about
the OpenSky Network, ADS-B, and meteorological data. Sec-
tion III presents the theory and methodology to estimate the
localization of persistent contrails and to mitigate them with
altitude diversions, and Section IV provides the analyses for
contrail formation and mitigation potential through altitude
diversions. Section V suggests some further research questions
and room for future analyses before we conclude in Section VI.
II. DATA SOU RC ES F OR FL IG HT S AND METEORO LO GY
A. The OpenSky Network
The OpenSky Network is a crowdsourced sensor network
that gathers surveillance data for air traffic control (ATC)
purposes. Its primary aim is to provide the public access to
real-world ATC data and to facilitate advancing and enhancing
ATC technologies and processes. Since 2013, the network has
been continuously collecting air traffic surveillance data. In con-
trast to commercial flight tracking networks like Flightradar24
or FlightAware, the OpenSky Network preserves the original
Mode S replies received by the sensors in a vast historical
database, which researchers and analysts from various fields
can access.
Initially, the network consisted of eight sensors located in
Switzerland and Germany. Today, it has grown to encompass
over 6,000 registered receivers situated worldwide. As of 2024,
OpenSky’s dataset contains over 10 years of ATC communica-
tion data. While the network initially focused solely on ADS-B,
it expanded its data range to include the complete Mode S
downlink channel in March 2017. Recently, it incorporated
other technologies such as FLARM [18] and VHF. The dataset
currently comprises more than 35 trillion Mode S replies.
Figure 1 displays the growth and evolution of the network
in recent years, which involved the inclusion of dump1090 and
Radarcape feeding solutions, as well as the integration of non-
registered, anonymous receivers. However, this practice was
discontinued in early 2019 to ensure the consistent quality of
the feeder data. In March 2020, the number of daily flights
decreased by approximately 30% compared to previous levels,
reflecting the reduction in air travel worldwide caused by the
COVID-19 pandemic. The processing of messages received by
the OpenSky Network has been refined to avoid the replication
of similar messages received by different receivers, resulting
in an artifact that could confuse the reader into thinking that
traffic volume did not return to pre-pandemic levels.
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
0G
5G
10G
15G
20G
25G
COVID-19
Stopped anonymous feeding
Number of daily messages received by the OpenSky network
Fig. 1: Daily messages received by OpenSky over time from 2013 to
2024
The global data reception of the OpenSky Network relies
entirely on its crowdsourced network of receivers, primarily
consisting of enthusiasts, academics, and other supporting in-
stitutions. The coverage provided by each individual sensor is
limited by the range of the antennas’ line of sight, typically
around 400–500 km for the best-performing antennas that reach
the radio horizon. The main areas of organic growth of any such
crowdsourced network effectively serve as a proxy for densely
populated and wealthier regions worldwide. Between 2018 and
2024, the network’s global coverage (see Fig. 2) reached a
saturation point that is typical of most crowdsourced networks,
with most new sensors significantly enhancing reception at
lower altitudes in areas already covered in Europe, the US,
and other developed countries. However, notable coverage
expansions can still be observed in the Middle East, South Asia,
and New Zealand.
In addition to the payload of each Mode S down link
transmission, OpenSky also stores supplementary metadata.
This metadata includes precise timestamps (suitable for mul-
tilateration), receiver location, and signal strength, depending
on the receiver hardware. For further details on the history,
architecture, and use cases of OpenSky, please refer to [2], [19],
[20], [21] or visit the website at https://opensky-network.org.
B. Space-based ADS-B
Geographical regions such as deserts and oceans lack ground-
based coverage due to physical constraints. To address this lim-
itation, commercial ADS-B providers partially rely on space-
based ADS-B and ADS-C data. With space-based ADS-B, a
constellation of low-altitude satellites attempts to receive and
decode ADS-B messages from aircraft in the troposphere and
forward positional information to ground-based stations.
Automatic Dependent Surveillance–Contract (ADS-C) also
uses satellite links to overcome communication limitations in
remote areas, allowing aircraft to share data with Air Traf-
fic Services Units (ATSUs) through negotiated agreements.
Unlike ADS-B, which broadcasts to all, ADS-C messages
are exclusively sent to the ATSUs involved in the specific
contract. Each ADS-C report includes essential information
like aircraft position, time, and precision level, with advanced
reports containing additional data outlined in the contract. The

Fig. 2: OpenSky’s global coverage in 2018 and 2024
Fig. 3: Example of the satellite trajectories on March 1st, 2022,
together with observed ADS-B flight data within 5 minutes.
OpenSky Network is also in the process of collecting and
integrating ADS-C data, with the first pilot projects underway
and limited data already available [22].
In this paper, we benefit from ADS-B satellite data collected
in March 2022 by a constellation of Spire satellites following
various orbits (equatorial, sun-synchronous, 37 degrees, 51.6
degrees, and 83/85 degrees inclination). Figure 3 shows the
trajectory of one of those satellites during a short interval of 5
minutes (in orange) together with the trajectories of a number
of aircraft flying eastbound. In practice, every aircraft is seen
by a satellite during an interval of about 1 minute before being
out of range of the satellite. Even with a constellation of many
satellites, necessary to adequately cover a large area such as
the Northern Atlantic region, it is not possible to get the same
granularity of trajectory data as with ground-based ADS-B.
Fig. 4: An Example of ADS-B data over the Northern Atlantic
region [6] (blue: OpenSky data, purple: space-based ADS-B)
Even though the trajectories we analyze over the Northern
Atlantic Ocean do not benefit from the same sampling rate as
we get from on-ground receivers, it is sufficient to get a clear
idea of its trajectory along the North Atlantic Tracks (NAT)
published on that day (Figure 4). For our analyses, we use
the traffic library [23] to merge trajectories collected from
both The OpenSky Network receivers and the Spire satellite
constellation and only used the satellite data to fill the gaps
over the Atlantic Ocean.
C. Meteorological data
Temperature and relative humidity from the meteorological
data are required to estimate contrail formation and persistence.
They are obtained from the European Centre for Medium-
Range Weather Forecasts (ECMWF) ERA5 dataset [24]. The
dataset has a resolution of 0.25 degrees horizontally. In this
study, data below the pressure altitude of 100 hPa (approxi-
mately 53,000 ft) are used.
To facilitate easy access to the ERA5 data, a previously
developed library, fastmeteo [17], is used to download and
process the data from the Google ARCO-ERA5 data storage.
The data is provided in a cloud-optimized format. This version
of ERA5 converts the commonly used GRIB data files to Zarr
data format, which is efficient in storing meteorological data
from ERA5 as compressed N-dimensional arrays [25].
The fastmeteo tool also provides a fast interpolated method,
which allows gigabytes of flight trajectory data to be processed
in a matter of tens of seconds. The interpolation provides
estimated meteorological conditions at any given position based
on data from the ERA5 grid.
III. ESTIMATION AND MITIGATION OF CONTRAILS
A. Estimation of persistent contrails
Generally, contrails form at low temperatures (-40 °C) and
high relative humidity conditions [13]. The specific atmospheric
conditions are governed by the Schmidt-Appleman criterion
(SAC) [26]. When an aircraft flies through temperatures below
the SAC, saturation with respect to liquid water occurs, leading
to the formation of contrails.
Not all contrails affect the climate equally. When meteoro-
logical conditions satisfy the SAC, a contrail may disappear

quickly. The persistence of contrail is determined by the
humidity of the air. In order to allow persistent contrails to
be formed, the ambient air also needs to be super-saturated
with respect to ice, which means the relative humidity over ice
should exceed 100%. These regions are identified as Ice Super
Saturated Region (ISSR).
Much of the theory behind contrail formation used in this
study is quite established and can be found in literature, such
as [13] and [26]. The implementation of the SAC, ISSR, and
determination of persistence of contrails based on flight data
and meteorological data are described in detail in [17]. Figure 5
below shows an example of determining the conditions of
persistent contrails under an example pressure (here 240 hPa).
0
5
10
15
20
25
Saturation water vapor pressures
over liquid water
over ice
ice supersaturation
0
5
10
15
20
25
critical temperature
Schmidt-Appleman
215.0 217.5 220.0 222.5 225.0 227.5 230.0 232.5 235.0
Temperature (K)
0
5
10
15
20
25
persistent contrail
non-persistent contrail
Fig. 5: Contrail forming criteria including the determination of SAC
thresholds and ISSR conditions at an example given pressure condi-
tion.
To briefly explain our inference process, the meteorological
information for all flights over the year 2022 is first estimated
from the ERA5 data using the aforementioned fastmeteo tool.
Then, the SAC and ISSR conditions associated with each data
point are determined based on temperature and humidity infor-
mation. We estimate whether persistent contrails could form at
each data point along the trajectory with these two criteria. We
only consider segments containing persistent contrail for more
than three minutes in the analysis.
In Figure 6, we illustrate an example flight and different
contrail formation conditions along the trajectory. We can see
Fig. 6: Contrail formation conditions at the original altitude of an
example flight on 10 January 2022, determined based on the mete-
orological data from ERA5. In this example, ISSR conditions and
persistent contrails overlap with unnoticeable differences.
a vast majority of the flight satisfies the Schmidt-Appleman
criterion due to the cold temperatures. However, only sections
of the trajectory experience ice supersaturation, which would
theoretically allow for persistent contrail formation.
B. Altitude diversions
Following the flight altitude-based contrail mitigation strat-
egy proposed in [10], we conduct a large-scale study to de-
termine whether small altitude diversions would help mitigate
some contrails in the region of interest. The percentage of
persistent contrail that could have been avoided is also studied.
The choice of the altitude diversion range is aligned with
[10], within 2,000 ft from the original altitude. This diversion
is assumed to be within the performance allowance of these
flights. Furthermore, it has been approved to cause marginal
safety risk and minor changes in fuel consumption by [10].
Eight alternative altitudes (separated by steps of 500 ft) are
considered to avoid the persistent contrails. At these different
Fig. 7: Potentially contrail mitigation strategy and results at two
different flight altitudes. In both cases, ISSR conditions and persistent
contrails mostly overlap.

Fig. 8: Regions prone to the formation of persistent contrails on 10 January 2022 at 08:00 UTC
flight altitudes, the contrail formation and persistent criteria
(SAC and ISSR) are evaluated, and the minimum deviation that
prevents the persistent contrail is obtained for each trajectory
point (sampled at 30 seconds). If there is no available alternative
altitude for the data point, the original altitude is maintained,
and we consider the contrail cannot be avoided with reasonable
altitude diversions.
Based on the example from Figure 6, we show the example
of potential mitigation at two different alternative altitudes in
Figure 7. Based on the persistent contrail distance indicated in
orange, the total contrails created by this flight could have been
significantly reduced at these new altitudes.
IV. ANA LYSI S
In this section, we perform an in-depth analysis explaining
the formation of contrails based on the large-scale flight data
containing all flights in 2022 for Northern America, Europe,
and the Atlantic regions. First, we illustrate one day of flights
to show the magnitude of contrail regions. After that, we
provide the overall statistics of the total distance of contrails
and mitigation potentials.
A. Contrail persistent regions and the estimated contrails
Persistent contrails often form in regions with higher hu-
midity and lower temperatures. Figure 8 illustrates the regions
prone to persistent contrail formations at 08:00 UTC on 10 Jan-
uary 2022. In this figure, most of the regions are located in the
coastal areas of North America and Europe, with examples of
four different altitudes. Knowing these regions at all altitudes,
we can infer the segments of flight that could have produced
contrails. The calculations are optimized as the computation
can be fully vectorized. As a result, Figure 9 shows estimated
contrail formation over the entire region for a single day: all
flight trajectories over this day are drawn in blue, and persistent
contrails are illustrated in red.
The contrail persistent regions change over the course of a
day and differ from other days. However, for this example day,
if we compare Figure 8 and 9, we can see a general alignment
between contrail formation regions (in purple) and the portions
of the flights (in red) with persistent contrails.
B. Total contrail distance over the one year
To better understand the statistics of contrails over the entire
year, we apply the same process to flights all year. We then
group the distance statistics by month to provide an aggregated
overview of potential contrail distances.
In Figure 11, we illustrated these statistics in bar charts.
Each line of bars shows the monthly total distance (in gray),
segments in SAC regions (in light blue), segments with ice
supersaturation (in dark blue), and segments determined to have
formed persistent contrail (in orange). Note that all colored bars
start at zero and partially overlap in each line.
It is also worth noting that the accuracy of calculated flight
distance depends on the coverage of the OpenSky network.
We employed algorithms to reconstruct complete flight trajec-
tories containing gaps of less than six hours, including most
transatlantic flights. Nevertheless, there is a small portion of
incomplete trajectories and missing flights in the data.
C. Altitude diversion for reducing persistent contrails
The contrail-sensitive regions usually have a shallow depth.
This provides a convenient way to mitigate the contrails ver-
tically without diverting from the original flight paths. In this
study, we calculated vertical diversion possibilities for all flights
in 2022 in the region of interest.

Fig. 9: The formation of persistent contrails over Europe and the United States for all flights on 10 January 2022 (blue: flight trajectories, red:
persistent contrails)
+/-500ft
22%
+/-1000ft
21%
+/-1500ft 14%
+/-2000ft
13%
remainder
31%
January
+/-500ft
22%
+/-1000ft
21%
+/-1500ft 13%
+/-2000ft
13%
remainder
30%
February
+/-500ft
22%
+/-1000ft
21%
+/-1500ft 13%
+/-2000ft
13%
remainder
32%
March
+/-500ft
22%
+/-1000ft
21%
+/-1500ft 14%
+/-2000ft
12%
remainder
32%
April
+/-500ft
20%
+/-1000ft
20%
+/-1500ft 13%
+/-2000ft
12%
remainder
34%
May
+/-500ft
24%
+/-1000ft
22%
+/-1500ft 15%
+/-2000ft
12%
remainder
28%
June
+/-500ft
25%
+/-1000ft
22%
+/-1500ft 16%
+/-2000ft
13%
remainder
24%
July
+/-500ft
26%
+/-1000ft
22%
+/-1500ft 16%
+/-2000ft
13%
remainder
24%
August
+/-500ft
23%
+/-1000ft
21%
+/-1500ft 14%
+/-2000ft
13%
remainder
29%
September
+/-500ft
22%
+/-1000ft
20%
+/-1500ft 14%
+/-2000ft
12%
remainder
32%
October
+/-500ft
22%
+/-1000ft
21%
+/-1500ft 14%
+/-2000ft
12%
remainder
32%
November
+/-500ft
19%
+/-1000ft
19%
+/-1500ft
12%
+/-2000ft
12%
remainder
37%
December
Fig. 10: The percentage of contrails that could have been prevented with altitude diversions over the months of 2022

0 10 20 30 40 50 60 70 80
Distance (106 nautical miles)
January
February
March
April
May
June
July
August
September
October
November
December
Month
total
in SAC
in ISSR
persistent
Fig. 11: Total flight distance and portions that are related to different
contrail forming conditions for all the months in 2022
Figure 10 shows the altitude diversion necessary to avoid cre-
ating persistent contrails following the methodology presented
in III-B. The reductions in persistent contrails in percentages
are calculated and grouped by diverted altitudes. Across the 12
months, we can see that around 70% of the persistent contrails
can be mitigated with an altitude diversion of less than or equal
to 2,000 ft (FL20). In other words, only about 30% of persistent
contrails cannot be avoided by applying altitude diversions
commonly applied in our current airspace structure. We can also
see more contrail mitigation potential in the summer months,
reflected by the lower percentage of unavoidable contrails. This
can be related to ISSRs being thinner in the summer, caused
by seasonal changes in the tropopause and temperature [27].
Table I provides a quantitative overview of the contrail
distance and mitigation potentials for the months of 2022. This
number reflects the illustration from Figure 11 and Figure 10.
The distances are rounded to 1,000 nautical miles in this table.
In the contrail column, we can observe that between 5% and
9% of the total flight trajectories are estimated to have produced
persistent contrails.
It is worth noting that Table I records the flight distance
through the contrail-forming regions. However, only segments
larger than three minutes are considered to avoid outliers. This
way of counting contrails is also why the numbers do not
strictly add up to 100% of the total flight distance.
D. Contrails in the North Atlantic region
Due to the limited receiver coverage over the Atlantic, the
reconstructed trajectories from OpenSky differ from the actual
flight trajectories. We can observe that in Figure 9, not all
strictly follow the North Atlantic Track system (by more than
the Standard Lateral Offset Procedure distances). To closely
study the formation of the contrails, space-based flight data
from Spire has been used to reconstruct the actual tracks over
the North Atlantic region. Figure 12 shows more accurate
positions of a subsample of flights in March 2022, containing
flights departing and arriving at the coasts of Europe and
North America.
Based on the visual inspection of this month’s trajectory
data, the North Atlantic Organised Track System can be clearly
observed in the figure. Using the same methodology proposed
earlier, we analyze the formation of contrails and study the
potential of mitigation with minimum altitude alteration of
existing tracks (trajectories).
Figure 13 shows the different contrail formation statistics
between trajectories reconstructed from only OpenSky data and
trajectories reconstructed with Spire data in March 2022. We
can see a slight underestimation of contrails with only OpenSky
data. The number of flights differs because only flights between
the coasts are included in the space-based subsample, and
hence, the percentage of contrail distances is shown in the
figure. Overall, persistent contrails count for around 5% and
6% of the total flight distance in these two subsets of data.
Similarly, based on the more accurate trajectories that include
space-based ADS-B observations, we have performed the same
altitude diversion strategy to study the contrail mitigation
TABLE I: Distances (in nautical miles) statistics related to contrails and mitigation through altitude diversions.
Total distance Contrail +/-500ft +/-1000ft +/-1500ft +/-2000ft remainder
January 523,229,000 32,041,000 [6%] -6,459,000 (-22%) -6,129,000 (-21%) -3,997,000 (-14%) -3,669,000 (-13%) 9,022,000 (31%)
February 512,474,000 28,071,000 [5%] -5,799,000 (-22%) -5,505,000 (-21%) -3,475,000 (-13%) -3,404,000 (-13%) 7,766,000 (30%)
March 628,990,000 39,601,000 [6%] -7,831,000 (-22%) -7,581,000 (-21%) -4,821,000 (-13%) -4,548,000 (-13%) 11,584,000 (32%)
April 614,694,000 44,673,000 [7%] -8,806,000 (-22%) -8,477,000 (-21%) -5,534,000 (-14%) -5,057,000 (-12%) 12,889,000 (32%)
May 672,770,000 59,662,000 [9%] -10,872,000 (-20%) -11,017,000 (-20%) -7,216,000 (-13%) -6,738,000 (-12%) 18,260,000 (34%)
June 657,438,000 37,515,000 [6%] -8,066,000 (-24%) -7,404,000 (-22%) -4,991,000 (-15%) -4,083,000 (-12%) 9,490,000 (28%)
July 649,078,000 29,365,000 [5%] -6,777,000 (-25%) -5,903,000 (-22%) -4,234,000 (-16%) -3,391,000 (-13%) 6,409,000 (24%)
August 681,109,000 40,032,000 [6%] -9,338,000 (-26%) -8,031,000 (-22%) -5,647,000 (-16%) -4,548,000 (-13%) 8,788,000 (24%)
September 670,247,000 39,792,000 [6%] -8,320,000 (-23%) -7,555,000 (-21%) -5,249,000 (-14%) -4,645,000 (-13%) 10,459,000 (29%)
October 692,919,000 52,276,000 [8%] -10,279,000 (-22%) -9,528,000 (-20%) -6,525,000 (-14%) -5,824,000 (-12%) 15,414,000 (32%)
November 655,203,000 46,078,000 [7%] -9,143,000 (-22%) -8,695,000 (-21%) -5,813,000 (-14%) -5,140,000 (-12%) 13,450,000 (32%)
December 550,760,000 48,790,000 [9%] -8,364,000 (-19%) -8,580,000 (-19%) -5,602,000 (-12%) -5,525,000 (-12%) 16,777,000 (37%)

Fig. 12: One month of trajectories over the North Atlantic region, reconstructed from OpenSky and space-based ADS-B from Spire. The North
Atlantic Organised Track System can be distinguished among the clusters of trajectories.
0 20 40 60 80 100
Distance (% of total flight distance)
osn
spire
March 2022
total in SAC in ISSR persistent
Fig. 13: Percentage of flight routes experiencing contrail forming
conditions along the North Atlantic Tracks, March 2022
potential. The result is shown in Figure 14. When comparing
the results from Figure 10, we can see that the statistics over
the Atlantic Ocean agree with the statistics in March for the
entire region in this study.
V. DISCUSSION
A. Selection and processing of flight data
For the analysis of this paper, we make use of the state vector
data from the OpenSky from the regions of the world with the
heaviest air traffic. All flights within the United States, Europe,
and between the United States and Europe are extracted for
analysis in this paper. To extract data efficiently, we first use
the aggregated flight lists from OpenSky to identify the flights
with corresponding departure and destination airports within
these regions. When flights are reconstructed, only those longer
than 30 minutes are kept for analysis.
For transatlantic flights, large sections of the flight trajecto-
ries over the Atlantic may be missing due to limited ground
coverage. An algorithm is developed to reconstruct the full
+/-500ft
21%
+/-1000ft
21%
+/-1500ft 13%
+/-2000ft
13%
remainder
32%
March 2022
Fig. 14: Contrail reduction potential for flights along the North Atlantic
Tracks, March 2022. Based on trajectories reconstructed with space-
based ADS-B data from Spire.
trajectory by considering the time gap in data segments (less
than six hours) and the aircraft identifiers. This way, the flight
states in the missing parts of trajectories during the cruise are
linearly extrapolated. This linear extrapolation may not fully
align with the actual trajectory and can cause some uncertainties
in contrail estimation. Besides space-based ADS-B data (often
proprietary and only commercially available), open ADS-C
data gathered by OpenSky [22] would potentially mitigate this
shortcoming in data gaps in the future.
Ample attention is also paid to data processing. Since many
flights span over two days, we extract the flight based on its
starting time (the first time observed by OpenSky receivers),
and we generate a daily dataset based on the starting time of
flights. This way, no flights are counted twice. We can also
ensure small segments (less than 30 minutes) of flights before
or after UTC 00:00 are discarded in the analysis. Lastly, based
on the traffic library, all flights are sampled at 30-second
intervals to provide a consistent resolution for the analysis.

B. Safety and emission considerations
One of the main debates in the aviation sustainability
community is the trade-off between contrail avoidance and
potentially induced fuel consumption and carbon emissions.
Furthermore, contrails should never be avoided at the cost of
hindering air traffic’s safety in terms of minimal separation.
We chose not to proceed with the safety analysis based on the
results obtained in an early study [10], where the potential loss
of separation induced by altitude diversion with coordination is
in the magnitude of single digits. The minimum safe separation
area is 1,000 ft vertical separation and 5 nm in the horizontal
plane. This means there is almost no extra workload for air
traffic controllers to solve extra conflicts.
From the same study [10], we also concluded that the
excess emission is marginal, which is between 0.25% and 2%,
considering the uncertainty in aircraft mass. This also agrees
with previous studies like [28], [29]. Hence, in this study, we
only focused on the potential of contrail mitigation, assuming
marginal and acceptable trade-offs for emissions and safety.
C. Limitations
This study relies on the theoretical model to determine
the formation and persistence of the contrails, especially the
Schmidt-Appleman criterion. The model relies on some engine
performance assumptions, which could lead to inaccuracy and
uncertainty in the contrail formation.
The persistent contrail is closely related to the estimation
of ice super-saturated regions. To obtain these in the chosen
airspace, we rely on information from the ECMWF ERA5 data,
which contains temperature and humidity and is assimilated
from different data sources. The inherent accuracy and bias
could affect the estimation of ISSR and the persistent contrails
based on flight data. The data assimilation tends to smoothen
out original measurements at a larger spatial and temporal scale.
The local variation can be missing from this data. To further
validate the estimated control, satellite remote sensing-based
techniques could be employed on a large scale to cross-validate
our estimations.
Furthermore, as contrails are not stationary, the dispersion
and transformation of contrails into cirrus clouds have addi-
tional impacts on the climate. These are not the focus of this
paper, which is focused on estimating the persistent contrail at
the formation based on available flight and meteorological data.
VI. CONCLUSION
This paper presents a new use case for large-scale ADS-B
data gathered by the OpenSky Network.
We present an analysis of the formation of potential persistent
contrails from flights carried out in 2022 for Northern America,
Europe, and the Atlantic regions. The entire year of state vector
data is obtained and used for this analysis, together with ERA5
meteorological data from ECMWF. Our analysis shows that
each month, between 5% and 9% of the total flight trajectories
could have produced persistent contrails, which count for quite
significant total contrail distances.
Mitigation strategies focused on altitude diversion are also
examined in this paper. A significant portion of contrails can
be mitigated by allowing flights to divert from their original
altitude by a maximum of 20 flight levels. In summary, around
70% of persistent contrails are avoidable with an altitude
change of less than or equal to 2,000 ft. This would significantly
reduce the total contrail distance to around 2% of the total flight
distance.
One month of space-based ADS-B data from Spire in the
Atlantic region is used to more accurately examine the contrail
formation and mitigation in the region where OpenSky lacks
coverage. We find a small underestimation of contrail forma-
tions without the space-based data.
Since we mainly focus on the use case of OpenSky data in
this paper, we also acknowledge some limitations, including the
reliance on theoretical models, which could differ from actual
flight conditions. Additionally, estimating ice super-saturated
regions relies on ECMWF ERA5 data, which may contain
inherent accuracy and bias issues. Despite these limitations, our
study provides a valuable contribution to the understanding of
contrail formation and mitigation through an open large-scale
flight data drive approach, which could benefit the ongoing
research and debates on flight contrails and the magnitude of
their climate impact.
ACK NOW LE DG EM EN T
The authors are grateful to the EC for supporting the present
work, performed within the NEEDED project, funded by the
European Union’s Horizon Europe research and innovation
programme under grant agreement no. 101095754 (NEEDED).
This publication solely reflects the authors’ view, and neither
the European Union nor the funding Agency can be held
responsible for the information it contains.
We also thank Spire Global for the sample data to evaluate
the potential of incorporating space-based ADS-B data, in
addition to the ground-based data, to improve the estimation
of contrail formation over the Atlantic.
REFERENCES
[1] M. Schäfer, M. Strohmeier, V. Lenders, I. Martinovic, and M. Wil-
helm, “Bringing up OpenSky: A large-scale ADS-B sensor network
for research,” in Proceedings of the 13th International Symposium on
Information Processing in Sensor Networks, 2014, pp. 83–94.
[2] J. Sun, L. Basora, X. Olive, M. Strohmeier, M. Schäfer, I. Martinovic,
and V. Lenders, “OpenSky Report 2022: Evaluating aviation emissions
using crowdsourced open flight data,” in Proceedings of the 40th Digital
Avionics Systems Conference, Sep. 2022.
[3] J. Sun, X. Olive, and M. Strohmeier, “Environmental Footprint of Private
and Business Jets,” in Proceedings of the 10th OpenSky Symposium, Delft,
The Netherlands, Dec. 2022. doi: 10.3390/engproc2022028013
[4] R. Dalmau, P. Very, and G. Jarry, “On the causes and environmental
impact of airborne holdings at major european airports,” Journal of Open
Aviation Science, vol. 1, no. 2, 2023. doi: 10.59490/joas.2023.7198
[5] H. Hardell, T. Polishchuk, and L. Smetanová, “Investigation of point
merge utilization worldwide using OpenSky Network data,” Journal of
Open Aviation Science, vol. 1, no. 2, 2023. doi: 10.59490/joas.2023.7218
[6] J. Sun, A. Tassanbi, P. Obojski, and P. Plantholt, “Evaluating transatlantic
flight emissions and inefficiencies using space-based ADS-B data,” in
Proceedings of the 13th SESAR Innovation Days, Sevilla, Spain, Dec.
2023. doi: 10.61009/SID.2023.1.41
[7] A. Salgas, J. Sun, S. Delbecq, T. Planès, and G. Lafforgue, “Compilation
of an open-source traffic and CO2 emissions dataset for commercial
aviation,” Journal of Open Aviation Science, vol. 1, no. 2, 2023. doi:
10.59490/joas.2023.7201

[8] R. Chevallier, M. Shapiro, Z. Engberg, M. Soler, and D. Delahaye,
“Linear contrails detection, tracking and matching with aircraft using
geostationary satellite and air traffic data,” Aerospace, vol. 10, no. 7,
2023. doi: 10.3390/aerospace10070578
[9] E. Roosenbrand, J. Sun, and J. Hoesktra, “Contrail altitude estima-
tion based on shadows detected in landsat imagery,” in Proceedings
of the 13th SESAR Innovation Days, Sevilla, Spain, Dec. 2023. doi:
10.61009/SID.2023.1.06
[10] E. Roosenbrand, J. Sun, and J. Hoekstra, “Contrail minimization through
altitude diversions: A feasibility study leveraging global data,” Trans-
portation Research Interdisciplinary Perspectives, vol. 22, Nov. 2023. doi:
10.1016/j.trip.2023.100953
[11] D. S. Lee, D. W. Fahey, P. M. Forster, P. J. Newton, R. C. Wit, L. L.
Lim, B. Owen, and R. Sausen, “Aviation and global climate change
in the 21st century,” Atmos. Environ., vol. 43, no. 22, Jul. 2009. doi:
10.1016/j.atmosenv.2009.04.024, 2009
[12] D. Lee, D. W. Fahey, A. Skowron, M. Allen, U. Burkhardt, Q. Chen,
S. Doherty, S. Freeman, P. Forster, J. Fuglestvedt, A. Gettelman, R. D.
León, L. Lim, M. Lund, R. Millar, B. Owen, J. Penner, G. Pitari,
M. Prather, R. Sausen, and L. Wilcox, “The contribution of global aviation
to anthropogenic climate forcing for 2000 to 2018,” Atmos. Environ., vol.
244, p. 117834, Jan. 2021. doi: 10.1016/j.atmosenv.2009.04.024, 2009
[13] U. Schumann, “On conditions for contrail formation from aircraft ex-
hausts,” Meteorologische Zeitschrift, vol. 5, no. 1, Mar. 1996. doi:
10.1127/metz/5/1996/4
[14] R. Teoh, U. Schumann, E. Gryspeerdt, M. Shapiro, J. Molloy, G. Koudis,
C. Voigt, and M. E. Stettler, “Aviation contrail climate effects in the north
atlantic from 2016 to 2021,” Atmospheric Chemistry and Physics, vol. 22,
no. 16, pp. 10 919–10 935, 2022.
[15] R. Teoh, U. Schumann, A. Majumdar, and M. E. Stettler, “Mitigating
the climate forcing of aircraft contrails by small-scale diversions and
technology adoption,” Environmental Science & Technology, vol. 54,
no. 5, pp. 2941–2950, 2020.
[16] A. M. Frias, M. Shapiro, Z. Engberg, R. Zopp, M. Soler, and M. Stettler,
“Feasibility of contrail avoidance in a commercial flight planning system:
an operational analysis,” Environmental Research: Infrastructure and
Sustainability, vol. 4, no. 1, p. 015013, 2024.
[17] J. Sun and E. Roosenbrand, “Fast contrail estimation with Open-
Sky data,” Journal of Open Aviation Science, vol. 1, 2023. doi:
10.59490/joas.2023.7264
[18] X. Olive, J. Sun, M. Strohmeier, and G. Tresoldi, “OpenSky Report 2023:
Low Altitude Traffic Awareness for Light Aircraft with FLARM,” in
Proceedings of the 42th Digital Avionics Systems Conference (DASC),
Barcelona, Spain, Oct. 2023. doi: 10.1109/DASC58513.2023.10311202
[19] M. Schafer, X. Olive, M. Strohmeier, M. Smith, I. Martinovic, and
V. Lenders, “OpenSky Report 2019: Analysing TCAS in the Real
World using Big Data,” in Proceedings of the 38th Digital Avion-
ics Systems Conference (DASC), San Diego, CA, Sep. 2019. doi:
10.1109/DASC43569.2019.9081686
[20] X. Olive, A. Tanner, M. Strohmeier, M. Schäfer, M. Feridun,
A. Tart, I. Martinovic, and V. Lenders, “OpenSky Report 2020:
Analysing in-flight emergencies using big data,” in Proceedings of
the 39th Digital Avionics Systems Conference (DASC), 2020. doi:
10.1109/DASC50938.2020.9256787
[21] J. Sun, X. Olive, M. Strohmeier, M. Schafer, I. Martinovic, and
V. Lenders, “OpenSky Report 2021: Insights on ADS-B Mandate and
Fleet Deployment in Times of Crisis,” in Proceedings of the 40th Digital
Avionics Systems Conference (DASC), San Antonio, TX, 2021. doi:
10.1109/DASC52595.2021.9594361
[22] M. Xapelli, T. Lüscher, G. Tresoldi, M. Strohmeier, and V. Lenders,
“A first look at exploiting the automatic dependent surveillance-contract
protocol for open aviation research,” Journal of Open Aviation Science,
vol. 1, no. 2, 2023. doi: 10.59490/joas.2023.7229
[23] X. Olive, “traffic, a toolbox for processing and analysing air traffic
data,” Journal of Open Source Software, vol. 4, no. 39, Jul. 2019. doi:
10.21105/joss.01518
[24] H. Hersbach, B. Bell, P. Berrisford, G. Biavati, A. Horányi,
J. Muñoz Sabater, J. Nicolas, C. Peubey, R. Radu, and I. Rozum, “ERA5
hourly data on pressure levels from 1940 to present. Copernicus Climate
Change Service (C3S) Climate Data Store (CDS),” 2018.
[25] T. A. Gowan, J. D. Horel, A. A. Jacques, and A. Kovac, “Using cloud
computing to analyze model output archived in zarr format,” Journal of
Atmospheric and Oceanic Technology, vol. 39, no. 4, pp. 449–462, 2022.
[26] U. Schumann, K. Graf, and H. Mannstein, “Potential to reduce the climate
impact of aviation by flight level changes.” 3rd AIAA Atmospheric Space
Environments Conference, 2011. doi: 10.2514/6.2011-3376
[27] K. P. Hoinka, M. E. Reinhardt, and W. Metz, “North atlantic air
traffic within the lower stratosphere: Cruising times and corresponding
emissions,” Journal of Geophysical Research: Atmospheres, vol. 98, no.
D12, pp. 23 113–23 131, 1993.
[28] D. Avila, L. Sherry, and T. Thompson, “Reducing global warming by
airline contrail avoidance: A case study of annual benefits for the contigu-
ous united states,” Transportation Research Interdisciplinary Perspectives,
vol. 2, p. 100033, 2019.
[29] B. Sridhar, N. Y. Chen, and H. K. Ng, “Fuel efficient strategies for
reducing contrail formations in united states airspace,” in 29th Digital
Avionics Systems Conference. IEEE, 2010, pp. 1–A.