Nighttime Contrail Characterization from Multisource Lidar and Meteorological Observations

Highlights

What are the main findings?

• A multisource approach combining nighttime lidar, ADS-B flight data, and ERA5 reanalysis enables robust detection and characterization of individual aircraft contrails.
• Optimized scattering-ratio, temporal, and altitude thresholds significantly improve contrail discrimination and reduce false detections.

What are the implications of the main findings?

• The methodology provides a reproducible framework for automated nighttime contrail monitoring, complementing passive satellite observations.
• The retrieved geometrical and optical properties support validation of satellite products and improvement of contrail parameterizations in climate models.

Abstract

The present study provides a comprehensive nighttime contrail characterization combining Raman lidar, ADS-B flight data, and ECMWF ERA5 reanalysis over southern France. Observations of different case studies of contrail formation and development throughout their lifetimes provide valuable insights into the contrails’ morphological, microphysical, and optical properties, persistence, and dispersion. We present a multisource methodology to detect and characterize nighttime aircraft contrails over the Observatory of Haute-Provence (OHP) in France. The determination of contrail signatures was performed by applying sensitivity analyses by spatiotemporal thresholding and clustering for contrail detection. Optimizing the thresholds permits the improvement of contrail detection and the reduction of unnecessary noise. The optimal combination of these thresholds, which best reduces false positives and negatives, was SR = 2.1, time = 7.2 min, and altitude = 0.3 km. Subsequent merging of the spots produces persistent contrail signatures at altitudes of 8.7–10.3 km, with thicknesses of 0.1–1.1 km, widths of 2–2.8 km, and optical depths of 0.05–0.40. Contrail optical depth correlates significantly with geometrical thickness and width, which highlights the interplay between contrail morphology and ambient thermodynamic conditions. Our methodology demonstrates the value of combining lidar and flight data for contrail characterization using lidar measurements, flight data, and meteorological information.

1. Introduction

Climate change nowadays is a hot topic among the scientific community. Aviation contrails are considered a significant contributor to radiative forcing and, consequently, to climate change, especially given the anticipated sharp rise in global air traffic [1]. Aviation contributes to non-CO₂ climatic impacts through emissions of nitrogen oxide (NOX), water vapor, sulfate, and soot particles [2]. Persistent contrails and contrail cirrus, together with sulfate aerosols, represent a major non-CO₂ component of aviation’s climate impact [3]. It is estimated that their contribution to net warming is twice that of CO₂ emissions, and the contribution of contrail cirrus alone is about 1.5 times that of aviation’s CO₂ emissions. Their combined effective radiative forcing (ERF) is estimated at 67 mW m⁻² (5–95% CI), corresponding to roughly two-thirds of aviation’s net ERF in 2018 (101 (55 ÷ 145) mW m⁻²) [2]. According to the Sixth Assessment Report (AR6), the global mean effective radiative forcing from combined contrails and contrail-induced cirrus is estimated at +40 mW m⁻², with a likely range of +10 to +70 mW m⁻² [4]. Recent studies attribute more than one-third of aircraft-related ERF to contrails alone, and current best estimates indicate that contrails and the aircraft-induced cirrus clouds they form account for about half of the anthropogenic ERF from aviation for the period 1940–2018. Given the significant contribution of contrails, systematic observations and analyses of their formation, persistence, and radiative properties are essential to reducing uncertainties in aviation-related climate forcing.

Aircraft contrails exhibit optical properties similar to those of cirrus clouds and typically form in the same altitude range, making their discrimination challenging [5]. Ground-based lidar systems have been widely used to investigate contrails of varying optical depths and to retrieve their geometrical and optical properties due to their high vertical and temporal resolution [5,6,7,8,9,10,11,12]. However, lidar-only approaches may suffer from ambiguities when distinguishing contrails from thin cirrus clouds with similar optical signatures.

Nevertheless, only the aged contrails with significant COD are observable by satellite imagery, such as the Spinning Enhanced Visible and InfraRed Imager (SEVIRI) onboard the Meteosat Second Generation (MSG) satellite, Advanced Very High-Resolution Radiometer (AVHRR) [12] onboard the Meteorological Operational (MetOp), Terra and Aqua Moderate-resolution Imaging Spectroradiometer (MODIS) [13], Geostationary Operational Environmental Satellite (GOES) [14], etc. However, satellite observation is hampered by contrails’ low optical depths and small widths, which are below the satellite spatial resolution. The combination of several of these tools enables the detection and quantification of contrail properties, as well as the investigation of their lifetime, as illustrated in a recent one-day case study by [15].

Several studies [16,17] classify contrails into three types: young, mature, and old, based on the morphology in terms of contrail width, shape (linearity), and edge sharpness (

Table 1. Categorization of contrails based on their stage of development. Different nomenclatures have been used to classify contrails by age.

Lifespan	Maturity	Persistence	Age	Width	Shape	Edges
short-lived	young	fresh	<10 min	>3 km	linear	sharp
long-lived	mature	persistent	<1 h	>7 km	linear	diffusive
	old	spreading	>1 h	>21 km	non-linear	diffusive

). The World Meteorological Organization has classified persistent contrails as the sole artificial cirrus clouds [18]. Persistent contrails gradually spread out, transitioning into a contrail cirrus that can merge with existing natural or other contrail cirrus. Their evolution can also be influenced by the long-range transport of air masses. These contrail cirrus can also be advected over long distances by large-scale atmospheric circulation, altering their spatial extent and optical properties; this also controls their persistence [19]. Contrail cirrus have optical depths two to three times greater than those of the contrails themselves [20]. These contrails can develop up to hundreds of kilometers long, but are nevertheless still shorter than natural cirrus, which extends up to thousands of kilometers, but can also be used as a proxy for climate impact studies.

These studies have identified contrails as vertically limited layers with substantial backscattering compared to surrounding cirrus clouds, with contrail altitudes concentrated mainly between 7 and 14 km [18,21,22,23]. Noise is considered as any isolated feature in the lidar profiles with a width less than 1 km and a thickness less than 60 m.

While satellite imagery contrail studies are more common [24,25,26], nighttime detection remains not fully explored. A more challenging point is that the persistence of the contrails is obtained mostly during the morning hours, when the relative and wind updraft are maximal, and less during our measuring time [27]. This study crosses this gap using synergistic lidar measurements and collocated information provided by flight data providers.

The goal of this study is to comprehensively investigate the microphysical and morphological evolution of individual contrails by combining collocated nighttime Rayleigh–Mie–Raman lidar observations, flight tracking, and the ERA5 dataset. This objective was reached through developing a contrail detection algorithm, retrieving contrail geometry, computing the contrails’ optical properties, and evaluating thermodynamic context governing their persistence criteria.

We present a nighttime contrail characterization over OHP that jointly optimizes lidar-based detection thresholds with spatiotemporal grouping, and cross-checks events against flight-track passages and ERA5 dynamics and thermodynamics. Beyond the lidar measurements, we (i) optimize detection/discrimination thresholds through conducting sensitivity analyses to select optimal thresholds of scattering ratio (SR), time, and altitude that maximize COD and contrail duration, which reduces false negatives/positives; (ii) derive their geometry (altitude, thickness, width, axis orientation) and optical properties (COD, SR_max) for each contrail signature; and (iii) contextualize these properties with ERA5 meteorology data (T, RH_ice, RH_liq, PV, etc.) to diagnose their persistent conditions. This multisource, threshold-optimized framework provides a reproducible pathway for contrail monitoring at night when passive imagery is unavailable, complementing recent ERA5-based climatologies and satellite radiative analyses.

The paper is organized as follows: Section 2 describes the methodology thoroughly, starting from the procedure of contrail retrieval and describing the instrumentation used, such as lidar, flight provider, and the ECMWF-ERA5 reanalysis dataset. Additionally, in this section, the contrail detection thresholds are explained. Section 3 presents the detailed results and discussions for all the case studies of the contrails. The conclusions of this study are presented in Section 4.

2. Materials and Methods

2.1. Contrail Retrievals

To detect contrails and investigate their geometrical properties, we use the collocated lidar and Automatic Dependent Surveillance-Broadcast (ADS-B) data as previously described [28]. This is done through the temporal synchronization and spatial matching of lidar measurements with flight passage times [29,30,31]. The integration of the multisource datasets is performed by temporally synchronizing and spatially matching the lidar profiles with flight passages and meteorological conditions. The lidar profiles (temporal resolution of 60 s) serve as the primary reference. For synchronization with flight data, we select lidar profiles only during periods where an ADS-B signal indicates a flight within a horizontal distance of ≤10 km from the observatory and within a temporal window of ±10 min relative to the lidar measurement time. The ERA5 reanalysis data, with a native horizontal grid of ~31 km and hourly temporal resolution, are bilinearly interpolated horizontally to the lidar site location. They are then temporally interpolated to each lidar time stamp and vertically interpolated to the lidar’s altitude grid (15 m vertical resolution), using pressure level data. This process ensures that each lidar profile is contextualized with coincident flight information and ambient meteorological parameters from ERA5.

The vertical extent (geometrical thickness) of a contrail is estimated directly from the altitude-resolved lidar backscatter profile for each time step where a signature is detected. The process starts once the contrail signature exceeds the optimized scattering ratio (SR) threshold. For that vertical column, the contrail’s base (z_base) and top (z_top) altitudes are defined as the lowest and highest altitudes, respectively, where the SR value remains continuously above the detection threshold. The geometrical thickness (CGT) is then calculated as follows:

$$\mathrm{C}\mathrm{G}\mathrm{T}={\mathrm{z}}_{\mathrm{t}\mathrm{o}\mathrm{p}}-{\mathrm{z}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}}$$

(1)

The mean contrail altitude is computed as the average of z_base and z_top. This process is repeated for all time steps belonging to the same aggregated contrail signatures.

The contrail’s geometric width is estimated by combining its observed temporal duration with ambient wind data. The duration (Δt) is determined from the lidar time series as the persistence interval of the aggregated contrail signature (Figure 1). The contrail’s horizontal width (W) is then calculated as the product of this duration and the cross-track component of the wind speed, obtained from the ECMWF ERA5 reanalysis, interpolated to the contrail’s specific altitude and location [15]:

$$\mathrm{W}={\mathrm{V}}_{\perp }\times \mathsf{\Delta }\mathrm{t}$$

(2)

This derivation assumes passive advection of the contrail by the ambient wind field, and its accuracy is contingent on the representativeness of the reanalysis wind data at the spatiotemporal scale of the observation.

The backscattering lidar profiles analyzed in this study have been provided by the Rayleigh–Mie–Raman lidar system (IPRAL, Laboratoire Atmosphères, Observations Spatiales (LATMOS), Institut Pierre-Simon Laplace (IPSL), Guyancourt, France), operated at the Observatory of Haute-Provence (43.9°N, 5.7°E; 673 m a.s.l.), which conducts only nighttime measurements, usually operating for about 6 h per session [7,8,32]. The lidar system employs a frequency-doubled Nd:YAG laser emitting at 532 nm (integrated laser source of the IPRAL system, LATMOS/IPSL, Guyancourt, France), operating at a repetition rate of 30 Hz with an average pulse energy of 300 mJ. Backscattered signals are collected by a telescope with an 800 mm diameter primary mirror (custom-built optical assembly, IPRAL system, LATMOS/IPSL, Guyancourt, France) and a field of view of 1 mrad [33]. The cirrus detection system consists of a primary lens and an interference filter (optical components integrated into the IPRAL lidar system, LATMOS/IPSL, Guyancourt, France). The vertical resolution of lidar measurements is 15 m, while the temporal resolution is 60 s, which enables the instrument to detect fresh contrails and contrail cirrus too [34].

The lidar measurements were carried out during the period between 2021 and 2023. The lidar scattering ratio profiles are used to identify signatures that suggest the presence of contrails [34]. Based on the information provided by the altitude-resolved profiles, the contrail optical and geometrical properties, such as the contrail mean altitude, geometrical thickness (CGT), contrail optical depth (COD), and their orientations, are determined.

The process of retrieving contrail optical depth using a lidar involves making certain assumptions. The COD was obtained from the SR profiles by the following expression given by [35]. So, the contrail optical depth at a certain altitude z is given by integrating the total volume scattering coefficient β(z).

$$\mathrm{C}\mathrm{O}\mathrm{D}\left(\mathsf{\lambda },\mathrm{z}\right)={\int }_{\mathrm{z}\mathrm{m}\mathrm{i}\mathrm{n}}^{\mathrm{z}\mathrm{m}\mathrm{a}\mathrm{x}}\mathsf{\beta }\left(\mathsf{\lambda },\mathrm{z}\right)\mathrm{d}\mathrm{z}$$

(3)

Further, the total volume scattering coefficient β(z) is given as the product of the total scattering cross section per molecule σ(z) and the molecular number density n_ar(z). The contrail optical depth is determined using the previous assumptions, as follows:

$${\mathsf{\beta }}_{\mathrm{R}\mathrm{a}\mathrm{y}\mathrm{l}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}}\left(\mathsf{\lambda },\mathrm{z}\right)={\mathrm{n}}_{\mathrm{a}\mathrm{i}\mathrm{r}}\left(\mathrm{z}\right)·{\mathsf{\sigma }}_{\mathrm{R}\mathrm{a}\mathrm{y}\mathrm{l}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}}\left(\mathrm{z}\right)$$

(4)

$$\mathsf{\tau }\left(\mathsf{\lambda },\mathrm{z}\right)=\mathsf{\eta }·\left(\mathrm{L}\mathrm{R}\right){\mathsf{\sigma }}_{\mathrm{R}\mathrm{a}\mathrm{y}\mathrm{l}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}}{\int }_{\mathrm{z}\mathrm{m}\mathrm{i}\mathrm{n}}^{\mathrm{z}\mathrm{m}\mathrm{a}\mathrm{x}}{\mathrm{n}}_{\mathrm{a}\mathrm{i}\mathrm{r}}\left(\mathrm{z}\right)\left[\mathrm{S}\mathrm{R}\left(\mathrm{z}\right)-1\right]\mathrm{d}\mathrm{z}$$

(5)

$$\mathsf{\tau }\left(\mathsf{\lambda },\mathrm{z}\right)=\mathsf{\eta }·\left(\mathrm{L}\mathrm{R}\right){\mathsf{\sigma }}_{\mathrm{R}\mathrm{a}\mathrm{y}\mathrm{l}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}}{\int }_{\mathrm{z}\mathrm{m}\mathrm{i}\mathrm{n}}^{\mathrm{z}\mathrm{m}\mathrm{a}\mathrm{x}}{\mathrm{n}}_{\mathrm{a}\mathrm{i}\mathrm{r}}\left(\mathrm{z}\right)\mathrm{B}\mathrm{S}\mathrm{R}\left(\mathrm{z}\right)\mathrm{d}\mathrm{z}$$

(6)

where the Rayleigh Scattering Cross Section σ = 5.4 × 10⁻³¹ m² [35,36,37,38].

The lidar scattering ratio, obtained from the Mie and Rayleigh scattering coefficients, is defined as the ratio of the particulate backscattering (excluding background aerosol contribution) to the total backscattering:

$$\mathrm{S}\mathrm{R}={\displaystyle \frac{{\mathsf{\beta }}_{\mathrm{a}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{s}\mathrm{o}\mathrm{l}}\left(\mathsf{\lambda },\mathrm{z}\right)+{\mathsf{\beta }}_{\mathrm{R}\mathrm{a}\mathrm{y}\mathrm{l}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}}\left(\mathsf{\lambda },\mathrm{z}\right)}{{\mathsf{\beta }}_{\mathrm{R}\mathrm{a}\mathrm{y}\mathrm{l}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}}\left(\mathsf{\lambda },\mathrm{z}\right)}}$$

(7)

In particle-free conditions, SR is equal to unity.

The contrail’s multiple scattering correction η is calculated based on the simplified Equation (6), [39,40,41,42]:

$$\mathsf{\eta }={\displaystyle \frac{\mathrm{C}\mathrm{O}\mathrm{D}}{\mathrm{e}\mathrm{x}\mathrm{p}\left(\mathrm{C}\mathrm{O}\mathrm{D}\right)-1}}$$

(8)

where COD is the contrail’s optical depth.

Nevertheless, η not only depends on the COD but also on the ice crystal effective radius, laser beam field of view, and penetration depth [43,44,45]. In a more specific formulation, also because of their ice content, contrails can be considered as thin cirrus clouds, even with their sharp difference in backscattering signal with the surroundings. Many studies have used different values of η to assess the cirrus cloud optical depth [46,47,48,49,50,51,52,53,54,55]. A value of 0.9 for subvisual cirrus clouds, 0.8 for moderately thick cirrus, and 0.6 to 0.7 for opaque cirrus was used by [47]. Another study, [48], suggested the following values for different temperature ranges: 0.54 (at −60 °C), 0.65 (at −40 °C) and 0.76 (at −20 °C). Multiple scattering factors, analyzed by [49], inferred from CALIPSO of cirrus clouds, suggested a η of 0.50 (at 240 K) and 0.80 (at 200 K). Similarly, studying cirrus clouds by CALIPSO products, ref. [50] suggested a η of 0.60. Lower values, 0.60 (thicker than 1 km) and 0.70 (thinner than 1 km), were used by [51,52], and ref. [53] used a higher η of 0.75. Regarding previous contrail studies, refs. [6,54,55] used a η of 0.70. This study used the same value of multiple scattering correction, 0.70, a value that can also be applied to contrails.

A variety of lidar ratios (LRs) are used in analyzing cirrus clouds. Hence, refs. [8,35] used an LR of 18 sr, refs. [56,57] used a higher LR (25 sr), while an even higher value of LR (31 sr) was used by [52]. Fewer papers have analyzed contrails using specific lidar ratios. Hence, Langford et al. [58], who observed contrails over Boulder, Colorado, used a variable SR of 13–40 sr. Additionally, an LR of 20 sr was used by [59], an intermediate LR of 25 sr was used by [6], and an even lower LR of 16 sr to investigate both cirrus and contrails was used by [60]. The chosen lidar ratio in this paper is LR = 25 sr, as in many similar studies [6,34,61].

2.2. Contrail Detection Thresholds and Their Properties

In total, seven cases corresponding to three nighttime observations are analyzed. Determination of the geometrical contours of the contrails is done through the optimization of effective thresholds.

In addition, ERA5 (the fifth generation of atmospheric reanalysis of the global climate provided by the Copernicus Climate Change Service (C3S) at the European Centre for Medium-Range Weather Forecasts (ECMWF), Reading, UK) is used to assess the environmental conditions around contrails [62,63]. Meteorological, atmospheric, and ERA5’s information about wind speed and direction help to obtain contrails’ geometrical width during their temporal lifespan. Additionally, mid-contrail temperature (T), relative humidity over liquid and ice (RH_liq and RH_ice), and potential vorticity (PV) have been analyzed [5]. It is assumed that, during the period when the contrail appears above the lidar, meteorological parameters remain constant.

We performed sensitivity analyses to optimize contrail detection based on the lidar scattering ratio (SR) threshold and to optimize their discrimination using temporal and altitude distance thresholds. Sensitivity analysis optimizes detection/discrimination via maximization of COD and contrail duration while preserving the observed event count. Robustness of the procedure is tested under the variations: SR ± (5–10%), LR (25–31 Sr), and η (0.6–0.8). These events are cross-checked against flight overpasses and ISSR persistence criteria (T < T_crit; RH_ice > 100%). This sensitivity analysis enables us to reduce false positives and false negatives in the detection of separate contrails and to avoid their misinterpretation. Sensitivity analyses by spatiotemporal thresholding and clustering for contrail detection were performed using R (v4.3.1).

For each individual lidar profile, potential contrail signatures are first identified by applying thresholds on the scattering ratio (SR) and minimum vertical thickness (Δz), thereby suppressing single-bin noise and isolated aerosol features. Detected bins that satisfy these criteria are treated as individual contrail “spots.” These spots are then aggregated in the time–altitude (t–z) domain using predefined proximity thresholds in time and altitude, such that spots separated by less than the specified temporal and vertical distances are grouped into a single continuous contrail signature. The aggregation process therefore envelopes the full space–time extent occupied by each contrail.

Once a contrail signature is contoured, its morphological and optical properties are derived. For each aggregated contrail’s signature, we calculate the mean altitude, vertical thickness, duration, and maximum scattering ratio (SR_max). The contrail width is estimated from the lidar observation timespan multiplied by the ERA5 cross-track wind speed, while the cloud optical depth (COD) is retrieved from the integrated backscatter. The major-axis orientation of each contrail is determined from the slope of the aggregated signature in the time–altitude space. This aggregation-based procedure enables automatic detection and characterization of individual contrail events.

A schematic representation of this procedure of contrail signature detection through varying SR, altitude, and time thresholds is presented in Figure 2. Aggregation of the spikes gives more insights into the identification of properties of the contrail events.

The objective of this study is primarily methodological, focusing on the development and demonstration of a novel contrail detection framework based on the combined use of three optimized thresholds (scattering ratio, temporal distance, and altitude distance) within a multisource observational context. By jointly exploiting high-resolution nighttime lidar observations, ADS-B flight tracking data, and ERA5 thermodynamic constraints, the proposed approach enables the physically consistent identification, aggregation, and characterization of individual contrail signatures. Rather than providing statistical detection performance metrics, the emphasis is placed on demonstrating how multisource information constrains contrail attribution and reduces ambiguities inherent in single-source approaches, particularly under nighttime and cirrus-contaminated conditions.

3. Results and Discussions

3.1. Contrail Geometrical Features, Case Studies

The investigation of the contrails’ morphological features and optical properties over OHP is focused on several case studies regarding young contrails under cloud-free and cloudy conditions. Among these properties, the mid-contrail’s altitude, vertical extent, and duration are the principal parameters taken into consideration.

3.1.1. Contrail Cases During 13 January 2023

A representative special case of multi-contrails formed under clear-sky conditions was identified during 13 January 2023 (Figure 3). The provided flight data hows that several flights over OHP coincide with the lidar-detected spikes. So, during the lidar measuring period from 17:25 to 21:25, four major spikes having potentially contrail signature origins were observed in the altitude profile of SR. Combining information from both sources, it can be seen that the four lidar spikes coincide with 2, 1, 4, and 2 flights over OHP at the respective intervals.

The flight characteristics corresponding to the four contrail cases identified on 13 January 2023, including time, altitude, speed, direction, distance from the lidar, aircraft type, engine configuration, and fuel type, are summarized in

Table 2. Flight characteristics during four potential contrail cases: contrail cases, time of occurrence (UTC+2), flight altitude (km), speed (km/h) and direction (grades), horizontal distance (km) from the lidar site, aircraft type, engines, and fuel used.

Case	Time	Altitude	Speed	Direction	Dist.	Type	Engines	Fuel
I	17:17	7.9	832	138	2.8	Airbus A320	2 turbofan	kerosene
	17:22	8.6	948	140	2.0	Airbus A319	2 turbofan	kerosene
II	17:43	10.3	1007	141	2.6	Dassault Falcon FA7X	3 turbofan	kerosene
III	18:24	10.2	893	137	0.5	Boeing B738	2 turbofan	kerosene
	18:29	10.3	941	141	2.4	Airbus A320	2 turbofan	kerosene
	18:46	8.6	943	140	3.1	Airbus A320	2 turbofan	kerosene
	18:48	9.8	963	142	3.4	Airbus A320	2 turbofan	kerosene
IV	19:03	8.9	832	142	3.0	Airbus A20N	2 turbofan	kerosene
	19:26	7.9	859	140	2.9	Airbus A319	2 turbofan	kerosene

.

There is no significant difference in the aircraft speeds. Also, flight direction is almost the same, around 140°. The flight altitudes range from the upper troposphere to the tropopause (7.9–10.3 km). The projected distance from the lidar is similar for all flights (~2–3 km), except for case III, which passes almost exactly over the OHP site (~500 m). Aircraft designs are different, dominated by Airbus models A319 and A320. All aircraft were twin-engine turbofan (Airbus and Boeing models), except for case III, a three-engine turbofan Dassault Falcon FA7X, which flew over the site. Kerosene was utilized as fuel in all these aircraft [64].

The first two neighbor signatures represent two contrail events, with lifespans of 17:47–17:53 and 17:55–18:04, respectively.

To simplify the identification of signal spikes in the lidar SR profiles, an aggregation procedure was applied to the nearby signatures. This procedure enables better identification of individual contrails. To determine the optimal thresholds of the lidar scattering ratio for contrail detection, as well as the temporal and spatial separation criteria for distinguishing single contrails, a sensitivity analysis was conducted. This approach helps to minimize false positive/negative detections.

To process the aggregation, the lidar scattering ratio threshold was chosen as the interval 1.5–5.0, by 0.5 steps. Temporal difference thresholds were chosen in the range of 5–30 min (increasing monotonically in 1 min steps), while the spatial difference thresholds ranged from 0.1 to 2.5 km (increasing monotonically in 200 m steps). Figure 4 shows the results of the sensitivity analysis.

Figure 4 indicates that lower SR thresholds result in higher mean COD values. This is because SR values lower than their threshold are not taken into account in the COD calculation, thus underestimating it. In addition, low SR thresholds count the weaker contrails, increasing the detection sensitivity. On the other side, high SR thresholds decrease the sensitivity but improve the detection significance, taking into account only undisputed contrail cases. In our case, there are prioritized combinations with high COD and high contrail counts, so balancing sensitivity and significance. Shorter time thresholds provide more precise aggregation but might fragment some long-lasting contrails (false positives), while longer time thresholds detect more continuous contrails but risk grouping no-contrail events (false negatives). Smaller space thresholds, despite the altitude precision, might split contrails into multiple groups (false positives), while larger space thresholds might aggregate different contrails (false negatives).

To obtain the optimal threshold values for each of the parameters, large and stable values of the mean COD and contrail durations across the variation of the temporal and spatial thresholds are required. However, specific constraints, such as the spatial and temporal differences between flight passages over the site during each special case, should be taken into account to obtain the optimal threshold values. After performing the sensitivity analysis to obtain the maximal value of contrail optical depth and duration and maintain the required number of groups, the optimal values of these thresholds result in one of the combinations (

Table 3. Summarized results of the sensitivity analysis regarding the thresholds of backscatter ratio, time, and altitude to detect and discriminate between the four contrail events. This procedure is based on the optimization of the contrail properties COD, duration, and number.

Time (min)	SR	Altitude (km)	Time (min)	COD	Duration (min)	Count
Range of thresholds	0.1–2.5	5–30	5.8–13.8	4	0.10–0.40	1.5–5.0
Optimal combinations	0.3–1.0	5–15	7.2	4	0.40	2.0–2.5
Selection	0.3	7.2	7.2	4	0.40	2.1

).

The best combinations of the thresholds are those that maximize the mean COD and the duration of the contrails. These combinations better configure the geometrical parameters of the four principal peaks that correspond to the contrails.

Among the several threshold combinations, we have chosen the best, which maximizes contrail COD and duration. This combination uses a contrail detection threshold of SR = 2.1 and contrail spatial–temporal discrimination thresholds of 0.3 km and 10–15 min. Because both temporal thresholds yield the same results, the 7.2 min temporal threshold is maintained due to the flight traffic frequency above the site (Figure 5). This case represents a clear-sky condition. However, in cirrus environments, higher scatter ratios should be used [65].

The timespan of the contrail detection by lidar is converted into the contrail’s width using wind direction and speed provided by ERA5. The wind directions have been projected perpendicularly to the contrail’s principal axis to provide their geometrical width. The timespan is different from the contrail lifetime, which is its life starting from its emission up to its total dispersion. Nevertheless, the lidar backscattering profiles do not provide total information about the contrail’s lifetime. To give more insight into the contrail properties,

Table 4. Mean parameters of each of the four aggregated cases during 13 January 2023. Altitude gives the contrail mean altitude. Maximal values of the scattering ratio (SR_max) indicate the density of the contrails. The contrail geometrical thickness is calculated as the maximal vertical extension on the contrail’s cross section. Contrail width is calculated by multiplying the time frame of the contrail’s lidar measurement by the wind projection perpendicular to the contrail principal axis. Orientation of the contrails is calculated as the slopes of their principal axes. In addition, IWC represents ice water content, and PV the potential vorticity over the contrail’s surroundings. The last column shows the contrail group according to its formation and persistence criteria, which in these four cases belong to persistent contrails (PCs).

Geometrical/Optical Parameters						Thermodynamic Parameters
Case	Alt. (km)	SRmax	Thick. (km)	Width (km)	Orient. (10−3)	COD	T (°C)	RHliq (%)	RHice (%)	PV (K·m2·kg−1·s−1)	IWC (kg·m−3)	Gr.
1	8.7	4.4	0.6	2	−1.1	0.10	−56.7	74.8	108.8	2 × 10−6	4.0 × 10−7	PC
2	9.2	7.4	1.1	9	−0.4	0.35	−60.3	84.5	126.9	4 × 10−6	4.0 × 10−7	PC
3	10.0	4.2	0.1	28	−0.3	0.05	−60.4	89.7	134.6	3 × 10−6	1.0 × 10−7	PC
4	10.3	7.9	0.8	18	0.9	0.28	−55.8	69.9	101.2	2 × 10−6	1.0 × 10−7	PC

provides the mean values of these properties for each group.

Table 4 provides insight into the geometrical and optical properties of the four contrail cases on 13 January 2023. The third case is by far the thinnest of the four but at the same time the widest. Previous works [66,67] regarding the relations between contrail vertical extent and age, suggested that all these cases represent contrails aged between 15 min and 30 min. The amplified width of the third case is due to the overlapping of two simultaneous very thin contrails over the site. Only the fourth case is characterized by a positive slope of the contrail principal axis, whereas the other cases present a negative slope. This suggests that the case may be characterized by uplift wind profiles. The relative humidity with respect to ice is significantly higher than 100% in all four cases. Hence, the environment during the four cases at the cruise altitudes results in supersaturation with respect to ice, giving perfect conditions for contrail persistence. However, to be formed, contrails need to have a temperature below the critical temperature and relative humidity with respect to water higher than its respective critical value. Critical temperatures for all cases range from −43.2 to −42.1 °C, whereas the actual temperature during these cases is well below these limits, ranging from −61.2 to −56.1 °C. To this end, this condition is clearly fulfilled. Also, RHliq ranges from 70–90%, which exceeds the critical values for all cases. Based on the categorization presented in Table 4, the circumstances of the four cases can be considered as persistent contrails (RHice > 100%). The contrails’ optical depths fit well with those reported by [68,69], except for case III, where a very optically thin contrail was observed.

3.1.2. Additional Suspicious Cases

Another three potential contrail signatures are examined in this section. A second young contrail case, this time under cloudy conditions, was observed on 3 May 2022. The indicated signature on the SR lidar profiles in Figure 6 shows the contrail spike. Additionally, two suspicious signatures under cloud-free conditions (25 November 2022) are shown in Figure 7.

Similar to Table 4,

Table 5. Similar to Table 4, the geometrical, optical, and thermodynamic parameters for the other three cases are presented. Case 5 occurred on 3 May 2022, while the other two cases occurred on 25 November 2022. Case 5 shows a persistent contrail (PC), while cases 6 and 7 show two no-contrail events (NoCs).

Geometrical/Optical Parameters						Thermodynamic Parameters
Case	Alt. (km)	SRmax	Thick. (km)	Width (km)	Orient. (10−3)	COD	T (°C)	RHliq (%)	RHice (%)	PV (K·m2·kg−1·s−1)	IWC (kg·m−3)	Gr.
5	8.7	13	0.3	1.4	0.2	0.33	−41.4	115.8	147.1	2.4 × 10−7	1.8 × 10−5	PC
6	6.9	35	0.3	10.7	−0.1	0.22	−44.3	28.5	43.6	3.0 × 10−6	1.6 × 10−5	NoC
7	6.7	56	0.7	13.6	−0.2	0.98	−44.3	28.5	43.6	3.0 × 10−6	1.6 × 10−5	NoC

presents the ensemble of the contrail parameters during the young contrail event under cloudy conditions (case 5), and during two other suspicious cases under cloud-free conditions (cases 6 and 7).

Given the Schmidt–Appleman and ISSR criteria, case 5 represents a persistent contrail, while cases 6 and 7 represent cases that do not fulfill the contrail formation and persistence criteria. The supersaturation condition (RH_ice > 100%) guarantees the persistence of the contrail of case 5. It is plausible that their respective spots during cases 6 and 7 result from non-contrail products. This assumption is also supported by the signature width, which in case 5 aligns reasonably with the contrail’s width, while in cases 6 and 7, their widths are largely wider than for usual contrails (>10 km). Additionally, there are many identified cases in which there are no contrail spots in the lidar SR profiles observed while aircraft overfly the OHP site. During these cases, the contrail formation criteria were obviously not fulfilled, and represent only NoC cases. No significant differences were found among cases 5–7 regarding the orientations of the signature axes, which were almost fully horizontal. Also, these cases’ altitudes, temperatures, geometrical thicknesses, and ice water contents were quite comparable.

The small slopes (close to zero) for case 5 indicate that a contrail signature nearly parallel to the isentropic surfaces, which is typical in the upper troposphere, where vertical wind shear is generally not significant, and atmospheric stability allows their persistence. The persistent contrail, case 5, was obtained at a lower altitude than that reported by [6], having a comparable COD with those of [26,68,69]. Its geometrical features, width, and thickness are inside the range suggested by [4] but significantly smaller than those reported by [6].

The threshold used to determine the contrail boundary affects estimates of its thickness and width. Taking all five contrail cases into consideration, their mean contrail width is comparable to that of the old contrails reported by [6], while their mean thickness is slightly higher than that of young contrails found from the same source.

The background cloud environment substantially alters contrail evolution. Thus, significant differences are observed between the four persistent contrails formed under clear-sky conditions on 13 January 2023 (cases 1–4) and the contrail observed beneath an overlying cirrus deck on 3 May 2022 (case 5). Case 5 is formed at 8.7 km in higher supersaturated conditions and beneath cirrus located ~3 km above, resulting in enhanced optical signatures (SR_max ≈ 13 and COD ≈ 0.33), but is thinner (0.3 km) and narrower (1.4 km). This alteration reflects radiative coupling with the cirrus layer, which reduces longwave cooling at the contrail top and stabilizes the environment, sustaining ice crystal growth [18,26]. Thus, while all cases satisfy formation and persistence criteria, the cirrus-overlying structure leads to denser contrails compared with those evolving in clean upper-tropospheric conditions.

The correlation-based visualization provided by the network correlation diagram (Figure 8) gives insights about the interrelationships among several geometrical, optical, and thermodynamic contrail properties. Thermodynamic parameters, such as temperature (T), relative humidity (RH_liq and RH_ice), and PV, tend to be more correlated. These parameters potentially share the same atmospheric conditions, like atmospheric stability and moisture availability, that govern these variables.

Also, optical parameters, such as optical depth (COD) and density (SR_max), show moderate correlations with geometrical properties (e.g., thickness and width). This suggests that the microphysical characteristics that impact the optical parameters are influenced by the contrail geometry.

Higher RH_ice values indicate favorable ISS conditions that stimulate ice crystal formation and growth within contrails. Consequently, the IWC results are positively correlated to RH_ice. In addition, IWC, which indicates the concentration of ice crystals, has positive feedback on COD.

Correlation coefficients of the mean altitude suggest that high-altitude contrails tend to be thinner and wider, so more horizontally dispersed. SR_max, contrail thickness, and optical depth are strongly correlated. COD is directly related to both contrail thickness and lidar spikes. However, contrails with higher particle densities in their centers result in greater thickness.

Contrails’ vertical and temporal extensions are generally expected to be positively correlated, implying that thicker contrails persist for longer durations. Interestingly, in this study, these parameters are found to be negatively correlated. The duration represents the horizontal contrail dimension, corresponding to the projection of its width along the wind direction. Hence, the inverse relationship between vertical and horizontal contrail dimensions suggests the conservation of contrail volume, whereas changes in shape may be primarily due to atmospheric dynamics. The more horizontally dispersed a contrail, the less geometrically and optically thick it tends to be. Previous studies have shown that dispersion, influenced by wind shear, affects both the geometric and optical properties of contrails, as well as their coverage [70,71].

Interestingly, the contrail orientation results are slightly positively correlated with their altitude and density. This may be because of the wind patterns in the upper tropospheric regions, which enforce contrail alignment [72].

However, these correlations do not provide a full picture of the situation. This is because contrail parameters depend strongly on the contrail’s density and on its current stage of life. Also, the orientation, represented here as the slope of the main contrail axis during its detection period, is influenced not only by the thermodynamic state of the atmosphere, but also by variability in aircraft flight altitudes, which prevents a clear dependence of orientation on the meteorological factors. Finally, due to the limited number of cases considered, the relationships obtained should be interpreted with caution.

Nevertheless, it is worth highlighting that these assessments are based on a limited sample size. Larger datasets would provide a more comprehensive understanding.

Several previous contrail detection studies have relied on single-source or dual-source approaches, each with inherent limitations. Satellite-only methods provide broad spatial coverage, but often miss optically thin or narrow contrails and generally lack reliable attribution to individual flights, particularly at night or during early contrail stages [13,20]. Ground-based lidar observations offer high vertical resolution and sensitivity to thin ice clouds, but are spatially limited and may suffer from ambiguity between contrails and natural cirrus when used alone [7,68]. Dual-source approaches improve either source attribution or physical consistency, but can still leave residual uncertainties. The multisource fusion strategy adopted here (lidar + ADS-B + ERA5) exploits the complementary strengths of each dataset by combining detailed vertical observations, flight attribution, and meteorological constraints on contrail formation and persistence [15,18,29].

The present analysis is based on a limited number of well-documented case studies concentrated in southern France during 2022–2023 and therefore does not aim to provide seasonal, synoptic, or climatological representativeness. Although a larger set of lidar observations and potential contrail events was examined during the study period, only the clearest and most representative cases are presented here, for which a consistent interpretation could be established using the combined lidar, ADS-B, and ERA5 information. The objective of this work is primarily methodological, demonstrating the performance of a three-threshold, multisource contrail detection framework under contrasting local atmospheric environments, including clear-sky, cloudy, contrail-forming, and non-forming conditions. The analyzed cases involve different aircraft models (e.g., Airbus A319/A320, Boeing B738, and Dassault Falcon FA7X), flight altitudes, and atmospheric states; however, the limited sample size does not allow a robust statistical assessment of regional or aircraft-type influences on contrail morphology and optical properties. Because the detection framework is grounded in physically based thresholds and explicitly incorporates thermodynamic constraints from ERA5, it is transferable to other regions and synoptic regimes. A systematic extension to longer time periods, multiple regions, and broader meteorological conditions, including dedicated analyses of aircraft and engine effects, will be addressed in future work.

4. Conclusions

This paper presents a study of contrail properties based on lidar measurements, flight data, and the ECMWF ERA5 reanalysis dataset. Among the seven events analyzed, five were confirmed as persistent contrails, fulfilling both the formation and persistence criteria, while the other two cases do not represent contrail cases.

The study demonstrates that a scattering ratio threshold of ~2.1 with temporal and spatial separations of ~7.2 min and ~0.3 km effectively identifies contrails and discriminates them from surrounding noise, such as cirrus clouds. All persistent contrails occurred within ice-supersaturated layers and at temperatures below −41 °C. Geometrical contrail properties, such as thickness and width, are 0.58 ± 0.40 km, 11.7 ± 11.3, having a quasi-horizontal orientation.

The proposed detection and grouping procedure can be applied for automatic processing, enabling routine automated identification and characterization of contrails from lidar data supplemented by flight tracks and ERA5.

Regarding the interrelationships between optical and geometrical properties, contrail optical depth clearly increases with both vertical thickness and horizontal width, suggesting that optical properties can serve as proxies for geometrical dimensions. This reflects the influence of contrail microphysical characteristics on geometry. Meteorological variables (temperature, relative humidity, and wind shear) are significantly correlated due to shared atmospheric conditions, emphasizing the importance of meteorological conditions in contrail evolution. Also, due to crystal growth, contrail optical depth results are correlated with RH_ice and ice water content. Contrail orientation shows a slight positive correlation with altitude due to upper-tropospheric wind patterns.

Another important result of this analysis is the difference between the geometric properties of the contrails under clear-sky conditions and those situated beneath cirrus clouds. Radiative effects of the present cirrus alter the environmental conditions, particularly contrail dispersion, impacting their persistency.

Having various implications for remote sensing and climate models, the multisource approach shows that combining lidar backscatter with flight data provider information yields reliable contrail detection and further characterization. Such a data combination can be used to validate satellite retrievals and improve contrail parameterizations in climate models.

The sample is relatively small and limited to a single region. Future work should include more extensive observational data from different locations and atmospheric conditions in synergy with satellite observations and additional tools, such as visible/thermal cameras. This will help to generalize the findings and reduce ambiguities related to the origin of the lidar signal signatures.