Skip to content

deepstupid/2024-06-contrails-workshop

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

47 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Contrails Analysis Workshop

All times in BST / UTC+1

Presentations

Wednesday, June 19, 2024

Observation of contrail formation (09:30 - 10:30)

  • Phillippe Very (Eurocontrol): The ContrailNet Portal: Stimulating Contrail Research by Sharing Data

    The ContrailNet initiative aims to integrate the work of different contrail-related projects and share valuable, sometimes costly, data to avoid duplication of effort and to stimulate collaboration and research in the field of contrail science. The scope has been extended to a global collaboration. The focus is on sharing high quality labelled data sets, especially those that require manual validation. After introducing the initiative, this presentation will focus on datasets of long sequences (1-2 hours) of human-labelled ground camera images that will be useful to better understand the evolution of persistent contrails.

  • Olivier Liandrat (Reuniwatt): Observation of clouds and contrails using sky imagers

    Reuniwatt's expertise is to monitor the presence and characteristics of clouds and to forecast their location in the next minutes, hours and/or days. We rely on three core technologies: our patented visible and thermal ground-based sky imagers, meteorogolical satellite observations and numerical weather prediction models. In this presentation, we will share our experience regarding the observation of clouds and contrails from the ground.

    https://www.reuniwatt.com

    https://reuniwatt.com/en/wp-content/uploads/sites/3/2023/08/MTI_SEPT2023_SkyImager_for_Contrail_Observation.pdf

  • Aaron Sarna (Google): Contrail Attribution in Geostationary Satellite Imagery

    I'll discuss the challenges of attributing contrails detected in geostationary satellite imagery to the flights that made them, and present a new algorithm for doing so. I'll also present a new approach for evaluating attribution algorithms, given the lack of abundant ground truth labels.

Observation of contrail formation (10:45 - 12:00)

  • Remi Chevallier (Airbus/ENAC): A Simple and Cost-Effective Method for Contrails Ground Observation: Application in Concept of Operations Feasibility Studies

    Condensation trails have a significant climate impact, and we are seeing a large number of initiatives to help mitigate this impact, proposing a large variety of concepts of operations (CONOPS). These are still today subject to a certain number of uncertainties (weather forecasts, contrail impact prediction models, potential flight delays, trajectory changes driven by air traffic control etc…). Each CONOPS may result in increased fuel consumption, greater airspace congestion, and additional constraints for airlines, such as the need for the ability to implement trajectory re-planning shortly before takeoff or even during flight.

    There is a tradeoff between the additional costs and constraints and the potential climate benefits offered by the contrail impact mitigation methods. Each source of uncertainty could indeed reduce the potential climate benefits of the CONOPS. It is crucial to identify and quantify all potential costs and constraints through simulations and observations. This validation process will help airlines and ANSP implement concepts of operations effectively. Airbus is working on both simulations and observations studies. This presentation will showcase one of our observation-related studies and demonstrate its application in the context of CONOPS validation.

    Geostationary satellite observations are currently used by several actors. However, they lack the capability to observe contrails formation because of their lack of spatial resolution. Ground-based observation could address this issue, but it requires a substantial number of observation sources to conduct an effective campaign. To meet this need, we developed a simple and cost-effective method for contrail observation using a basic smartphone camera. Our approach includes a lightweight contrail detection model optimized for inference on affordable devices. Combining these ground-based observations with air traffic data enables the identification of aircraft producing contrails, the study of contrail formation conditions, and comparisons with contrail prediction models. This could prove valuable in the CONOPS validation process.

  • Wouter Vandenneuker (MUAC): Prototyping a Contrail Camera Network: From Concept to Reality

    Satellites track long-lasting contrails but lack real-time feedback and aircraft identification. At Eurocontrol MUAC, we built a contrail ground camera to support contrail mitigation trials. We present valuable insights from over a year of observations with webcams and machine learning in our airspace.

  • Luc Busquin (Alaska Airlines): Using Global Meteor Network Cameras for Contrail Observation

    In this presentation, I will introduce a project focused on leveraging the Global Meteor Network (GMN) for contrail observation. The GMN, with its existing network of approximately 1000 cameras worldwide, offers a unique opportunity to enhance contrail research.

    This project explores the potential of adding contrail observation capabilities to the GMN camera stations. By utilizing the existing GMN infrastructure, we aim to develop a cost-effective and efficient method for continuous day and night contrail observation without interrupting meteor research.

    I will cover the technical aspects of adapting GMN cameras for contrail monitoring. Preliminary results from our project will be shared, demonstrating the effectiveness of this approach in capturing contrail data alongside meteor observations.

    By integrating GMN resources into contrail research, we can significantly improve our observational capabilities, especially regarding the issue of contrail attribution.

    https://youtu.be/DUIoQJUn1YU?si=lDnodF6PhqrCzC-o

    https://youtu.be/6JxqNqtUJQc?si=lgmpSxUWxiT7MpBP

  • Peter DeBock (ARPA-E), Miad Yazdani (RTX), Saikat Majumder (GE): ARPA-E PRE-TRAILS (Predicting Real-time Emissions Technologies Reducing Aircraft Induced Lines in the Sky)

    ARPA-E launched a program to develop novel high accuracy sensing mechanisms, data fusion and observation mechanisms to predict aircraft induced cirrus with high accuracy. Five teams were funded with diverse approaches to develop and demonstrate these technologies.

    https://arpa-e.energy.gov/technologies/exploratory-topics/aviation-contrails

Evaluating forecasts (13:00 - 14:20)

  • Vincent Meijer (TUDelft): Contrail forecast evaluation using satellite data

    We construct a dataset of flight waypoints matched to contrails and not matched to contrails using CALIOP LIDAR observations of contrails collocated with GOES-16 ABI data. We also develop a contrail forecast evaluation framework that can directly capture the potential benefits of contrail avoidance and use the constructed dataset to evaluate these benefits for two numerical weather prediction products and a satellite-based nowcasting system.

  • Scott Geraedts (Google): ContrailBench: datasets for forecast evaluation

    Showcasing how we can use observations to compare contrail forecasts to each other, and proposing a public model comparison framework.

  • Greg Thompson, Adam Durant (Satavia): On the fidelity of high-resolution numerical weather forecasts of contrail-favorable conditions

    The potential climate-warming impact from aircraft contrails may be similar in magnitude to the direct effect from carbon dioxide emissions across all aviation. The warming impact may be mitigated through pre-tactical flight trajectory optimization to avoid ice supersaturation regions (ISSRs) while also considering aircraft performance and CO2 emissions. The ability to perform such deviations depends on accurate predictions of water vapor in the upper troposphere and lower stratosphere (UTLS). Herein we evaluated the performance of two leading global numerical weather prediction (NWP) models: the Global Forecast System (GFS) developed in the USA and the Integrated Forecast System, (IFS) developed in Europe, and a research mesoscale model, Weather Research and Forecasting configured by SATAVIA (S‐WRF) to predict UTLS moisture and ISSR. We compared humidity forecasts to observations from 383 aircraft flights and 3480 radiosonde profiles comprising approximately 1.5 million measurements over Europe and the Middle East for 10 months in 2022. Neither GFS nor IFS properly reproduced the observed distribution of relative humidity with respect to ice (RHice). Moreover, in addition to not being usable for prospective flight planning, the ERA5 reanalysis only slightly improved the outcome of the IFS. Only the S‐WRF model with multi-moment cloud physics and high spatial resolution (5 km grid spacing) closely reproduced the observed relative frequency distribution of RHice. Furthermore, ISSR validation using near equal-area neighborhoods when computing Matthews Correlation Coefficient and F1 score showed that S‐WRF scored higher (F1 = 0.66) than the IFS (F1 = 0.62), while the GFS had near zero score due to its near complete lack of predictions of RHice greater than 100% in stark contrast to observations. In fact, S‐WRF also correctly predicts 92% of the time when conditions were not conducive to contrail formation. Ultimately, the S‐WRF model could be used to alter flight plans to deviate above or below nearly certain contrail formation regions to reduce non-CO2 climate impacts of aviation.

  • Sebastian Eastham (Imperial): Using LIDAR to analyze and evaluate contrail models

    One of the major sources of uncertainty in the climate impact resulting from contrails is how the contrail develop over time, in large part because it is so challenging to measure the physical and optical properties of contrails. Similarly, the high uncertainty in meteorological data makes it difficult to evaluate whether or not discrepancies between observed and modelled contrails are due to the input conditions or due to inaccuracy in the contrail model. We use observational data from satellite LIDAR to test two contrail models, first using ERA5 data but then using the models to infer the most likely meteorological conditions. Our results show large differences in the radiative forcing estimated by different models for the same conditions, but that this is in part driven by differences in the sensitivity of those models to the local conditions. This divergence in sensitivity to parameters such as local relative humidity suggests that current approaches to identify high- or low-impact flights may be unreliable and that more work is needed to establish robust, high-speed contrail models which can support such analyses.

Operational Evaluation (14:40 - 16:00)

  • Christiane Voigt (DLR): New results on contrails, SAF and lean burn engine technology from recent aircraft campaigns

  • Aarón Sonabend (Google), Tom Dean (BE): Evaluating preliminary avoidance trials

  • Louis Robion, Prashanth Prakash (MIT): Leveraging contrail observations and measurements from flight campaigns

    The presentation will focus on approaches to leverage the growing contrail observation data and associated techniques and measurements from flight campaigns to improve models and bound modeling uncertainties. The presentation will also briefly cover the use of the MIT Contrail Avoidance Support Tool (MCAST) as a research tool to further our understanding of problems in contrail observations.

  • Dennis Piontek, Kai Widmaier (DLR): First results for the AKKL 100-flights-trial & CONCERTO

    We present recent work on two ongoing projects: the German AKKL 100-flights-trial and the SESAR project CONCERTO. AKKL focusses on developing and testing procedures and strategies to operationally avoid contrail-sensitive airspaces by airspace users. Four airlines and two flight trajectory optimizers are taking part in this project, and the first contrail avoidance flights were conducted in spring 2024. We present first post-flight analysis data. The SESAR project CONCERTO focusses on developing indicators and tools to select flight with potentially high climate impact and high mitigation potential. The development focusses on air navigation service providers and in particular flow managers. Here we present recent studies on CoCiP uncertainties carried out in this context.

Reflecting on Day 1's learnings (16:00 - 17:00)

  • Shelagh McLellan (Google): Breakout groups - small group discussion with guided questions

Thursday, June 20, 2024

New forecast and planning approaches (09:00 - 10:20)

  • Axel Seifert (DWD): A two-moment cloud ice scheme in the ICON model for predicting ice supersaturation

    I will present the new two-moment cloud ice scheme of the ICON model that has been developed to explicitly predict ice supersaturation in the upper troposphere. By predicting ice crystal number concentration the model has a better representation of the phase relaxation time and, hence, ice supersaturation. This provides an alternative approach to the parameterization in ECMWF's IFS model, which allows only supersaturation in the cloud-free area. Verification with radiosondes shows that both models have a similar skill in predicting ice supersaturation. The makes this new ICON version a viable alternative for providing meteorological input for contrail prediction models.

  • Zane Dedekind (Environment Canada): Reducing the Impact of Aircraft Induced Clouds on Climate – Development of the Contrail Avoidance Tool (CoAT)

    Aviation's rapid global growth presents significant sustainability challenges in alignment with the Paris Climate Agreement (Grewe et al., 2021). In addition to CO2 emissions, non-CO2 effects, notably from contrail cirrus and nitrogen oxides, contribute to the globally averaged climate forcing share of ~ 3.5% (Lee et al., 2021), with the majority arising from non-CO2 effects (Kärcher, 2018). Contrail cirrus exerts a net warming radiative effect (Forster et al., 2012; Teoh et al., 2022a), contributing an effective radiative forcing (ERF) of 57 mW m−2 in 2018, the largest among aviation-induced emissions (Lee et al., 2021). However, the impact of contrail cirrus within these non-CO2 effects carries a large uncertainty. To address the uncertainty, we are developing a Contrail Avoidance Tool (CoAT) based on Environment and Climate Change Canada (ECCC) operational numerical weather prediction (NWP) modeling systems using the Global Environmental Multiscale (GEM) atmospheric model (Côté et al., 1998; Girard et al., 2014). The core of CoAT is based defining contrail formation regions using the Schmidt-Appleman Criteria (Schumann, 1996) and then calculating the properties of young contrails (Unterstrasser, 2016) which is then advected by the NWP model. The CoAT is being tested on the High-Resolution Deterministic Prediction System (HRDPS; Milbrandt et al., 2016) at a horizontal resolution of 1 km x 1 km, before being employed over the pan-Canadian domain, which is run four times daily with 48-hour forecasts. CoAT will then be adapted for the Global Deterministic Prediction System (GDPS; Buehner et al., 2015), which is run twice daily with 10-day forecasts, thereby providing coverage for Canadian airspace and the Atlantic and Arctic Ocean regions with the densest flight routes.

  • Zeb Engberg (BE), Tharun Sankar (Google): A hybrid contrail forecast model

    We present the development and application of a generalized 4D contrail impact forecast. The contrail forecast is the hybrid of two individual models: (1) Google Research forecast of regions with high probability of contrail formation; and (2) Breakthrough Energy forecast of regions with highly warming contrails.

    Google Research has developed a data-driven forecast system for contrail formation, utilizing contrail detections in satellite imagery. This approach leverages machine learning algorithms and integrates weather data as inputs to predict the probability that an aircraft generates a contrail visible in the GOES satellite imagery as it passes through a specific region of the atmosphere. Breakthrough Energy has developed a physics-based forecast of contrail climate forcing on a regular grid based on the Contrail Cirrus Prediction (CoCiP) model. The gridded CoCiP model operates deterministically, relying on three inputs: meteorology, aircraft performance, and epistemic model parameters describing the governing atmospheric physics.

    The combined hybrid forecast model generates real-time expected values of contrail climate forcing provided on a regular grid in standard meteorological formats (NetCDF, GRIB). We present an overview of each individual model and the approach to synthesizing the models to produce a single probablistic forecast of contrail formation and climate forcing.

  • Alejandra Frias-Martin (Flightkeys): Implementation of contrail avoidance in commercial flight planning

    How is contrail avoidance being implemented in flight planning? What has changed in the last year and what is predicted to change in the year to come? What do airlines think? All these questions will be answered in this presentation.

Observations of Climate Forcing (11:00 - 12:00)

  • Ollie Driver (Imperial): Factors influencing contrail observability in satellite images

    Not every contrail is observable in satellite imagery, due to either sub-pixel structure or low optical depth. In this study, the contrail observability threshold is assessed as a function of the contrail’s properties and the imager used. Synthetic satellite images were created using radiative transfer simulations of simple linear contrails in clear sky. Their detectability was tested using a detection algorithm tuned to a labelled dataset to derive a threshold for detectable contrails in terms of their properties.

    The analysis was combined with a modelled global contrail population to determine the fraction of contrails, and of contrail radiative forcing, that is observable. It was shown that the most strongly warming contrails are also more easily detectable than other contrails, but that significant fractions of climate-relevant contrails are not detected using current techniques and instruments. The detectability of contrails as their properties evolve is also explored, finding that most contrails are observable at some point in their lifecycle, with the onset of detection typically within the first two hours. This work applies to future design of experimental trials—optimising for strongly-observable outcomes—as well as being readily expandable to future detection algorithms, backgrounds, and observations.

  • Xinyue Wang (Climaviation): Estimating the radiative effect of contrail outbreaks by using geostationary satellite observations

    Estimation of the perturbation to the Earth's energy budget by contrail outbreaks is required for estimating the climate impact of aviation and verifying the climate benefits of proposed contrail avoidance strategies such as aircraft rerouting. Here we identified two successive large-scale contrail outbreaks developing in clear-sky conditions in geostationary and polar-orbiting satellite infrared images of Western Europe lasting from 22–23 June 2020. Their hourly cloud radiative effect, obtained using geostationary satellite cloud retrievals and radiative transfer calculations, is negative or weakly positive during daytime and positive during nighttime. The cumulative energy forcing of the two outbreaks is 7 PJ and −8.5 PJ, with uncertainties of 3 PJ, stemming each from approximately 15–20 flights over periods of 19 and 7 hr, respectively. This study suggests that an automated quantification of contrail outbreak radiative effect is possible, at least for contrails forming in clear sky conditions. We are currently working on enhancing the method for more complicated cloud systems, such as 2-layer cases with a uniform liquid cloud layer below and contrails above, coexisting with surrounding single-layer liquid clouds and ice clouds. These cases are being tested with new data obtained from EUMETSAT.

    https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2024GL108452

  • Kevin McCloskey, Aarón Sonabend (Google): 12hr satellite-driven contrail flux via causal inference

    In this talk we discuss a new methodology we are using to estimate contrail top-of-atmosphere flux. Other methods used so far have either depended heavily on simulations, or need to solve the challenging "contrail tracking" problem; the new method doesn't need to do either. We start with an overview of the COllocated Irradiance Network (COIN) that gives shortwave and longwave flux on 10-minute refresh with ~2km pixel size, which is sufficient to observe formation and evolution of contrail radiative forcing. We then show a naive method of estimating warming based on COIN, but note that it does not account for radiative confounders. We pull from the causal inference literature to address radiative confounders, giving first a brief intro example of framing problems with a causal directed acyclic graph. Finally, we outline a series of tests devised using synthetic data that boost confidence in the results of the method.

Labs

Thursday, June 20, 2024

Session 1 (13:00 - 14:00)

  • Scott Geraedts (Google): Using ContrailBench

    In this lab the participant will load public, observational datasets, and use them to evaluate different contrail forecasts.

Session 2 (14:00 - 15:00)

  • Nick Masson, Tom Dean (BE): Contrails Observer app hackathon

    We'll go spelunking in a handful of compelling datasets, with our initial bearing centered on images captured by the Contrails Observer App (iOS and Android), a mobile phone app for community-generated observations of contrails.

    Curated datasets include:

    • Contrail App images (where people see contrails!)
    • ADS-B waypoints (where the airplanes are at!)
    • pycontrails CoCiP polygons (where we expect contrails should be)
    • GOES contrail-detections (where satellites + computer vision see contrails)
    • GOES full-disk and mesocale images (where satellites are collecting imagery)

    https://github.com/contrailcirrus/2024-06-contrails-workshop/blob/main/labs/observer_app_hackathon/INTRO.md

  • Seb Eastham (Imperial), Tristan Abbott (BE): Getting set up with APCEMM

    This lab will provide an introduction to simulating contrails using the Aircraft Plume Chemistry Emission and Microphysics Model (APCEMM, https://github.com/MIT-LAE/APCEMM), an intermediate-complexity model that aims to bridge the gap between Gaussian plume models and computationally-expensive large eddy simulations. We will begin by using a case study with idealized meteorology to demonstrate how input data for an APCEMM simulation is formatted, how an APCEMM simulation is run, and how APCEMM output files are organized. We will them demonstrate how a new pycontrails interface to the APCEMM model can be used to run APCEMM simulations for a real flight using real meteorological fields. Any remaining time can be used to discuss potential use cases and pycontrails or APCEMM features needed to support them. Attendees will leave this lab with the ability to configure and run APCEMM simulations and parse APCEMM output for comparison with other contrail models or observations.

Session 3 (15:00 - 16:00)

  • Nick Masson (BE), Shelagh McLellan (Google): User stories: Designing effective data, interfaces, and products

    "If you build it, they will come". While this might fly for Kevin Costner, it does poorly to describe adoption of technology in the "real world." In this interactive discussion based lab, participants will write user stories — a common tool in product development for designing more user-centered technology. We’ll also explore barriers to adopting contrail forecasts/nowcasts, and monitoring, reporting and verification (MRV) tools for managing and complying with targets for non-CO2 climate impacts. By the end of this lab you’ll gain greater empathy for users, and have a toolkit that you can apply in your own work to better drive user engagement.

  • Zeb Engberg, Tristan Abbott (BE): Finding flights in Landsat and Sentinel

    Validating contrail prediction models requires observations of contrail evolution from formation to dissipation. This lab introduces new pycontrails tools for observing the earliest stages of the contrail lifecycle by intersecting flights with high-resolution (10-30 m) Landsat and Sentinel-2 satellite imagery. We will begin with an overview of the Landsat and Sentinel-2 platforms, then show how to use BigQuery to efficiently find satellite scenes that contain commercial aircraft. We will then use a small number of scenes as case studies to demonstrate that aircraft and newly-emitted contrails can readily be geolocated and identified in Landsat and Sentinel-2 imagery. Attendees will leave this lab able to efficiently search and download Landsat and Sentinel-2 data and visualize young contrails in single-band and composite imagery.

About

2024 June 19 - 21 Contrails Workshop

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 100.0%