The Contrails API uses two different models to forecast contrails: a machine
learning(ML)-based model that predicts contrail-likely zones and a
physics-based model, Contrail Cirrus Prediction (CoCiP), that predicts warming.

## ML-based model

The ML-based Contrail Likely Zone (CLZ) forecasting model predicts the
probability of contrail formation. The model is a deep neural network that
takes weather features as inputs and predicts CLZs based on satellite
contrail detections (*Geraedts et al. 2023*).

Its inputs consist primarily of HRES weather features: specific humidity,
temperature, u component of wind, v component of wind, vertical velocity,
relative vorticity, fraction of cloud cover, specific cloud ice water content,
specific snow water content, and divergence. Relative humidity is calculated
using specific humidity and temperature. Inputs to the model also include
local solar time, day of year, and altitude of flight waypoints.

For some geographic areas (e.g., United States), the model is fine-tuned to use
latitude, longitude, and altitude as input features, whereas the model's global
fallback uses only altitude (and does not use latitude or longitude).

The model
achieves state-of-the-art performance when evaluated against observational
contrail data.

## CoCiP model

The Contrail Cirrus Prediction model predicts the
energy forcing of the contrail, which is a measure of the climate impact of the
contrail.

Energy forcing is defined as:

\\\[ EF \[J\] = \\int_{0}\^{t} RF'(t) \\times L(t) \\times W(t)dt \\\]

meaning the instantaneous radiative forcing of the contrail integrated over its
lifetime (*Teoh et al. 2020*). Energy forcing is normalized by flight
distance, leading to its units of (J/m).

CoCiP is a physics-based model that simulates contrail formation, evolution, and
impact using atmospheric conditions, aircraft type, flight path, and other
features (*Schumann 2012* ; *Schumann et al. 2012* ). The model uses 10 ensemble
members from ECMWF's high-resolution forecast ensembles (HRES ENS) as inputs
for advecting the flight waypoints where contrails have formed forward in time
(*Hersbach et al. 2020*). The CoCiP model also uses cloud microphysics theory to
determine which contrails persist, accounting for initial downdraft, fall, and
sublimation. Given the simulated evolution of the contrail, CoCiP calculates
energy forcing based on the contrail properties and the surrounding weather
conditions.

In addition to CoCiP's estimate of energy forcing, a climatological
estimate of energy forcing is computed by averaging a year of
CoCiP outputs, binned by time of day, season, and latitude.

The final energy forcing quantity is an average of the energy forcing from the
CoCiP ensemble members with nonzero EF and the climatological average, which is
always nonzero. Including climatology in the average ensures an
estimate of contrail impact, even when CoCiP does not predict the formation of
a contrail using any of the weather ensemble members.

## Expected effective energy forcing

Expected effective energy forcing is then calculated as the product of the
probabilities of contrail formation computed from the ML model and effective
energy forcing of these contrails computed from the CoCiP model.

## Contrail forcing index values

The computed expected effective energy forcing values can then be mapped to
a scale of 0 to 4 to produce Contrail forcing index values. Contrail forcing
index values are inspired by [turbulence forecast values](https://www.weather.gov/source/zhu/ZHU_Training_Page/turbulence_stuff/turbulence/turbulence.htm#:%7E:text=TURBULENCE%20INTENSITY,attitude%20or%20a%20slight%20bumpiness).

> [!NOTE]
> **Note:** Contrail forcing index values are computed with `applied_erf_over_rf_ratio = 0.42` (i.e., an effective radiative forcing over radiative forcing ratio of 0.42) ([*Lee et al., 2021*](https://www.sciencedirect.com/science/article/pii/S1352231020305689)).

The Contrails API translates `expected_effective_energy_forcing` to the
`contrails` severity index through clipping and a linear scale:

    ds["contrails"] = ds["expected_effective_energy_forcing"].clip(min=2e7, max=2e8)
    ds["contrails"] = ((ds["contrails"] - 2e7) / (2e8 - 2e7)) * 4

For more background information on this mapping, see
[Energy Forcing Interpretation](https://apidocs.contrails.org/ef-interpretation.html).

Additionally, you can read the [Contrails API release notes](https://developers.google.com/contrails/release_notes)
and subscribe to announcements to be notified about changes to this scale.

## CO2-equivalents

Expected effective energy forcing can also be interpreted in terms of climate
impact of contrails comparable to carbon dioxide, i.e., their
$\\text{CO}_2$-equivalent.

To calculate the $\\text{CO}_2$-equivalent of forecasted contrail warming, use
the following conversion:

\\\[ \\text{CO}_2\\text{eq \[kg\]} = EEEF \\text{ \[J\]} \\times \\text{GWP Factor} \\\]

where:

- **EEEF** is Expected Effective Energy Forcing
- **GWP Factor** is the Global Warming Potential Factor; the inverse of the total integrated energy forcing produced by 1 kg of $\\text{CO}_2$ over a specific time horizon (H, in years). Can be either GWP 20, 50, or 100.

### GWP Factor offerings

Based on time horizons of 20, 50, and 100 years, the GWP Factors offered:

| Time horizon | GWP Factor (kg-$\\text{CO}_2$/J) |
|---|---|
| **GWP20** | 2.497e-9 |
| **GWP50** | 1.173e-9 |
| **GWP100** | 6.779e-10 |

### GWP Factor computation

These GWP Factors are derived from the Absolute Global Warming Potential (AGWP)
for $\\text{CO}_2$ using [IPCC AR5 factors](https://www.ipcc.ch/site/assets/uploads/2018/07/WGI_AR5.Chap_.8_SM.pdf)
and the following calculation:

\\\[ \\text{GWP Factor} = \\frac{1}{AGWP_{\\text{CO}_2, H} \\times A_{\\text{earth}} \\times S_{\\text{year}}} \\\]

where:

- $AGWP_{\\text{CO}_2, H}$: Absolute Global Warming Potential of $\\text{CO}_2$ for the chosen time horizon, sourced from [IPCC AR5](https://www.ipcc.ch/site/assets/uploads/2018/07/WGI_AR5.Chap_.8_SM.pdf)
- $A_{\\text{earth}}$: Earth's surface area ($5.101 \\times 10\^{14} \\text{ m}\^2$)
- $S_{\\text{year}}$: Number of seconds in a year ($31,536,000 \\text{ s}$)

## License

The data exposed by the Forecast API is licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).

## What's next

- Read the [ContrailWatch attributions overview](https://developers.google.com/contrails/v1/ContrailWatch-description) to understand how historical flight-level contrail attributions are used to improve contrail forecasting.

## References

- Geraedts, Scott, Erica Brand, Thomas R. Dean, Sebastian Eastham, Carl Elkin, Zebediah Engberg, Ulrike Hager, et al. 2023. "A Scalable System to Measure Contrail Formation on a per-Flight Basis." arXiv \[physics.ao-Ph\]. arXiv. http://arxiv.org/abs/2308.02707.
- Hersbach, Hans, Bill Bell, Paul Berrisford, Shoji Hirahara, András Horányi, Joaquín Muñoz-Sabater, Julien Nicolas, et al. 2020. "The ERA5 Global Reanalysis." Quarterly Journal of the Royal Meteorological Society 146 (730): 1999-2049.
- Schumann, U. 2012. "A Contrail Cirrus Prediction Model." Geoscientific Model Development 5 (3): 543-80.
- Schumann, U., B. Mayer, K. Graf, and H. Mannstein. 2012. "A Parametric Radiative Forcing Model for Contrail Cirrus." Journal of Applied Meteorology and Climatology 51 (7): 1391-1406.
- Shapiro, Marc, Zeb Engberg, Roger Teoh, Marc Stettler, and Tom Dean. 2023. Pycontrails: Python Library for Modeling Aviation Climate Impacts. https://doi.org/10.5281/zenodo.825291
- Teoh, Roger, Ulrich Schumann, Arnab Majumdar, and Marc E. J. Stettler. 2020. "Mitigating the Climate Forcing of Aircraft Contrails by Small-Scale Diversions and Technology Adoption." Environmental Science \& Technology 54 (5): 2941-50.
- Teoh, Roger, Zebediah Engberg, Ulrich Schumann, Christiane Voigt, Marc Shapiro, Susanne Rohs, and Marc E. J. Stettler. 2024. "Global aviation contrail climate effects from 2019 to 2021." Atmospheric Chemistry and Physics 24: 6071--6093. https://www.researchgate.net/publication/384141259_Global_aviation_contrail_climate_effects_from_2019_to_2021.