Authors: A. Singh & J.P.R. van der Laarse
Publication Date: December 4, 2021

For our final case study, we focused on predicting CO2 emissions from flights taking off and landing at Schiphol Airport, using AI. 

Why this is an interesting case 

Schiphol has promised to be CO2 neutral by 2030. This promise, however, only considers the airport itself, and not the actual flights. Think for example of the cars that drive along the taxiways. 

Parallel to this promise, we see an annual increase in flight movements that would largely cancel out the reduced CO2 emissions of the airport.

Inverse Surveillance AI 

If we have a look at our definition of Inverse Surveillance AI the case study should follow the following rules: 

(1) Citizens are surveilling governments and bigger organizations (2) in order to control and influence and (3) thus promote transparency and equality, and by doing so democratizing power.

In this case, the AI predicting CO2 emissions of flights will be used by citizens to surveil a bigger organization, Schiphol Group, in the context of their promise to reduce emissions. 

Flight Data 

For this case study, the data we need to surveil this larger organization is freely available, namely flight data. For this we can use the following: 

Via the developers portal from Schiphol airport, we can use their API and request all sorts of data, especially flight data: https://developer.schiphol.nl/login

We can also make use of this Schiphol Airport Flight Data API Python Wrapper on Github 

CO2 emission of flights

We can use the following links for making CO2 emission assumptions and calculations:

CO2 Emission of commercial aviation (per airplane type) https://theicct.org/sites/default/files/publications/CO2-commercial-aviation-oct2020.pdf 

CO2 Emission of aviation calculations by Carbonindependent
https://www.carbonindependent.org/22.html 

CO2 footprint calculation based on the distance between airports
https://github.com/acircleda/footprint 

Github links – CO2 Prediction

Base to start with
https://github.com/deoudit/Predicting-CO2-Emissions-using-Multiple-Linear-Regression

Different type of regressions:
https://github.com/pranavtumkur/Predicting-CO2-emission-using-ML-Regression-models

https://github.com/Strifee/co2_predict

https://github.com/tannyamishra/CO2-emission-prediction-in-cars-regression

Overview of time series forecasting
https://www.kaggle.com/vijaikm/co2-emission-forecast-with-python-seasonal-arima

https://machinelearningmastery.com/time-series-prediction-lstm-recurrent-neural-networks-python-keras/

Panopticon Effect 

Now that citizens can predict CO2 emissions using AI, and thus surveil Schiphol Group by doing so, this surveillance would not work if we do not include the panopticon effect. For this, we need to publicly share the output of this AI where citizens, activists, and journalists can freely see and export this information and use it against Schiphol Group in their protests, and (social) media. By doing so, we can hold Schiphol Group accountable, poke through their emission-free story by showing that the real problems are the actual flights and substantiate (political) arguments against Schiphol Group. By doing so, citizens will be empowered to stand up against the Schiphol Group and change the power dynamic more in their favor.

International Use 

Similar cases can be developed for other airports. Although for the panopticon effect we currently rely on a free society where protests are allowed and have impact on politicians and thus regulations.