FlightClim 2.0: Advancing the simplified estimation of per-flight climate effects

Kurzfassung

This study introduces FlightClim 2.0, a functional enhancement of FlightClim for the simplified estimation of the climate effect of individual flights. As input data, the tool only requires information based on origin and destination airports and the maximum takeoff weight of the flight`s aircraft. Based on this information, the formulas allow a simplified estimation of the climate effect of CO2- and non-CO2-effects per flight and per passenger according to current scientific understanding. The enhanced tool is based on a new dataset that is representative of global passenger aviation in 2019. The dataset relies on emission inventories developed within DLR-project ELK, as well as climate effect estimates from the climate-response-model AirClim. It undergoes extensive data preparation including a clustering according to climate effect characteristics. Because of the distribution of the climate effect values over several orders of magnitude, an adjusted error metric is designed as a key improvement, tuned to fit the used dataset. It is utilized in the derivation of symbolic regression formulas, which are chosen as estimation models. Starting with the choice of hyperparameters, data normalization, and selection of regression variables, several settings for the genetic algorithm-based regression approach are tested and evaluated in an iterative process. The final regression formulas show a considerable improvement in estimation quality and uniformity compared to the models of the previous tool versions. The regression model is integrated into a Python tool. It features an application programming interface, as well as a graphical user interface and combines the results of this study.