Aviation H2O and NOx climate cost functions based on local weather

Aviation H2O and NOx climate cost functions based on local weather

van Manen, J.

Grewe, V. (mentor)

Aerospace Engineering

Control & Operations / Aircraft Noise and Climate Effects (ANCE)

Aerospace Engineering

2017-04-25

Aviation contributes significantly to anthropogenic global warming, and one promising possibility for mitigation is climate-optimised routing. For the REACT4C project a novel approach was used to simulate the variation of aviation water vapour and NOx emission climate impact with location and weather patterns, but this is too computationally expensive to apply beyond initial research. Results showed about 10% climate impact reduction from a 1% cost increase. For implementation of climate-optimised routing, algorithms are needed which will allow climate impact to be estimated in real-time from weather predictions. This research focuses on formulating algorithmic approximations of aviation water vapour and NOx emission climate impact based on local weather data by systematically examining correlations between climate impact data and weather data at the time of emission in the REACT4C dataset. The methods and models used for generating the REACT4C data are assessed in detail down to their first publications and potential errors and omissions are identified. The analysis is split into direct water vapour, short-lived ozone from NOx, and methane from NOx climate impact. Long-lived ozone and stratospheric water vapour from methane effects are neglected. The water vapour and NOx ozone and methane Climate Cost Function (CCF) results from REACT4C are reverse-engineered to the original grid they were emitted from to prevent inflation of statistical power. Weather and chemistry data at the time of emission are interpolated to the same grid for regression analysis. Literature reviews are used to identify causal predictors and derived variables. A variety of statistical tools are applied to assess variability of the CCFs and search for the best predictors. Four algorithms are developed for each species, using zero-dimensional instantaneous regression analysis. A tailored trade-off framework is applied to choose the best algorithm for application. The chosen algorithmic CCF for water vapour emissions is linear with potential vorticity and has an adjusted R2 of 0.59. Both the mean and the variance of the water vapour climate impact appear strongly determined by the altitude of an emission relative to the tropopause. The relationship between water vapour CCF results and emission altitude is investigated to critically reflect and expand on results from a previous publication. The chosen algorithmic CCF for ozone is bilinear with geopotential and temperature plus their interaction and has an adjusted R2 of 0.42. Ozone climate impact appears moderately determined by altitude and temperature of the emission location. The relationship between ozone CCF results, background NOx concentration and latitude during emission is investigated to critically reflect and expand on results from a previous publication. The chosen algorithmic CCF for methane is bilinear with geopotential and the solar incidence, and has an adjusted R2 of 0.17. Methane climate impact has low variability and is relatively independent of weather at the time and location of emission. The relationship between methane CCF results and background NOx concentrations during emission is investigated to critically reflect and expand on results from a previous publication. Methane climate impact can be more accurately predicted by using simulated ozone climate impact, but the variance left unexplained by the ozone algorithm would lead to worse results in application. The correlation between methane and ozone is weaker than in previous studies. Chemical concentrations, lightning frequency, and lightning NOx production at the time and location of emission do not predict aviation NOx climate impact beyond the extent of basic meteorology unless a large amount of predictors are included in the regression. The chain of models and assumptions from basic climate science to algorithmic CCFs is assessed to identify relative effects on uncertainty of the results. Several steps are identified that should be revisited and several opportunities for future data analyses to increase understanding and certainty of algorithmic CCFs. Future steps for research into and application of algorithmic CCFs depend on upcoming verification activities for the results presented here.

http://resolver.tudelft.nl/uuid:597ed925-9e3b-4300-a2c2-84c8cc97b5b7

Student theses

master thesis

(c) 2017 van Manen, J.