Climate impact of a contrail cirrus outbreak and a novel prediction of relative humidity for weather forecast

Kurzfassung

Contrails form in the upper troposphere (UT) from aircraft aerosol emissions and can per-sist in ice supersaturation conditions. Contrail cirrus constitute the largest net warming radiative forcing (RF) component to the total aviation effect on climate. However, micro-physical properties and radiative effects of contrail cirrus and natural cirrus in the same meteorological conditions remains unclear, and the distribution of water vapor in the UTLS is subject to large uncertainties. The contrail cirrus radiative impact is still not completely resolved in remote sensing and climate modelling. An extended cirrus region perturbed by aviation in the North Atlantic region (NAR)during the Midlatitude Cirrus (ML-CIRRUS) experiment is then investigated by combining airborne and geostationary Meteosat Second Generation satellite (MSG) observations. A new method for in situ measurements was used to distinguish between contrails, contrail cirrus and natural cirrus based on ice number and NO gas concentrations. Contrail effective radii (Reff ) reach at most 11 µm, while contrail cirrus Reff can be as large as 51 µm. Contrail and contrail cirrus mean Reff is 18% smaller than that of natural cirrus in this case. A new method to estimate top-of-atmosphere instantaneous RF in the solar and thermal range is developed based on radiative transfer model simulations exploiting in situ and lidar measurements, satellite observations and reanalysis data. For a larger spatial area around the flight path, the contrail cirrus outbreak is warming in the night and early morning and cooling during the day. This study could be applied to a broader data set for the assessment of radiative impacts of cirrus and contrail cirrus. To improve the humidity prediction at the cruise levels, a novel approach is developed utilizing artificial neural networks (ANN). The model combines relevant parameters in-cluding modelled thermodynamic conditions and dynamical states from ERA5 reanalysis data as input, and measured humidity from the In-Service Aircraft for Global Observing System (IAGOS) aircraft observations as output. Correlation analyses indicate that the combination of previous (-6, -2 hour) and current atmospheric states within ±2 ERA5 pressure layers around the IAGOS flight latitude shows the high predictive skill for the current humidity status. Compared with ERA5, the ANN approach improves the consis-tency between predicted and observed humidity from IAGOS in cloudy and cloud-free UT and lower stratosphere (LS), with an increase in correlation by up to 0.25 and a decrease of mean absolute error up to 9.07% and 0.03 g/kg for relative humidity and specific humid-ity, respectively. The predicted relative humidity from the ANN approach shows better agreement to independent airborne measurements. The increase of the accuracy for ice supersaturation forecast is about 43% over NAR in 2020. An increased specific humidity in the UT but a decrease in the LS induce a stronger warming effect at the surface and a locally weaker warming effect of water vapor in the UTLS. The simulated contrail cirrus radiative effects better compares to MSG observations than the results without humidity forecast correction. In a test case, optical thickness and contrail RF increase for the im-provements of humidity prediction. For contrail mitigation, the diversion of flights to avoid ice supersaturation can be implemented with higher efficiency when the ANN method is applied to weather forecast products.