Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.781194
Title: Multi-model ensemble predictions of atmospheric turbulence
Author: Storer, Luke N.
ISNI:       0000 0004 7966 8272
Awarding Body: University of Reading
Current Institution: University of Reading
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Atmospheric turbulence is a major aviation hazard, costing the aviation industry millions of dollars each year through aircraft damage and injuries to passengers and crew. In this thesis we compare reanalysis data to climate model simulations to understand how well climate models predict the location of Clear-Air Turbulence (CAT). We then model how climate change will impact CAT on a global scale, in all four seasons and at multiple flight levels. This provides ample motivation for the second half of the thesis which aims to improve aviation turbulence forecasting by testing a multi-model ensemble forecast by combing the Met Office Global and Regional Ensemble Prediction System (MOGREPS-G) and the European Centre for Medium Range Weather Forecasting (ECMWF) Ensemble. The main results found are that climate models are able to skilfully predict the location of CAT, with the main uncertainty of the location of CAT coming from which turbulence index is the best and not from the use of a climate model. We also found CAT will increase globally in the future with climate change, for multiple aviation-relevant turbulence strength categories, at multiple flight levels and in all seasons. For ensemble forecasting we started with a single-diagnostic equally weighted multi-model ensemble and found it is at least as skilful as the single-model ensembles. This lack of significant improvement in the forecast skill could be because when increasing the forecast spread, we capture more turbulence events but also more false alarms. The relative economic value of the forecast is improved for the multi-model ensemble, particularly at low cost/loss ratios. Through combining two ensembles we gain consistency, gain more operational resilience and create one authoritative forecast whilst maintaining skill and increasing value. Extending this work further, it is found that these results apply more generally for multiple turbulence diagnostics, as the multi-model ensemble was more skilful than either of the single-model ensembles. When combining the predictors, the multi-diagnostic multi-model ensemble was more skilful than the two single-model ensembles. It was also found that an optimised 12-member ECMWF and MOGREPS-G multi-diagnostic ensemble was more skilful than the 51-member multi-diagnostic ensemble. What this therefore indicates is that a smaller ensemble spread for the individual diagnostics within a multidiagnostic ensemble is important for optimising operational forecasts in the future, which could reduce computational costs for turbulence forecasting.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.781194  DOI:
Share: