Use this URL to cite or link to this record in EThOS:
Title: Properties of the ensemble Kalman filter for convective-scale numerical weather forecasting
Author: Vetra-Carvalho, Sanita
Awarding Body: University of Reading
Current Institution: University of Reading
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Atmospheric data assimilation has now started to deal with high model resolution scales of O(lkm) where dynamical properties of the atmosphere exploited in larger scale models may no longer be valid. This leads to a problem in high-resolution data assimilation systems since balances such as the hydrostatic balance are still used to model forecast errors. From scale analysis arguments we recognise that such balances do not necessarily need to be valid at small scales and in this work we use the convective scale Met Office Global and Regional Ensemble Prediction System (MOGREPS) to show that indeed the hydrostatic balance at a horizontal resolution of 1.5 km ceases to be valid in the ensemble perturbations in regions where convection is present while it is valid in regions with no convection. We show that the horizontal threshold at which the hydrostatic balance becomes valid as a vertical average in the ensemble perturbations regardless of the presence of convection is 22 km. We also make use of ensemble methods to establish their applicability (0 convective scale models. In particular we apply (he ensemble square root filter (EnSRF) to a one-dimensional idealised column model wilh a parameterized cloud scheme and a discontinuous rain scheme. We show that the ensemble filter can caprure the true solution within a linear ('No cloud') model regime and non-linear ('Cloud') regime; however, if many good quality observations are used the ensemble fails to capture the true solution within the discontinuous CRain') regime. Interestingly, this can be alleviated if only a portion of the state space is observed. Moreover, having fewer spatial observations also improves the ensemble estimate for the ~mperature in the 'Rain' regime, while the estimate of state variables is slightly degraded in the 'No cloud' and 'Cloud' regimes.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available