Dr Helen Buttery
Helen is currently researching the use of the forecast sensitivity to observations (FSO) technique in the UK 1.5-km model (UKV).
Areas of expertise
- 4D-Var assimilation in the North
Atlantic European model (NAE)
- 3D-Var assimilation in the UK4
- Radar-derived precipitation rates
and their errors
- Doppler radial winds
- Radar reflectivities
Helen is currently researching the use of the FSO in the Met Office Numerical Weather Prediction models. The FSO technique allows the sensitivity to all observations to be calculated simultaneously by making use of data assimilation techniques, rather than calculating the impact of each observation separately using a data-denial technique. Previously the FSO technique has been used successfully in determining the sensitivity of the forecast to observations in global models. This new work aims to extend its use to the Met Office Numerical Weather Prediction models.
Helen has carried out trials of the indirect assimilation of radar reflectivities in the UK 4-km model (Met Office Numerical Weather Prediction models). The indirect assimilation of reflectivity data involves using 1D-Var to create columns of pseudo-observations of temperature and relative humidity, combining the reflectivity data and the model background. These are then assimilated with 3D-Var or 4D-Var during a standard data-assimilation cycle. Reflectivity data is available from all of the UK radar network as well as two radars in the Republic of Ireland: a total of 18 radars; it is provided on four or five scan levels, at one degree by 300-metre resolution every five minutes; the maximum range of the scans is 250 km. In the trials carried out one full set of scans was used at three hourly intervals in the Met Office Numerical Weather Prediction models. Initial trials were promising, but further work is required. The aim is to improve the forecast of precipitation rates - especially in the early stages of the forecast (0-6 hours).
Helen has also carried out trials of the 3D-Var assimilation of Doppler radial winds in the Met Office Numerical Weather Prediction models. Doppler radar measurements are currently available at four radar stations in the South of the UK (Chenies, Clee Hill, Cobbacombe and Dean Hill) and two radar stations in Scotland (Drium-A-Starraig and Hill of Dudwick). They provide radial wind observations out to a maximum radius of 100 km at various elevations (the exact elevations vary with radar station, but they are all between one and nine degrees). Observations are provided every one degree azimuthally and every 600-metre radially, every five minutes, and thus give radial wind measurements with high spatial and temporal resolution. It is hoped that assimilation of this data will improve the local representation of wind velocity and therefore also the location of convergence/divergence and hence of local rain features, and it is thus very important for high-resolution nowcasting and short-range forecasting. The trialling Helen carried out, with assimilation of the data once every three hours and thinned to the Unified Model grid (4 km), has shown an overall neutral impact in forecast skill, but, importantly, an improvement in precipitation location.
Helen previously worked on the 4D-Var assimilation of radar-derived instantaneous precipitation rates. Accurate knowledge of the location and rate of precipitation is essential for flood prediction. Met Office radar data provides high-resolution (5 km resolution every 15 minutes, and now 2 and 1 km resolutions every five minutes) precipitation-rate data covering the entire UK. At present radar precipitation rates are incorporated into the model by latent-heat nudging, rather than being assimilated directly with 4D-Var. By assimilating the precipitation data into the model using 4D-Var, a better analysis of the state of the atmosphere should be obtained, as other model variables adjust to the measured rainfall rates. The study Helen carried out involving different cases with diverse synoptic conditions suggested that 4D-Var precipitation assimilation in its current state provides an improvement over latent-heat nudging of precipitation data when there is predominantly large-scale precipitation. Latent-heat nudging remains superior in the largely convective situations. This is true both for the location of the precipitation and the average precipitation rate.
Prior to joining the Met Office, Helen achieved a double-first physics degree (BA) and MSci at Cambridge University. In 2004 Helen completed a PhD in Astrophysics, also at Cambridge, studying high redshift clusters of galaxies. She then worked as a post-doctoral researcher at Arcetri Observatory in Florence studying very high redshift galaxies, and subsequently as an assistant editor of the journal Advanced Materials at Wiley-VCH in Weinheim.
She joined the Met Office in December 2007 and has previously worked on the 4D-Var assimilation of radar-derived precipitation rates in the Met Office Numerical Weather Prediction models, the 3D-Var assimilation of Doppler radial winds in the Met Office Numerical Weather Prediction models and the indirect assimilation of radar reflectivities.