Using convection permitting model simulations to improve radar retrievals over hills and mountains

Accurate observations and forecasts of the spatial distribution and intensity of rainfall over orography are critical for generating accurate flood forecasts. These are critical for providing early warning of flood events and hence minimising the risk to human life and the economic impact on property and infrastructure. Radar observations of rainfall provide crucial spatial information on the distribution and intensity of rainfall. Radar however do not observe rainfall at the surface, but at some height about the ground. Particularly over orography this can be problematic as the radar may not detect low level enhancement of the rainfall due to the orography. Over the UK this is often assumed to be through the seeder-feeder mechanism where rain falls through low level cloud, and collects additional liquid water leading to higher surface rainfall. Currently the Met Office uses a simple model based on Alpert and Shafir (1989) to predict the enhancement in radar observed rainfall using profiles of wind speed, direction, temperature and relative humidity from the MetUM. There is evidence that this scheme improves precipitation estimates overall, but not in all situations or for all orography. One suggestion is that the scheme is being applied in situations where it may not be appropriate, for example in convective situations or where there is low level flow blocking. It is also possible that errors in the model profiles used to derive the enhancement may lead to errors in the radar rainfall estimates. The current scheme has been largely tested over the higher hills in England (the Lake District and Pennines), and so it may be that this tuning means it is less appropriate over the higher mountains of Scotland, or lower hills such as Dartmoor.

Independent observations to validate radar retrievals or model simulations are difficult over orography. Traditional rain gauges only provide point observations and are often not situated over high ground. In additional to available gauge data we will use hydrological river flow data to provide catchment scale measurements of total precipitation. We will also look at case studies from recent deployments of the NCAS X-band radar in Scotland and Cumbria. These will, for example, provide vertical scans not available from the operational network which may aid in detecting enhancement and in validating the model. The recent upgrade of the operational radar network to dual polarization has made additional data available, for example on the shape and size of hydrometeors, which may help to identify different enhancement regimes.

This project aims to use high resolution simulations of orographic precipitation over the UK in a range of different situations to assesses the appropriateness of the current radar rainfall enhancement scheme and to suggest improvements. As a secondary aim the comparison of model output with a range of observations will also help test the capabilities of the operational UKV in capturing extreme orographic rainfall events and identify the impact of increased resolution or model enhancements, such as the inclusion of CASIM, on rainfall prediction. CASIM (Cloud-AeroSol Interacting Microphysics) is a new microphysics scheme including a more detailed representation of aerosol and cloud processes than is used in the operational MetUM scheme. The more detailed microphysics may be important for the seeder-feeder mechanism. We will initially make use of operational UKV simulations to allow the student to get used to analysing MetUM data, but anticipate running higher resolution simulations for interesting case studies, possible also testing the impact of CASIM in these cases.

One challenge in this work is that both model and observations will have errors. To avoid the impact of model errors on radar retrievals one possibility is to generate virtual radar data from the model and then apply the radar retrieval to this synthetic data. The retrieved synthetic radar rainfall can then be directly compared with the model rainfall without having to apply any correction for model errors.

Likely questions addressed in the project included:

1) Is the use of a seeder-feeder radar rainfall correction scheme (Alpert and Shafir) appropriate in all synoptic situations / over all scales of UK orography?

2) Can a combination of radar and model data be used to identify the different flow regimes?

3) Can the Alpert and Shafer scheme be improved with the incorporation of additional model or observational information (for example identification of the bright band, or additional hydrometeor shape information from the dual-polarization upgrade)? This might include accounting for low level flow blocking.

4) Could a more appropriate correction scheme improve retrievals in situations where the seeder-feeder process is not important e.g. convective situations?

The availability of new data from the dual polarization upgrade and the NCAS X-band radar combined with developments in high resolution modelling capabilities and improved model microphysics make this a good time to revisit the problem of radar retrievals over orography.

The ultimate aim of the project will be to improve real time radar observations of precipitation over orography. These observations are important to provide more accurate real-time inputs for flood forecasting models. Improved radar rainfall observations are also valuable for verification of rainfall from the UKV and should contribute to improved UKV forecasts through assimilation of the radar observations into the model.