Convective-scale predictability

New data assimilation (DA) techniques and the exploitation of new observations have great potential, but it is important to realize that if predictability is very low then even a substantially improved representation of the initial state does not necessarily lead to significant improvements in the forecast. Experiences with convective-scale models have demonstrated that the processes of error growth, and so the extent to which improved forecasts are possible given an improved initial state, are sensitively hinged on the environment in which convection develops.

Large forecast uncertainties are characteristic of non-equilibrium convective situations that depend on details of model physics and the small-scale flow whereas more predictable synoptic-scale forcing controls equilibrium convection. This project aims to develop methods and diagnostic tools to assess from the environment alone the extent to which a forecast is likely to benefit from better specification of the initial state at the convective-scale. It looks forward to the prospect of adaptive forecasting strategies, in which convective-scale DA would be prioritized in situations where it is valuable, whereas perhaps larger ensembles might be prioritized otherwise.

Links for this work

Papers

1. A paper on the impact of a new stochastic boundary layer perturbation scheme.
2. A paper providing an overview of results from the flood forecasting programme of which this project is a part.
3. A paper on convective-scale error growth and its dependence on convective regime.
4. PhD thesis (by David Flack) on the convective-scale error growth.
5. A paper on a UK climatology of the convective adjustment timescale.

Talks:

1. Stochastic boundary layer perturbations in the context of a wider class of stochastic parameterizations presented at the ECMWF Model Uncertainty workshop in May 2022.
2. Convective regime occurence within the UK presented (by David Flack) at the FRANC project meeting in April 2015.

Posters:

1. Results from the stochastic boundary layer parameterization presented (by Peter Clark) at the ECMWF Model Uncertainty workshop in May 2022.
2. The dependence of predictability on convective regime presented (by David Flack) at the AMS mesoscale meeting in San Diego, July 2017.
3. Sensitivity of adjustment timescale to its method of calculation presented (by David Flack) at the FRANC project meeting in June 2014.
4. Overiew of FFIR project presented at the MOAP poster conference in February 2014.

Others:

1. Working Group Report from the ECMWF Model Uncertainty workshop in May 2022.
2. Outreach exhibition at which David Flack and others from FFIR won a public prize for their exhibit about flash floods.
3. Departmental blog entry (by David Flack) about presenting some of this research at the EGU.
4. Main FFIR project page
5. Description of FFIR project in the MOAP newsletter