Abstract/Summary

Forecasts of lightning in the Met Office’s Unified Model (MetUM) are known to overforecast the total number of lightning flashes in the UK. One of the difficulties in understanding why this happens comes from the dependence of the lightning forecast on the convective forecast. This problem can be mitigated using a more physically representative forecasting method to compare against and by comparing lightning production in modelled thunderstorms to that in observed thunderstorms. In order to provide a more physically representative forecasting method a new, explicit thunderstorm electrification and lightning scheme is implemented within the MetUM. This scheme uses non-inductive collisional charging to represent the charge generated on hydrometeors and produce a charge density distribution. From this, the magnitude of the electric field is calculated, with appropriate thresholds selected to allow initiation of lightning events. It is shown that this scheme accurately represents observed thunderstorm charge magnitude and structure. Results from the new electrification scheme are compared with those from the existing lightning parameterisation within the MetUM, and to natural lightning observations in two case studies. The new electrification scheme performs well in both a scattered, fair weather convection case study in the UK and an organised, deep convection case study in the US. It shows realistic lightning coverage and reproduced the daily lightning flash accumulation relatively accurately. The collision-separation efficiency is found to be a key parameter and therefore a potential source of uncertainty in the scheme. Through comparison with the new scheme, the existing MetUM parameterisation is shown to be producing lightning in a manner that is too closely dependent on the rainfall accumulation, which it is suggested is related to its poor performance in the UK case study. Observations of single cell thunderstorms are used to investigate the production of lightning in thunderstorms in the UK. It is found that prior to the onset of lightning production, single cell thunderstorms show an increase in storm core area. Model simulations of similarly intensifying thunderstorms show that, during these intensifications, the updraft velocity and area both increase, as does the graupel mass in the storm core. It is shown that the new electrification scheme can reproduce the increase in updraft area and graupel mass in intensifying storms, whereas the existing lightning parameterisation does not reproduce any of these parameters. The use of this new electrification scheme, whilst not operationally feasible with existing computer power, provides a research tool with which to further improve lightning forecasting.