The use of kilometre‐scale ensembles in operational weather forecasting provides new challenges for forecast interpretation and evaluation to account for uncertainty on the convective scale. A new neighbourhood‐based method is presented for evaluating and characterizing the local predictability variations from convective‐scale ensembles. Spatial scales over which ensemble forecasts agree (agreement scales, SA) are calculated at each grid point ij, providing a map of the spatial agreement between forecasts. By comparing the average agreement scale obtained from ensemble member pairs (SijA(falsemm¯)) with that between members and radar observations (SijA(falsemo¯)), this approach allows the location‐dependent spatial spread–skill relationship of the ensemble to be assessed. The properties of the agreement scales are demonstrated using an idealized experiment. To demonstrate the methods in an operational context, SijA(falsemm¯) and SijA(falsemo¯) are calculated for six convective cases run with the Met Office UK Ensemble Prediction System (MOGREPS‐UK). SijA(falsemm¯) highlights predictability differences between cases, which can be linked to physical processes. Maps of SijA(falsemm¯) are found to summarize the spatial predictability in a compact and physically meaningful manner that is useful for forecasting and model interpretation. Comparison of SijA(falsemm¯) and SijA(falsemo¯) demonstrates the case‐by‐case and temporal variability of the spatial spread–skill, which can again be linked to physical processes.

New techniques have recently been developed to quantify the location‐dependent spatial agreement between ensemble members, and the spatial spread–skill relationship. In this article a summer of convection‐permitting ensemble forecasts are analysed to better understand the factors influencing location‐dependent spatial agreement of precipitation fields and the spatial spread–skill relationship over the UK. The aim is to further investigate the agreement scale method, and to highlight the information that could be extracted for a more long‐term routine model evaluation. Overall, for summer 2013, the UK 2.2 km grid spacing ensemble system was found to be reasonably well spread spatially, although there was a tendency for the ensemble to be overconfident in the location of precipitation. For the forecast lead times considered (up to 36 h), a diurnal cycle was seen in the spatial agreement and in the spatial spread–skill relationship: the forecast spread and error did not increase noticeably with forecast lead time. Both the spatial agreement and the spatial spread–skill were dependent on the fractional coverage and average intensity of precipitation. A poor spread–skill relationship was associated with a low fractional coverage of rain and low average rain rates. The times with a smaller fractional coverage, or lower intensity, of precipitation were found to have lower spatial agreement. The spatial agreement was found to be location dependant, with higher confidence in the location of precipitation to the northwest of the UK.

With movement toward kilometer-scale ensembles, new techniques are needed for their characterization. A new methodology is presented for detailed spatial ensemble characterization using the fractions skill score (FSS). To evaluate spatial forecast differences, the average and standard deviation are taken of the FSS calculated over all ensemble member-member pairs at different scales and lead times. These methods were found to give important information about the ensemble behavior allowing the identification of useful spatial scales, spinup times for the model, and upscale growth of errors and forecast differences. The ensemble spread was found to be highly dependent on the spatial scales considered and the threshold applied to the field. High thresholds picked out localized and intense values that gave large temporal variability in ensemble spread: local processes and undersampling dominate for these thresholds. For lower thresholds the ensemble spread increases with time as differences between the ensemble members upscale. Two convective cases were investigated based on the Met Office United Model run at 2.2-km resolution. Different ensemble types were considered: ensembles produced using the Met Office Global and Regional Ensemble Prediction System (MOGREPS) and an ensemble produced using different model physics configurations. Comparison of the MOGREPS and multiphysics ensembles demonstrated the utility of spatial ensemble evaluation techniques for assessing the impact of different perturbation strategies and the need for assessing spread at different, believable, spatial scales.

The Convective Precipitation Experiment (COPE) was a joint U.K.–U.S. field campaign held during the summer of 2013 in the southwest peninsula of England, designed to study convective clouds that produce heavy rain leading to flash floods. The clouds form along convergence lines that develop regularly as a result of the topography. Major flash floods have occurred in the past, most famously at Boscastle in 2004. It has been suggested that much of the rain was produced by warm rain processes, similar to some flash floods that have occurred in the United States. The overarching goal of COPE is to improve quantitative convective precipitation forecasting by understanding the interactions of the cloud microphysics and dynamics and thereby to improve numerical weather prediction (NWP) model skill for forecasts of flash floods. Two research aircraft, the University of Wyoming King Air and the U.K. BAe 146, obtained detailed in situ and remote sensing measurements in, around, and below storms on several days. A new fast-scanning X-band dual-polarization Doppler radar made 360° volume scans over 10 elevation angles approximately every 5 min and was augmented by two Met Office C-band radars and the Chilbolton S-band radar. Detailed aerosol measurements were made on the aircraft and on the ground. This paper i) provides an overview of the COPE field campaign and the resulting dataset, ii) presents examples of heavy convective rainfall in clouds containing ice and also in relatively shallow clouds through the warm rain process alone, and iii) explains how COPE data will be used to improve high-resolution NWP models for operational use.