This dataset consists of 3D spatial grids of weather radar reflectivity, which have 5-minute temporal, 1km horizontal, and 500m vertical resolution. They are constructed from UK weather radar network scans, provided by 16 radars in England, Scotland, Wales, Northern Ireland, and the Channel Islands.

In addition to the 3D, there are some 2D grids of fields derived from the vertical grid columns, including maximum column dBZ and vertically integrated liquid water. Please see descriptions below.

Note – this dataset contains non-operational data products, with this time-limited dataset provided primarily to aid use within the ParaChute research programme.

The interpolation method used to arrive at the multi-radar gridded values is similar to that described in Zhang (2005). The reasons for choosing this method over another more recent one (Scovell and al-Sakka, 2016) can be found in Stein et al. (2020).

The horizontal domain spans X=[-405000, 1320000], Y=[-625000, 1550000] metres on the UK National Grid (EPSG:27700) projection. This is regularly spaced, with 2175 rows x 1725 columns, and is the same as the “Nimrod” grid used by RadarNet (Harrison et al., 2009). Grid points are located at the centres of each grid box (at X/Y coordinates ending in 500). The vertical is comprised of 24 evenly spaced 500m height levels in the range h=[250,11750] metres AMSL, with the first at 250m AMSL.

The data are temporally continuous, at 5-minute resolution, from 2023-06-01 00:00 UTC to 2023-08-31 23:55 UTC. An exception being for two periods of network outage, which are 2023-06-12 17:00-19:00 UTC, and 2023-08-14 08:00-09:00 UTC.

The 3D radar grids are formed using scan data following the operational scanning strategy of the UK. This favours low elevation angles, to aid with surface quantitative precipitation estimation. Thus, at higher altitudes, coverage can be sparse (Scovell and al-Sakka, 2016) and the observations are of relatively poor quality, being at long range. No 3D grid point has a data value that has been extrapolated beyond 2.5km range horizontally. Thus, there are large data voids ~10km, at the highest altitude levels. Smaller gaps can appear at lower altitudes. At the lowest levels, and at long range from a radar site, there may sometimes be no coverage. This is unavoidable, due to the curvature of the Earth.

DATASETS
The data are stored in an HDF5 file format, with the standard HDF5-native gzip compression. The stored attributes and datasets are based on, but do not strictly adhere to, the ODIM data model specification (Michelson et al., 2008).
The following ODIM quantities encoded:
• DBZH: 3D reflectivity composite
• ZDR: 3D ZDR composite
• RHOHV: 3D Fisher-Z (arctanh) -transformed RHOHV composite
• MAXDBZ: 2D “column maximum” , derived from DBZH. In numpy these are computed with np.max ( reflectivity, axis = 0)
• VIL: 2D Vertically Integrated Liquid water, as in Green and Clarke (1972)
• TOP45, TOP18: echo top heights (highest height level) for DBZH > 45/18
• POH: Probability of Hail; equal to f * ( TOP45 – height of T=0C isotherm ), as in DeLobbe and Holleman (2003).
• VII, CRIT_IND: Vertically Integrated Ice and (lightning) Criterion Index, as defined in Mosier et al. (2011), and Haklander (2014).
• SHI, POSH, MEHS: these are hail and lightning indices derived from formulae in Witt et al. (c. 1998)
The following caveats apply to the ODIM formatting:
• Unofficial non-compliant ODIM attributes have been added to allow storage of 3D information in the ODIM HDF5 format.
• The metadata describing the 3D grids are not complete.

See the online resources section for full citations used on this record.