Large data sets created within the MaRIUS project (Managing the Risks, Impacts and Uncertainties of drought and water Scarcity).
Three time periods are covered: historical baseline (BS: 1900-2006), near-future (NF: 2020-2049) and far-future (FF: 2070-2099), whereby the RCP8.5 is assumed for future time slices. The 30-year baseline to which future time slices should be compared to are years 2075-2004 of the historical baseline, since oceanic variability in future time slices is based on these years.
One set of 100 time series is available for the historical baseline, while 5 sets of about 85-100 time series are available for each future time slice. The five sets sample climate model uncertainty by sampling a range of sea surface temperature warming patterns as follows: nf and ff use the median warming pattern of CMIP5 climate models. The four others sets are called nf-pYYz (near future) and ff-pYYz (far future) whereby 'YY' stands for the percentile of global mean SST warming (YY=10 or 90) and 'z' stands for the North Atlantic SST gradient (z=n for minimum, z=x for maximum). All natural forcings (e.g., volcanoes) in the future time slices are replications of years 1975-2004, while anthropogenic forcings (e.g., CO2 concentrations) follow the high-emission scenario RCP8.5.
The time series were created from single-year simulations that have been stitched together while ensuring continuity in the slowly-evolving variability (see Guillod et al., Hydrol. Earth Syst. Sci., 2017 for details). Therefore, for each set (e.g., near_future_p10n), a 'stitching table' is provided alongside with data from individual yearly simulations, which contains the simulation ID ('umid') for each year and each individual time series (i.e., each column within the table list the umid for each year of the time series).
Data are gridded NetCDF V3 files, provided on a rotated longitude-latitude grid at a 0.22 degree resolution (European CORDEX grid) over a domain encompassing the United Kingdom. A subset of variables are available at daily, 5-days, and monthly temporal resolutions, based on a calendar with 360 days a year (30 days per month). Included are only data from HadRM3P, the regional climate model within weather@home2. After unpacking yearly *tgz files for a given time slice and scenario (e.g., near_future_p10n) within a directory with the relevant name, the data are structured as follows:
timeslice / year / umid / temporalResolution / WAH_umid_variable_temporalResolution_g2_year.nc
where:
- The temporal resolution ('temporalResolution') available depends on the variable, but it can be 'daily', 'monthly', or '5days'
- 'timeslice' can be one of: 'baseline', 'near_future', 'near_future_p10n', 'near_future_p10x', 'near_future_p90n', 'near_future_p90x', 'far_future', 'far_future_p10n', 'far_future_p10x', 'far_future_p90n', 'far_future_p90x'.
- 'year' is the year of the simulation and data.
- 'umid' is the ID of the simulation, to be read from the stitching table.
- 'variable' is the variable, following CMOR conventions (see also Table 3 in Guillod et al., Hydrol. Earth Syst. Sci., 2017).
Further details about the data, including validation of the climate time series and analyses of some changes in the future projections, can be found in the PDF documentation as well as in Guillod et al., 2017b. A detailed validation of the model (weather@home2) can be found in Guillod et al., 2017a (see list of references below).