************************************************************************ README_data.txt Petri Räisänen, Finnish Meteorological Institute, 28 Aug 2024. ************************************************************************ This file contains information on the contents of data files related to the manuscript Räisänen et al. (2024) Räisänen, P., Luomaranta, A., and Jylhä, K. (2024): Future snow scenarios for Northern Europe based on Coupled Model Intercomparison Project phase 6 data. International Journal of Climatology, submitted. Data are provided in 11 zip files named as follows. They are listed in the order their contents are described below. yearly_historical+ssp126.zip yearly_historical+ssp245.zip yearly_historical+ssp370.zip yearly_historical+ssp585.zip yearly_historical+ssp245_ACCESS-ESM1-5.zip yearly_historical+ssp245_CanESM5.zip yearly_historical+ssp245_MIROC6.zip yearly_ERA5-Land.zip trends.zip areal_means.zip composite.zip ******************************************************************************* DATA FORMAT: The data are in direct access binary format. This format can be read by the GrADS software (http://cola.gmu.edu/grads/) and (e.g.) by fortran programs, but it is not self-describing. Therefore, one has to know in which order the data records are written, in order to be able to read the data properly. The order of the records can be seen from the GrADS control files (.ctl) provided in the zip packages. ****************************** 1. yearly_historical+ssp126.zip yearly_historical+ssp245.zip yearly_historical+ssp370.zip yearly_historical+ssp585.zip ****************************** These zip files contain yearly snow season metrics calculated for the CMIP6 models for snow years 1951-2100 (i.e., winters 1950/1951 -> 2099/2100). The files combine data from historial experiments (years 1951-2014) with data from scenario experiments (years 2015-2100) for four emission scenarios: SSP126, SSP245, SSP370 and SSP585 (15-17 models depending on the scenario). Only a single realization is included for each model and emission scenario. The files are provided both at the native horizontal resolution of each model and interpolated to a uniform 0.50 x 0.50 degree grid. The files are named as output_yearly-{MODEL}.grads output_yearly-{MODEL}-0.50x0.50_deg.grads and the corresponding GrADS control files are output_yearly-{MODEL}.ctl output_yearly-{MODEL}-0.50x0.50_deg.ctl A total of 17 variables are included in these files, only a subset of which are actually used Räisänen et al. (2024). The list of variables can be seen from the .ctl files: VARS 17 tas_month 12 99 Monthly-mean 2-m air temperature (Sep -> Aug) swe_month 12 99 Monthly-mean SWE (Sep -> Aug) snowday_month 12 99 Monthly-mean snow occurrence (0-1) (Sep -> Aug) tas_year 0 99 Yearly-mean 2-m air temperature (Sep -> Aug) swe_year 0 99 Yearly-mean SWE (Sep -> Aug) snowday_year 0 99 Yearly-mean snow occurrence (0-1) (Sep -> Aug) nsnowday 0 99 Number of days with snow per "snow year" nsnowseason 0 99 Indicator for a snow season with "good statistics" dayfirst 0 99 Day of year for the first day with snow daylast 0 99 Day of year for the last day with snow daystart 0 99 Day of year for the start of the longest snow period dayend 0 99 Day of year for the end of the longest snow period length 0 99 Length of the longest snow period (days) swemax 0 99 Maximum value of SWE dayswemax 0 99 Day of year for maximum SWE daysnowoff1 0 99 Snow-off day of year, based on maximum SWE daysnowoff2 0 99 Snow-off day of year, based on longest snow period ENDVARS Here, the first three variables are given for 12 months of the year, treated here for convenience as the vertical dimenions. The remaining 14 variables have only a single value for each year and grid point. (The value "99" for the GrADS units field is a place holder with no practical meaning). ******************************************** 2. yearly_historical+ssp245_ACCESS-ESM1-5.zip yearly_historical+ssp245_CanESM5.zip yearly_historical+ssp245_MIROC6.zip ******************************************** These zip files contain yearly snow season metrics for snow years 1951-2100 calculated for large single-model ensembles (40-50 realizations) for the ACCESS-ESM1-5, CanESM5 and MIROC6 models. The files combine historical runs for years 1951-2014 with the SSP245 scenario for years 2015-2100. These files are provided both at the native resolution of each model and interpolated to a 0.50 x 0.50 deg grid. The files are named as output_yearly-{MODEL}_{REALIZATION}.grads output_yearly-{MODEL}_{REALIZATION}-0.50x0.50_deg.grads and the corresponding GrADS control files are output_yearly-{MODEL}_{REALIZATION}.ctl output_yearly-{MODEL}_{REALIZATION}-0.50x0.50_deg.ctl The list of variables is otherwise identical to that shown above, except that the two temperature-related variables (tas_month and tas_year) are not included. ************************* 3. yearly_ERA5-Land.zip ************************* Yearly snow season metrics computed from ERA5-Land data for snow years 1951-2023, at original 0.1 deg resolution and conservatively averaged to 0.50 x 0.50 deg grid. The files are named as output_yearly-ERA5-Land.grads output_yearly-ERA5-Land-0.50x0.50_deg.grads and the corresponding GrADS control files are output_yearly-ERA5-Land.ctl output_yearly-ERA5-Land-0.50x0.50_deg.ctl All 17 variables listed above are included in these files. **************** 4. trends.zip **************** This zip file contains trends in snow season metrics for CMIP6 models and ERA5-Land reanalysis data. Five directories are included: historical+ssp126 historical+ssp245 historical+ssp370 historical+ssp585 ERA5-Land The first four directories contain trends for CMIP6 models for years 1951-2023 and 2023-2100, combining historical experiments with four alternative emission scenarios. The last directory contains trends in ERA5-Land data for 1951-2023. The trends are computed based on snow season metrics represented in a uniform 0.50 x 0.50 deg grid. The data files (.grads) and GrADS control files (.ctl) are names as follows: (i) For CMIP models, with a single realization per experiment trends-{MODEL}-1951-2023-0.50x0.50_deg.grads trends-{MODEL}-2023-2100-0.50x0.50_deg.grads trends-{MODEL}-1951-2023-0.50x0.50_deg.ctl trends-{MODEL}-2023-2100-0.50x0.50_deg.ctl (ii) For the three CMIP6 models with large single-model ensembles (ACCESS-ESM1-5, CanESM5 and MIROC6; for the historical+ssp245 experiments only) trends-{MODEL}_ensemble-1951-2023-0.50x0.50_deg.grads trends-{MODEL}_ensemble-2023-2100-0.50x0.50_deg.grads trends-{MODEL}_ensemble-1951-2023-0.50x0.50_deg.ctl trends-{MODEL}_ensemble-2023-2100-0.50x0.50_deg.ctl (iii) For ERA5-Land data trends-ERA5-Land_1951-2023-0.50x0.50_deg.grads trends-ERA5-Land_1951-2023-0.50x0.50_deg.ctl As indicated by the .ctl files, trends were computed for five snow season metrics: VARS 5 nsnowday 9 99 Number of days with snow per "snow year" daystart 9 99 Day of year for the start of the longest snow period dayend 9 99 Day of year for the end of the longest snow period length 9 99 Length of the longest snow period (days) swemax 9 99 Maximum value of SWE ENDVARS Three options were employed for calculating the trends 1) Ordinary linear regression 2) Ordinary linear regression with pre-whitening to account for lag-1 autocorrelations, based on Eq. (A11) in Wang, X. L., and V. R. Swail, 2001: Changes of Extreme Wave Heights in Northern Hemisphere Oceans and Related Atmospheric Circulation Regimes. J. Climate, 14, 2204–2221, https://doi.org/10.1175/1520-0442(2001)014<2204:COEWHI>2.0.CO;2. 3) Theil-Sen regression, with Mann-Kendall test to determine statistical significance In practice, the trends calculated using these three options agree closely. For simplicity, results for the ordinary linear regression are used Räisänen et al. (2024). For each of these options, three parameters are provided: the intercept parameter, the slope parameter and the Z-value related to statistical significance. Hence, for each of the five variables listed above, nine (3 times 3) numerical values are provided for each grid point. ******************** 5. areal_means.zip ******************** This zip file contains yearly areal-mean values of snow season metrics computed for CMIP6 models for years 1951-2100 for two regions: Southern Finland ("south"; 60.5-62N, 23-27E) and Finnish Lapland ("north"; 67-69N, 24-28E). Also included are 20-year running-mean statistics computed from these values. Four directories are included, each combining data for the historical experiments with one of the four SSP emission scenarios: historical+ssp126 historical+ssp245 historical+ssp370 historical+ssp585 The areal-mean files and corresponding GrADS control for CMIP6 models with a single realizaztion are named as avg_South_yearly-{MODEL}.grads avg_North_yearly-{MODEL}.grads avg_South_yearly-{MODEL}.ctl avg_North_yearly-{MODEL}.ctl For the historical+ssp245 experiments, areal-mean values are also provided for the large single-model ensembles available for three models (ACCESS-ESM1-5, CanESM5 and MIROC6). These files are names as avg_South_yearly-{MODEL}_ensemble.grads avg_North_yearly-{MODEL}_ensemble.grads avg_South_yearly-{MODEL}_ensemble.ctl avg_North_yearly-{MODEL}_ensemble.ctl Based on the yearly areal-mean values, statistics were computed both for the ensemble of 11 CMIP6 models approved for developing snow scenarios in Räisänen et al. (2024) (with a single realization per model) and for the three large single-model ensembles (ACCESS-ESM1-5, CanESM5, MIROC6). For the set of 11 CMIP6 models, both unweighted statistics and statistics weighted based on the model Transient Climate Response (TCR) were computed (see the section on Model weighting in Räisänen et al. 2024). These files are named as statistics_South_CMIP6_unweighted.grads statistics_North_CMIP6_unweighted.grads statistics_South_CMIP6_TCR-weighted.grads statistics_North_CMIP6_TCR-weighted.grads and the corresponding GrADS control files are statistics_South_CMIP6_unweighted.ctl statistics_North_CMIP6_unweighted.ctl statistics_South_CMIP6_TCR-weighted.ctl statistics_North_CMIP6_TCR-weighted.ctl The files for the three large single-model ensembles (in the directory "historical+ssp245") are named as statistics_South_{MODEL}.grads statistics_North_{MODEL}.grads statistics_South_{MODEL}.ctl statistics_North_{MODEL}.ctl where {MODEL} stands for ACCESS-ESM1-5, CanESM5 or MIROC6. The following statistics are included in these files, for the 17 snow season metrics for the CMIP6 ensembles and 15 snow season metrics for the single-model ensembles: - ensemble-mean 20-year running-mean values - standard deviation of the 20-year running mean values - minimum and maximum of the 20-year running mean values - 5th, 25th, 50th, 75th and 95th percentage point of the 20-year running mean values - ensemble-mean difference of 20-year runnning-mean values to the reference period 1981-2010 - standard deviation of the differences of 20-year running mean values to the reference period 1981-2010 - minimum and maximum of the differences of 20-year running mean values to the reference period 1981-2010 - 5th, 25th, 50th, 75th and 95th percentage point of the differences of 20-year running mean values to the reference period 1981-2010 Note that the differences for the metrics involving the snow water equivalent, that is swe_month, swe_year and swemax, relative differences are given (in per cent), while for the other metrics, absolute differences are given. ******************* 6. composite.zip ******************* This directory contains results of the composite analysis of SWE and 2-m air temperature in the snow melt season (from 60 days before to 10 days after the snow-off date) for CMIP6 models and for the ERA5-Land reanalysis. Data files are provided both in the native grid of each CMIP6 model and ERA5-Land reanalysis, and in a uniform 0.50 x 0.50 deg grid. The data (.grads) and control file (.ctl) names for CMIP6 models are output_composite-{MODEL}.grads output_composite-{MODEL}-0.50x0.50_deg.grads output_composite-{MODEL}.ctl output_composite-{MODEL}-0.50x0.50_deg.ctl The corresponding files for ERA5-Land data are output_composite-ERA5-Land.grads output_composite-ERA5-Land-0.50x0.50_deg.grads output_composite-ERA5-Land.ctl output_composite-ERA5-Land-0.50x0.50_deg.ctl These files provide mean values of SWE, 2-m air temperature, and probability of snow on ground (defined as SWE > 1 kg m-2) composited with respect to the time before the snowoff-day (first day with SWE < 1 kg m-2) after the winter's main snow period. Two definitions of the "main snow period" were used. 1) The period in which the winter's maximum SWE occurs. 2) The winter's longest period with SWE > 1 kg m-2. The latter definition was used in Räisänen et al. (2024) (in Supporting information Figure S1) but both give very similar results. Thus the variable list is as follows: VARS 6 tas1 0 99 average temperature (K), snow-off based on SWE maximum snw1 0 99 average SWE (kg m-2), snow-off based on SWE maximum psnow1 0 99 probability of snow present, snow-off based on SWE maximum tas2 0 99 average temperature (K), snow-off based on longest snow period snw2 0 99 average SWE (kg m-2), snow-off based on longest snow period psnow2 0 99 probability of snow present, snow-off based on longest snow period ENDVARS