Numerical data analysed to produce Figure 3a of Nature Climate Change submission "Five challenges for subseasonal to decadal prediction research " by Merryfield et al.

NetCDF4-formatted files containing daily sea ice concentration data from Environment and Climate Change Canada's CanSIPSv2 seasonal forecasting system described in Lin et al. (2020) https://doi.org/10.1175/WAF-D-19-0259.1 

• 2 models, CanCM4i and GEM-NEMO
• 10 ensemble members for each model, each in separate files as indicated by suffixes _1 to _10
• initialized May 1, 1980 to 2021
• 840 files total (42 predicted years x 10 ensemble members x 2 models)
• model outputs interpolated to common 1-degree grid

The calibrated probabilistic forecast map shown in Figure 3a is based on the nonhomogeneous censored Gaussian regression (NCGR) method described in Dirkson et al, (2021) https://doi.org/10.1175/WAF-D-20-0066.1 and produced using scripts available at https://github.com/adirkson/sea-ice-timing 

The procedure uses as inputs

• freeze-up dates calculated from the provided model outputs as described in Sigmond et al. (2016) https://doi.org/10.1002/2016GL071396

• NOAA/NSIDC Climate Data Record of Passive Microwave Sea Ice Concentration, Version 3: https://nsidc.org/data/G02202/versions/3