What input data does SD-GCM need?
Answer: For future downscaling, SD-GCM needs three aligned datasets: observed station data, historical GCM data that overlap the observations, and future CMIP5 RCP or CMIP6 SSP data. For historical model evaluation alone, the observation and historical GCM datasets are sufficient. The variable, time step, date interpretation, and final units must be compatible across inputs.
Observed station data
Provide a dated local time series for one or more stations, with station names and coordinates where gridded extraction is required. The observations define the local historical climate used to calibrate and evaluate the correction.
Historical and future GCM data
The historical model run must overlap the observation record. The future run should use the same GCM, variable, ensemble member, grid conventions, and compatible time step; SD-GCM can load gridded NetCDF data or prepared tabular model series supported by its workflow.
Review the SD-GCM input workflow for CMIP5/CMIP6 observations, historical runs, and scenario data.