Spin-up simulation behaviors in a climate model to build a basement of long-time simulation

Abstract:

It is essential to develop start-up information when conducting long-time climate simulation. In case that the initial condition is already available from the previous simulation of same type model this does not necessary; however, if not, model needs spin-up simulation to have adjusted and balanced initial condition with the model climatology. Otherwise, a severe spin may take several years. Some of model variables such as deep soil temperature fields and temperature in ocean deep layers in initial fields would affect model’s further long-time simulation due to their long residual memories. To investigate the important factor for spin-up simulation in producing an atmospheric initial condition, we had conducted two different spin-up simulations when no atmospheric condition is available from exist datasets. One simulation employed atmospheric global circulation model (AGCM), namely Global Forecast System (GFS) of National Center for Environmental Prediction (NCEP), while the other employed atmosphere-ocean coupled global circulation model (CGCM), namely Climate Forecast System (CFS) of NCEP. Both models share the atmospheric modeling part and only difference is in applying of ocean model coupling, which is conducted by Modular Ocean Model version 4 (MOM4) of Geophysical Fluid Dynamics Laboratory (GFDL) in CFS. During a decade of spin-up simulation, prescribed sea-surface temperature (SST) fields of target year is forced to the GFS daily basis, while CFS digested only first time step ocean condition and freely iterated for the rest of the period. Both models were forced by CO₂ condition and solar constant given from the target year. Our analyses of spin-up simulation results indicate that freely conducted interaction between the ocean and the atmosphere is more helpful to produce the initial condition for the target year rather than produced by fixed SST forcing. Since the GFS used prescribed forcing exactly given from the target year, this result is unexpected. The detail analysis will be discussed in this presentation.