| Summary: | The values of many of the parameters in climate models are often not known with any great precision. We describe the use of automatic differentiation to examine the sensitivity of an uncoupled dynamicthermodynamic sea-ice model to various parameters. We also illustrate the effectiveness of using these sensitivity derivatives with an optimization algorithm to tune the parameters to maximize the agreement between simulated results and observational data. 1 Introduction The sea-ice cover in polar regions plays an important role in modeling the Earth's climate system. For example, sea-ice acts as a powerful insulating boundary layer, reducing the atmosphere-ocean heat exchange and reflecting incoming solar energy on the surface of sea-ice covers. Also sea-ice is a major factor in the thermohaline circulations [MMM90],[Jac97]. Thus, seaice is capable of profoundly amplifying and modulating the climate variations in regional and global-scale climate systems. For this reason, sea-ice .
|
|---|