Title: Challenges of assimilating observations for producing high-resolution sea-ice analyses.

Authors: Alain Caya (Environment Canada)
Mark Buehner (Environment Canada)
Michael Ross (Environment Canada)
Tom Carrieres (Environment Canada)

Assimilation of observations with different resolutions for producing high-resolution sea-ice analyses offers some interesting problems. Satellite borne passive microwave sensors are the primary source of information and complementary to higher resolution data obtained from other sensors (e.g. Synthetic Aperture Radar). Our analysis grid has a higher resolution than the passive microwave data and therefore an appropriate observation operator is required to avoid contaminating high resolution analysis details. A spatial averaging footprint operator is used to map fields from the analysis grid to the passive microwave observation locations. Another interesting aspect of sea-ice analysis is that the ice concentration analysis variable is bounded between zero and one. This violates the assumption of Gaussian distributed errors. Moreover, the sea-ice concentration field is discontinuous and often undergoes abrupt transitions between zero and one at the ice edge. This makes the background-error horizontal correlation particularly difficult to model. The problem is demonstrated using a simple Gaussian correlation function. A technique using displacement error as a control variable is proposed to partially overcome this challenge.

GMAO Head: Michele Rienecker Global Modeling and Assimilation Office NASA Goddard Space Flight Center

Curator: Nikki Privé Last Updated: May 27 2011