The assimilation of Synthetic Aperture Radar (SAR) data in an operational data assimilation system for the
purpose of estimating sea ice properties is a relatively unexplored area. One of the areas where SAR data can
provide key information regarding the details of the ice cover is in the marginal ice zone. However, automatic
interpretation of the images in this area is challenging due to sensitivity of the backscattered signal to changing
surface and environmental conditions and speckle noise. In this paper the impact of denoising on the assimilation
of SAR texture features is examined. Two cases have been considered; one assimilating texture features without
denoising and the other one having a denoising step prior to the assimilation of texture features. The contribution
of the denoising to the analysis is evaluated.