Despite the improvements in SIC and SIT, we note that sea ice volume shows significantly different time variability among the different reanalyses (Figure 6). For instance, our model simulations and the PIOMAS reanalysis show a more substantial seasonal variation than the TOPAZ4 data, especially before the year 2011. INTAROS-opt changes the total sea ice volume in the summer season throughout the 10 years and in the winter season after the year 2010 when additional SIT observations are available. Despite the different assimilation methods and numerical models, we see that the state-of-the-art ocean-sea ice models reproduce the spatiotemporal-varying SIC successfully. We note that SIT is also improved either through rectifying mechanisms by improvements on SIC or by assimilating additional SIT data in the winter season. Beneficial effects of assimilating satellite SIT are more clearly visible in the TOPAZ4 reanalysis after January of the year 2014.
To examine the relative importance of SIC and SIT on the residual sea ice volume, we replace each component in INTAROS-opt with observations (blue line and black dots in Figure 6). Replacing SIT with observations (black dots in Figure 6) achieves a better match with the observed sea ice volume than replacing SIC with observations (blue line in Figure 6), indicating that the sea ice volume improvement mostly results from the SIC assimilation and that SIT needs to be further improved for further reducing sea ice volume error.