This study presents parameter estimation experiments utilizing the CESM model to assimilate Sea Surface Temperature (SST) and Temperature/Salinity (T/S) profiles for initializing ENSO prediction. The results demonstrate the potential of parameter estimation over state estimation, revealing enhanced ENSO prediction skills achieved through more accurate parameter estimates. This work is interesting and worthy of publication. However, some minor revisions are necessary before acceptance.

Major Comments:

Starting from line 141, the authors conducted sensitivity experiments by perturbing multiple parameters to assess the model's temperature and salinity variables' sensitivity to those parameters. It is noted that parameters were perturbed simultaneously. Have the authors considered perturbing these parameters individually? Could the sensitivity of variables to different parameters differ?

From line 188, the authors mention that during the first phase of Parameter Estimation (PE), parameters were perturbed but only State Estimation (SE) was used, lasting for a year. The state variables employed for PE and the motivation for this approach need clarification.

In Figure 8, the authors present Root Mean Square Error (RMSE) without specifying the reanalysis data it pertains to. Despite earlier indications of similar results from different reanalysis datasets, it is advisable to explicitly mention the data used. Additionally, line 219 asserts that the maximum error occurs at the depth most sensitive to parameters, which is not immediately apparent. It is recommended to include a subfigure depicting parameter sensitivity along the equatorial range, using a logarithmic depth coordinate. Similar concerns are noted in Figure 9.

Minor Suggestions:

In lines 56-58, apart from the atmosphere, ocean, land, and sea ice, CESM encompasses other modules as well. The authors should use "as well as other modules" to accurately depict the model.

Is Equation (1) valid outside the Banda Sea region, using a value of 1.0 within the Banda Sea? Clarify this description for improved understanding.

Regarding line 108, the authors mentioned "daily profiles were merged and assigned to the final day of each sequence". Is there any other references employed the same approach to process the data?

Line 109 mentions interpolation to 31 layers. Could the specific depths of these layers be provided?

Please find the attachment.