This paper describes the East Asia Reanalysis System (EARS), a regional reanalysis covering all of East Asia, with 3-hourly products provided at a horizontal resolution of 12 km and 74 levels in the vertical. The paper covers the methodology, the use of observations, and a variety of performance aspects. The paper is well organized and well written, with explanations in clear language.

In recent years, CMA has made great strides in developing an ambitious reanalysis program, which has already delivered a global atmospheric reanalysis (CRA40) and now also a unique regional reanalysis product. As the authors point out, EARS is the first regional reanalysis covering all of East Asia. This fills an important gap. 

According to the paper, all reanalysis data as well as many of the observations used will be accessible via the China Meteorological Data Service Centre at data.cma.cn. (I was not yet able to find the data when I tried during this review). It is very gratifying and good news for the global reanalysis community that CMA is making their data products and observation data available.

The work on observations that has been done in preparation of the ERAS production is significant and potentially very valuable. As the authors point out, many of the observations have not been used before, either for global numerical weather production or for reanalysis. It is very good news for the scientific research community if CMA is indeed able to share these data openly. It would be good to have more information (possibly in a separate paper) describing the observations and their quality control.

Overall, I think this is a good paper about an important dataset that can be highly valuable for large groups of users around the world. I have many questions and suggestions to the authors for additional work, but I don't think there is need for a major revision. My recommendation is therefore to publish after minor revision. 

Here are my comments and questions about the details:

If I understand correctly, the background fields used for the reanalysis are WRF short forecasts, which are initialized from ERA-Interim data, and using ERA-Interim data for lateral boundary conditions. There is a 6-h spin-up. Do you have any diagnostics (or have you investigated) the size of the spin-up for different variables, and whether this spin-up depends on the interval (e.g. 6-h vs. 12-h vs. 24-h)? Spin-up can be especially significant for precipitation and cloud, especially because the model used to generate ERA-Interim data is very different from the WRF model.

Can you provide statistics of the analysis increments (defined as: EARS analysis minus WRF forecast)? This will help to expose biases in the system, due to biases in the model and/or in the observations.

Can you provide more information about the nudging scheme used to introduce surface observations? Does the scheme depend on estimates of uncertainty of the observations?

Can you provide more information about the quality control steps used to prepare the input observations, especially the older observations recovered from analogue sources?

Can you provide details on any bias corrections applied to the observations?

What kind of automated quality control is applied in GSI for the upper-air analysis? Do you have any statistics on the rejection rates etc.?

Can you provide more information about the characteristics the background error covariances used in the GSI analysis?

Can you provide more information about the assimilation of radar data? Has there been any pre-processing of the radar data?

Many of the validation results in the paper refer to the improvements in EARS relative to ERA-Interim. Those are mostly good results, but they are not very surprising given the higher resolution and use of many additional observations. I think that it would be very useful to show more diagnostics that focus on the use of observations specifically, such as time series of observation-minus-background statistics.  These can be very informative and can be used to identify issues and problems with the observations and/or the data assimilation scheme, that could possible be addressed in a future reanalysis.

Fig 5: Is this for a single sounding? What does the shift signify? There is no description of the x-axis. 

Fig 11: I don't understand the grey shapes in this figure.

【General comments】

Yin et al. developed the East Asia Reanalysis System (EARS) and constructed a 39–year (1980–2018) reanalysis data over East Asia via the system with multi-source observations assimilated. The EARS and used observations are described in detail and principal work is conducted to validate the reliability of reanalysis datasets. This work is necessary and the conducted data has important potential applications for regional weather and climate studies. This manuscript is generally in a good shape. However, several minor revisions are still required before publication, listed as follows.

【Minor comments】

(1) The EARS covers a large domain. However, the observations out of China were not used in the validation. Although the results are reasonable and representative, it is advisable to give a detailed explanation in the text.

(2) Did the authors compare EARS with other regional and/or global reanalysis data, such as ERA5, CFSR, JMR, and others? This may be beyond the scope of this paper as the main purpose of this paper is to present EARS and preliminary results. If not, please specify this issue, which may encourage readers to conduct potential associated work.

(3) Given the present results, the EARS datasets are encouraging and promising. This paper is to report the progress of the project. I suggest the authors try to share all the EARS data to the public as soon as possible for potential applications.

(4) Lines 100-102: changing “intending to produce a high-resolution 100 regional atmospheric reanalysis dataset for East Asia, with high quality for mesoscale weather system study and regional climate analysis” to “intending to produce a high-resolution 100 regional atmospheric reanalysis dataset with high quality for mesoscale weather system study and regional climate analysis over East Asia ”

(5) Line 71: Please provide the horizontal resolution of China’s first generation of global atmospheric reanalysis (CRA40) for general information.

(6) Line 176: changing “regular” into “conventional”.

(7) lines 306 and 311: missing “the” before RMSE.

(8) Line 324: modifying “that WRF downscaling at a high resolution has significant performance gains in downscaling” to “significant performances have been gained in WRF downscaling at a high resolution”.

(9) Line 397: Please provide references for “previous studies and with operational predictions”.