|I am a new reviewer. The study is interesting and of potential great significance to relevant research on weather forecast and climate modeling. After reading through the manuscript and the authors’ response to the comments from two Anonymous Referees, I find the authors do a good job in addressing most of the reviewers’ comments. However, the manuscript might still need some revisions before accepted for publication in ACP. I fall between minor revision and major revision and suggest a major revision such that there will be enough time for the authors to revise.|
(1) I think some of Anonymous Referee #2’s comments suggest a fair evaluation. For example, “Major comments 3b: Figure 4 and 8 … whereas at V4km the precipitation magnitudes seem too positively biased over a greater area and longer time…”. The authors did reply to this on timing and paid less attention to the magnitude. I understand it is very important to capture the locations of precipitation, but overestimation of precipitation is also important to note as it may cause “false alarm” in weather forecast. Please adjust some words to mention the overestimation in the model. In addition, mean bias and root mean square error are good metrics to evaluate the magnitude. The authors can provide the metrics as in Table 2.
A fair evaluation also lies in other aspects:
(a) the spatial correlation is calculated for the selected domain and indicates the mean metric for the whole domain. However, some analysis is shifted to the rain belt. To find the locations of rain belts which the authors point out in the text, please denote the outlines of rain belt (e.g., > a selected value) or heavy rain belt (e.g., > a selected value) using contour lines.
(b) From Figure 4(8) to Figure 3(7), a temporal mean is derived. Zonal mean results (magnitude) may depend on the longitude range used for the calculation. The author should explicitly mention the domain extent (I suppose the domain extends from 114 E to 122.5 E) and justify why this range is used. In addition, if zonal mean is derived, the temporal variation of precipitation can be derived and the temporal shift of precipitation in the models can be clearly shown. I suggest temporal variations can be provided (the authors can decide where to put it: in main text or supplement).
(2) Section 2.1.2: Is MPAS ran in weather-forecast mode? In particular, what are the surface boundary conditions (such as SST and surface temperature/soil moisture) used for the model run? Please clarify whether it is fair to compare MPAS and GFS.
(3) Section 2.2: I am wondering how many stations are used and what is minimum/mean/maximum distance between a station and surrounding stations. It is possible that the observations miss some extreme precipitations, which may contribute partly to the model biases relative to the current observation (e.g., overestimation mentioned earlier). In addtion, I am wondering whether data control is done on hourly data? In my opinion, evaluation of high-resolution model needs high-resolution observations. Please provide some information on the station observations.
(1) The authors should consistently set the orders of experiments in Figures 11-13 as in previous Figures 5-10: U60km.WSM6, V30km.WSM6, V16km.WSM6, V4km.WSM6, V4km.Thompson.
(2) Section 2.1.1, Lines 218-220: Since there are two options of PBL scheme, which one is used in the study?
(3) Section 2.2, Lines 307-308: I cannot open https://www.ncdc.noaa.gov/data-access/model-data/model-datasets/global-forcast-system-gfs.
(4) Section 3.1, Lines 337-338: “with only negligible impacts from different convective parameterizations” is not clear.
(5) Section 3.2.3, Figure 10: It will be useful if the authors can provide a map with geographical locations of provinces, mountains, plains mentioned in the main text. The map can be put in Figure 10.
(6) Section 3.2.3, Lines 574-575: Change "in that simulations" to "in those simulations"?
(7) Conclusions, Lines 609-663: The sentences can be shorten be one-third or one-half.