| 25 Sep 2026
Development and evaluation of the GRIMs–NEMO fully coupled model v1.0 for simulating the East Asian Summer Monsoon
Abstract. This study introduces the newly developed GRIMs–NEMO fully coupled model version 1.0 for the first time and assesses its capability to simulate the boreal summer climate, with a particular focus on the East Asian summer monsoon (EASM). The model couples the GRIMs atmospheric model, NEMO ocean, and SI3 sea ice models through OASIS3-MCT, providing a computationally efficient framework for climate simulation. Benchmark experiments show that GRIMs–NEMO requires substantially fewer computational resources than does CESM2, suggesting its potential utility for long-term integrations and ensemble-based studies. The model reasonably reproduces the large-scale features of precipitation, atmospheric circulation, and temperature, and its precipitation skill is broadly comparable to those of CMIP5 and CMIP6 models. Nevertheless, several systematic biases are identified during boreal summer, including excessive rainfall over the equatorial western Pacific, cold surface temperature biases, weakened Tibetan Plateau heating, and circulation errors over the western North Pacific. The model also captures key modes of climate variability associated with the EASM, including the El Niño–Southern Oscillation, the Pacific–Japan pattern, and the boreal summer intraseasonal oscillation, despite remaining biases in their spatial structure, amplitude, and periodicity. A comparison between the coupled and atmosphere-only simulations indicates that air–sea coupling improves sea surface temperature–precipitation feedbacks and the seasonal evolution of the EASM rainband. Overall, GRIMs–NEMO shows reasonable capability as a computationally efficient, fully coupled climate model, while the identified biases highlight the need for further improvements in atmospheric physical parameterizations and the ocean model component.
General comments:
The authors present their newly developed coupled ocean-ice-atmosphere model GRIMS-NEMO v1.0 and evaluate its performance. The paper is generally well presented and the figures are of good quality. The authors find that the model has a dry bias (in contrast to CMIP5/CMIP6 models) but of similar magnitude to those models, and a global cold bias compared to observations/reanalysis. The authors evaluate three modes of variability that have an impact on the East Asian Summer Monsoon and find generally good performance albeit with some important caveats. They also find that air-sea coupling improves model performance in simulating tropical pacific precipitation and SSTs. As discussed in my comments below, there are several points that need clarification – as the paper is now, some explanations appear “out of the blue” and need more justification (e.g. from literature or further analysis), and I am confused by some of the interpretations of the figures. However, if the authors can address these issues then I think the paper certainly holds promise for publication and will be a valuable contribution to the modelling community.
Specific comments:
L39-40 seems a bit simplified, please at least include references to back up this claim.
L160-165: It says twice here that precipitation is underestimated in high latitudes, I would suggest removing one of those sentences. This also does not stand out to me as a feature – perhaps a difference plot would make this more obvious (see comment below)? Looking at the plots for a longer time, I think you might mean the North Pacific and Southern Ocean – it would help to name a few specific examples. Also, South America really stands out as a land region with large precipitation biases (perhaps due to the land model and lack of vegetation-atmosphere feedbacks?). I appreciate South America is not the focus of this paper but I think it would still be interesting to point this out, as it is a global model which others might want to use for studying other regions.
Figure 1: A difference plot would be really helpful here to identify regions with large precip biases. Also, why not just use a linear colorbar?
L179 – 185: The geopotential height bias is discussed for the Tibetan Plateau and linked to surface temperature biases here, but from the plots it appears that the geopotential height biases are quite zonally uniform. Perhaps a difference plot would show the Tibetan Plateau stand out more clearly? Does figure 1f actually show the cold bias over the Tibetan Plateau (as it is a zonal mean) or did you mean Fig 1e, because it looks like other regions e.g. Northern Africa would also be contributing to the zonal mean cold bias of surface temperature. I’d also be curious to see a timeseries plot (could be in supporting info) of how quickly the cold bias develops in the model from the initialised state. Can you explain a bit more about the land-surface model and why you think it can explain the cold bias?
L193: It is mentioned here that the temperature and geopotential height biases would affect the precipitation distribution (i.e. via the circulation/dynamics), but would the cold bias not also imply lower specific humidity (thermodynamic) issues which might at least contribute to the negative precipitation bias in the model?
Fig 2 caption: Please state explicitly in the caption what the “difference” is in panels e and f, i.e. which one is subtracted from which one. Presumably CPL minus ERA5?
L206: Please include references here to studies that show impact of ENSO, BSISO and PJ on EASM. Are these the most important modes of variability for EASM? What about IOD, PDO,…? I am not suggesting to analyse all of these, but it would be helpful for the reader to include a stronger motivation for why the three mentioned above were chosen.
L240: Please include relevant citation for definition of the BSISO life cycles.
L243-245: The authors attribute the dry bias over India to a “warm SST bias in the Indian Ocean”, however from Fig 2 it actually looks like there is a cold SST bias in the Indian ocean? And a warm bias over land (the latter could perhaps be at least partly due to lack of rainfall, and thus enhanced sensible instead of latent heat release). This would be expected to strengthen the land-sea thermal contrast and enhance monsoon circulation… In addition, it would be interesting to expand this discussion on what impacts the BSISO biases have on the East Asian Monsoon, since this region is the focus of the current study.
Fig.5 : Missing labels in caption (which column is ERA5 and which is the coupled model). Also, I would suggest extending the latitude range to ~ 60N (as in Fig 7). Currently we can barely see the EASM region and half of it is cut off, which is odd given the focus of the paper.
L258, L290 and others: The authors mention “the negative relationship between SSTs and precipitation”, but there are also large areas of positive feedback (Fig 6) so I would be a bit more specific about this in the text throughout that you are referring to specific regions where the correlation is negative. At the moment, it sounds like a blanket statement.
L255-257: I would change these two sentences to present tense as the analysis is yet to come. I also would not say that this analysis (figure 6) is “over the East Asian region”, but rather over the North Pacific (nothing is actually plotted over East Asia).
L262-266: The authors discuss that the ATM model does not capture regions of negative correlation between precipitation and SSTs, and then discuss that the coupled model does capture this relationship. These two arguments need to be separated, because currently it sounds like because ATM does not capture this relationship, CPL does. Please rephrase, otherwise it’s confusing to the reader.
L272-274: I don’t think you need to say again here that ATM “was conducted using the GRIMs atmospheric model, a component of Coupled GRIMs–NEMO”
Figure 6: I think including a subplot for ATM_CSST would be interesting here to isolate the air-sea feedback (as done in Fig 7). Also, please include precipitation contours because currently we can’t see whether regions of negative correlation are actually associated with strong precipitation. I would also argue that the sign of the correlation in CPL compared to the observations is actually the opposite in key regions – why is this the case? This might be related to different precipitation distribution.
L294-298: I think the statement that in CPL the rainband migration is “similarly reproduced” as in IMERG is a bit generous. There is a weak signal of precipitation that appears to be moving northward but nowhere near as strong and organised as IMERG. To me CPL and ATM_CSST look very similar to each other and ATM_CSST even looks like it is capturing the rainfall later in the season a bit better than CPL. Also why did you choose to take an average over the region 120-140E, which is off the coast of Eastern China? I think it’s important to include land regions of the EASM in this hovmoeller plot (see e.g. Chen and Bordoni 2013 (DOI: 10.1175/JCLI-D-13-00479.1), Fig 9), especially given the focus of the paper.
L330, L337: The authors refer to a “warm SST bias” in the coupled model, yet in Fig 2e it very much looks like a cold bias almost everywhere except over India (and the high latitudes)?
L335: “the negative relationship between precipitation and SSTs”: As mentioned above, there is more subtlety here, not all regions have a negative correlation. I understand the argument about cloud feedbacks, but there are several things that impact the correlation between precipitation and SSTs (generally precipitation is concentrated in regions of high SSTs, so would expect a positive correlation). I would at least say something like “the negative relationship between precipitation and SSTs in some regions”
Manuscript Title: I appreciate the region of interest for this paper is the EASM, but is there anything specific to the model that would imply that simulation of EASM would be better than any other region of the globe? The title of the paper implicitly suggests that the model has been developed specifically for studying the EASM, but there doesn’t seem to be anything in the model that would make performance in the EASM better than anywhere else. I would suggest changing the title to make it clearer that a global model has been developed, and is here tested on the EASM. Perhaps: “Development of the GRIMs–NEMO fully coupled global model v1.0 and evaluation of its performance in simulating the East Asian Summer Monsoon”
All figures: I would find it easier to look at the figures if the labels were added as titles to the subplots/in legends, rather than having to skip between the figure caption and the subplots to see which one is CPL, ATM, IMERG, ERA5 etc.
Reproducibility: I note that the model has been uploaded to zenodo but is currently with restricted access - presumably the authors are planning to make it available upon acceptance of the manuscript (?)