|I am not an expert in land-use and land-cover change modelling, but do have some experience in modelling of human-environment / social-ecological modelling. I read the current version of the manuscript and the reviews of referees #1 and #2 incl. the responses of the authors.|
In the context of modeling feedbacks between human and natural systems the authors discuss coupling approaches of four models where land-change models have been coupled to natural-system models. From these four modelling cases the authors draw general lessons learned and ways forward for the future of coupled Earth system modelling.
I find the overall scope of the paper valuable. I especially like the conceptualization of coupling approaches and feedback types presented in Figure 1 and 2. However, I suggest to revise the paper to clarify and strengthen its arguments.
Overall, I suggest to clarify terms used in the paper. For one thing, these are the terms ‘Input’, ‘Output’, ‘Data’ and ‘Coupler’. I refer to Figure 1 and the model description figures (Fig. 3-7). By the way, there is no Figure 4 in the paper. Do you mean with ‘Data’, externally observed data forcing a model? But how can such ‘Data’ be updated (c.f. Fig 3) If not, how then ‘Data’ is different from model ‘Input’ and ‘Output’. And how is ‘Data’ different from a ‘Coupler’? What is a ‘Coupler’? And what is the difference between a normal and an integrated assessment coupler? And how can you couple a coupled model without a coupler (c.f. Fig 5). These questions may seem very basic, but I feel clarity in those terms would make the article much more accessible to a wider audience.
Another set of terms that deserves more clarity is ‘model’, simulation environment’, ‘computational laboratory’ and ‘framework’. Initially I thought that four models are about to be discussed. But the first “model” is a “computational laboratory” (l.232). The second is a framework (l. 301). This is surely no major issue, however I suggest to clarify these terms, especially “framework”, since the article continues with a discussion about another kind of framework.
Here, the article needs more clarity with respect to the framework of coupled models. In the abstract I read: “However, a common framework and set of guidelines to model human-natural systems feedbacks are lacking.” (l. 41) Further below, the text says “Novel integrative modelling methods are being developed to create technical frameworks for, and intersecting applications between, these two communities (e.g., Hill et al. 2004, Lemmen et al. 2017, Peckham et al. 2013, Robinson et al. 2013, Collins et al. 2015, Barton et al. 2016)” (l.101)
Either these two sentences are contradictory, or two kinds of frameworks are being discussed. Either way, I suggest to clarify.
A third alternative could be that the authors feel the lack of a COMMON framework. If this is the case, I would like to read how their proposed framework differs from the ones mentioned and how it is better to eventually become the COMMON framework.
Moreover, is the framework that is being envisioned in Sec. “A way forward” a third kind of framework, or how does it relate to the frameworks mentioned so far?
“New coordinating frameworks for next generation coupled modeling of human and Earth systems are being developed within a number of relevant organizations” (l.756)
Scanning recent article of the respective special issue in ESD, Donges et al. submitted “Earth system modelling with complex dynamic human societies: the copan:CORE World-Earth modeling framework” which states to be a “open source software library that provides a framework for developing, composing and running World-Earth models, i.e., models of social-ecological co-evolution up to planetary scales”. I wonder and invited the authors to discuss whether and how they would consider such an attempt within their line of desired modelling framework.
Concerning the lessons learned, from my point of view, some of them are quite specific to coupled human-natural systems modelling (lesson 4), others apply to modelling in general (lesson 1). While I absolutely agree that also general modelling lessons are of great importance I suggest to discuss and perhaps order the lessons according to how specific they apply to coupled human-natural models. Also, the paragraph for lesson 7 needs an argument, why the observed and described problems could be solved with open-access code.
L. 209: “The examples are situated at different points along the three dimensions of configuration, frequency of communication, and coordination.”
I suggest to illustrate these points, where the examples sit, directly into Figure 1 and 2. In doing so the reader would gain a better initial overview of the examples and how they relate to the given context (framework?).
“One-sided approaches are prone to generating biased results” (l. 67) I encourage the authors to discuss the amount of natural vs. human system components within the presented four cases. Are the selected four cases all in good balance or do (at least) some of them tend to be one-sided. I suggest to use color in the model figures to indicate human, natural and technical model components (couplers, etc.). This would take up this discussion form the introduction which I missed in the rest of the manuscript.
L.114: “The potential gains from greater coupling are threefold.”
Please indicate the three gains more clearly. For example, “the use of many of the Earth’s
resources by humans alters the state and trajectory of the Earth system“ sounds rather as an argument why one should couple models, than a gain from coupling.
Lastly, I still agree with referee #2 that the manuscript could benefit from shorting sentences.