A useful presentation of the available oceanographic data within the Indonesian Seas.  Serves as a convenient starting point to access data for researchers interested in studying the Indonesian Seas and the Indonesian Throughflow. They also cite publications that provide scientific overview of the Indonesian Seas. 

General Evaluation 

This manuscript delivers a valuable and comprehensive collation of two centuries of oceanographic observations in the Indonesian Seas, tracing how spatial coverage of key variables has evolved over time. The detailed source descriptions and clear maps provide a solid foundation for researchers seeking to understand both the strengths and the remaining gaps in this dataset. Overall, the work represents an important reference for the community and will spur targeted efforts to fill critical historical voids.

However, a few aspects should be addressed before this paper is suitable for publication in Earth System Dynamics.

Main points:

1. There is some structural redundancy that hinders readability. In particular, Section 4 feels very repetitive, as it revisits content already discussed in the Results and Methods, merely adding further depth. I recommend dissolving Section 4 and blending its unique insights into the appropriate Methods or Results sections. This consolidation will streamline the manuscript, enhance readability, and prevent readers from navigating through repetitive content.
2. The title promises to assess historical trends is undermined by ambiguity. It is unclear whether the authors refer to trends in data coverage or to actual temporal changes in the ocean variables. Indeed, Figure 8 (the only one showing trends) is hard to interpret: the dot colors lack reference to a specific temporal window, the colorbar uses absolute units without temporal context, and the red contour line is unexplained. The manuscript should define the temporal windows used for each trend calculation, state trend units and magnitudes (e.g., °C decade⁻¹), and detail the statistical methods (linear regression, significance testing) used to compute and characterise these trends. 
3. The paper is generally well written but contains a few grammatical errors and incomplete sentences that sometimes hinder the understanding. I recommend a thorough spell-check and language edit to ensure clarity throughout.

Addressing these points will sharpen the paper’s focus and ensure that its key contributions,  both in data synthesis and trend analysis, are presented with maximum clarity and impact.

Other additional points:

Lines 49-50: Rewrite to “These modes of variability influence weather and regional climate via air-sea interactions…”

Lines 56-57: Can you elaborate on how the lack of observations affects projections? These are simulations run with coupled global climate models that are not initialized from observations.

Line 65: What do you mean by “has been emphasized”?

Line 66: Please rewrite. You mention twice the word challenge but in both instances it is unclear what specific challenges you refer to.

Lines 74-75: The phrasing is unclear. What do you mean by “as fast as possible in (near) real-time”?

Lines 132-133: What potential collaborations do you refer to? Also, it is hard to reconcile the first part of the sentence with the second. You mention that there is no governmental data available and at the same time say that you use open-source data. If this later is not from governmental sources you should specify it.

Lines 147-148: This sentence has no verb.

Lines 170-171: Did you also check if the actual values were the same as a final proof that the data was actually duplicated?

Lines 215-227: You should use verb tenses consistently throughout the paragraph

Lines 244-246: Note that some word are capitalized but they shouldn’t (i.e. A, Calculation)

Lines 262-268: You define a category for empty cells and Q1-Q5 quantile bins, which does not seem appropriate for separating in 6 quantiles. I suggest separating “No Data” (or call it Q0) and then defining six true quantile bins Q1-Q6 (0.1-16.7%, 16.7-33.3%, …, 83.3-100%). This will make it semantically clear and align each bin with its exact percentile range.

Lines 330-334: Could you elaborate, or at least hypothesize, on why measurements have dropped so drastically in the most recent period? I find this surprising, given the dense observational network active from the 1950s through the 1990s and the fact that global measurement counts have actually risen over the last decade.

Lines 348-351: It would be more useful if you discuss global numbers in parallel with those specific to the ISS region. Indeed, these global numbers do not illustrate the evolution in the ISS region, which indeed experienced a decrease in stations over the last 2 decades.

Lines 368-369: Some of this information is repeated from the last sentence in the previous paragraph. Please avoid repetition.

Lines 368-377: Instead of repeating the numbers from the plot for individual countries, it would be more interesting to aggregate some together, for example, for the different continents or geographical areas (Europe, North America, Australia, Southeastern Asia...).

Figure 6: The choice of colors hinders the readability of the plots, as several regions have very similar color shades, and is therefore very difficult to distinguish them.

Figure 7: Could you be a bit more specific on what the plot is actually showing? I can’t even tell if the first two sentences refer to the plot

Figure 8: Can you clarify if you are showing total counts of casts per grid point over the whole observational period?

Lines 609-611: The sentence as it is written implies that scarce observations actually cause model errors. In reality, model biases in ITF variability stem from incomplete or imperfect representation of physical processes. The primary impact of limited high-resolution data in Makassar and Lombok Straits is that it prevents rigorous evaluation and validation of those model simulations, rather than introducing the errors themselves. You might rephrase to clarify that observational gaps constrain our confidence in model performance, rather than being the root cause of simulation inaccuracies.