The paper by Hemming and coauthors describes an application of the multi-decadal 

in-situ observational dataset of marine temperatures (AMDOT) 

for the detection of extremes (AMDOT-EXT).

While I appreciate the importance of extended observational records for 

assessing long-term changes, I am not sure to understand how to interpret MHW/MCS numbers

with less-than-daily frequency data. From the images included, 

it seems that high-impact events can be detected only over the last ~20 years

(compare Table 3 and Figure 4), even if at l100 an event in 1944 is described.

I find Table 1 and the duration of the MHW/MCS records not really clear.

Please explain (with a table or figure) the periods for which high-frequency extreme

statistics are available.

Moreover, some information on the changes in climatology should be provided, since

using the whole record likely lead to more events in recent years.

Background on other possible data sources in the area should be given.

For example, no long-term measurement is available for Northern Australia?

Data for the last ~20 years are likely available for more sites.

The authors rely on Roughan 2022b for too many details. More information

on the observational methods need to be provided. For example, no discussion 

on the long-term consistency of these records is provided.

Familiarity with the Hobday's terminology is also assumed, please explain what 

technical terms (e.g., cumulative intensity) mean. Please carefully check the units.

The authors mention 'optimal depths', but it is unclear to me how these are defined.

The depth-resolved information seems an asset, please include some illustration of 

an event detected at multiple depths.

'Code and data availability' and A1 contents look too similar. Revise.

The provided codes can be useful to know how to open the data but seems quite basic.

I do not understand how you deal with event statistics: categories can be defined at daily 

frequency, but statistics are for individual events (e.g., cumulative intensity).

How do you save them as a function of TIME and DEPTH? The same values are repeated for all 

the days during a MHW/MCS?

Specific comments (also by line number)

l27 please give details on the depth they analysed

l30 the last part of the sentence is a repetition

l37 if from the early 40s, why not ~80 years?

Fig 1 please explain what isolines are (bathymetry? Derived from?)

Table 1 for 2m depth, I would show not the relative (i.e., always 100%) 

but the absolute available fraction (maybe on monthly or weekly basis).

This would help interpreting other relative fractions.

Moreover, I do not like the fact that the table is an image - please check

the editorial guidelines.

l55 please provide information on the instrumentation, I guess it varied over 

time leading to varying uncertainty. How these levels are related to local 

conditions such as the mixed layer depth?

l66 define CTD

l68 how L3S differs from L3? Define 'QC'

l75 'free' => 'open source'

l104 how do you identify the MLD?

l105 missing citation

l123 and elsewhere (e.g. Table 2) I would use italics or a typewriter font for variable names,

to distinguish them from MHW, MCS,... - check journal guidelines

l130 Excel is not really needed for csv files, they are simply text files

(as they are not spreadsheets).

Moreover, Excel is not open nor free - suggest providing open-source alternatives.

l132, 157, 221 what is 'onedrivelink'? Why not using the website you cite?

l149 I would suggest to compare the events detected in the last ~20-30 years when 

using your full-record climatology and a more standard 30-years baseline.

l161 explain 'cumulative intensity'

l165 remove space between the circle sign and C

l170 without context, Table 3 is not very useful. Please add some references.

l172 and elsewhre, I think units should be 'C * day', not 'C / day'. These are the 

units for onset/decline rates instead. This seems correct in the netCDF files.

l221 are you also using zenodo?

l225 NetCDF is not necessarily a python format, it is actually aimed at

portability.

Please rephrase 'appendix section XX' => 'appendix XX'

General Comments

This paper presents an observational dataset documenting extreme ocean temperatures observed at four coastal stations in the vicinity of Australia, referred to as AMDOT-EXT. This dataset serves as an extension of the previously published AMDOT dataset, encompassing both surface and subsurface temperature records.

The study adheres to established standards and utilizes publicly available algorithms to identify and characterize heatwaves.

Given the global concern regarding the warming of the Earth's oceans, the dataset presented herein bears potential significance for climate impact studies and warrants consideration for publication, notwithstanding its inherent limitations. However, a substantial area in need of improvement lies in the discussion of uncertainties.

Specific Comments

• The unique and pertinent attributes of this dataset, despite its limited coverage of only four stations, could be emphasized, particularly in Section 2.
• Notably, Table 2 reveals a significant discrepancy in the "relative % record available" between station PHA and the other stations (PHB/PH100, MAI, and ROT). For PHA, data availability remains relatively uniform throughout the water column and generally falls below 100% below the surface. Conversely, for the other stations, data availability increases significantly at greater depths compared to the surface, with ROT, for instance, exhibiting 172% data availability at a depth of 29 meters relative to the surface. An explanation for this peculiar behavior would be valuable.
• The PHA dataset primarily comprises monthly and weekly data which is not consistent with the methodology adopted for the identification of heat waves. It is also located relatively close to the PHB/PH100. It is advisable to provide a more detailed discussion if there is a specific rationale for including this record in the dataset.
• In reference to HS and CS (Page 6, Lines 87-88), my interpretation aligns with temperature exceeding the 90th percentile from above and the 10th percentile from below. Consequently, Table 2's columns pertaining to HSs and CSs appear non-uniform. For PHA, the number of events appears to be associated with a timescale of weeks or months (>5 days), while at the other sites, peaks are expected to encompass events lasting less than 5 days. Furthermore, it remains unclear whether "#Days Sampled" for PHS refers to the number of (weekly?) samples or the total number of days.
• While the paper acknowledges the impact of employing climatology periods of varying durations (Page 8), it does not provide an analysis of the consequences of selecting different climatology periods. Such an analysis is particularly crucial for this dataset since continuous daily records may not always be available (Figure 2), and the presence of long-term trends cannot be disregarded. In this context, the evaluation of designating the entire 80 year period as a reference climatology remains lacking. A plausible approach could involve generating different realizations of MHW and MCS to reflect data discontinuities as errors, for instance, in the count of heatwaves. Consequently, the statement asserting that the "data products highlight the challenges encountered when detecting extreme temperature events using sporadic long time series" receives weak support from the limited uncertainty analysis presented in the paper.
• In the context of uncertainty assessment, it is also important to account for the influence of secular trends, as exemplified in recent work on marine heatwaves in the Mediterranean by Rosselló, P., Pascual, A., & Combes, V. (2023), titled "Assessing marine heat waves in the Mediterranean Sea: a comparison of fixed and moving baseline methods," published in Frontiers in Marine Science, Volume 10, Page 1168368.
• Information regarding potential future updates to the dataset is not provided. Even the acknowledgment of an absence of clear plans for future updates constitutes valuable information.

Techincal Corrections

• On Page 6, Lines 87-88, for clarity, it is advisable to specify that HS and CS refer to temperature exceeding the 90th percentile from above and the 10th percentile from below.
• The final paragraph on Page 6 (Lines 102-105) is not sufficiently clear and warrants rephrasing.
• The presence of a question mark at Line 105 is likely a typographical error.