General comment: This data manuscript describes a unique and valuable microstructure dataset sampled with a glider over several years and seasons in the western Mediterranean. The manuscript provides and exhaustive overview of the data processing, which  will be quite useful for anyone using this dataset in the future.

Few sections can be shortened without trying to produce a review or give arguments that such a dataset contains key parameters. All measured variables in the ocean are valuable and  need to be publicly available, whether they are scarce (such as microstructure observations) or widely collected (such as CTDs).

You will find all my comments, questions and suggestions in the commented PDF attached to my review.

With kind regards.

Summary of the paper

The paper “Advancing Turbulence Essential Ocean Variable: A Reference Glider-Based Microstructure Dataset from the Western Mediterranean” presents a decade-long (2015–2024) dataset of turbulence measurements collected by a Slocum Deep Glider equipped with a Rockland MicroRider. The missions repeatedly crossed the Sardinia–Balearic transect, producing over 3,000 high-quality profiles of turbulent kinetic energy dissipation (ε) and thermal variance dissipation (χ) down to 1000 m depth. The dataset, processed through standardized L0–L4 levels and validated against strict QC metrics, represents one of the largest autonomous microstructure archives in the Mediterranean.

Review summary

I am OK to publish the paper with minor revisions. The dataset is of high value and well-documented. The main improvements should focus on clarifying and tightening the introduction, replacing subjective or informal wording, and adding representative turbulence spectra to better demonstrate data quality and satisfy the expectations of turbulence specialists.

Fig. 1 & L70–110 – Replace “expected” ε profiles with observed average data; clarify whether values are measured or estimated.

L83–100 – Add citations for regional circulation and mesoscale dynamics; clarify that the author is not an expert in the region.

L105–110 – Rephrase “abyssal peace” and “natural mixed layer”; use neutral scientific language.

L120–135 – Include citations for turbulence parameterization and EOV framework; clarify what “expected dissipation levels” mean.

L125–130 – Add explicit mention of spectral methods and noise floor corrections.

L165–195 (Data & Methods) – Define Level 0 clearly here; improve flow—some content fits better under “Processing.”

L175–180 – Replace “averaging involved” by “segmentation or windowing”; clarify that raw data are not averaged prior to spectra.

L195–200 – “On the contrary” instead of “At the contrary.”

L200–225 – Explicitly state which spectral model and integration limits are used for ε and χ (Nasmyth vs Batchelor); mention the electronic noise floor.

L230–240, Fig. 2 – Cite Lueck et al. 2024 explicitly when stating the workflow follows it “exactly.”

L250–265 – In Table 1, note that this is a list of routines, not processing “steps.”

L480–495 – Simplify data structure description; too detailed for ESSD, could be moved to Supplement.

L545–570 (QC section) – Explain rationale for focusing on QC=2 rather than others; the purpose and statistical relevance are unclear.

L560–565 – Use FOM = 1.15 as the nominal threshold (per Lueck et al. 2024) and note that users can apply their own.

L565–570 – Remove statements implying that removal is “unnecessary”; just report the data and FOM values.

Fig. 3–4 – Add a few examples (panels or zoomed insets) illustrating good vs problematic QC spectra.

Minor editorial – Fix “than” → “then” (L580), typos like “enventually,” “affixable,” and consistent use of italics for ε and χ.

Additional suggestion – Include representative shear and thermistor spectra in the Results or Supplement (always appreciated by turbulence experts).