the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jul 2026
Coral Reef In Situ Product (CRISP): a global compilation of coral reef temperature time series
Abstract. Coral reef ecosystems are increasingly affected by climate-driven marine heatwaves, but the thermal environments experienced by corals at the reef-scale are still not well resolved by satellite-derived sea surface temperature (SST) products. In situ temperature loggers can capture much of this fine-scale variability, including diel temperature ranges, short-term extremes, and depth-dependent structure. However, these datasets are often fragmented across sources, inconsistently formatted, or not readily accessible for synthesis and reuse. To address these gaps, we developed the Coral Reef In Situ Product (CRISP), a global aggregation of temperature time series collected from coral reef environments. At the time of writing, CRISP includes 2,548 logger deployments assembled from a mix of public repositories, peer-reviewed literature, gray literature, and community contributions. Each deployment record is associated with available metadata fields stored in a consistent manner across all loggers. The entire dataset is provided in NetCDF format to support use in common scientific workflows (e.g., MATLAB, Python, R). By bringing together these otherwise fragmented observations, CRISP enables analyses of fine-scale thermal variability across coral reef systems. The dataset adheres to the FAIR data principles by emphasizing findability, accessibility, interoperability, and reusability. CRISP can be used for a variety of scientific applications including coral bleaching research, thermal stress assessment, reef climate monitoring, model validation, and investigations of environmental variability across reef systems.
- Preprint
(1994 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on essd-2026-522', C. Mark Eakin, 24 Aug 2026
-
RC2: 'Comment on essd-2026-522', Anonymous Referee #2, 13 Sep 2026
1. General comments
This manuscript presents the Coral Reef In Situ Product (CRISP), a global compilation of in situ temperature records from coral reef environments, comprising 2,548 logger deployments and more than 595 million observations. The effort involved in assembling, standardising and making these disparate records accessible is substantial, and the resulting archive has clear potential value for studies of reef thermal environments, coral bleaching, model validation and broader comparative analyses.
However, in its present form I do not think the manuscript provides sufficient information to assess the quality, comparability and appropriate use of the compiled records. This is particularly important given the considerable heterogeneity among the underlying datasets in sampling interval, deployment depth, logger model, deployment configuration and metadata completeness. The manuscript currently places considerable emphasis on the motivation for compiling in situ temperature observations, but substantially less on demonstrating how the resulting data product has been quality controlled and how users should account for differences among records. The description of quality control is especially limited, relying heavily on visual assessment without sufficiently explicit or reproducible criteria, while the procedure used to infer and correct timezone offsets also requires much more justification.
I therefore think CRISP could become a useful community resource, but the manuscript requires substantial revision before users can confidently interpret and combine records across the archive. In particular, the authors should provide a much more detailed and reproducible description of quality-control procedures, quantitatively characterise the major sources of heterogeneity within the dataset, and provide clearer guidance on the suitability of different records for resolving thermal variability across different temporal and spatial scales.
2. Specific comments
2.1. Major comments
- The quality-control procedures require substantially greater documentation and, ideally, quantitative characterization. CRISP combines a very large number of records collected using different instruments, deployment approaches, sampling intervals and metadata standards, yet the present description of QC relies heavily on subjective visual assessment. For a data product intended for broad reuse, users need to be able to understand precisely how observations were assessed, what criteria were used to flag or modify records, how ambiguous cases were treated, and how many observations or deployments were affected. This is particularly important for the treatment of anomalous values and inferred timezone corrections. I think a reproducible QC protocol, accompanied by a clear description of QC flags and summary statistics on their application, is essential.
- More generally, the manuscript needs to engage much more directly with the substantial heterogeneity among records and what this means for their comparability and fitness for different analyses. Sampling intervals span several orders of magnitude, deployment depths vary greatly, and logger models and deployment configurations differ in ways that can materially affect the temperatures recorded. These are not simply ancillary metadata differences: they determine whether a record can resolve diel variability, short-lived extremes, internal-wave events or other processes that motivate the development of CRISP in the first place. The manuscript should therefore provide a much more quantitative characterization of these sources of heterogeneity and clearer guidance on how users should filter or stratify the dataset for different applications. Without this, there is a risk that the apparent standardization of the archive implies a degree of comparability among records that the underlying observations do not necessarily support.
2.2. Detailed comments
- L11: The word “still” implies that satellite SST products might ultimately be expected to resolve in situ reef temperatures at these spatial and temporal scales. I do not think that implication is warranted; these are fundamentally different measurements with different spatial and temporal representations.
- L37: I do not think it is accurate to state that NOAA Coral Reef Watch “monitors bleaching events globally”. CRW monitors remotely sensed thermal conditions associated with bleaching risk; it does not directly observe bleaching events. The wording should be revised accordingly.
- L42: This discussion should also acknowledge the well-established limitations of DHW as a predictor of observed bleaching severity. DHW captures only part of the spatial and temporal variability in bleaching response, particularly where local hydrodynamics and subsurface thermal structure strongly modify reef temperatures. Relevant literature demonstrating these limitations should be cited here.
- L53-55: This passage warrants referencing.
- L62-63: The statement that discrepancies “can exceed 0.5 °C” substantially understates the magnitude of the differences documented in some reef environments. Several studies report departures from satellite SST of several degrees during processes such as upwelling and internal-wave forcing. The wording should better reflect the range of observed discrepancies.
- L67: Again, an important limitation is that, regardless of the underlying data source, DHW is a temporally coarse metric, integrating thermal stress over a 12-week (~3-month) period. It is therefore unlikely to adequately capture short-duration heatwave events and other high-frequency thermal variability relevant to coral bleaching.
- L102: What was the rationale for using Rich et al. (2024) as the basis for identifying the 314 publications screened for data? That study does not appear to constitute a comprehensive census of the reef thermal-environment literature. Please explain how the literature search and selection criteria were defined and what important datasets or study classes might consequently have been omitted.
- L173–177: The QC procedure requires substantially more detail for the resulting data product to be reproducible and its reliability assessable. “Careful visual observation” is inherently subjective. Please specify the criteria used to identify anomalous observations, whether thresholds or rules were applied, how deployment/retrieval artefacts were distinguished from genuine environmental extremes, who performed the assessment, whether classifications were independently checked, and how many observations or records were ultimately flagged. Equivalent detail is also needed for the timezone QC procedure described below.
- L177: “No raw data were deleted” appears inconsistent with the preceding statement that measurements “were removed”. Presumably the observations were retained but flagged/excluded from downstream use; please clarify the terminology.
- L180–184: Inferring timezone offsets from the phase of the diel temperature cycle seems potentially problematic in environments where temperature variability is driven primarily by tides, internal waves, upwelling or advection rather than local solar heating. Please provide the decision criteria used for these corrections, indicate how ambiguous cases were handled, and ideally retain a flag distinguishing reported timestamps from inferred/corrected timestamps.
- L205–207: The sampling resolution should also be stated here (i.e., in the figure description/caption). Temporal resolution is perhaps one of the most important factors determining whether an in situ record is meaningful for resolving the thermal variability of interest.
- L243-244: This discussion needs greater nuance. Solar heating of an exposed logger may represent a measurement bias relative to bulk water temperature, but corals themselves can also experience substantial radiative heating. The manuscript should therefore be explicit about what environmental quantity CRISP is intended to represent (e.g., bulk water temperature versus the thermal environment experienced by benthic organisms), and about the extent to which shading or sensor-specific radiative responses affect that interpretation. More broadly, I am not entirely convinced by the framing in Rich et al. (2024), where exposure to solar radiation is treated principally as a source of logger bias; in shallow reef environments, eliminating radiative heating from the measurement does not necessarily provide a better representation of the thermal conditions actually experienced by corals.
- L239-261: I think that, as noted above, sampling interval is perhaps one of the most important limitations when interpreting these data, together with logger model and deployment depth. Sampling intervals in CRISP span several orders of magnitude, meaning that records differ fundamentally in their ability to resolve diel variability, short-lived extremes and high-frequency oceanographic processes. Likewise, some logger models and deployment configurations may yield substantially less robust measurements than others. These issues warrant considerably greater treatment in the limitations section, including their implications for comparing or combining records across the dataset.
Citation: https://doi.org/10.5194/essd-2026-522-RC2
Data sets
Coral Reef In Situ Product (CRISP): a global compilation of coral reef temperature time series Angela G. Larson, Avi Strange, Oliwia Jasnos, Emily Johnson, Abigail G. Heck, Patrick L. Colin, and Thomas M. DeCarlo https://doi.org/10.5281/zenodo.21045962
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 248 | 107 | 116 | 471 | 163 | 123 |
- HTML: 248
- PDF: 107
- XML: 116
- Total: 471
- BibTeX: 163
- EndNote: 123
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
CRISP is a great idea and will be of great value to the coral reef community.
Providing the data in netCDF format is great for scientists skilled in data analysis. However, a user-friendly interface should be provided. Many scientists and most coral reef managers have no idea how to handle NetCDF data but may wish to access particular datasets.
Line 33: Mortality can result from SEVERE or prolonged heat stress.
Line 36: You really should cite recent papers on the third and fourth global coral bleaching events here.
Line 40: Actually, CRW has a suite of products. DHW is only one of them. This should be noted here.
Line 59: Exposure to higher DTRs MAY experience improved thermal tolerance -- this is not a given.
Line 66: It isn't just the coarse resolution but also the thin layer of surface water that is measured by satellites.
Lines 69-70: Pat may have said in situ loggers are the "gold standard" but they only measure what is happening at a very precise point. Therefore, they are only the "gold standard" for adjacent corals. Unless someone plans to deploy a fine mesh of loggers, loggers don't address the spatial variability you describe earlier in this paragraph. In situ loggers are a great tool, but they aren't perfect either! Don't oversell them.
Line 80: Accessible to people trained in the use of NetCDF but not to a broad audience unless a friendly user interface is created as well.
Line 100: NOAA is a big organization that does a lot more than coral reefs. Which parts of NOAA? While not as big as NOAA, AIMS is a big organization that does a lot more than coral reefs. Which parts of AIMS? NOAA Coral Reef Watch has a compilation of logger data that may not be in NCEI. Did you reach out to them?
Lines 110-112: CRISP should include a list of these inaccessible data in case others wish to pursue them.
Lines 132-133: Using the mid-point of depth ranges is a good solution.
Line 161: UTC is the correct universal time based on the atomic clock standard and should be used instead of the now archaic GMT. This should be corrected in both the dataset and the paper.
Line 205: Was this a single 9.2 year logger deployment? Those HoboTemp units had a 6-year nominal life and a non-replaceable battery. Files should indicate when units were swapped out or maintenance was performed. Neither is noted here, implying a continuous record.
Conclusion: You fail to indicate the future of CRISP. Is this a one-and-done exercise? Is there a plan to continue to update CRISP with new data? An effort like this one really deserves to have a long-term future.
Line 317: Storing the data in NCEI is the right place.
Line 319: Well that's odd. You just said they were at NCEI? Where are they really?