the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A peat-depth database for Canada (PDD-Can-1)
Abstract. Canadian peatlands hold an estimated ~25 % of the world's peatland carbon (C) stock. However, a more accurate calculation of C stock requires high-quality empirical datasets to refine numerical techniques, perform validation steps, and calculate the resulting uncertainties in prediction. Here, we present two databases (i) a compilation of 88,763 survey points of basal peat-depth across Canada, extracted from 844 publications (government reports, environmental impact assessments and scholarly journals), and (ii) a compilation of 436,923 survey points containing no peat. Many of these peat-depth survey points were digitized here for the first time. We showcase the peat-depth data by presenting a timeline of peat studies over the past ~100 years and discuss the motivation behind much of this peatland work across the country. Measurements of peat-depth across Canada range from 0–1200 cm and from relatively shallow wetlands in the Arctic tundra to deep fens and bogs in the southern temperate forests. Importantly, our compilation includes peat-depth measurements from regional soil or surficial material surveys, largely from environmental impact assessments. These datasets tend toward shallower peats in the range of 50–100 cm, and therefore represent transition areas, regions of shallow peat and/or forested peatlands that were not previously widely documented. There are five data-quality considerations: (i) some data have low precision geographic (latitude and longitude) coordinates; (ii) across the database, there are apparent issues with rounding, vague methodology and unclear definition of the mineral-peat interface; (iii) some peat-depth data were measured 100+ years ago, so there may be mismatches between the present-day condition of the peatland and the original measurements; (iv) there are notable spatial gaps and biases across Canada, including few data from the Hudson Bay Lowlands, overall sparse coverage north of ~55° and a patchwork of peat-depth data that is sometimes abruptly truncated at provincial/territorial borders; (v) we do not include a classification of peat-depth data into fens or bogs because this information is lacking in the original data or was unreliable. We finish by discussing potential uses for these databases along with future improvements. The databases are available at https://doi.org/10.5281/zenodo.17409850 (Dalton et al., 2025).
- Preprint
(3339 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 03 Dec 2026)
-
CC1: 'Comment on essd-2026-169', Rod Smith, 03 Jul 2026
reply
-
CC2: 'Reply on CC1', April Dalton, 13 Nov 2026
reply
Response from the authors of the manuscript to the community comment
We thank Rod Smith for this insightful and candid community comment. Most of the above comments concern the shothole drillers’ datasets, an exceptionally detailed source of information on the surficial geology of the Mackenzie River area. We are grateful for the additional information the commenter provided about this database, and we have expanded the manuscript in several places to add context for the reader. However, in alignment with the commenter’s suggestion, we ultimately judged these records unsuitable for inclusion in our compilation because the reliability of many records is uncertain and their quality control is unclear.
We did not intend to misconstrue the shothole drillers’ datasets, and we are grateful for the opportunity to make the suggested revisions. We also appreciate the commenter’s agreement that the shothole drillers’ log-derived muskeg depths can be suspect.
The commenter recommends using the final database compilation (Smith 2011 – GSC OF 6833, ~344,000 records) rather than version 2 (OF 6049, ~275,000 records). We agree that we cited the incorrect version, thank the commenter for pointing this out, and will correct the relevant parts of the manuscript. Our conclusion is unchanged, however: the shothole drillers’ datasets are not sufficiently vetted, so they remain excluded from the analysis and compilation in our manuscript. Based on the commenter’s remarks about the age of the dataset and the training of the drillers, it is unclear whether vetting would be possible at all.
The commenter also notes that shothole data were previously used to map peatland thickness by Smith & Lesk-Winfield 2010b (OF 6410). We thank the commenter for this reference and will revise the manuscript to acknowledge this prior work.
The commenter recommends using the “positive” muskeg/peat records as “presence data” instead of relying on the specific depth measurements because peat/muskeg thickness were not recorded consistently. The commenter suggests that thickness measurements may be usable, but that some are clearly incorrect (for example, values exceeding 10 m), and that the only way to assess them is to return to the original drillers’ logs to determine which are accurate. We are happy to clarify this in the manuscript and to highlight the dataset as a strong source of peat presence confirmation. However, because our compilation is a peat-depth dataset, we cannot integrate this presence/absence information into our work. We appreciate the pointer to the Banks Island compilation; we reviewed our notes and confirmed that we examined the publication of Smith and Farineau 2015 – GSC Open File 7322 and excluded this information because it was presence/absence instead of depth data. Our compilation is specifically intended to compile data for training machine-learning models, which require actual depth values and cannot make use of presence/absence records.
The commenter also cautions against the use of non-peat records as zeros in our dataset. In the first version of the manuscript we had intended to include the null (0 m peat) information from this dataset, but satellite imagery showed that some null points fall on what we consider to be peatlands, which made us wary of relying on them. We therefore have already excluded them from our work. We are glad to make these clarifications for the reader and we sincerely appreciate the additional information the commenter provided about the shothole drillers’ dataset.
We appreciate the commenter’s feedback on our use of the word “haphazard” to describe the data collection. We agree this wording may have been too strong, and we will instead use “inconsistent” to describe this data collection.
We will also correct the manuscript to say that the lithostratigraphic materials and the drill logs were created by drillers, NOT geologists.We also appreciate the additional information about the lack of instructions on stratigraphic details given to these drillers, and the fact that the drill logs were recorded in several different formats over the ~50 years compiled in the dataset – we will use all of this information to update the text of the manuscript to better contextualize the database.
Citation: https://doi.org/10.5194/essd-2026-169-CC2
-
CC2: 'Reply on CC1', April Dalton, 13 Nov 2026
reply
-
RC1: 'Comment on essd-2026-169', Anonymous Referee #1, 02 Nov 2026
reply
See attached file
-
AC1: 'Reply on RC1', April Dalton, 14 Nov 2026
reply
Summary
Dalton, Garneau and Melton present PDD-Can-1, the first national-scale compilation of peat-depth measurements for Canada. The database contains almost 90,000 basal peat-depth survey points drawn from almost 900 sources, including government reports, research publications, and environmental impact assessments. An additional dataset detailing over 400,000 survey points where peat is absent is also included for comparison. A considerable amount of work went into digitizing and georeferencing historical hand-drawn peatland survey maps, many of which are made available in digital form for the first time, and into recovering peat-depth data from the soil and borehole components of EIAs, a source rarely used in peatland science. The manuscript is clearly written, describes the compilation approach, presents an interesting timeline of peatland investigation in Canada, and is candid about the limitations of the data, including coordinate precision, inconsistent definitions of the peat–mineral interface, the age of some measurements and the spatial gaps in coverage. This dataset fills an important gap and will be of direct value for peatland mapping, carbon stock estimation and model validation, and the authors should be commended for the scale of the effort. My comments below are mostly requests for clarification: some wording around peatland classification, the potential bias introduced by excluding deeper measurements from EIA and Ducks Unlimited data, some inconsistencies in the time periods used in the text and figures, and whether the database will be maintained in future. I recommend minor revisions.
>>>Reply from authors>>>We sincerely thank the reviewer for the generous and encouraging comments on our manuscript. We are especially grateful that our candor about the limitations of the data was noted. Working with such varied datasets was challenging, but we saw the value in including them, and we are glad that the care we put into writing these sections was recognized.
Abstract
Line 34: Swamps can also be peatlands, as arguably can marshes under some classification systems (not one I personally agree with, but it does exist). To avoid this wording generating a classification controversy, it may be simpler to state that a classification of land cover or peatland type was not included. This would also apply to the ‘no peat’ locations.
>>>Reply from authors>>>Thank you for this excellent suggestion. Peatlands are indeed classified in different ways, and we agree that describing swamps and marshes in the abstract could create confusion about classification. We have therefore removed the classification wording from the abstract, as suggested. However, we retain information about peatland type in the Methods section to give the reader a better sense of the type of environments in which the data were located and how the literature was searched to populate the database.
Introduction
Line 38: I was surprised that this section does not mention that peatlands represent long-term carbon sinks because of their ability to accumulate organic matter. I realize this may not hold for all peatlands at present, but where there is peat, they have been sinks historically.
>>>Reply from authors>>> We agree this context is important for the introduction. We will gladly add this to the Introduction and are grateful to the reviewer for pointing it out.
Literature Review
Line 74: It sounds like industry reports were included through the Impact Assessment Registry. It would be helpful to add wording that makes this clearer.
>>>Reply from authors>>> Thank you for this helpful suggestion. We agree this was unclear. In the revised manuscript, we clarify that data were also obtained from environmental impact assessments. We did not name a specific registry because it goes by different names at the federal and provincial levels.
Quality Control
Lines 124–129: Could the removal of EIA survey points deeper than 220 cm, bias the dataset toward shallower peat and result in an underestimation of peat depth in some regions? It is currently unclear what proportion of the peat depths in a given area comes from which source. This would be a concern if industry were to use these shallower values for decision making when they reflect a bias in data selection. It would be helpful to report how many points were excluded for this reason and to discuss how this affects the comparison shown in Figure 8.
>>>Reply from authors>>> Thank you for raising this important point. We compiled only measured peat depths. As a result, readings reported only as minimums, such as “greater than 220 cm” or “greater than 50 cm”, were not included. We agree that excluding them probably biases depths downward in certain areas. However, because these points were never compiled, we cannot report how many were excluded or quantify the effect. In the revised manuscript, we acknowledge this issue in three different places so that this is clear to the reader.
>>>Reply from authors>>> Here is the new text: “Only measured peat depths were compiled, so readings reported solely as minimums (e.g., “greater than 220 cm” or “greater than 50 cm”) were not included. This likely biases depths downward in some regions (see Section 6.1). Because these points were never compiled, we cannot report how many were excluded”
>>>Reply from authors>>> We also add “In addition, because environmental impact assessment points deeper than ~220 cm (the length of the sampling rods) were excluded, peat depths from these sources may be biased toward shallower values. Users should treat depths in areas where the database is dominated by environmental impact assessment data as conservative estimates and should be cautious about using them alone in decision making”
>>>Reply from authors>>> We have also revised the Figure 8 caption to include: “The reader should keep in mind that some reports did not give peat depths greater than ~220 cm because that was the length of the sampling probe.”
Timelines
The time periods used in the figures do not always match those used in the text. For example, Section 4.2 spans 1970 to 1990, but the Figure 5 caption states 1979 to 1990. Section 4.3 covers 2005 to present day, but the Figure 8 caption describes the EIA data as starting in the mid-1990s. The Figure 8 caption also places the second inventory period between 1970 and 1980, whereas Section 4.2 gives 1970 to 1990. Please make these consistent.
>>>Reply from authors>>> Thank you to the reviewer for carefully identifying these inconsistencies. We will revise the text and figure captions so that the time periods are consistent. In some places we deliberately used shorter time windows to describe what was happening; for example, impact assessments rose distinctly after 2005. Where the periods differ intentionally we will state this explicitly so that the text and figures align.
General Comment
Do the authors intend to update the database as new data become available? If so, please describe how future versions will be released (e.g., new DOIs or versioning) and who will maintain them. If not, please state that it is a static release.
>>>Reply from authors>>> Thank you for raising this question. In the revised manuscript, we clarify that this release is version 1. By making it publicly available with versioning, we hope that others will build on it and contribute updates as needed.
Citation: https://doi.org/10.5194/essd-2026-169-AC1
-
AC1: 'Reply on RC1', April Dalton, 14 Nov 2026
reply
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 322 | 127 | 142 | 591 | 152 | 124 |
- HTML: 322
- PDF: 127
- XML: 142
- Total: 591
- BibTeX: 152
- EndNote: 124
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Having read Dalton et al’s paper, I would like to draw attention to a number of items they have misconstrued, confused, or omitted. I do not, however, dispute their main point that the shothole drillers’ log-derived muskeg depths can be suspect.
First, the seismic shothole drillers’ log database was a project iteratively compiled and published over several years of effort trawling through Industry’s archives (see discussion in Smith (2015)). Dalton et al. have used the version 2 database (Smith and Lesk-Winfield 2010a – GSC Open File 6049) which contained approximately 275,000 records, and not the final database compilation (Smith, 2011 – GSC Open File 6833) which contained ~344,000 records; although they cite both publications, they appear to only use the actual data from OF 6049.
The use of these records to decipher regional extents and thicknesses of muskeg is also something that I had already published – based on the same version 2 database that they used – Smith and Lesk-Winfield 2010b – GSC Open File 6410, so, Dalton et al.’s “application” of the shothole data in this example is not original (our number of “muskeg” records queried from the version 2 database differ only slightly – 13,140 in OF 6410 and 13,014 for Dalton et al. (there’s some “complexed” term records they have missed) . Note, in this OF 6410 publication, I had also included the records from S. Smith et al. (2005) (digital borehole geotechnical database), which Dalton et al. has also I presume discovered and included. I too share Dalton et al’s profound sense that these geotechnical borehole records are far more accurate, and indeed I had suggested and used these as reliable “anchor” points to ground regional interpretations of the shothole records (although there are also limitations to the geotechnical records (and more broadly of hand auger/probe records with refusal at base of active layers) that also need to be understood).
Derivative publications based on the lithostratigraphic shothole drillers’ log data were created and published for 13 different thematic/subject areas at various stages of the database compilations – muskeg thickness included, although if you understand the nature of the data and its collection, it’s probably one of the more suspect records. Indeed, in OF 6410 I made the following comments in the Discussion:
"As with all other aspects of the shothole database, caution is required when interpreting this data. The most obvious constraint is that muskeg presence and/or thickness was not consistently recorded by drillers, or in all areas. Muskeg is clearly more extensive than the distribution of points illustrated here conveys. The data presented here should instead be used to deduce generalized regional trends, principally as they pertain to thickness. More specific inferences can be made where a proposed development coincides directly with actual “muskeg” records. Users seeking further information on peatlands along the Mackenzie corridor are directed to Aylsworth and Kettles (2000; and references therein)."
Note, my comments published in OF 6410 clearly identify a major contention I have with how Dalton et al. have used this data. Quite simply, you cannot use the “null” hypothesis with the shothole data – i.e., just because a driller did not record the presence of a particular unit (in this case, surface muskeg presence), you cannot ascribe a non-presence to that data point. What you can do at a minimum I would argue (with a reasonable sense of confidence), is use the records that do identify muskeg as a simple record of “presence,” separate from any considerations or concerns of unit thickness. This requires a re-querying of the database (already done in OF 6410…but please consider using instead the final database in OF 6833) to include complexed drillers logs where “muskeg” is identified in combination with other lithostratigraphic records (e.g. 0-10 m, muskeg, clay, rocks), but its individual thickness is not identified – for OF 6410, this includes an additional 8331 records (identified in the publication as a separate GIS layer).
Dalton et al use the word “haphazard” to describe the data collection – I have chosen to use “inconsistent” to better describe what can be discerned from this data collection…and specifically the caution I have urged users to employ with this data. To correct Dalton et al., the lithostratigraphic materials and the drill logs were created by drillers, NOT geologists, and there is no way to go back and interrogate seismic or petroleum companies as to what instructions were given to drillers for guiding lithostratigraphic logging – this was never something I’d ever encountered that was written down. There are about 5 different formats (with several different variations) that drill log records were written up/recorded through the 50 or so years of records in the shothole database. Different companies may well have instructed their drillers to note particular items, perhaps, including muskeg, however there is no way to know this from the archival records.
Dalton et al. also identify the “stepped” nature (modal depth) of reported unit thicknesses – something I had also recognized and discussed in the shothole publications. They are quite correct, but for this, a greater appreciation of how these shotholes were drilled, and how the data was recorded is needed. The principal reason these shotholes were drilled was to seat explosive charges, preferably in bedrock, for petroleum seismic exploration. Because it mattered in terms of static interferences and seismic propagation, it was most critical for the drillers to note whether the charge was seated in bedrock, and at what depth that occurred (this then was something more accurately recorded), and then if not in bedrock, what the nature of the overlying sediments were (lithologies and depths). Any other lithostratigraphic attributes that the drillers recorded over and above that is considered baseline geoscience “gravy”. Some of this may well have been collected to assist in development/access considerations (i.e. muskeg thickness was problematic for any contemplated summer petroleum well drilling operations; seismic shothole drilling was always conducted in the winter months), but any conversations I had with petroleum or exploration companies wouldn’t support this – indeed, this kind of use/insight for the shothole data was almost wholly unrecognized previously. Shothole drilling rigs (albeit the records would reflect changes in technology used over a 50-year time period) typically used air-rotary drills with 5- to 10-foot-long auger flights (1.5 to 3 m). It can be surmised that some drillers may have been “homogenizing” lithostratigraphic records in accordance with individual auger flights, or pairs of flights …i.e. “in the upper 5 feet, it was mostly muskeg, but maybe a foot or so of clay and rocks, so I’m going to write down 0-5’ muskeg.” This is not always the case, and indeed you can interrogate the records to see where individual drillers along a particular seismic line have much more precisely broken-down depths by the foot (i.e. not correlating to drill stem intervals of 5’, 10’, 15’ or 1.5, 3, 4.5 m etc.); any hope for reporting precision beyond the foot or half metre is simply not realistic to the drilling and recording practices employed. This is, however, one way I have previously queried out records I am more confident in, while regarding others as more generalized, or regarded as “maximum thickness” estimates – look for records with specific depths that don’t match drill-stem length intervals. You can also look where there are crossing/proximal seismic lines from different years reporting very much the same thing – adds weight to their reliability.
Can drillers’ logs be completely wrong or misleading …absolutely! I similarly regard the exceptional muskeg thickness records (>10 m…indeed probably anything >5 m) as highly dubious. Can’t provide the explanation for these, other than to go back and re-interrogate the original archival record to see if there was a transcription error (tested at <0.5% for the overall database during pre-publication QA/QC).
So…can the shothole drillers’ log muskeg thickness records be used – perhaps with some greater caution and data interrogation, and probably best when anchored by what are considered more “reliable” records (e.g. geotechnical borehole data). Can they be used to identify trends in muskeg thickness across a broader region – probably. At a minimum, drillers’ log records citing the presence of muskeg, can be used to simply identify presence …at that specific shotpoint…but CANNOT be used in the null sense. If you choose to use the data for simple presence…then consider also that there is a separate shothole drillers’ log compilation for Banks Island that also includes muskeg records (Smith and Farineau 2015 – GSC Open File 7322).
Regards,
Rod Smith
Smith 2015 http://dx.doi.org/10.1016/j.grj.2015.01.005
Smith and Farineau 2015 http://dx.doi.org/10.4095/296215
Smith 2011 http://dx.doi.org/10.4095/288754
Smith and Lesk-Winfield 2010a http://dx.doi.org/10.4095/261497
Smith and Lesk-Winfield 2010b http://dx.doi.org/10.4095/261784