the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 01 Oct 2024
The SahulCHAR Collection: A Palaeofire Database for Australia, New Guinea, and New Zealand
Abstract. Recent global fire activity has highlighted the importance of understanding fire dynamics across time and space, with records of past fire (palaeofire) providing valuable insights to inform current and future management challenges. New records from the recent increase in palaeofire studies from Australia and surrounds have not been captured in any database for broader comparisons, and Australasia is poorly represented in current international databases used for global modelling of palaeofire trends. These problems are addressed by SahulCHAR, a new collection of sedimentary charcoal and black carbon records from Sahul (Australia, New Guinea, and offshore islands) and New Zealand. Data are stored in the OCTOPUS relational database platform, with a structure designed for compatibility with the existing Global Paleofire Database. Metadata are captured at site-level and observation-level, with observations including age determinations and charcoal or black carbon data. SahulCHAR Version 1 contains 687 records of charcoal or black carbon, including digitized data, unchanged and modified records from the Global Paleofire Database, and original author-submitted data. SahulCHAR is a much-needed update on past regional palaeofire compilations that will also provide greater representation of records from Sahul and New Zealand in future global syntheses.
- Preprint
(1914 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 25 Nov 2024)
-
RC1: 'Comment on essd-2024-328', Patrick Bartlein, 17 Nov 2024
reply
General comments:
This paper describes a late-Quaternary sedimentary charcoal database from the Sahul region (Australia, New Guinea, and New Zealand). The data set will be a welcome update for a region where the variations of fire are particularly important for our overall understanding of the controls and history of biomass burning. A considerable amount of effort has been expended in harmonizing metadata and retrieving data from original sources. For the most part, the manuscript is well written and clear. However, there are a few misapprehensions or omissions in the text, and the relationship between the specific data in this database and “legacy” databases (e.g. GPD/paleofire.org) needs to be reviewed and perhaps reconsidered.
My first specific comment below illustrates some of the general issues in the initial development and propagation of databases: misapprehension of the heritage and contents of existing databases (and uncritical acceptance of those contents). This is not a simple matter of attribution, but these issues can have impacts on the scientific results they are used to generate if they are not appreciated by uncritical users. I think the current manuscript and database do a good job of avoiding those issues, but they are not completely free from them.
Consequently, I think a little light editing and an “internal review” of the GPD-to-SahulCHAR data transfer would suffice to make the manuscript publishable, and it would certainly be timely, given the apparent community-wide desire to ingest charcoal data into Neotoma.
Specific comments:
p. 3, line 70: “However, the last major compilation and synthesis of sedimentary charcoal records from Australasia was Mooney et al. (2011, 2012), containing 224 sedimentary charcoal records, primarily derived from the Global Paleofire Database (GPD, formerly known as the Global Charcoal Database; Power et al., 2010).”
There are several issues here: Mooney et al. (2011) used data from the Global Charcoal Database version 2 (GPDv2, Daniau et al. 2012): (“We extracted 196 sedimentary charcoal records this broadly defined Australasian region (20N - 50S and 100E to 177W) from Version 2 of the Global Charcoal Database (GCD-V2; Daniau et al., in preparation,) compiled by the Global Palaeofire Working Group (GPWG: http://gpwg.org/). These data were supplemented by an additional 27 sites, chosen to increase the number of long records and to improve the spatial coverage (GCD-V2.5).”) Mooney et al. (2012) used the same data set (“GCDv2.5”), which was a preliminary version of GCDv3 (Marlon et al., 2013, 2016).
Neither paper therefore used data from the Global Paleofire Database (as represented by the R package GCD (e.g. GCDv4.0.7), or the web interface at paleofire.org, neither of which were in existence in 2011 or 2012). There is a distinction between the various GCD’s of Power, Daniau, and Marlon, and the derivative GPD (“GPDv4” available as the R package GCDv4.0.7 or via the web interface at paleofire.org). Although the new database (SahulCHAR) does not directly depend on any of the GCD databases, it does so indirectly through the GPD, which in some instances degraded the data in the antecedent GPDv3 (see comment for p. 3, line 86 below). This kind of fuzziness contributes to mispprehension of the actual sources of the “legacy” data in subsequently produced databases, and the extent to which those data may or may not have problems that may be inherited by subsequent databases. This is a community-wide issue, and not specific to SahulCHAR, but it would be good to straighten things out here before publication.
p. 3, line 82: “These records are replicated in the RPD [RPDv1b] …” “Replication” here might imply “copy”, which was not quite the case (Harrison et al., 2022). Many records were given new chronologies, and in some cases the GPD data were replaced by data obtained from original authors.
p. 3, line 86: “As noted by Harrison et al. (2022), current limitations of the GPD include potential duplicates of sites, missing metadata and age data, and necessary updates to incorporate newly published records.” In addition, the GPD lost precision in locational metadata and in key variables such as sample depth and charcoal values relative to the parent GCDv3 data. For example, Galway Tarn’s location in the GPD (e.g., in GCDv4.0.7), which is inherited here, is given to two decimal places, while in GCDv3 it was given to six. This seems like a negligible difference, but resolves to a distance of about kilometer, which could become significant in matching different data sets or extracting modern climate data for sites.
In the case of sample depth, loss of precision would influence the calculation of influx from concentrations (or the reverse), and this would be compounded if the moribund R paleofire package was used, which defaults to a “spreadsheet” approach for calculating sample thickness. (Explicit sample thickness values that were present in GCDv3 were dropped in GCDv4.) Consequently, it would be important to not uncritically accept data from the GPD, but instead to review what has been copied.
Along those lines, it would be optimal to add an additional field to the metadata, perhaps called “HOWCOPIED” that might have the levels “verbatim” (to indicate data copied without modification), “repaired” (for instances where, for example, depth was recorded in centimeters, when it was actually in meters), “replaced” (for instances when data in the source database that might have been degraded in precision is replaced by author-provided data), and so on. (I don’t know exactly what was done in migrating data from the GPD, so there may be more levels/categories.)
Tables 1-3: It would be good at some point to point out that the tables are for discussion here, and are not constituent tables in the database.
p. 9, line 161: “… during data migration to the GPD.” I would have reservations about doing that. I think that the SahulCHAR database has several advantages over the current GPD, and should stand alone, or become a constituent of a newer database.
p.9, line 173: “Char data were last accessed from PANGAEA…” The R code unique(sahulchar_sf$DATASOURCE) applied to the sahulchar sf/data.frame returns “ "Author", NA, "Digitised", "Paleofire database" ”. Are the PANGAEA data folded into one of those categories?
p. 9, line 173: “While the RPD contains Australasian char data, these records are derived from the GPD and therefore the RPD was not used for 175 SahulCHAR data compilation.” But RPDv1b does contain alternative chronologies relative to those originally published (as well as other fixups), and so the RPDv1b data are not a simple mirror of the GPD data.
p. 12, lines 188-192: This discussion explains why migrating SahulCHAR to GPD is a bad idea.
Table 3: “CHARCOUNTS” Inspection of the data (and the example in the table) shows that the values in this field are not always counts (which should be integer values), but include influx and concentration values, ratios, etc. To avoid future misapprehensions of what the database contains, I suggest renaming this to “CHARVALUES” or something (but not just “CHAR” which could be ambiguously interpreted as “charcoal accumulation rate”).
Figs. 3 and 4: Pie charts are denigrated for the kind of application here.
Fig. 4d: Why the strange class intervals?
Technical comments:
I made several attempts to access the data:
1) Via the web interface at https://octopusdata.org/:
I was able to open the interface, but when attempting to export data, got the following message, which implies the available data online are only a subset of the database intended to illustrate the kind of metadata available:
“Sedimentary Charcoal & IPPD: The data provided here is a subset, including only overview information at the collection-unit and dataset levels. For complete data access, refer to the online documentation and use WFS (Web Feature Service).”
2) Via WFS and QGIS, using the link: http://geoserver.octopusdata.org/geoserver/wfs
Using the QGIS instructions at https://octopus-db.github.io/documentation/usage.html, I was able to open the “Sedimentary Charcoal Records: Sahul & NZ (all data)” collection, but could not find data such as the sample depths and charcoal values, only metadata. This is consistent with the above message.
3) Using the R instructions at https://octopus-db.github.io/documentation/usage.html
I was not able to open the collection, getting the error message:
“Cannot open "http://geoserver.octopusdata.org/geoserver/wfs?service=wfs&version=2.0.0&request=GetFeature&typename=Sedimentary%20Charcoal%20Records%3A%20Sahul%20%26%20NZ%20%28all%20data%29&srsName=EPSG%3A900913"; The file doesn't seem to exist.”
(I’m guessing that upon acceptance of the paper, these three approaches will “go live”. If not, some fixups or better instructions would be in order.)
4) Via the downloadable shapefile at http://dx.doi.org/10.25900/KKDX-XH23 and R:
I was able to able to open the file, and export the data as a .csv file, using the following code:
library(sf)
shp_file <- "/Users/bartlein/Dropbox/Docs/Rehn/sahulchar/sahulchar.shp"
sahulchar_sf <- st_read(shp_file)
csvfile <- "/Users/bartlein/Dropbox/Docs/Rehn/sahulchar.csv"
write.csv(sahulchar_sf, csvfile)This yielded a .csv file that (it’s fair to say) is a little messy. However, simple sorting on site-name and sample-depth fields (or parsing the sf/data.frame using R) would yield data that could be easily incorporated into an analysis.
P.J. Bartlein
Citation: https://doi.org/10.5194/essd-2024-328-RC1
Data sets
SahulCHAR Emma Rehn et al. http://dx.doi.org/10.25900/KKDX-XH23
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 239 | 30 | 64 | 333 | 6 | 6 |
- HTML: 239
- PDF: 30
- XML: 64
- Total: 333
- BibTeX: 6
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1