the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 31 Aug 2026
How Dutch subsurface professionals find, understand, and share information: a sector-wide survey
Abstract. Safe and effective subsurface governance depends on information that is transparent, accessible, and independently verifiable. However, while much research has examined how geoscientists communicate with the public, little is known about how information moves among the professionals who research, regulate, conduct, and contest subsurface activities. We surveyed 135 professionals from more than 55 organizations across the Dutch subsurface sector, spanning research institutes, universities, national and regional government, advisory bodies, industry, and consultancy, on how they find, understand, trust, and share subsurface information, and where each breaks down. In general, finding or understanding information was not perceived as particularly difficult, but respondents still reported barriers to both: fragmentation across sources was by far the most frequently cited obstacle, and information was often insufficiently documented and explained. Intellectual property concerns and limited time and capacity mostly constrained sharing. Information flowed mainly within clusters of professionally similar organizations, with (environmental) NGOs and community organizations largely disconnected from the rest of the sector. Respondents strongly valued transparent communication of uncertainty but reported that it was insufficiently practiced and poorly taught within their organizations. Knowledge institutions scored highest on source credibility, including trustworthiness, while organizations at the sector's periphery were rated lower. The Dutch subsurface information system thus works reasonably well for those at its center but poorly reaches those at its edge. We identify four recommendations for policy and practice: a curated and standardized national information platform, the extension of information flows to civil society, guidance on communicating uncertainty, and greater transparency from less-credible sources.
- Preprint
(969 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 26 Oct 2026)
-
RC1: 'Comment on egusphere-2026-4774', Anonymous Referee #1, 30 Sep 2026
reply
-
AC1: 'Reply on RC1', Helena Schmidt, 11 Oct 2026
reply
We sincerely thank the reviewer for carefully reading our manuscript and for the constructive, detailed comments. They have helped us sharpen the definitions of our key concepts, be more precise about what our sample and design allow us to conclude, and improve the transparency of our statistical reporting. Below, we respond to each comment in turn and describe the changes we intend to make. Line numbers refer to the discussion version of the manuscript.
Comment 1: Definitions of credibility, trustworthiness, competence, and transparency
"The terms 'credibility', 'trustworthiness', 'competence' and 'transparency' should be defined more clearly. They are related but not identical, and clearer definitions would help readers interpret better the results."
We agree that these concepts deserve explicit definitions, and we thank the reviewer for pointing this out. We will make three changes.
First, we will add a short paragraph to Section 2.1 (after L135) defining source credibility as the overarching concept and explaining how the four measured dimensions relate to it. The proposed text reads:
"We conceptualized source credibility as the overall evaluation of an organization as a believable provider of information. We measured four dimensions: clarity (the extent to which the information an organization provides is clear and easy to understand), transparency (the extent to which an organization is open about its methods, data sources, and limitations), competence (the expertise an organization demonstrates in the subsurface topics it communicates about), and trustworthiness (the extent to which respondents trust the information an organization provides)."
Second, we will add the exact wording of the four items to Table 1 (Block 5), so that readers can see precisely what respondents rated:
- Clarity: "The information the organizations provide about the subsurface is clear and easy to understand."
- Transparency: "The providing organizations are transparent about their methods, data sources, and limitations."
- Competence: "The organizations demonstrate expertise and competence in the subsurface topics they communicate about."
- Trustworthiness: "I trust the information these organizations provide about the subsurface."
Third, we will use these terms consistently throughout the manuscript. We will use "trustworthiness" for the measured construct (including in Table B4, where the heading currently reads "Trust in information"). We will also distinguish more clearly between a source's transparency, as defined above, and transparency about uncertainty as a characteristic of information (Section 3.2.4).
Comment 2: Self-selected sample and the term "sector-wide"
"The manuscript should more clearly acknowledge the limitations of the sample. [...] the expression 'sector-wide survey' may give the impression of broader representativeness than the sample allows."
We agree and thank the reviewer for this suggestion. We will change the title to "How Dutch subsurface professionals find, understand, and share information: a cross-sector survey".
We will also state the nature of the sample more prominently and earlier in the manuscript, rather than only in the Limitations. In Section 2.2 (after L167), we will add: "Because participation was voluntary and recruitment relied on existing contacts and forwarded invitations, the study used a non-probability sample. The results therefore describe the views of the participating professionals and are not statistically representative of the Dutch subsurface sector as a whole." In Section 3.1 (after L199), we will add that the sample covers a broad range of organizations (Table A1) but does not represent the sector proportionally.
Comment 3: Cross-sectional design and causal interpretation
"In some parts of the Discussion, the authors seem to move from observed associations to possible causal explanations, for example regarding information flows, source credibility and actors' positions within the sector. [...] More cautious wording would strengthen the manuscript."
We fully agree that the cross-sectional design does not allow causal conclusions. We describe the design as cross-sectional in Section 2 (L124) and explicitly note that it precludes causal inference in Section 4.2.4 (L648–649), regarding the relationship between source credibility and actors' positions in the sector, and in Section 4.3 (L660–662), regarding information flows and trust. Where we offer possible explanations for observed differences, we have phrased them as interpretations, using hedges such as "likely", "probably", and "could".
We were not certain which further statements the reviewer had in mind. We would be grateful if the reviewer could point us to any specific passages they consider problematic, so that we can address them directly in the revised manuscript.
Comment 4: Limited participation of civil society (n = 2)
"The limited participation of civil society (n = 2) should be more prominently acknowledged when interpreting patterns of trust, credibility and information exchange. Any conclusions about this group should be very cautious."
We agree and thank the reviewer for raising this point. In revising, we will make one distinction explicit that we believe is important for readers. We excluded the two civil society respondents from all group comparisons (L177–179). Our findings about civil society therefore do not reflect civil society's own views, but how professionals from the other stakeholder groups engage with and perceive (environmental) NGOs and community organizations. The ratings of these organizations as information providers were given by respondents who reported consulting them at least sometimes, which concerns more than the two civil society respondents but still a relatively small number of raters.
We will make the following changes:
- Section 2.3 (after L179): add that the exclusion concerns the views of civil society respondents themselves, whereas respondents from the other stakeholder groups rated these organizations as information providers.
- Section 3.2.4 (L335–343): note that the ratings of NGOs and community organizations rest on relatively few raters. The confidence intervals we will add in response to Comment 6 will make the lower precision of these estimates visible.
- Abstract (L15–17) and Conclusions (L695–699): replace "largely disconnected from the rest of the sector" with wording that reflects what we observed, namely that these organizations were rarely consulted or addressed by the other respondents. We cannot observe the reverse direction.
- Sections 4.2.2 and 4.2.4: add a caveat that we can describe how other professionals engage with civil society, but not how civil society actors themselves seek and share subsurface information. We will also frame the recommendation to extend information flows to civil society as following from the reported behavior of the professionals in our sample, while noting that civil society's own needs remain to be investigated (as already stated in Section 4.3, L678–681).
Comment 5: Testing strategy, correction procedures, and effect sizes
"The text reports many statistical comparisons. A clearer explanation of the testing strategy, choice of tests and correction procedures would improve transparency. Emphasizing effect sizes and general patterns, rather than isolated p values, would make the results easier to interpret."
We thank the reviewer for this helpful suggestion. Section 2.3 (L179–186) already describes which test was applied to which type of item and why: Fisher-Freeman-Halton exact tests for multi-select items, because they are robust to sparse cell counts; Kruskal-Wallis tests with Bonferroni-adjusted pairwise comparisons for single ordinal items; and repeated-measures ANOVA with Greenhouse-Geisser correction and Bonferroni-adjusted pairwise comparisons for batteries of items rated by all respondents. We will strengthen this section and the reporting of results in four ways:
- Post-hoc procedure. We will specify (after L180) how we determined which stakeholder groups were over- or underrepresented after a significant Fisher-Freeman-Halton test, namely by inspecting adjusted standardized residuals (|z| > 1.96).
- Correction for multiple testing. We will apply corrections at two distinct levels. As before, pairwise comparisons following a significant Kruskal-Wallis test will be Bonferroni-adjusted within each item. In addition, we will apply the Benjamini-Hochberg false discovery rate correction (q = .05) to the omnibus tests within each item family (for example, topics, purposes, and barriers). Tables B1 and B2 will indicate which results remain significant after this correction, and the Results will focus on those. We will update the corresponding limitation in Section 4.3 (L658–659).
- Effect sizes. We will report Cramér's V for the Fisher-Freeman-Halton tests (Table B1) and rank-based eta squared (η²H) for the Kruskal-Wallis tests (Table B2), alongside the partial eta squared values already reported for the repeated-measures ANOVAs (Table B3). Section 2.3 will state which effect size measures we used and how we interpret them.
- Patterns rather than isolated p-values. We will revise the paragraphs on stakeholder differences in Sections 3.2.1, 3.2.2, 3.3.1, and 3.4.1. Each will open with the overall pattern (for example, that differences in topics and purposes largely followed organizational mandates), illustrate it with the largest effects, and refer to the appendix tables for the full test statistics.
Comment 6: Confidence intervals
"Adding confidence intervals for key descriptive results would help readers judge the precision of the estimates, especially for smaller stakeholder groups."
We agree and will add 95% confidence intervals to all means reported in Table 5 and Table B4. Table B4 contains the ratings of information providers, including the organization types rated by the fewest respondents, so the intervals are most informative there. We will note that, given the non-probability sample, the intervals indicate the precision of the estimates within this sample rather than allowing inference to the population.
We have decided not to add confidence intervals to the percentages in Table 3. This table reports nearly 150 descriptive frequencies of multi-select items, and adding intervals to each would considerably expand the table, which would run counter to the reviewer's request to shorten the manuscript (Comment 8). The comparisons between stakeholder groups for these items are reported with test statistics and, following Comment 5, effect sizes in Table B1.
Comment 7: Preferences versus evidence on uncertainty communication
"Respondents' preference for a best estimate with a numerical range, visual formats and multiple scenarios is useful. Still, the text should more clearly separate preferences from evidence that these formats improve understanding or decision making."
We thank the reviewer for their positive assessment of these findings and agree that the distinction should be clearer. We will rewrite L627–631 in Section 4.2.3 as follows:
"Our respondents' format preferences are consistent with this evidence: they favored a best estimate with a numeric range, visualizations, multiple scenarios, and percentiles over vaguer qualitative terms. However, these are stated preferences, and our data do not show whether these formats improve respondents' own understanding or decision-making. Moreover, much of the cited evidence stems from studies with lay audiences rather than professionals. Our findings do show a gap between how strongly respondents value transparent uncertainty communication and how adequately they find it practiced and supported in their organizations. Practical support, such as guidance on formats and terminology grounded in the existing evidence (van der Bles et al., 2020; Erharter et al., 2024; Juanchich et al., 2025), could help close this gap."
In line with this comment, we will also tone down our conclusions about organizational guidance. Respondents' mean rating of the adequacy of guidance (M = 3.62) lies between "neither agree nor disagree" and "agree". We will therefore replace "poorly taught" in the Abstract (L18) and Conclusions (L697), and "little guidance" and "limited guidance" in Sections 4.2.3 (L615) and 4.1 (L573), with "only moderately supported by organizational guidance". We will also correct the verbal label for this item in Table 5 to "Neither nor – agree", as it was reported incorrectly.
Comment 8: Length
"The manuscript is quite long. Please consider streamline some sections to improve focus and readability."
We agree and will shorten the manuscript, which will also offset the additions made in response to the other comments. Specifically, we will:
- Condense the summary of results in Section 4.1 (L523–578), which partly repeats Section 3, to a short overview that leads into the cross-cutting themes in Section 4.2.
- Condense the paragraphs on stakeholder differences in Sections 3.2.1, 3.2.2, and 3.3.1 into pattern summaries, referring to the appendix tables for item-level results (see also Comment 5).
- Merge Section 3.6 (preferred tools and formats) into Section 3.5, as both address respondents' suggestions for improvement and partly overlap.
- Remove overlap between the "other" responses in Section 3.2.3 (L302–305, L309–312) and the open-ended themes reported in Section 3.5.
- Tighten the Introduction, particularly the description of the Groningen case (L31–48) and the three conditions (L49–80).
Comment 9: Aleatory versus aleatoric
"L.403: The text uses 'aleatory uncertainty' and Table B1 uses 'aleatoric uncertainty'. Ensure consistent use of the term."
Thank you for spotting this. We will use "aleatory" throughout, including in Table B1. We will also harmonize the item labels between Table 3 and Table B1, some of which currently differ.
Comment 10: Acronyms
"Ensure all acronyms are explicitly defined."
We agree. We will define all acronyms at first use in the text and in each table note, so that tables can be read independently. This includes, among others, NAM, SodM, KNMI, and TNO (first used at L38, but currently defined later or not at all), KEM, CCS, NLOG, DINOloket, ThermoGIS, BRO, SCAN, LLM, and the organization abbreviations in Table 5 (RVO, EZK, EBN, PBL, TNO-GDN). We will also add the missing abbreviation ADV to the note of Table B1.
We thank the reviewer again for their time and thoughtful comments, which we believe will substantially improve the manuscript.
Citation: https://doi.org/10.5194/egusphere-2026-4774-AC1
-
AC1: 'Reply on RC1', Helena Schmidt, 11 Oct 2026
reply
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 185 | 91 | 47 | 323 | 49 | 45 |
- HTML: 185
- PDF: 91
- XML: 47
- Total: 323
- BibTeX: 49
- EndNote: 45
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents original research on how Dutch subsurface professionals seek, interpret, trust and share information, showing structural barriers such as fragmented data sources, limited information flows between actor groups and difficulties in communicating uncertainty.
General Comments:
The paper addresses highly relevant scientific and policy questions that fall clearly within the scope of GC.
The topic is relevant to subsurface governance, particularly given the increasing importance of geothermal energy, subsurface energy storage, Carbon Capture and Storage and other underground activities.
The manuscript is logically structured, the empirical methodology is appropriate and the writing is clear. It is an important and very interesting contribution that provides well‑supported recommendations for improving subsurface information practices and governance.
A few points, however, would benefit from clarification and revision.
Specific Comments:
- The terms “credibility”, “trustworthiness”, “competence” and “transparency” should be defined more clearly. They are related but not identical, and clearer definitions would help readers interpret better the results.
- The manuscript should more clearly acknowledge the limitations of the sample. The respondents were self-selected, so the results should not be presented as statistically representative of the Dutch subsurface sector as a whole. For example, the expression “sector-wide survey” may give the impression of broader representativeness than the sample allows. Cross-sector survey or multi-stakeholder survey or another equivalent expression may be a more precise description.
- The cross-sectional design should also be considered more clearly when interpreting the findings. In some parts of the discussion, the authors seem to move from observed associations to possible causal explanations, for example regarding information flows, source credibility and actors’ positions within the sector. These interpretations are plausible, but the cross-sectional design does not allow causal relationships to be established. More cautious wording would strengthen the manuscript.
- The limited participation of civil society (n = 2) should be more prominently acknowledged when interpreting patterns of trust, credibility and information exchange. Any conclusions about this group should be very cautious.
- The text reports many statistical comparisons. A clearer explanation of the testing strategy, choice of tests and correction procedures would improve transparency. Emphasizing effect sizes and general patterns, rather than isolated p‑values, would make the results easier to interpret.
- Adding confidence intervals for key descriptive results would help readers judge the precision of the estimates, especially for smaller stakeholder groups.
- The findings on uncertainty communication are very interesting. Respondents’ preference for a best estimate with a numerical range, visual formats and multiple scenarios is useful. Still, the text should more clearly separate preferences from evidence that these formats improve understanding or decision‑making.
- The manuscript is quite long. Please consider streamline some sections to improve focus and readability.
- The disclosure about using Claude and Grammarly is appropriate. The authors clearly state these tools were used only to improve readability, grammar, and structure.
Technical Corrections:
- L.403: The text uses ‘aleatory uncertainty’ and Table B1 uses ‘aleatoric uncertainty’. Ensure consistent use of the term.
- ensure all acronyms are explicitly defined
Conclusion:
Overall, this manuscript is very interesting and well written, bringing an important insight into geoscience communication research and contributing to the advancement of the field. The results are well aligned with the research questions and provide empirical basis for the recommendations proposed.
Should the authors find it relevant, the main revisions I suggest concern the clarification of some concepts, a more explicit acknowledgement of sampling and generalizability limitations, some caution when interpreting causal relationships, and a clearer distinction between respondents’ preferences and evidence of communication effectiveness.