Corpus 1,272 assessed · 1,173 scored · 643 reproduced ≥75 · 168 flagged ·∅ 74.1/100
← New search

GlASS - Global Aggregation of Stream Silica.

Sci Data · 2025
L1 89/100 PQI 88
Why this verdict

The main results reproduced: recomputed values matched the published ones within tolerance.

Reproduced on the brainbox compute brainarbeit.com
Scoring basis — itemised

Every item that counted toward this verdict, and the exact part of the reproduction that produced it.

Supporting (toward a concern)
Content-critical question only partially held
+2 pts
From: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q8 · Severity of the miss (overall human judgment) 🟡
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +3
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • No authors-side cause for any deviation
  • Any deviation was negligible
  • The central claim held under reproduction
What did not (or only partly)
  • 🟡A deviation arose in the data or preprocessing
  • 🟡Reported values were not (fully) derivable from the shared data
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
89/100
Reproducibility score
0.8 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 77% of all assessed papers rank 246 of 1173 scored

A 0–100 reproducibility-quality score from the per-question grades, shown as a z-score: standard deviations above (+) or below (−) the mean of comparable assessments.

Reproduction agent’s raw note

Described well enough and reproduces 1:1. Data is a USGS ScienceBase descriptor (10.5066/P138M8AR); BRIEF's Zenodo DOI was wrong (it pointed to an unrelated land-cover product) and was corrected. Recomputed the descriptor's headline summary stats directly from the published data records on «our HPC»/«infra»: 13/17 claims exact (river count 421, obs 605718, NH4 196 sites, P 339 sites, all four temporal ranges, 24 networks, years-of-record 4-55, per-record range 1-178, latitude -77.6..70.2), 2 within-tolerance (NO3/NOx 395 vs 397; watershed ~400-421), 1 partial (median obs/yr 12.8 vs 14.2, ~10% low), 1 mismatch flagged for human audit (climate zones: 10 names/19 codes in v2 vs reported 11; likely a version/aggregation difference, not evident fabrication). NOT attempted (optional ~20%): re-running the full R/WRTDS harmonization from the 24 raw upstream network feeds, which are not all publicly downloadable and are unnecessary to verify the descriptor's claims.

💻 Code ↗ 🗄 Data: 10.5281/zenodo.8239305

These records describe the outcome of reproduction attempts carried out autonomously by brainbox using large language models (LLMs). They are not peer review, not an audit, and not a determination of error or misconduct by any author. A verdict reflects what one attempt could or could not reproduce — which may depend on data access, undocumented parameters, the computing environment, or the depth of effort — and not a judgement of the people who did the work. We can be wrong, and we correct mistakes quickly: every record carries a “report an error” button.

Assessment versions

Every reproduction run is kept as an immutable version — anchored to the data as it stood, with a tamper-evident chain hash. A rerun (e.g. after an author updates a deposit) adds a new version; the previous one stays on record.

  1. v1 current initial assessment Score 89
    assessed: 2026-06-15 ⛓ 97bfd02c8ed5
✎ I am an author of this paper

Updated or fixed a deposit, or is there an erratum? Ask us to re-run the metrics. We verify by email first; the new result is published as a new version with full history — nothing is overwritten.

Reason for the rerun

We email you a confirmation link first. The rerun is an objective re-measurement — it cannot change the verdict in your favour, only ask us to look again.

Provenance — full disclosure

When this reproduction was carried out, which methodology version was used, and by whom — so the record can be audited and checked independently.

Reproduced
2026-06-15
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-16
no human curator yet
Last updated
2026-08-05

Provisional, curator- or AI-assessed, and independently checkable. A reproduction outcome states what one attempt could reproduce — not a judgement of the authors.

Deep full-text extraction

Model: opus
Founding hypothesis

The paper describes the construction of the Global Aggregation of Stream Silica (GlASS) database to assess changes in river silicon (Si) concentrations and fluxes, their relationship to other nutrients (N and P), and to evaluate the mechanisms driving Si availability across global rivers.

Core claims
  • GlASS harmonizes dissolved Si, dissolved inorganic N, and dissolved inorganic P concentrations, daily discharge, and watershed characteristics for 421 rivers spanning multiple climate zones from 1963 to 2024 resource
  • The database contains over 600,000 individual nutrient chemistry observations across 421 rivers, each paired with daily discharge data enabling load estimation resource
  • Rivers deliver >80% of annual Si loads to global oceans, linking terrestrial weathering, nutrient, and carbon cycles via siliceous diatom primary production finding
  • Using a portion of this dataset, the majority (62%) of 60 rivers examined showed significant increases in DSi yields over the past two decades, most markedly in alpine and polar regions finding
  • A subset of GlASS identified five distinct seasonal Si regimes across the Northern Hemisphere, with most rivers exhibiting multiple regimes over time finding
  • A reproducible QA/QC and harmonization workflow standardizes disparate datasets into common units, formats, and naming with public R code method
  • Watershed-scale climate and ecosystem productivity factors (snow cover, temperature, green-up day, evapotranspiration) are associated with variation in seasonal Si regimes mechanism
Experimental setups
Assay System Perturbation Readout Platform
Dissolved silica (DSi) concentration aggregation 421 rivers across 11 Koeppen-Geiger climate zones (−77S to 70N) none DSi concentration (mg Si/L)
Dissolved inorganic nitrogen (DIN) concentration aggregation 397 sites (NO3/NOx) and 196 sites (NH4) none NO3-N, NOx-N, NHx-N concentration (mg/L)
Dissolved inorganic phosphorus (DIP) concentration aggregation 339 river sites none SRP or PO4-P concentration (mg/L)
Daily discharge data aggregation 421 rivers none discharge (m3 s-1), gap-filled <30 day gaps via linear interpolation
Watershed delineation river watersheds (<1 km2 to ~4 million km2) none watershed boundary polygons HydroBASINS (Lehner and Grill)
Land cover characterization watersheds none proportion of each land cover class (forest, grassland, wetland, cropland, etc.) global land cover data at 30 m resolution
Lithology characterization watersheds none volcanic, sedimentary, plutonic, metamorphic, carbonate/evaporite classes PANGEA dataset
Elevation/slope and climate characterization watersheds none elevation, slope, precipitation, air temperature, snow-covered area, ET, NPP, green-up day, permafrost probability WorldClim (30 arc-second SRTM DEM); remotely sensed/modeled sources
Key results
  • 62% of 60 rivers examined displayed significant increases in DSi yields over the past two decades 62% of 60 rivers
  • Rivers deliver more than 80% of annual Si loads to global oceans >80%
  • Diatoms represent ~20% of photosynthetically fixed CO2 each year ~20%
  • Five distinct seasonal Si regimes identified across Northern Hemisphere rivers 5 regimes
  • Database contains over 600,000 individual nutrient chemistry observations across 421 rivers >600,000 observations
  • Interpolated discharge data represent less than 0.01% of the total dataset <0.01%
  • Median number of observations per year per stream was 14.2 (range 2.4 to 64) median 14.2
Key statistics
  • count 421 (total individual river sites in dataset (all with DSi and discharge))
  • count over 600,000 (individual nutrient chemistry observations)
  • percent 62% (of 60 rivers examined showed significant increases in DSi yields over two decades)
  • percent >80% (fraction of annual Si loads to global oceans delivered by rivers)
  • count 14.2 (mean/median number of samples per site per year)
  • count 397 sites NO3/NOx; 196 sites NH4; 339 sites P (number of sites reporting each nutrient species)
  • other <0.01% (proportion of discharge dataset that was interpolated/gap-filled)
  • count 24 (number of different observation networks providing chemistry data)

Statistical methods review

Model: sonnet

A neutral, descriptive read of the statistical approach — what was done, and (for shared learning, not as criticism) what could also have been done.

This is a data descriptor paper; no inferential statistical tests are performed within the manuscript itself. The paper documents the construction, harmonization, and quality assurance of the GlASS database (421 rivers, >600,000 observations). Descriptive statistics (counts, medians, ranges) characterize data coverage, and linear interpolation is used to fill short (<30-day) discharge gaps. Downstream analyses are cited as separate publications that used GlASS subsets.

Replicationna Sample size421 sites described; mean 14.2 samples per site per year (range 2.4–64); periods of record 4–55 years per river GroupsNo groups compared; observational database Pairingna Randomization/blindingna Dispersionrange
Approaches that could also have been used
  • Discharge gaps of fewer than 30 days were filled by linear interpolation
    Could also: Spline interpolation, autoregressive (AR) time-series imputation, or physically-based hydrograph recession models could also be used to fill short gaps — For streamflow, which often follows a smooth recession curve, spline or recession-based methods may better preserve hydrograph shape; reporting the method used (as the paper does) is the key transparency requirement regardless of choice
  • Time-varying watershed covariates (e.g., evapotranspiration on an 8-day step, snow cover at daily resolution) were summarized to annual mean values
    Could also: Seasonal or monthly summaries, or principal-component reduction of the full temporal profile, could also be derived and provided — Annual means compress seasonal dynamics that may be mechanistically important for Si cycling; seasonal summaries would preserve that information for users interested in within-year drivers
  • Land cover years between five-year increments (1985–2000) were linearly interpolated
    Could also: A step-function (hold previous value until next observation) or nearest-neighbour assignment could also be applied — Linear interpolation assumes a smooth trajectory of land-cover change, which may not reflect abrupt disturbances (fire, harvest); a step or nearest-neighbour approach makes the assumption of constancy explicit and may be more defensible for categorical land-use transitions
  • A single globally consistent spatial data source was chosen for each watershed covariate to ensure cross-site comparability
    Could also: A hierarchical approach using locally available, higher-resolution data where present and global products as fallback could also be used — Higher-resolution local data often reduce spatial mismatch error for smaller catchments; the paper notes this trade-off explicitly, and a hierarchical approach would let users query which data source underlies each site's covariate
  • Outlier screening was described as part of QA/QC but specific criteria (e.g., ±N SD, IQR fences, expert review) are not detailed in the main text
    Could also: Explicit, reproducible outlier criteria (e.g., Grubbs test, ±3 SD within site-year, or Hampel identifier) could also be documented and applied programmatically — Transparent, algorithmic outlier rules make the QA/QC step fully reproducible and allow users to apply alternative thresholds; the associated R code on GitHub partially addresses this, but inline documentation of the criterion aids interpretability
  • Central tendency of sampling frequency was reported as the median number of observations per site per year
    Could also: The full distribution (e.g., histogram or percentile table) or a site-level coefficient of variation in sampling density could also be reported — The median alone does not convey whether the spread in sampling frequency is driven by a few intensively sampled sites or by broad heterogeneity, which matters for load estimation uncertainty across sites
Software: R

Citation network

Where this publication sits in the reproducibility-weighted citation graph — what it is built on, and what is built on it. Citation data from OpenAlex.

Citations
0
Impact: low
Foundation confidence
None of its references are in our reproducibility record yet — its foundation cannot be assessed.
Topics

No assessed neighbours yet — the network grows as more papers are assessed.

Data lineage

The datasets this paper uses (text-mined from the full text via Europe PMC), and which other assessed papers stand on the same data. A shared dataset is a factual link — not a judgement.

10.5281/zenodo.8239305 DOI in References (http://purl.org/orb/References)
no other assessed paper uses this yet

What was reproduced

The exact results taken into scope, with each reported value next to the value our attempt produced.

n_rivers
Reported
421
Reproduced
421
exact
n_observations
Reported
over 600,000
Reproduced
605718
exact
sites_NO3_or_NOx
Reported
397
Reproduced
395
within tolerance
sites_NH4
Reported
196
Reproduced
196
exact
sites_P
Reported
339
Reproduced
339
exact
yr_DSi
Reported
1964-2023
Reproduced
1964-2023
exact
yr_DIN
Reported
1964-2023
Reproduced
1964-2023
exact
yr_DIP
Reported
1969-2023
Reproduced
1969-2023
exact
yr_discharge
Reported
1963-2023
Reproduced
1963-2023
exact
n_networks
Reported
24
Reproduced
24
exact
yrs_of_record
Reported
4 to 55
Reproduced
4 to 55
exact
range_obs_record
Reported
1 to 178
Reproduced
1 to 178
exact
range_obs_per_yr
Reported
2.4 to 64
Reproduced
2.74 to 63.92
within tolerance
median_obs_per_yr
Reported
14.2
Reproduced
12.8
partial
lat_range
Reported
-77S to 70N
Reproduced
-77.62 to 70.23
exact
n_sites_watershed
Reported
400
Reproduced
421/407
within tolerance
n_climate_zones
Reported
11
Reproduced
10 named / 19 coded
did not match

Assessments & scoring basis

Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.

🤖 AI curator · claude (ai-curator room) · v1.0 L1 89/100

An automated assessment. It can flag an open question for review but can never, on its own, record a discrepancy verdict (C5) against a paper.

🟢1. Data identity
🟢2. Endpoint comparability
🟡3. Location of the main deviation
🟢4. Cause of the deviation
🟡5. Derivability / plausibility
🟢6. Severity of the deviation
🟢7. Core claim
🟡8. Severity of the miss (overall human judgment)
Scoring basis — itemised

Every item that counted toward this verdict, and the exact part of the reproduction that produced it.

Supporting (toward a concern)
Content-critical question only partially held
+2 pts
From: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q8 · Severity of the miss (overall human judgment) 🟡
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +3

This Sci Data descriptor reproduces essentially 1:1: all headline counts (421 rivers, 605,718 observations, per-solute site counts, 24 networks, every temporal and latitude range) recompute exactly from the public USGS ScienceBase v2.0 deposit, with no fabrication signal. Two minor deviations remain on our/data side — median obs/yr 14.2→12.8 (~10%) and climate-zone count 11 vs 10 names/19 codes — most plausibly a v1-vs-v2 version/aggregation difference, not an authors' defect. A notable provenance issue is that the manifest's Zenodo DOI pointed to an unrelated land-cover product and had to be corrected. Overall a solid reproduction with small, explainable discrepancies → yellow.

🤝
Reproduced automatically — and fairly

Automated reproduction checks whether a published result can be regenerated from the paper’s described methods and shared data. When something does not reproduce, that is not a claim of error or misconduct — most often it reflects under-described methods, software or environment differences, or gaps in data access, and some of the pre-print papers in the queue may carry issues their authors had no part in. The goal is shared awareness that rigorous, fully-described methods help everyone — never a judgement of any author.

Are you an author? We would genuinely like to hear from you — to clarify the record, add data or code, re-run the pipeline after an accession update, and publish your response right next to the assessment. Everything here is open and auditable.

🚩 Report an error in this record

Spotted something wrong — a verdict you’d contest, a data or value error, or a private detail that slipped through? Tell us, with a short justification. Authors and readers are equally welcome to write in; we review every report.

Prefer email, or the form below not working? Contact us at support@doesitreproduce.com.

Reproduction footprint

claude-opus-4-8

Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.

100.3 k
tokens (I/O) · 5.9 M incl. cache
15 min
runtime · 0.01 CPU-h
1.9 GB
peak RAM
1
HPC jobs
hummel
machine