Description: This dataset accompanies the study "Locked in heat: present-day thermal constraints on Summer Olympic host cities and the limits of operational adaptation" (Defrance & Gadais). It provides Wet-Bulb Globe Temperature (WBGT) indicators characterising summer heat-stress constraints on Summer Olympic competition, derived from the ERA5-Land hourly reanalysis. Source data. All indicators are computed from the ERA5-Land hourly reanalysis (Muñoz-Sabater et al., 2021), 0.1° × 0.1° (~9 km) spatial resolution, distributed by the Copernicus Climate Change Service (C3S). Source variables: 2 m air temperature ( t2m ), 2 m dewpoint temperature ( d2m ), 10 m zonal and meridional wind components ( u10 , v10 ), and surface solar radiation downwards ( ssrd , de-accumulated to instantaneous hourly mean flux). Temporal coverage: July and August only, 2006-2025 (20 years), the months encompassing the modern Summer Olympic Games. WBGT computation. WBGT is reconstructed using the Stull (2011) psychrometric approximation for the natural wet-bulb temperature and the Hajizadeh et al. (2017) empirical regression for the black-globe temperature. Two formulations are provided: outdoor (sun-exposed), WBGT = 0.7 Tw + 0.2 Tg + 0.1 Ta (Yaglou & Minard 1957 weighting), with full solar load; and indoor/shaded, following the ISO 7243 shade formulation with incoming shortwave radiation set to zero. Contents. Global gridded indicators ( global_days_gt28_2006-2025.nc , NetCDF): mean number of July-August days per season with at least one hour exceeding WBGT thresholds, on the full ERA5-Land land grid. Variables: days_gt28_outdoor , days_gt28_indoor , days_gt32_outdoor , days_gt32_indoor . Thresholds: 28 °C (high heat-stress risk) and 32 °C (extreme risk), per international sport-federation guidelines. City-scale hourly series ( hourly_<City>.parquet , 8 cities): hourly time series in mean local solar time, July-August 2006-2025, with WBGT (outdoor and indoor) and its components ( Ta , RH , Tw , Tg_sun , ssrd ). City-scale derived indicators (CSV): city_summary.csv (per-city constrained-day counts at both thresholds and formulations, peak-hour WBGT decomposition, 2006-2025 linear trends); diurnal_<City>.csv (percentage of hours exceeding each threshold by local hour); yearly_<City>.csv (annual constrained-day counts). Analysis code is archived together with the data and at [GitHub URL]. Study cities (nearest valid ERA5-Land land cell): Doha (QA), Ahmedabad (IN), Tokyo (JP), Los Angeles (US), Paris (FR), Rio de Janeiro (BR), Brisbane (AU), Cape Town (ZA). Scope. Present, observed climate only; no future projections. WBGT is a first-order indicator at the regional near-surface scale and does not resolve intra-urban or venue microclimates.
Sea surface temperature retrievals produced by NOAA/NESDIS/Office of Satellite and Product Operations (OSPO) office from Visible and Infrared Imager/Radiometer Suite (VIIRS) sensor cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sea_surface_temperature (sea surface sub-skin temperature, degree_C) graphics (graphics overlay planes)
Sea surface temperature retrievals produced by NOAA/NESDIS/Office of Satellite and Product Operations (OSPO) office from Visible and Infrared Imager/Radiometer Suite (VIIRS) sensor cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sea_surface_temperature (sea surface sub-skin temperature, kelvin) graphics (graphics overlay planes)
This dataset quantifies the uncertainty in mapping late-successional and old-growth (LSOG) forest across the approximately 4.2 million hectares of Maine's unorganized townships, and tests whether LSOG is rapidly disappearing. Three to four independent, credible mapping methods are compared on a common 100 m grid: (M1) a reproduction of the Hagan et al. (2026) airborne-LiDAR canopy random forest, rebuilt from their public Zenodo deposit; (M2) a logistic model of the FIA field-structure LSOG class on Potapov (GEDI-calibrated) canopy height; (M3) a direct canopy-height threshold; and (M4) the FIA structural class imputed to every pixel via USFS TreeMap (2016, 2020, 2022). Version 1.2.0 additions. This version adds the materials behind the formal Ecosphere Comment on Hagan et al. (2026): (a) a cross-validated accuracy assessment (AUC) of each mapping approach on the original authors' own training plots, showing that high training accuracy does not transfer to agreement among independent maps; (b) an FIA design-based estimate of older forest with sampling-error confidence intervals, the unbiased ground reference the original analysis lacked, putting older forest at about 3.9 percent (3.3 to 4.6) and rising, including on private commercial timberland; (c) a threshold-sensitivity sweep and a 20-seed reproduction ensemble; (d) an ownership-resolved breakdown (private commercial versus public); (e) a hex-scale (8 km) summary of cross-method disagreement; and (f) the Comment manuscript and Supporting Information. Headline findings. Credible methods disagree by roughly 2.8 times on how much LSOG exists and on the location of most LSOG hectares, while agreeing closely on the rare, well-defined old-growth core. Protecting the top 5 to 20 percent of hectares by one map versus another overlaps on only 16 to 30 percent of the ground, so single-map patch-level prioritization for large expenditures is fragile. The design-based FIA estimate and TreeMap imputation both show older forest stable to increasing rather than rapidly declining; the apparent loss reported elsewhere is a gross harvest flux, not a net stock decline. Contents. Derived 100 m GeoTIFFs (reproduced Hagan class, v5.1-GEDI probability, TreeMap class, a per-cell method-consensus layer), summary tables (area by method, pairwise agreement, concordance, prioritization fragility, AUC by approach, design-based older-forest trend with CIs, ownership breakdown, and FIA validation), the analysis R scripts, quick-look figures, the Ecosphere Comment manuscript and Supporting Information, and a full methods-and-findings report (PDF). Privacy. No FIA plot coordinates are included; all products are derived rasters or aggregate summary tables. Caveats: the robust temporal signal is direction rather than precise rate; FIA stand age is modeled, so a structural large-tree domain is reported alongside the age domain; cross-validated intervals are best read as lower bounds because plots are spatially dispersed but not independent. See the README and report for full methods, provenance, and limitations. Version 1.12.0 additions. The cross-map comparison is refined to independent remote-sensing operationalizations only. (a) A three-map remote-sensing ensemble over Maine on a common 100 m grid: reproduced Hagan airborne-LiDAR (any-LSOG 21.9 percent), an FIA-structure class on Potapov GEDI-calibrated spaceborne canopy height (14.0 percent), and the ORNL/Bruening national old-growth stratum (36.1 percent); the three span a 2.6-fold range and agree on only 2.7 percent of flagged hectares, with the ORNL stratum spatially uncorrelated with the structure maps. The USFS TreeMap imputation is reclassified as a second FIA-anchored accounting, reported with the design-based estimate rather than as an independent map. (b) A design-based estimate of LSOG itself: integrated any-LSOG 14.1 percent (12.9 to 15.3) and strict four-axis true LSOG 3.1 percent (2.5 to 3.7) of Maine forestland, the airborne map exceeding even the inclusive ground estimate. (c) A balanced-model LSOG probability surface at 100 m, with a binary class calibrated to the design-based area to bound over-prediction. (d) A multi-objective support vector regression pilot tracing the Pareto front of total versus systematic (attenuation) error. Derived rasters, the three-map agreement layer, the ORNL stratum reprojected to the study grid, tables, R scripts, the updated Comment, and the companion manuscript are included. No FIA plot coordinates are included. Version 1.13.0. Final consolidated release. Adds: a rare-class remedy menu for the reproduced random forest (default vs class weighting vs balanced sub-sampling vs voting-threshold; old-growth detection 0.24 to 0.82, mapped old-growth area 1.0 to 2.3 percent); the definitive five-model LSOG probability map for Maine with across-model uncertainty and reference reserves (MNAP/TNC network, Baxter, Big Reed) over real state and county boundaries; design-based 95 percent confidence intervals for forest-type and ecoregion representation and for disturbance shares; a full robustness/stress-test matrix; and the copy-edited, sole-authored Comment, companion manuscript, and Maine Forest Products Council technical report. Authorship updated to Aaron R. Weiskittel.
Daily Merge seasurfacetemperature from NOAA-20 Visible and Infrared Imager/Radiometer Suite (VIIRS) cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sea_surface_temperature (sea surface sub-skin temperature, degree_C) graphics (graphics overlay planes)
Daily Merge seasurfacetemperature from NOAA-20 Visible and Infrared Imager/Radiometer Suite (VIIRS) cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sea_surface_temperature (sea surface sub-skin temperature, kelvin) graphics (graphics overlay planes)
Sea surface temperature retrievals produced by NOAA/NESDIS/Office of Satellite and Product Operations (OSPO) office from Visible and Infrared Imager/Radiometer Suite (VIIRS) sensor cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][rows][cols]): l2p_flags swath_latitude (Latitude, degrees_north) swath_longitude (Longitude, degrees_east) sea_surface_temperature (sea surface subskin temperature, degree_C) sses_bias (SSES bias estimate, degree_C) sses_standard_deviation (degree_C) graphics (graphics overlay planes)
This archive contains a global compilation of marine net primary productivity (NPP) measurements derived from bottle incubation experiments using radioactive or stable carbon tracers ( 14 C or 13 C). All NPP estimates are reported as depth-integrated values within the euphotic zone. The database includes sampling date, geographic coordinates, integration depth, depth-integrated NPP, measurement method, data source, and reference information. The dataset was compiled to support the evaluation and development of satellite-based and data-driven marine productivity models and to improve observational constraints on the global distribution of ocean primary productivity. Chen, X., Huang, Y*., Liu, H., Cassar, N., Liu, X., Wang, W., Chai, F., Kang, J., & Huang, B. (2026). Reduced estimate of global marine primary productivity and hemispheric redistribution over the satellite era revealed by an expanded global observational database. Submitted to Communications Earth & Environment .
U.S. West Coast Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua High Resolution Sea Surface Temperature (SST) Climatology Fields (July 2002 - March 2014). This suite of Chlorophyll-a (CHLA) and SST climatology and anomaly data products are derived from daily, 0.0125 degree x 0.0125 degree, MODIS Aqua CHLA and SST fields that cover the California Current System (22N - 51N, 155W - 105W) for the 11-year period July 2002 through June 2013. These daily fields, obtained from the NOAA CoastWatch West Coast Regional Node website, were processed using a successive 3x3, 5x5 and 7x7 grid cell hybrid median filtering technique. This technique was found to effectively reduce noise in the daily fields while maintaining features and detail in important regions such as capes and headlands. The resulting median filtered daily fields were then linearly interpolated to a 0.025 degree x 0.025 degree grid and averaged to create 132 monthly mean fields. The seasonal cycles at each 0.025 degree x 0.025 degree grid cell were obtained by fitting each multiyear time series of monthly means to a nine-parameter regression model consisting of a constant plus four harmonics (frequencies of N/(1-year), N-1,4; Risien and Chelton 2008, Journal of Physical Oceanography (JPO)). Even with the median filtering and the temporal averaging of the daily fields, the highly inhomogeneous nature of the MODIS fields still resulted in regression coefficients that were excessively noisy. We therefore applied the same successive 3x3, 5x5 and 7x7 hybrid median filtering technique, described above, to the regression coefficients before finally spatially smoothing the coefficients using a loess smoother (Schlax et al. 2001, JTECH) with filter cutoff wavelengths of 0.25 degree latitude by 0.25 degree longitude. The seasonal cycles were then calculated from the filtered regression coefficients for each 0.025 degree x 0.025 degree grid cell using the mean and all four harmonics. It is important to note that THIS SUITE OF DATA PRODUCTS IS HIGHLY EXPERIMENTAL and is strictly intended for scientific evaluation by experienced marine scientists. cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sst (Sea Surface Temperature, degree_C)
Data archive supporting the manuscript " Multilayer canopy model outperforms big-leaf model for evapotranspiration predictions under high water and heat stress conditions " (Authors: Raghav, Liu, Kumar, Bisht) This archive provides, for each of 28 eddy-covariance sites, the hourly time series of modeled and observed evapotranspiration (ET) together with the meteorological drivers used to force the models. ET is reported as latent heat flux (LE), in W m⁻² , the native model and tower-measurement unit (to convert to a water flux, divide by the latent heat of vaporization λ ≈ 2.45 × 10⁶ J kg⁻¹: 1 W m⁻² ≈ 0.00147 mm h⁻¹). The two model configurations are run from the same CLM-ml code and differ only in the number of within-canopy layers; in this archive both configurations use identical below-ground (root) and soil parameters , so that differences reflect the above-ground canopy representation alone. Contents File Description <SITE>_ET_hourly.csv` Per-site hourly time series (28 files; columns below). all_sites_ET_hourly.csv All 28 sites concatenated (same columns, plus `Site`). sites_metadata.csv Site list with latitude, longitude, number of hours, and date range. model_forcing_netcdf/<SITE>_forcing.nc Complete model forcing for each site (all driver variables and the gap-filled/closure-corrected flux products; see below) Sites (28) CA-Cbo, CA-Gro, CA-TP3, CA-TPD, CH-Lae, CZ-Lnz, CZ-RAJ, CZ-Stn, DE-Hai, FR-Bil, FR-Hes, IT-Cp2, IT-SR2, US-Bar, US-Me2, US-Me6, US-NC1, US-NR1, US-Oho, US-UMB, US-UMd, US-xAB, US-xBR, US-xDL, US-xHA, US-xJE, US-xTA, US-xTR. Coordinates and record lengths are in sites_metadata.csv . Columns in <SITE>_ET_hourly.csv Column Definition Units TIMESTAMP Time at the start of the hour, as provided in the model forcing (site local-standard-time convention) YYYY-MM-DD HH:MM:SS Site Site identifier - ET_obs_LE_gapfilled_Wm2 Observed latent heat flux, gap-filled by the marginal-distribution-sampling (MDS) method (FLUXNET/ONEFlux `LE_F_MDS`). W m⁻² ET_obs_LE_corrected_Wm2 Observed latent heat flux after energy-balance-closure correction (Bowen-ratio-preserving). This is the target used to evaluate the models. W m⁻² ET_MLCAN_Wm2 Modeled latent heat flux from the multilayer canopy (MLCAN) configuration. W m⁻² ET_1L_Wm2 Modeled latent heat flux from the single-layer / big-leaf (1L) configuration. W m⁻² H_obs_gapfilled_Wm2 Observed sensible heat flux, MDS gap-filled (`H_F_MDS`). W m⁻² H_obs_corrected_Wm2 Observed sensible heat flux after energy-balance-closure correction. W m⁻² SW_IN_Wm2 Incoming shortwave radiation (model driver, `FSDS`). W m⁻² TA_degC Air temperature (model driver, `TBOT`, converted from K). °C VPD_kPa Vapor pressure deficit, computed from observed relative humidity and air temperature. kPa SWC_m3m3 Volumetric soil water content (same across all soil layers). m³ m⁻³ LAI_m2m2 Effective leaf area index used to drive the models (`ELAI`). m² m⁻² Note: Missing values are written as empty fields. The fraction of finite, energy-balance-corrected observed ET per site is given in `sites_metadata.csv` (`ET_obs_corrected_pct_finite`). Model forcing NetCDF files (`model_forcing_netcdf/`) Each `<SITE>_forcing.nc` contains the complete set of driver variables used to run both configurations and the full observed-flux products, at the same temporal resolution. Key variables (units as stored): - Meteorology: `TBOT` (K), `RH` (%), `WIND` (m s⁻¹), `FSDS` (incoming shortwave, W m⁻²), `FLDS` (incoming longwave, W m⁻²), `PSRF` (surface pressure, Pa), `PRECTmms` (precipitation, mm s⁻¹), `CO2MF` (CO₂ mole fraction), `ZBOT` (reference height, m). - Vegetation / soil: `ELAI`, `ESAI` (effective leaf/stem area index), `SWC` (volumetric soil water), `H2OSOI`, `TSOI` (soil-profile moisture and temperature). - Observed fluxes: `LE_F_MDS`, `H_F_MDS` (MDS gap-filled), `LE_c`, `H_c` (energy-balance corrected), `GPP_DT`, `GPP_NT` (daytime/nighttime partitioned gross primary productivity). - Temporal coverage: each site's record covers 1 July - 31 December of each year (the study's analysis window). Methods (brief) Eddy-covariance processing . Half-hourly fluxes were computed with EddyPro and quality-controlled; latent and sensible heat were gap-filled using the MDS algorithm and then corrected for the surface-energy-balance closure gap (Bowen-ratio-preserving). Models. CLM-ml (Bonan et al., 2021; https://doi.org/10.1016/j.agrformet.2021.108435) was run in two configurations viz. multilayer (MLCAN) and single-layer (1L) that share identical code, leaf-level formulations, and soil/root parameters and differ only in the number of within-canopy layers. Each configuration was independently calibrated to the energy-balance-corrected observed ET. Provenance, license, and citation The **original** half-hourly eddy-covariance observations for each site are distributed by AmeriFlux, the ICOS Drought-2018 and Warm-Winter-2020 collections, and NEON; the per-site dataset DOIs are listed in the Supplementary Information file of the associated manuscript. The non-gap-filled (raw) observed LE can be obtained from those original datasets. The modeled ET, the processed (gap-filled and corrected) observed ET, and the assembled forcing in this archive are the data generated by this study . - License: Creative Commons Attribution 4.0 (CC-BY-4.0).
This dataset provides a curated and ready-to-use collection of gridded climate indicator maps for the ADRION domain, derived from the Copernicus Interactive Climate Atlas/C3S Atlas. It includes reanalysis-based climatological indicators for the 1991-2020 reference period and climate-projection change anomalies for selected future periods relative to the 1981-2010 baseline. Data are provided as GeoTIFF (*.tif) files, with each file representing one climate indicator map for a specific combination of spatial domain, input dataset, temporal aggregation, climate scenario, and, where applicable, future time horizon. The files are organised to support local climate hazard and risk evaluation, climate-service applications, and adaptation-related assessments over the ADRION area. This dataset does not replace the original Copernicus source data, but offers a harmonised subset extracted and structured for easier use within the ADRION context. Users should acknowledge the Copernicus Climate Change Service (C3S) and the Copernicus Interactive Climate Atlas / C3S Atlas when using these files.
U.S. West Coast Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua High Resolution Sea Surface Temperature (SST) Climatology Fields (July 2002 - March 2014). This suite of Chlorophyll-a (CHLA) and SST climatology and anomaly data products are derived from daily, 0.0125 degree x 0.0125 degree, MODIS Aqua CHLA and SST fields that cover the California Current System (22N - 51N, 155W - 105W) for the 11-year period July 2002 through June 2013. These daily fields, obtained from the NOAA CoastWatch West Coast Regional Node website, were processed using a successive 3x3, 5x5 and 7x7 grid cell hybrid median filtering technique. This technique was found to effectively reduce noise in the daily fields while maintaining features and detail in important regions such as capes and headlands. The resulting median filtered daily fields were then linearly interpolated to a 0.025 degree x 0.025 degree grid and averaged to create 132 monthly mean fields. The seasonal cycles at each 0.025 degree x 0.025 degree grid cell were obtained by fitting each multiyear time series of monthly means to a nine-parameter regression model consisting of a constant plus four harmonics (frequencies of N/(1-year), N-1,4; Risien and Chelton 2008, Journal of Physical Oceanography (JPO)). Even with the median filtering and the temporal averaging of the daily fields, the highly inhomogeneous nature of the MODIS fields still resulted in regression coefficients that were excessively noisy. We therefore applied the same successive 3x3, 5x5 and 7x7 hybrid median filtering technique, described above, to the regression coefficients before finally spatially smoothing the coefficients using a loess smoother (Schlax et al. 2001, JTECH) with filter cutoff wavelengths of 0.25 degree latitude by 0.25 degree longitude. The seasonal cycles were then calculated from the filtered regression coefficients for each 0.025 degree x 0.025 degree grid cell using the mean and all four harmonics. It is important to note that THIS SUITE OF DATA PRODUCTS IS HIGHLY EXPERIMENTAL and is strictly intended for scientific evaluation by experienced marine scientists. cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sst (Sea Surface Temperature, degree_C)
Jet Propulsion Laboratory data from a local source. cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][level][latitude][longitude]): sst (Analysed SST with land and ice masks applied, degree_C)
Sea surface temperature retrievals produced by NOAA/NESDIS/OSPO office cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][latitude][longitude]): sea_surface_temperature (sea surface sub-skin temperature, degree_C) sses_bias (SSES bias estimate, degree_C) sses_standard_deviation (degree_C) l2p_flags sst_count (Count of input L3U SST pixels) quality_level (quality level of SST pixel) sst_source (Source of major (highest weight) contribution to sst) satellite_zenith_angle (degrees) dt_analysis (deviation from SST reference, kelvin) wind_speed (m s-1) sst_dtime (time difference from reference time, seconds) sst_gradient_magnitude (SST gradient magnitude value, kelvin/km) sst_front_position (Binary SST front position indicator)
Agreement Between Large Language Models and Humans in Research Proposal Review - Data and Code This repository contains the data and code required to reproduce the analyses, statistical results, and figures presented in the associated manuscript. Files are organized by function and described below. All research proposals are anonymized and labeled with non-identifying identifiers (A, B, C, ...). Reviewer identities were never provided to the authors. To prevent inadvertent disclosure, all free-text review content from both human reviewers and large language models (LLMs) has been removed; only the numerical evaluation data required to reproduce the reported analyses are included. Data files Human_raw_scores.csv Individual numerical scores assigned by human reviewers, one row per reviewer × proposal × criterion. Used to compute panel-level summary statistics and the inter-reviewer reliability metrics reported in the manuscript. Human.csv Proposal-level human reference scores (one row per proposal) used as the human benchmark against which LLM scores and rankings are compared. LLM_data_combined_clean_filtered.csv All numerical scores generated by the evaluated LLMs. Preprocessed to remove evaluations in which a model failed to return one or more required numerical scores (see Data provenance and known limitations for counts). review_criteria.txt The evaluation criteria and rating scales. See the note under Data provenance regarding the 2023 vs. 2024 criterion naming. Data dictionary Human_raw_scores.csv Column Description Applicant Anonymized proposal identifier (A, B, C, ...). Cycle Review cycle the proposal belongs to (2023 or 2024). Label Scoring criterion: Intellectual merit , Potential for impact , Collaborative Potential , or Overall ranking . Reviewer_Seq Reviewer index within a proposal (1, 2, 3, ...). Identities are unknown; this index only links a single reviewer's ratings across criteria for the same proposal, in source-file order. It is not consistent across proposals (reviewer 1 for proposal A is not reviewer 1 for proposal B). Rating Numerical score. Criterion ratings use a 1-5 scale; Overall ranking uses a 1-3 scale (3 = fund, 2 = fund with modifications, 1 = do not fund). Human.csv Column Description Name Anonymized proposal identifier (matches Applicant above). Cycle Review cycle (2023 or 2024). IM , Impact , Collab , Overall Panel-mean scores for the four criteria. Score Weighted composite panel score, computed as 0.4·IM + 0.3·Impact + 0.3·Collab , matching the weighting applied to the LLM composite scores. LLM_data_combined_clean_filtered.csv Column Description Name Anonymized proposal identifier (matches Human.csv ). Type Input given to the model: Abstract or Full_Proposal . Model Model identifier. For models with controllable reasoning depth, the tier is appended as a suffix ( _low , _medium , _high ). The portion before the first underscore is the root model. Prompt Prompting strategy ( OneShot or CoT ). Seed Requested random seed. For models that did not support seed specification at the time of execution (the reasoning models listed in the Methods), the API ignored this value and it functions only as a replicate index; output is not reproducible from it for those models. Temp Sampling temperature (0.1, 0.5, 0.9). IM , Impact , Collab Criterion ratings (1-5 scale). Overall Overall recommendation (1-3 scale). See limitation note on out-of-scale values. Score Weighted composite, 0.4·IM + 0.3·Impact + 0.3·Collab . Category Reasoning/architecture category of the model. Year Evaluation wave in which the run was performed (proposals were re-evaluated as new model generations were released); this is not the proposal's submission cycle. Use Cycle in the human files for submission cycle. Data provenance and known limitations We document the following so that users can interpret the data accurately. Two review cycles, combined. The 28 proposals come from two internal seed grant cycles: 15 from 2023 and 13 from 2024 ( Cycle column). For 2024 proposals, proposal-level means in Human.csv are the official institute panel means; for 2023 proposals they are computed from the individual ratings in Human_raw_scores.csv . Reviewers per proposal. Panels ranged from 3 to 6 reviewers. Reviewer identities were never provided; the human inter-reviewer reliability is therefore estimated with a one-way random-effects model (ICC(1,1)), which is the appropriate model when each proposal is rated by a different, unidentified set of reviewers. Six 2024 reviews not present at the individual level. For six 2024 proposals (G, R, T, W, X, Z), one reviewer's scores were submitted without written comments and are not included in Human_raw_scores.csv . For these proposals, Human.csv carries the official institute panel means, so the panel mean in Human.csv and the mean recomputed from Human_raw_scores.csv differ slightly. The reproducibility check in 06_Human_data.ipynb confirms exact agreement for all proposals with complete individual records and reports the expected small differences for these six. Out-of-scale LLM Overall ratings. A small number of responses (≈1.2% of reviews, almost entirely from gpt-3.5-turbo) rated the overall recommendation on a 1-5 scale rather than the requested 1-3 scale, in a format the parser could not distinguish. The analysis code masks values outside the valid range before computing any Overall -based result; the composite Score does not use Overall and is unaffected. Excluded LLM evaluations. Evaluations in which a model failed to return one or more required numerical scores were removed prior to analysis. The file provided here is the post-exclusion (analyzed) dataset. Code notebooks Two groups of notebooks are provided: (i) the LLM evaluation pipeline and (ii) statistical analysis and figure generation. LLM evaluation pipeline (Notebooks 1-4) Documentation of the methodology used to generate the LLM evaluations. These use synthetic examples and contain no confidential data, API credentials, or real proposal text. 01_pipeline_overview.ipynb - architecture, configuration, criteria, output format, evaluation matrix. 02_prompting_strategies.ipynb - one-shot and chain-of-thought prompting; example selection; text vs. vision input. 03_response_parsing.ipynb - regex extraction of ratings and comments; error handling; decimal ratings. 04_example_evaluation.ipynb - end-to-end workflow on synthetic data. Statistical analysis and figures (Notebooks 5-6) 05_Data_Processing.ipynb - ANOVA and effect sizes; empirical absolute score differences; ICC(2,1) and Spearman correlations vs. the human panel; Monte Carlo comparisons; scatter, slope, and bias figures. 06_Human_data.ipynb - ICC(1,1)/ICC(1,k) for human reviewers with bootstrap confidence intervals; Monte Carlo single-reviewer vs. leave-one-out panel rank agreement; consistency check of Human.csv against Human_raw_scores.csv . Reproducibility Running 05_Data_Processing.ipynb and 06_Human_data.ipynb against the included data reproduces the statistical results and figures in the manuscript. Notebooks 1-4 document the evaluation pipeline using synthetic examples. Requirements pandas numpy scipy statsmodels scikit-learn matplotlib The LLM evaluation pipeline notebooks (1-4) additionally use the packages listed in requirements.txt . Citation If you use this data or code, please cite the associated manuscript and this archive: Gorski, C., Leo, N., Gayah V. Agreement Between Large Language Models and Humans in Research Proposal Review - Data and Scripts. 2026. Zenodo. https://doi.org/10.5281/zenodo.18187034 License Data are released under CC BY 4.0; code is released under the MIT License.
Jet Propulsion Laboratory data from a local source. cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][level][latitude][longitude]): sst (Analysed SST with land and ice masks applied, degree_C)
General Information This repository contains the comprehensive collection of empirical lacustrine time series, sedimentary core records, standardized data matrices, and execution scripts required to fully replicate the figures, network topologies, and statistical null models presented in the associated manuscript. I. Software and Environment Requirements Python (v3.8+): Required libraries include numpy , pandas , matplotlib , scipy , statsmodels , and pymannkendall . R (v4.2+): Required libraries include wsyn , igraph , zoo , parallel , pbapply , ggplot2 , and cowplot . Geospatial Platform: Esri ArcGIS Desktop (v10.8) or ArcGIS Pro (for cartographic rendering and vector layer manipulation). Computation Note: To mitigate boundary artifacts and edge effects inherent in chronological sliding-window operations, the final 12-24 data points of the generated time series are systematically excluded from the final trend evaluation. II. File Inventory and Component Descriptions 1. Empirical and Core Datasets ( .csv ) Global_Lake_Chlorophyll_a_Time_Series.csv Description: Long-term, multi-decadal monthly gridded chlorophyll- a concentration time series across global limnological cohorts, featuring unique lake identification codes as columns and sequential temporal intervals as rows. Global_Lake_Water_Color_Time_Series.csv Description: Normalized global lake water color time series quantified via the Forel-Ule Index (FUI) framework, structured identically to the chlorophyll dataset for multi-proxy comparison. Global_Lake_Sediment_Pigments.csv Description: Stratigraphic sedimentary core pigment records providing long-term retrospective evidence of historical limnological synchronization and baseline shifts. Figures_data.CSV Description: Consolidated and curated data matrices containing the exact values, coordinates, and regional groupings utilized to plot the core text figures. 2. Statistical Analysis and Mathematical Scripts ( .py & .R ) Pairwise correlation -based synchrony.py Description: Computes macro-scale spatial synchrony trends across lacustrine nodes using the standard pairwise correlation matrix stream following frequency-domain decomposition. Loreau φ Metric for lake synchrony.py Description: Execution script utilizing the classic Loreau-de Mazancourt φ metric to calculate multi-lake population-level synchrony across global and latitudinal cohorts. Sliding window sensitivity.py Description: Explores scale dependency and temporal robustness by executing the analytical data stream across varying sliding temporal windows (e.g., 2, 5, 8, 10, 15 time steps). Network and modularity analysis.R Description: Implements the wsyn continuous signed-power soft-thresholding paradigm and leverages igraph to partition similarity networks into topological communities, calculating decadal modularity . Permutation-based significance of synchrony trends.py Description: Performs Mann-Kendall trend tests on sliding-window synchrony series and runs empirical hypothesis testing to extract true directional shifts. Permutation-based significance of synchrony trends-Null_Model_Generator.py Description: Harnesses multi-core parallel processing to shuffle network weights 1,000 times, constructing empirical null distributions to validate the significance of observed network community dissolution. 3. Geospatial Visualization & Documentation ( .rar & .txt ) Figure 1.rar Description: Compressed archive containing all raw geospatial project databases, vector layer shapefiles ( .shp ), metadata tables, and cartographic layout definitions ( .mxd ) used to generate the global geographic distribution map of sample lakes (Figure 1). Compiled within Esri ArcGIS 10.8. Data_Sources_and_References.txt Description: A comprehensive standalone text file documenting the complete bibliographic literature sources, historical baselines, and corresponding DOI attributions compiled within the empirical datasets. III. Execution and Replication Workflow Data Cleaning & Detrending: Feed raw time series through Python/R scripts to execute STL harmonic regression models and filter low-frequency background signals. Synchrony Calculations: Execute the pairwise and Loreau metric scripts to plot continuous synchrony variations over time. Network Configurations: Run the R network script to output the high-resolution PCA community plots and decadal modularity comparisons. Significance Evaluation: Launch the null model generator to confirm that network configuration shifts significantly exceed random stochastic expectations ( P < 0.001 ). Spatial Reconstruction: Extract Figure 1.rar into your local GIS directory to access, modify, or re-export the multi-layered baseline global sampling maps.
Daily Merge seasurfacetemperature from S-National Polar-orbiting Partnership (NPP) Visible and Infrared Imager/Radiometer Suite (VIIRS) cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sea_surface_temperature (sea surface sub-skin temperature, kelvin) graphics (graphics overlay planes)
Daily Merge seasurfacetemperature from S-National Polar-orbiting Partnership (NPP) Visible and Infrared Imager/Radiometer Suite (VIIRS) cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): sea_surface_temperature (sea surface sub-skin temperature, degree_C) graphics (graphics overlay planes)
U.S. West Coast Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua High Resolution Chlorophyll-a (CHLA) Climatology Fields (July 2002 - March 2014). This suite of CHLA and Sea Surface Temperature (SST) climatology and anomaly data products are derived from daily, 0.0125 degree x 0.0125 degree, MODIS Aqua CHLA and SST fields that cover the California Current System (22N - 51N, 155W - 105W) for the 11-year period July 2002 through June 2013. These daily fields, obtained from the NOAA CoastWatch West Coast Regional Node website, were processed using a successive 3x3, 5x5 and 7x7 grid cell hybrid median filtering technique. This technique was found to effectively reduce noise in the daily fields while maintaining features and detail in important regions such as capes and headlands. The resulting median filtered daily fields were then linearly interpolated to a 0.025 degree x 0.025 degree grid and averaged to create 132 monthly mean fields. The seasonal cycles at each 0.025 degree x 0.025 degree grid cell were obtained by fitting each multiyear time series of monthly means to a nine-parameter regression model consisting of a constant plus four harmonics (frequencies of N/(1-year), N-1,4; Risien and Chelton 2008, Journal of Physical Oceanography (JPO)). Even with the median filtering and the temporal averaging of the daily fields, the highly inhomogeneous nature of the MODIS fields still resulted in regression coefficients that were excessively noisy. We therefore applied the same successive 3x3, 5x5 and 7x7 hybrid median filtering technique, described above, to the regression coefficients before finally spatially smoothing the coefficients using a loess smoother (Schlax et al. 2001, JTECH) with filter cutoff wavelengths of 0.25 degree latitude by 0.25 degree longitude. The seasonal cycles were then calculated from the filtered regression coefficients for each 0.025 degree x 0.025 degree grid cell using the mean and all four harmonics. It is important to note that THIS SUITE OF DATA PRODUCTS IS HIGHLY EXPERIMENTAL and is strictly intended for scientific evaluation by experienced marine scientists. cdm_data_type = Grid VARIABLES (all of which use the dimensions [time][altitude][latitude][longitude]): chla (Concentration Of Chlorophyll In Sea Water, mg m-3)
Scholarly works
unavailable this run
Patents
0 matches
No patents in the bounded preview corpus for this query.