Skip to content

LandClim

LandClim supplies primary pollen context through normalized site sequences and REVEALS grid cells. It supports landscape and vegetation interpretation around other evidence without becoming direct evidence for an archaeological feature or ancient-DNA sample.

Checked-In Evidence

The current governed summary records:

Surface Count Meaning
normalized site sequences 490 pollen-context points retained after family-specific normalization
sequences with numeric BP intervals 368 records eligible for bounded temporal comparison at site-sequence level
sequences with refused BP intervals 122 112 negative-BP and 10 partial source intervals retained without canonical numeric bounds
aggregate REVEALS grid cells 77 discovery summaries across the captured reconstruction windows
temporal REVEALS grid features 2,515 dataset-specific cell and modeled time-window combinations
distinct modeled windows 25 explicit windows from 0–100 BP through 11200–11700 BP
dataset citations / publication references 3 / 4 governed bibliography linked to normalized evidence

Counts describe the checked-in snapshot, not exhaustive coverage. The family contract identifies data/landclim/raw/ as captured material, data/landclim/normalized/ as normalized evidence, the cross-family evidence stage matrix as review, and regional pollen layers as publication.

Curation Lineage

Boundary Governing material Decision preserved
source capture raw/landclim_sources.json and the checked-in workbooks or archive which LandClim release material entered the repository
sequence normalization normalized/nordic_pollen_site_sequences.geojson stable site geometry, source identity, and site-level temporal fields
model normalization normalized/nordic_reveals_temporal_grid_cells.geojson one dataset, grid cell, and published time window remain a separately filterable areal model feature
bibliography normalization normalized/landclim_bibliography.json dataset citations, method papers, and feature-level reference keys remain connected
family summary normalized/landclim_summary.json denominators for sequence, interval, and grid-cell claims
source review review/spatiotemporal_review.json dataset-by-dataset temporal coverage, unresolved intervals, and bibliography linkage
cross-family review data/source_spatiotemporal_posture_registry.json whether the family can participate in spatial or temporal comparison

The split between sequence and model normalization is a scientific boundary, not a storage convenience. It prevents a modeled grid value from acquiring the identity or evidentiary meaning of a sampled pollen sequence.

Multi-Artifact Preparation Receipt

The captured LandClim family is assembled from unlike source artifacts: site metadata, REVEALS results, grid-cell quality, taxa-to-functional-type mappings, productivity and dispersal parameters, land-cover classifications, and archived model outputs. Preparation must preserve the join that brought a value into a sequence or cell rather than citing the family name alone.

Prepared claim Minimum retained lineage
site-sequence identity and geometry source artifact, source-native dataset or sequence key, site metadata row, and normalization rule
numeric sequence interval native bounds and basis, parsing rule, normalized BP values, and comparability posture
model-cell value grid-cell identity and geometry, model release, time window, taxon or functional-type mapping, and applicable quality evidence
cross-window summary member cell, exact contributing windows, aggregation method, and missing-window posture
flowchart LR
    Metadata["site and grid metadata"] --> Member["typed sequence or cell identity"]
    Results["REVEALS results"] --> Member
    Mappings["taxa, functional types, and parameters"] --> Value["interpretable model value"]
    Member --> Value
    Quality["grid quality evidence"] --> Value
    Value --> Product["qualified pollen-context feature"]

The receipt matters when a workbook or archive changes independently. A stable cell count cannot demonstrate unchanged model meaning if its taxon mapping, time-window definition, or quality evidence changed.

flowchart LR
    Capture["source workbooks and archive"] --> Sites["normalized site sequences"]
    Capture --> Grid["dataset × cell × time-window features"]
    Sources["dataset and method citations"] --> Grid
    Sites --> Time["site-sequence BP intervals"]
    Sites --> Review["spatiotemporal review"]
    Grid --> Review
    Sources --> Review
    Time --> Review
    Review --> Product["time-filterable pollen-context layers"]

What LandClim Supports

  • broad vegetation and palaeoenvironmental interpretation;
  • pollen context around lakes, samples, sites, and regions;
  • chronology-aware comparison where a normalized site-sequence interval is present;
  • comparison between local sequence evidence and modeled landscape context;
  • reproducible Nordic pollen layers whose records retain family identity.

Site sequences and grid cells have different meanings. A grid cell summarizes modeled landscape context; it is not another observation at the cell center.

Temporal Interpretation

LandClim has two explicit time-bearing units. Site-sequence intervals describe the captured temporal coverage of a pollen sequence. REVEALS intervals belong to one modeled estimate for one grid cell and published window. Neither is an event date, and neither automatically aligns with every sample inside the same interval.

The temporal grid keeps 2,515 dataset-cell-window features separate: 309 from the Marquer reconstruction, 331 from LandClim I, and 1,875 from LandClim II. Together they cover 77 distinct aggregate cells and retain 25 distinct numeric windows where supplied by the source datasets. Each feature carries numeric BP bounds and bibliography keys. Atlas time filtering therefore selects published estimates rather than treating a cell's full Holocene coverage as one continuous value. Source combinations without a published estimate remain absent; they are not converted to zero cover.

The ten sequences without numeric intervals remain spatial pollen context. They are not assigned synthetic dates to make the family appear uniformly time-resolved.

Read A LandClim Member

First identify which normalized unit is visible. A site sequence and a REVEALS grid cell can occupy the same map but require different interpretations.

Question Site sequence REVEALS grid cell
what is represented? a normalized pollen sequence at a source-linked site modelled vegetation context for a grid area
what does geometry mean? the sequence site's governed point the spatial support of the model cell, not an observation at its center
what does time mean? the sequence interval when numeric bounds are present the model's declared temporal context
what can be compared? pollen context within compatible place and time precision landscape-level model context under the grid contract
what must not be inferred? direct evidence for a nearby sample or event a measured pollen sample at the rendered coordinate
flowchart LR
    Feature["visible LandClim feature"] --> Kind{"sequence or grid?"}
    Kind -->|sequence| Site["site identity and BP posture"]
    Kind -->|grid| Model["cell identity and model semantics"]
    Site --> Claim["qualified pollen context"]
    Model --> Claim

A reusable claim retains the feature kind, source identity, normalized member identifier, spatial basis, temporal posture, and publication scope. “LandClim point” is not specific enough to preserve those distinctions.

Worked Sequence And Grid Trace

Aal Præstesø demonstrates the site-sequence path. Its stable normalized member ID is 897303:Aal Præstesø:55.637778:8.257222; the record points to PANGAEA dataset 897303, classifies the site in Denmark, and carries a 100–350 BP interval. Those fields support a bounded pollen-context claim for that sequence. They do not make the interval a date for every archaeological or biological record near the site.

The REVEALS cell 10.000000,55.000000,11.000000,56.000000 demonstrates the model path. The aggregate grid can show that the polygon has captured reconstructions, but temporal interpretation uses separate dataset-window features such as LandClim II 0–100 BP. That feature retains the published estimate, standard error, quality evidence, dataset DOI, and method references. Its midpoint is useful for navigation, but the polygon remains modeled areal context—not a pollen observation at the center and not a continuous value between published windows.

The retained LandClim II archive has 25 mean and 25 standard-error members, with 10,500 rows in each result family. Ingestion validates all 10,500 row pairs before spatial filtering. Exactly 150 pairs—six published grid IDs in all 25 windows—carry the provider's paired #N/A longitude and latitude and are accounted as nonspatial exclusions; their values are not converted to zero. Three finite mean tokens in two otherwise spatial rows are slightly negative (-0.0117533696932253 twice and -0.000677087305195939 percentage points). They remain exact source values, while standard errors must remain finite and nonnegative. The atlas treats estimates outside the display domain as unavailable instead of clipping them. These rules establish source fidelity and exclusion accounting, not ecological accuracy or raw-input reproduction.

Reproduction Evidence And Its Boundary

A separate, locally supplied LandClim II author-example bundle (SHA-256 464a05e3a62c0454dfbadb9fdbe786ccddc42c88e051abc0ddd56938621e219c) was replayed through the original R preprocessing steps 1–4. All 17 paired example sites (nine in grid 512 and eight in grid 562) reproduced their archived prepared files byte for byte. An independent count check reconciled 501,713 raw pollen counts = 202,053 excluded or outside the declared windows + 299,660 prepared counts. The exclusions remain explicit; they are not zero cover or missing observations. For those two example grids, the R and Python site and regional cover and standard-error results also agree on the declared input closure, including missing-value masks.

A separate method-interface run joined the first time bin (50) of author grid 512 to analytical LOVE on the committed Python engine c150668d0d878b5d6a5f5f06224c628a43436856. Five source sites contributed to the 31-taxon regional estimate. The same 1,000 archived site-bootstrap index rows produced 1,000 regional cover draws, which were supplied to one constructed positive-geometry 200–300 m local LOVE ring. All 31,000/31,000 resulting local draw cells were finite, and a deliberately truncated 999-draw input was refused. 8,312/31,000 local cells were negative (minimum -0.46953): the analytical equation leaves mathematical outputs unclipped, so finite output here must not be read as valid physical cover. The named landclimii-2016-gpm-neutral-analytical-v1 profile preserved the author's Filipendula fall speed of 0.006 m/s; the default profile still refuses that value. The governed interface result records the source identities, regional draw hash, local-cell summary and reproduction command without depending on an old run directory. An earlier source-R comparison of the same grid/bin's regional draw method matched 31,000 cells within 1.8874e-13 absolute difference on a separately frozen candidate. The local site, BROADOWN, is also in the regional cohort; the constructed local ring is a method control. These outputs are neither an independent held-out reconstruction nor a measured-vegetation accuracy result.

This establishes reproduction of the supplied examples. The full published study identifies 1,128 historical site records whose raw counts, selected chronologies, revisions, and reuse rights are not yet authenticated as one input closure here. Its complete raw-input run, article figure and table comparison, and accuracy against independently measured vegetation remain unverified. The published REVEALS estimates are model outputs, not measured vegetation reference labels.

A bounded recovery check illustrates this gap. The publication metadata lists Amsoldingersee as an ALPADABA record with filename code E#69, 23 pollen samples and an author chronology. The current Neotoma site has three pollen cores. One has 23 sample IDs but no chronology record; the other two have 24 and 52 sample IDs with radiocarbon chronologies. None of the three public core responses establishes an E#69 identity or the historical author age model. This one publication row remains excluded from raw-input reproduction; a sample-count match alone cannot supply its missing lineage.

The ESSD article says 12 plant functional types, while the PANGAEA deposit says 11. The captured parameter table assigns 31 taxa to 12 distinct codes with 11 distinct descriptions. The archived result headers have 13 functional-type positions, including populated ISTS, which has no matching parameter-table code. This count and mapping conflict remains unresolved; ISTS is preserved rather than silently merged or discarded.

flowchart LR
    Aal["Aal Præstesø sequence"] --> SiteInterval["100–350 BP site context"]
    Grid["10°E–11°E, 55°N–56°N cell"] --> Windows["separate modeled windows"]
    SiteInterval --> Compare["qualified comparison"]
    Windows --> Compare
    Compare -. does not merge .-> Observation["single observation identity"]

This pair exposes a crucial denominator rule: one normalized sequence member, one aggregate grid cell, one dataset-cell-window feature, and the source observations behind them are different units. A comparison must name which unit it counts.

Bibliography And Attribution

The governed bibliography names the three captured PANGAEA datasets and four publications that establish the reconstruction and method lineage. Temporal grid features link back to the applicable keys, so a map export does not lose the evidence needed to cite or interpret its values.

Dataset Captured temporal role Principal linked reading
PANGAEA 900966 309 Nordic grid-window estimates across 13 cells and 25 published windows Marquer et al. (2017) and Sugita (2007)
PANGAEA 897303 LandClim I grid estimates across five published windows Trondman et al. (2015) and Sugita (2007)
PANGAEA 937075 LandClim II estimates and standard errors across 25 published windows Githumbi et al. (2022), Trondman et al. (2015), and Sugita (2007)

The review packet keeps all three datasets visible separately. It also records the 54 site-sequence rows associated with dataset 900966, preventing its site intervals and 309 modeled grid-window estimates from being collapsed into one denominator.

Compare Pollen Families Without Flattening Their Units

A LandClim sequence can participate in a site-level comparison with a Neotoma member. A REVEALS cell cannot enter that comparison as though it were another site. It can only contribute a separate, areal model-context relation.

Comparison edge Admissible statement Required receipt
LandClim sequence to Neotoma site the two governed site-level members meet the declared spatial and, when admitted, temporal rules both member IDs, geometry basis, distance rule, both temporal postures, and lineage-independence result
LandClim sequence to REVEALS cell the sequence lies within or relates to a declared model cell sequence ID, cell ID, containment or distance rule, and compatible time-window rule
REVEALS cell to Neotoma site the site is interpreted against modeled landscape context cell ID, site ID, spatial relation, model window, and site temporal posture
any pollen member to aDNA or archaeology pollen supplies environmental context around direct evidence both identities, relation rule, time posture, and an explicit refusal of direct association
flowchart TD
    Left["LandClim member"] --> Kind{"member kind"}
    Kind -->|site sequence| SiteEdge["site-level comparison edge"]
    Kind -->|REVEALS cell| ModelEdge["model-context edge"]
    Other["Neotoma, archaeology, or aDNA member"] --> SiteEdge
    Other --> ModelEdge
    SiteEdge --> Receipt["typed comparison receipt"]
    ModelEdge --> Receipt
    Receipt --> Claim["qualified cross-family claim"]

The comparison receipt, rather than a shared map coordinate, establishes what was compared. It also prevents one relationship from silently changing type: a point-in-polygon result is not a sequence match, and interval overlap is not evidence that two records derive from independent observations.

Relationship To Other Families

Compared with LandClim contributes Other family contributes
Neotoma broad sequence and REVEALS-oriented landscape context site-centered palaeoecological comparison
SEAD or RAÄ environmental setting archaeology context
SVAR pollen setting around candidate basins lake and hydrographic identity
aDNA surrounding environmental context direct sample-owned biological evidence

Proximity between a LandClim record and another feature is a declared spatial relation. It does not establish shared chronology or causal association unless those dimensions are supported separately.

LandClim and Neotoma may also carry shared data lineage. Some captured Neotoma dataset notes attribute contributions to the LandClim project or the European Pollen Database. A matching or nearby member across the two normalized families is therefore not automatically independent corroboration. Independence review must compare dataset identity, contributor, site, and underlying sequence lineage before treating the two records as separate support.

Choose LandClim For The Question

Question Use Retain with the claim
What pollen sequence context exists near this place? normalized site sequences member identifier, distance rule, and sequence temporal posture
What modeled vegetation context covers this area? REVEALS grid cells cell identifier, model semantics, and declared time context
Can this record be compared with a dated sample? only sequence rows with compatible admitted intervals both intervals, units, precision, and overlap rule
Does nearby pollen prove association with an aDNA or archaeology feature? neither LandClim surface alone a separate association design and evidence would be required

Review A Refresh By Observation Unit

A LandClim refresh is not one undifferentiated layer diff. Review site sequences and REVEALS grid cells as separate populations. For sequences, compare source identity, site identity, coordinate, sequence extent, and time posture. For grid cells, compare cell identity, geometry, model meaning, and declared time slice. Only then assess changed publication membership and cross-family relations. A site-to-cell reassignment is a semantic change even when the rendered map looks unchanged.

Governing Surfaces

  • data/landclim/raw/landclim_sources.json records source capture context;
  • data/landclim/normalized/landclim_summary.json records family counts;
  • data/landclim/normalized/nordic_pollen_site_sequences.geojson governs normalized sequence points;
  • data/landclim/normalized/nordic_reveals_grid_cells.geojson governs aggregate grid discovery;
  • data/landclim/normalized/nordic_reveals_temporal_grid_cells.geojson governs separately filterable reconstruction windows;
  • data/landclim/normalized/landclim_bibliography.json governs dataset and publication citations;
  • data/landclim/review/spatiotemporal_review.json audits temporal coverage and bibliography linkage;
  • data/source_spatiotemporal_posture_registry.json records comparison posture.

Public renderings are derived from these surfaces and cannot strengthen their spatial or temporal claims.