SEAD¶
SEAD contributes environmental-archaeology sites to the Nordic evidence atlas. Those sites are not static. Many are connected to dating ranges or relative periods in SEAD's relational database, while others still have no usable numeric chronology. The repository preserves that unevenness instead of forcing every site into one temporal category.
This source family stays an archaeology context layer: it is strong contextual evidence for nearby activity and chronology review, but it is not sample-owned proof of one lake event.
The practical rule is simple: use the governed Sweden discovery surface in the Nordic Atlas, then return to the normalized temporal-evidence and site-inventory products when auditing its source. The discovery surface keeps every Swedish site, preserves each eligible linked interval, and represents missing numeric chronology explicitly rather than as a static or timeless point.
Current Evidence State¶
The current governed full-evidence run is sead-full-evidence-39bfff6a-ce80714e (sha256:ce80714e4c9e9974b24913e5da50f49854670ed642879c1ca5076499e1d56725). Its raw-parent admission SHA-256 is 168ad1efe6fa68cdb7789246a6cce16670377cdbacd14d1673beaab8ddd1ebd1. Its denominators are:
| Governed population | Count | Interpretation |
|---|---|---|
| source tables | 61 | complete captured relational table set |
| sites in the Nordic bounding-box review | 2,195 | country-decision denominator |
| assigned four-country sites | 2,069 | SE 1,925, DK 59, NO 45, FI 40 |
| sites requiring country review | 103 | retained outside assigned publication membership |
| unassigned sites | 23 | retained without a governed country assignment |
| atlas SEAD features | 10,868 | 2,069 four-country site features plus 8,799 Swedish chronology-discovery features; not a distinct-site count |
| chronology claims | 25,109 | 13,170 comparable, 11,238 context-only, 60 refused by the numeric BP contract, 641 unresolved |
| source-native observations | 177,763 | quantitative observation denominator |
| source-native taxon relations | 1,974 | preserved source taxonomy, not accepted cross-source classification |
| dimension relations | 2,639 | explicit source-native measurement dimensions |
| eligible / refused propagation events | 0 / 177,763 | refused: source_classification_not_accepted |
Site, feature, claim, observation, relation, and event counts are different units. The atlas may display SEAD chronology and source-native detail, but it must not turn the refused event population into migration or propagation evidence.
The source-to-map custody replay checks every assigned site ID and UUID against its country decision, mapped site feature and Nordic map traceability record. It keeps the 126 review and unassigned decisions outside the published site set.
The country discovery and retrieval ledger also reconciles all 2,195 discovered IDs to source retrieval and normalized site rows. The source supplied no country field, so its per-country counts are governed boundary assignments; the 126 pending decisions remain excluded.
The four-country chronology account groups all 25,109 reviewed claims by country, age basis and display eligibility. Its 853 sites with claims are distinct from the 2,069 assigned site denominator; none of those claims is admitted as an ecological event.
The country retrieval and evidence account finds zero terminal site losses among the 2,069 assigned sites and zero recorded query failures across the 61-table capture. All 177,763 source-native observation dispositions are refused for ecological events; no event is eligible for propagation. The 126 review and unassigned site decisions remain outside these assigned-country denominators.
The country source-unit account separately follows assigned sites to captured sample groups, physical samples, analysis entities, linked datasets and analysis values. The excluded 126 sites have unknown downstream unit counts. These relations support source audit, not a pollen cohort or an approved ecological event.
From A Relational Database To A Map Point¶
A SEAD date belongs to a chain of records, not directly to a coordinate. The collector follows that chain and retains the intermediate identities before deriving a site envelope.
flowchart LR
Site["site"] --> Group["sample group"]
Group --> Sample["physical sample"]
Sample --> Entity["analysis entity"]
Entity --> Value["analysis value"]
Value --> Dating["analysis dating ranges"]
Entity --> Relative["relative dates and ages"]
Entity --> Model["modelled entity ages"]
Entity --> Geochron["geochronology"]
Entity --> Dendro["dendrochronology"]
Site --> Reference["site, sample-group, dataset, and period references"]
Reference --> Bibliography["bibliography"]
Dating --> Evidence["record-level temporal evidence"]
Relative --> Evidence
Model --> Evidence
Geochron --> Evidence
Dendro --> Evidence
Evidence --> Timeline["time-filterable chronology layer"]
Evidence --> Envelope["coarse site envelope"]
Envelope --> Point["site-inventory layer"]
The site envelope remains a publication convenience for ranking and overview.
The temporal-evidence layer is the appropriate product for chronological
navigation: each feature keeps one interval and the identifiers of all source
rows grouped into it. For sample-level reasoning, continue into
raw/nordic_sites.json; a mapped chronology feature is more precise than a
site envelope but is still contextual evidence, not proof of a lake event.
Site Posture And Chronology Posture¶
Numeric interval and context¶
A site has a normalized time_start_bp and time_end_bp derived from
captured temporal relations. It may also retain the original relative-period
labels and uncertainty notes. The optional site-inventory layer displays this
envelope as context; its points are never time-filtered. Only eligible linked
record intervals in the Swedish discovery layer enter the atlas BP filter.
For example, Agerod V (4237) is published with a site envelope of
7000–10000 BP. That envelope does not make the site present in every
overlapping map window. An Agerod V discovery feature appears in a selected
window only when its own eligible source-record interval overlaps.
Source label without numeric support¶
A site can have source period language but no stable numeric interval accepted
by the repository. A self-encoded label such as CAL_1242_AD- can be converted
because the bounds and era are present in the source value. A broad label such
as Quaternary cannot be placed on the slider by name alone. In the current
snapshot, such sites are part of the unresolved site population rather than a
separate numeric chronology layer.
Unresolved¶
The captured relations do not support either an eligible numeric envelope or
a useful normalized period label. The site remains valid spatial context. A
null time value means “not resolved under this contract,” not 0 BP, “modern,”
or “undated in the upstream database.”
flowchart TD
Relations["captured temporal relations"] --> Numeric{"eligible numeric bounds?"}
Numeric -->|yes| Timed["numeric interval and context"]
Numeric -->|no| Label{"source period label?"}
Label -->|yes| Context["contextual label only"]
Label -->|no| Unresolved["unresolved"]
Timed --> Display["site-envelope display context"]
Relations --> Record["eligible linked record interval"]
Record --> Slider["numeric timeline filtering"]
Context --> Reading["popup and source interpretation"]
Unresolved --> Reading
Numbered CE/AD and BCE/BC dates cannot use year zero. The source parser
refuses such an interval while retaining its original values and labels; an
invalid endpoint cannot inherit the other endpoint. Genuine 0 BP remains
valid. Native dates after 1950 retain their signed BP conversion, while their
admission to the current canonical timeline remains governed separately.
How The Atlas Timeline Treats SEAD¶
The Nordic Atlas exposes a governed Sweden archaeology discovery layer and an optional four-country SEAD site-inventory layer. The discovery layer contains all 1,925 assigned Swedish SEAD sites as 7,244 linked numeric chronology features plus 1,555 explicitly unresolved site features. Once a reader narrows the time window:
- numeric features remain visible only when their own linked interval overlaps the selected window; and
- unresolved features are withheld because overlap is unknown.
This is different from declaring unresolved sites absent from the selected period. The interface is refusing a comparison it cannot make. The optional 2,069-site inventory remains spatial context at any BP window. Its derived site envelopes are displayed for orientation and never used for interval filtering or counted as evidence within the selected period. Static layers such as boundaries are likewise unaffected.
The normalized temporal-evidence product remains available for audit and reuse. It is not loaded beside the discovery layer because that would duplicate the same record intervals.
What The Current Capture Cannot Resolve¶
Of the 2,069 assigned sites, 1,219 have no defensible numeric site envelope:
1,217 are unresolved and two retain contextual labels only. The captured
bibliography offers a directly followable DOI or URL for three sites; 2,066
require the source site page or further primary evidence. The per-site
dispositions and affected UUIDs are retained in
data/sead/review/temporal_review.json,
data/sead/review/access_model.json, and
data/sead/review/recovery_requirements.json. These are different gaps: a
bibliography relation does not supply a sample date.
Earlier backlog counts of 1,268 sites without numeric chronology and 2,192 without a direct bibliography link belong to a different captured population. Their differences from the current counts are not evidence that individual sites were recovered. A new date or link needs an identified source relation, stable source page, or citable primary record; otherwise the row stays contextual with its reason visible.
Compare SEAD With LANDCLIM And AADR Carefully¶
All three families can appear on one numeric BP timeline, but their observation units remain different.
| Source family | Timed map unit | What interval overlap supports | What it does not support |
|---|---|---|---|
| SEAD temporal evidence | grouped linked chronology interval | one or more identified source chronology rows overlap the selected window | the chronology proves a lake event or every record at the site is contemporaneous |
| LANDCLIM | pollen site-sequence interval | the sequence covers part of the selected window | direct association with a nearby archaeological site |
| AADR | dated human sample or governed locality descendant | the sample chronology overlaps the window | identity between a sample and a nearby site |
Temporal overlap is therefore a candidate relation for investigation, not a join key. Keep source identity, spatial distance, interval basis, and observation unit with any cross-source statement.
The 126 Captured Non-Members¶
The difference between 2,195 bounding-box review rows and 2,069 assigned site points is a country-membership decision. The 126 non-members comprise 103 rows requiring country review and 23 unassigned rows. They retain source identities and coordinates without being silently promoted into one of the four governed publication countries.
flowchart LR
Capture["2,195 captured sites"] --> Coordinate{"inside a governed country geometry?"}
Coordinate -->|assigned| Map["2,069 four-country site features"]
Coordinate -->|review| Review["103 country-review rows"]
Coordinate -->|unassigned| Retained["23 unassigned rows"]
Capture --> Chronology["25,109 chronology claims"]
Chronology --> Comparable["13,170 comparable claims"]
Chronology --> Ineligible["11,238 context-only + 60 refused + 641 unresolved"]
This is not deduplication or evidence deletion. A boundary or publication scope change must reevaluate the same 23 identities.
Appropriate Claims¶
SEAD supports:
- finding environmental-archaeology sites near a lake, pollen sequence, or aDNA locality under a declared distance rule;
- navigating 13,170 comparable chronology claims through interval-preserving features;
- retaining relative-period language for human interpretation without inventing numeric bounds;
- identifying sites whose bibliography or deeper relational evidence merits inspection; and
- comparing regional evidence coverage while keeping capture and publication denominators separate.
SEAD does not by itself support:
- treating proximity as direct association;
- treating a site envelope as a sample or event date;
- assigning dates to unresolved sites from neighbours or broad historical expectations;
- interpreting database density as past population or activity; or
- merging SEAD and RAÄ into one archaeology truth set.
Continue with Sweden archaeology site discovery for the complete population, readiness order, artifact set, and interpretation contract.
Governing Surfaces¶
| Surface | Responsibility |
|---|---|
data/sead/raw/nordic_sites.json |
captured site rows, relational inventories, source counts, and acquisition lineage |
data/sead/normalized/nordic_environmental_sites.geojson |
mapped features, temporal fields, popup evidence, and country membership |
data/sead/normalized/nordic_temporal_evidence.geojson |
mapped record-level chronology groups used by the atlas time filter |
data/sead/derived/sweden_archaeology_site_discovery.json |
complete Swedish site registry and evidence-readiness contract |
data/sead/derived/sweden_archaeology_site_discovery.geojson |
Atlas-ready exact intervals and explicitly unresolved Swedish sites |
data/sead/review/temporal_review.json |
row-level comparison posture and capture denominators |
data/sead/review/access_model.json |
mirrored versus upstream-only access boundary |
data/sead/review/evidence_legibility_review.json |
interpretability and publication risk |
data/sead/review/recovery_requirements.json |
remaining evidence gaps and their satisfaction signals |
data/source_spatiotemporal_posture_registry.json |
cross-source summary of allowed spatial and temporal use |
The SEAD handbook provides a step-by-step reading method. The SEAD exports guide explains which artifact to use for analysis, review, and publication.