# Historical atlas terrain

Built by `scripts/build-atlas-terrain.cjs` followed by `scripts/build-atlas-levant.cjs` and `scripts/build-atlas-surface.cjs`. Terrain assets are served locally; no third-party map requests are made by the browser.

- Land colors: Natural Earth II, 1:50m, version 3.2.0, public domain. Idealized environmental land cover. https://www.naturalearthdata.com/downloads/50m-raster-data/50m-natural-earth-2/
- Optional modern boundaries: Natural Earth 1:50m Admin 0 boundary lines, v5.1.2 repository release. De facto boundaries follow the source dataset; disputed and indefinite lines are dashed. Generated by `scripts/build-atlas-boundaries.cjs`. https://www.naturalearthdata.com/downloads/50m-cultural-vectors/50m-admin-0-boundary-lines-2/
- Shorelines: Natural Earth 1:10m land polygons, public domain. https://www.naturalearthdata.com/downloads/10m-physical-vectors/10m-land/
- Elevation: Mapzen / Tilezen Terrain Tiles, Terrarium PNGs at zoom 7, hosted by AWS Open Data. https://registry.opendata.aws/terrain-tiles/
- Levant detail: zoom 9 elevation, 32–38° E, 29–35.5° N, 601 × 651 grid. Blended and stitched into the overview at the boundary.
- Inland lakes: Natural Earth 1:10m lake polygons (Sea of Galilee and Dead Sea), public domain. https://github.com/nvkelso/natural-earth-vector/blob/master/geojson/ne_10m_lakes.geojson
- Source attribution and license details: https://github.com/tilezen/joerd/blob/master/docs/attribution.md
- SRTM and GMTED2010 terrain data courtesy of the U.S. Geological Survey.
- ETOPO1 terrain data: DOC/NOAA/NESDIS/NCEI, National Centers for Environmental Information, NOAA, U.S. Department of Commerce. U.S. Government material is not subject to copyright protection within the United States.
- Europe terrain data produced using Copernicus data and information funded by the European Union, EU-DEM layers.

## Processing

Bounds: 8–45° E, 24–46° N. Elevation tiles are resampled from Web Mercator into a north-up equirectangular grid (741 × 441). Heights below sea level are clamped to zero; vertical relief is deliberately exaggerated for readability. The map is not a surveying or navigation tool. Modern land cover is context, not a historical reconstruction.

The surface builder uses the same public-domain Natural Earth II and 1:10m land polygons (pinned vector release v5.1.2), plus the already cached Tilezen zoom 7/9 elevation. No additional geographic sources or runtime services are used.

Land color is a low-frequency 2048-pixel texture. Saturation is independently tunable through `ATLAS_SATURATION` (default 1.0, preserving source saturation); a 12% warm parchment blend softens the palette. The earlier 22% blend was reduced to retain the source's Nile vegetation and desert/rock color variation. Inland color is propagated beyond the coast with a nearest-seed dilation, so bilinear sampling never introduces the source raster's blue shoreline pixels.

Coastlines are rasterized into signed Euclidean distance fields: overview 4096 pixels wide with a ±0.3° range; Levant 3072 pixels wide with a ±0.04° range over exactly 32–38° E, 29–35.5° N. The shader blends the patches at their edges and uses screen derivatives for a roughly one-device-pixel land/sea transition. SDF textures use linear filtering without mipmaps. Shallower-looking coastal water is decorative distance shading, not measured bathymetry. The SDF land/sea transition and lake polygon edges define the visible shores; no separate elevated coastline stroke is drawn.

Fine hillshade is computed from bilinearly sampled zoom 7 elevation (2400 pixels wide) and zoom 9 elevation (1201 pixels wide). Shading is deliberately restrained and complements the exaggerated elevation mesh; the material's fine-shade multiplier is 1.15. The five runtime surface textures total 1,523,567 bytes (1.45 MiB), replacing the older baked coast and bump textures. Original `relief.png` / `levant-relief.png` remain reproducible intermediate assets but are no longer downloaded by the atlas.

North Africa visual cross-check: NASA/JSC, STS040-078-025 (June 1991), describing the green Nile Delta, surrounding desert, and darker rocky Red Sea/Sinai terrain: https://eol.jsc.nasa.gov/Collections/EarthFromSpace/photoinfo.pl?PHOTO=STS040-78-25 . Consulted as a geographic reference only; no NASA imagery was imported. The refinement preserves the existing measured heights and source colors, without invented desert mountains or vegetation.

Map markers restore the original locally generated illustrative sculptures (scrolls, fragments, tablets, stelae, cylinders, prisms, ossuaries, pools and codices), with existing reference photographs on supported surfaces. These are stylized illustrations, not archaeological 3D scans or measured reconstructions. Original photographic credits remain in the associated articles and image credits page; this change imports no new photographs or datasets. The evidence panel retains photographs; its separate 3D inspector remains removed. Sculptures replace the circular medallions, with compact centered labels. Approximate locations remain explained in the Key and evidence panels.

## Terrain typography

Cardo Regular and Italic by David J. Perry and Inter by Rasmus Andersson are bundled locally under the SIL Open Font License 1.1. Original license files are included in `public/fonts/atlas/`. Sources: https://github.com/google/fonts/tree/main/ofl/cardo and https://github.com/google/fonts/tree/main/ofl/inter . The SDF text renderer uses these local fonts; its default and Unicode fallback URLs are also local, never a third-party CDN.

Ancient geographic names are manually placed orientation labels, adapted from the existing atlas gazetteer and general geographic locations. They do not assert political boundaries or simultaneous territorial control. Modern names use the existing Natural Earth-derived country label positions and appear only when modern boundaries are enabled.

Region names use stronger warm ink with a soft parchment halo and tighter tracking for small regions. Terrain-plane text targets a 12 CSS-pixel projected cap height for small regions, with a 4× scale cap near the horizon. Broad labels retain their larger geographic sizing with a 9-pixel cap-height floor; sea/modern/city floors are 8/9/10 pixels. The collision pass runs after this scale adjustment, and context yields to both artifact silhouettes and chips. Modern boundaries remain optional, in thin warm grey; rivers use a separate blue stroke. No ancient political outlines were inferred or added.


## Jerusalem city plate (5 km)

- Historical scan: Charles Wilson, Ordnance Survey of Jerusalem, survey 1864–65; digitized by the Eran Laor Cartographic Collection, National Library of Israel. The NLI-donated Commons scan is explicitly marked public domain / Public Domain Mark 1.0: https://commons.wikimedia.org/wiki/File:Ordnance_Survey_of_Jerusalem_-_The_Survey_was_made_in_the_Years_1864-5_by_Captain_Charles_W._Wilson,_R.E.jpg . Original publication predates 1931; Wilson died in 1905. Downloaded original 2185 × 2835 JPEG from https://upload.wikimedia.org/wikipedia/commons/0/02/Ordnance_Survey_of_Jerusalem_-_The_Survey_was_made_in_the_Years_1864-5_by_Captain_Charles_W._Wilson%2C_R.E.jpg .
- Four manually identified control points: Jaffa Gate, Damascus Gate, Zion Gate, and the Dome of the Rock. Reference coordinates are OpenStreetMap node 561157424, way 616186088, node 5003920348, and way 4709536. The cached source response is `/images/historical/jerusalem/control-points-osm.json`. Pixel pairs, WGS84 coordinates, affine coefficients and residuals are recorded in `data/jerusalemInset.json`; `scripts/build-jerusalem-inset.cjs` performs the least-squares fit and bilinear resampling. Fit RMS is approximately 0.58 m **at these four control points**, not a claim of absolute positional accuracy; manual identification, historic survey distortion and extrapolation remain limitations.
- The scan does not extend south to North Talpiot. Outside its actual coverage, the plate uses locally cached modern street context from OpenStreetMap API 0.6, requested in eight bounded extracts over 35.2055–35.2585 E, 31.7455–31.7660 N. Attribution: © OpenStreetMap contributors, Open Database License 1.0. https://www.openstreetmap.org/copyright and https://opendatacommons.org/licenses/odbl/1-0/ . Original selected street geometries and IDs are downloadable at `/images/historical/jerusalem/osm-context.json`; that extract and the OSM-derived coordinate data in `data/jerusalemInset.json` are available under ODbL 1.0. Modern streets are contextual, not a reconstruction of the ancient city.
- The inset's traced wall/Temple Mount outlines and Kidron/Hinnom valley guides are interpretive tracings of the 1865 scan, not archaeological boundary surveys. The 1865 Old City walls are labelled by date. Find coordinates and evidence content are taken unchanged from `historicalNodes.ts`. Callout offsets never modify those coordinates. All map imagery, coordinates, street geometry and fonts are served locally; there are no third-party runtime map requests.

## Persistent artifact labels and sizing

The November 2026 label refinement changes placement, projected sculpture sizing, and the existing Rome fan only. It adds no imagery, datasets, fonts, or external runtime requests. Recorded find coordinates and evidence content are unchanged; geographic leader lines still return the fanned objects to their recorded location.

## Interactive Jerusalem terrain (November 2026)

The dedicated 3D city view uses Tilezen / Mapzen Terrarium elevation tiles at zoom 14 from `https://s3.amazonaws.com/elevation-tiles-prod/terrarium/14/{x}/{y}.png`. Source and attribution: https://registry.opendata.aws/terrain-tiles/ and https://github.com/tilezen/joerd/blob/master/docs/attribution.md . SRTM terrain data courtesy of the U.S. Geological Survey (public domain). `scripts/build-jerusalem-terrain.cjs` resamples the tiles offline to a 161 × 161 grid, about 31.25 m spacing, covering the existing 5 km city plate centred at 35.232° E, 31.768° N. Runtime data is the locally bundled `data/jerusalemElevation.json` (about 104 KB); no tile service is contacted by readers.

Terrain represents modern elevations, with 1.5× vertical relief. Wall and Temple Mount footprints reuse the interpretive 1865 Wilson traces documented above; extruded heights are illustrative, not surveyed. The small regional city gateway uses those same wall outlines. The tunnel symbol is an original schematic illustration, not an archaeological reconstruction. The existing artifact illustrations and photographs are retained.

Ground rings use unchanged evidence coordinates. A stable display fan separates the densely packed city-centre objects; solid leaders return them to their recorded ground points. Four shared-coordinate accounts use one explicitly approximate marker and four individually accessible list entries. The scene is not a reconstruction of any single ancient period. Wilson survey, OSM control-point attribution and limitations above still apply; the original survey plate remains available as a reference.

## Geographic context and controls refinement

The additional Sahara, Caucasus, Northern Sinai, Syrian Desert and Negev labels are manually placed physical/geographic orientation names, using the same approach as the existing ancient gazetteer. They do not delineate political claims or reconstruct an era. The close-view Mediterranean label has separate desktop and narrow-screen anchors within the sea; the Dead Sea reuses the existing lake geography. No new external datasets, imagery, fonts or runtime requests were introduced in this controls/typography pass. Geographic captions remain anchored on the map but face the reader, with a restrained halo; background captions yield to artifacts and their names.

Siloam photograph restored at its existing node.image path: Félix Bonfils, circa 1870–1880, via roni manor / PikiWiki, https://commons.wikimedia.org/wiki/File:PikiWiki_Israel_71717_the_siloam_pool.jpg . Commons identifies the file as public domain (PD-author; structured data records a public-domain dedication). Original download: https://upload.wikimedia.org/wikipedia/commons/1/1c/PikiWiki_Israel_71717_the_siloam_pool.jpg . Resized to 1200 px wide, JPEG quality 85, to restore the image already credited in the site's WordPress article backup. This is the historic northern/Byzantine pool, not a photograph of the southern Second Temple pool's modern excavation. The existing node image path, caption, coordinates and evidence content are unchanged; no runtime third-party image request is added.

## Caption and model detail refinement

No new source data or external runtime service is used. Cyrus, Taylor prism and
Ketef Hinnom surface details are cropped/repeated from their existing local `node.image`
photographs; these are illustrative texture mappings, not scans of a complete
object. Pool masonry, curled silver foil, ossuary relief, and the Levant collection
arrangement are original symbolic modeling. They do not assert surveyed geometry.

The Jerusalem gateway reuses the credited Wilson wall/precinct traces with an
original illustrative roofscape, towers and parapets. Those buildings and heights
are not surveyed footprints or a period reconstruction. The city ground blends
linework/building tint from the already georeferenced local Wilson scan onto the
modern elevation. Outside the scan, the existing ODbL street extract supplies
subdued present-day context. Decorative stone courses, coping and battlements
add depth without claiming archaeological masonry detail. Geographic locations
and original node content are unchanged; selection reveals the recorded anchor
beneath any cartographic display offset.

Ancient geographic captions now target 14px capitals for broad regions and 12px
for secondary/local regions, with tighter tracking. Their representative label
placements do not define ancient borders. Modern boundaries remain optional.

## Manuscript presentation and regional spacing

The Papyri symbol reuses the existing local `/images/wp/Papyrus_52.jpg` asset.
Its face is cropped to the photographed fragment on the left, excluding the
modern English explanatory text on the right. The thin display frame is original
illustrative modeling, not an ancient binding or a claim of a complete codex.
The original image, evidence card and attribution remain unchanged. The physical
fragment/manuscript distinction is consistent with CSNTM's P52 catalogue:
https://manuscripts.csntm.org/Manuscript/View/GA_P52 (reference only; no downloaded
asset or external runtime request).

Pilate Stone, Mesha, Lachish and Tel Dan gain small screen-space presentation
offsets. Recorded coordinates, location notes and geographic selection anchors
are unchanged; these offsets are not corrections to discovery locations.

Hydrology reference for interpreting the existing Jordan linework: NASA,
“The Dead Sea,” https://www.nasa.gov/image-article/dead-sea-2/ and
https://earthobservatory.nasa.gov/IOTD/view.php?id=6136 . The Jordan feeds the
Dead Sea from the north; the basin has no river outlet. These references supply
no new map geometry or runtime service.

River drawing is now split at intersections with the existing Natural Earth lake
polygons. Lacustrine centerlines are not drawn over lake surfaces; the Jordan's
bundled Dead Sea centerline and southern-basin connector are also suppressed so
the exposed land bridge does not imply a river outflow. Source river and lake
datasets are retained unchanged.

## Survey-guided Jerusalem roofscape

`scripts/build-jerusalem-roofscape.cjs` processes only the previously credited
local Wilson 1865 survey. It samples the pink built-area tint on a 24 m grid,
clips against the traced Old City perimeter, excludes the Temple Mount court,
and combines neighboring samples into coarse masses. The 17 KB generated helper
is `lib/jerusalemRoofscape.ts`; no additional image or service is requested.
These masses convey the survey's built-area density and openings. They are not
measured individual building footprints. Heights, roof articulation, parapets,
vaulted accents, gate details and warm stone colors are original illustrative
modeling. Main-scene gateway proportions are exaggerated for recognition. City
terrain still uses the existing elevation; the restrained valley tint is a
cartographic aid, not a vegetation reconstruction. Recorded find-spots are kept
clear of decorative masses.

Alternative sea-caption anchors stay within their named basins and are merely
label placements. Only one candidate for each sea is displayed; these anchors
add no geographic or political boundary data.

## Jaffa Gate district architectural study

This single, approximately 350 × 300 m study reuses the already credited local
Wilson 1865 georeferenced scan and elevation. No new external asset, dataset,
font or runtime service is used. `lib/jerusalemDistrict.ts` stores hand-authored
generalizations in the 1800 × 1800 scan's pixel coordinates. The bent entrance
lane and broad openings were visually referenced to that plate. Individual
house footprints, court arrangements, branching lane widths, heights, vaults,
arcade, stairs, masonry basin and facade openings are original illustrative
architecture, not excavated features or measured 1865 buildings. The gate
opening is anchored to the existing Jaffa Gate control point; its modeled
facade is schematic, not an architectural reconstruction.

The district replaces only the coarse roof masses intersecting its boundary.
Paving follows the bundled modern elevation; facade height uses the same
miniature exaggeration as the scene. Original evidence coordinates and content
remain unchanged. The shared approximate accounts symbol uses a continuous
close-view display fan to keep it above the study area and inside phone framing;
selecting it still reveals its recorded anchor and solid connector.
At the phone's whole-city scale, Siloam has a little extra presentation spacing
from Hezekiah's Tunnel so their retained caption tiers do not intersect after
returning from a close-up. This smoothly tapers out on zooming in and does not
change either recorded location.

### District proportions and materials refinement

The Jaffa Gate study now uses connected courtyard wings, two stepped upper
rooms, lower roof parapets, a modeled open stone arch and an asymmetric gate
return. These are original illustrative design choices on the same survey-guided
layout, not additional surveyed or archaeologically documented structures.
Timber shutters, restrained cloth shades and one small courtyard tree are also
illustrative; they do not assert a historical shop, planting or furnishing at
these locations. The existing gate control point and evidence coordinates remain
unchanged.

`components/historical/districtMaterials.ts` generates original stone-course,
paving and plaster variation in the material shader. No photograph, downloaded
texture, additional dataset or runtime service is used. Colors distinguish pale
limestone, weathered masonry, roof plaster, dusty ground, paved lanes and dark
timber. Course sizes are exaggerated for the miniature view and fade below
screen resolution; these are visual material cues rather than measured masonry.
The surrounding coarse roofscape receives only lower parapets and quieter
colors. The detailed architectural study has not been expanded beyond its
existing boundary.

### Gate-to-courtyard composition and evidence reader

The same district study now uses original wall-shell geometry with recessed
window and doorway openings, connected street-facing wings, stepped upper rooms,
a roofed courtyard threshold and a larger open arcade. The gateway has an open
pointed portal and dressed surrounds on both faces. The short curved approach,
cloth shades, bench, vessels and planting are illustrative scale cues, not newly
mapped roads, archaeological objects or documented historic furnishings. The
existing Wilson survey and gate control point remain the geographic basis; no
new scan, photo texture, architectural dataset or external runtime request is
introduced. This is a nineteenth-century survey-guided illustration, not a
reconstruction of first-century Jerusalem.

At district scale the shared approximate accounts symbol uses a stable local
display anchor beside the study. The transition depends continuously on camera
distance and target proximity; the recorded accounts coordinate, selection
pinpoint and content are unchanged. The full evidence reader reuses the atlas's
existing photographs and attribution. Its optional index lists all nine entries;
the four accounts remain individually accessible from the shared marker.

### Connected entrance frontage study

The gate and four adjoining building footprints have been revised within the
existing district. The covered lane threshold, timber balcony, shop bays,
awnings, stair and dressed facade details are original illustrative geometry;
they are not documented buildings or archaeological features. The continuous
approach and blended ground are also presentation geometry, not new mapped
roads. Existing evidence coordinates, photographs and survey control points are
unchanged. No new external asset or runtime request is introduced.

Visual reference catalogues consulted: Library of Congress, American Colony
photographs [Jaffa Gate](https://www.loc.gov/pictures/item/2019698090/) and
[street scene inside the gate](https://www.loc.gov/pictures/item/2019698387/).
No image from these records is reproduced or bundled. They inform the later
city's architectural vocabulary, not a claim that the authored frontage matches
a measured historic facade or a first-century reconstruction. Stone grain,
plaster, color variation and surface relief remain original procedural shaders.


### Gate scale and Jerusalem location audit — 11 November 2026

This supersedes earlier notes about unchanged coordinates and a district-local
accounts anchor. No new runtime request or bundled image was added.

**Architecture.** Library of Congress, *The Jaffa Gate, Jerusalem, Holy Land*,
1890–1900, Photochrom collection, [catalogue 2002724999](https://www.loc.gov/item/2002724999/),
[digital reference ppmsca.02688](https://www.loc.gov/resource/ppmsca.02688/).
Public-domain historical photograph; inspected during authoring, not reproduced
as a texture. The compact doorway below a taller pointed blind recess, offset
fortified shoulder and restrained parapet informed original geometry. Dimensions
are proportional estimates, not an architectural survey. The six attached
entrance houses, shades and balcony remain illustrative, within the existing
study. Their smaller plots replace four oversized wings; the city is not expanded.
Do not present this Ottoman-period reference as first-century Jerusalem.

**Caiaphas tomb.** [Greenberg and Keinan, 2009, gazetteer entry 703](https://www.academia.edu/219444/Israeli_Archaeological_Activity_in_the_West_Bank_1967_2007_A_Sourcebook_The_West_Bank_and_East_Jerusalem_Archaeological_Database_Project)
records Old Israel Grid 171730/129570, IAA permit A-1748/1990. The old six-digit
northing is 1129570 in EPSG:28193. Offline conversion using PROJ through pyproj
3.8.0, `TransformerGroup(28193,4326,always_xy=True)`, returns longitude
35.22844908655, latitude 31.75879115551. Its selected operation is inverse Israeli
CS + Palestine 1923 to WGS84 (1), nominal transformation accuracy 2 m; this is
NOT the accuracy of the excavation gazetteer or object findspot. Stored rounded
coordinate: 31.75879,35.22845 (previously 31.749,35.226). The gazetteer supplies
a tomb-area point, not a position inside a burial chamber.
[Excavator Zevi Greenhut's 1991 account](https://www.jerusalemperspective.com/2592/)
independently places the 1990 discovery in the Peace Forest near North Talpiyyot.
[CRS definition](https://epsg.io/28193). These copyrighted publications were
consulted for attributed geographic facts; their text and images are not bundled.

**Ketef Hinnom.** [University of Michigan collection record](https://image-database.nes.lsa.umich.edu/items/show/96)
places the amulets in a tomb across the Hinnom Valley southwest of the Old City.
The [photographed site marker and location record](https://www.hmdb.org/m.asp?m=63881)
locate the caves beside St Andrew's Church. The approximate site-area marker is
now 31.7689,35.2253, replacing 31.768,35.224. It is NOT the precise repository in
Cave 24. The photographed sign's coordinate is not treated as the amulet findspot.
Only attributed location facts are used; no copyrighted image or text reproduced.

**Bethesda, Siloam and tunnel.** Bethesda remains the original approximate pool
complex coordinate (31.7814,35.2364), with no claim of surveyed trench precision.
Siloam and tunnel coordinates remain unchanged pending an excavation-plan check.
[City of David's pool description](https://cityofdavid.org.il/en/siloam-pool-opened-eng/)
and its [2022 visitor map](https://cityofdavid.org.il/wp-content/uploads/2023/03/Traveler_COD_Eng_2022_lowrez.pdf)
distinguish the Second Temple pool from the tunnel outlet and locate the
inscription near the southern end of the tunnel. A waterworks-area marker must
not be described as the exact inscription discovery point. These copyrighted
resources are cited, not bundled or traced into new geometry.

**Four accounts.** Their common coordinate is a representative collection
anchor, not a findspot. [1 Corinthians 15:6](https://www.biblegateway.com/passage/?search=1+Corinthians+15%3A6&version=ESV)
does not name the location of the appearance to the five hundred. Medical
corroboration is analysis of texts; the other accounts likewise do not establish
a shared exact point. Their stored coordinates and evidence text remain unchanged,
but coordinate basis and reader notes now make the distinction explicit. The
city collection no longer uses a selected ground pinpoint or moves beside Jaffa
Gate for framing. No biblical text was newly copied.

**Presentation.** City display offsets taper with view distance and are capped
at 25 m at street scale (up to 1.2 km only at whole-city scale for label separation).
They never change geographic anchors. Selecting a physical find shows its stored
site-area point; provenance and remaining uncertainty appear in its reader.

### Connected gate street — 11 November 2026

`jerusalemGateStreet.ts` is original authored geometry replacing the immediate
four generic frontage plots with two connected facades, setbacks and an open
court. Openings, shutters, awning, bench, jars, paving and vine are illustrative
scale cues, not surveyed buildings or archaeological reconstructions. The gate
continues to use the public-domain LOC Photochrom reference credited above;
no additional external asset, dataset or runtime request was introduced.

The limestone, plaster, timber and paving colors and weathering are original
vertex colors/procedural materials. The street follows the existing terrain;
it does not assert a measured historical street level. Its setting remains an
Ottoman-era visual study, not a depiction of the city at the evidence dates.
All evidence coordinates and location provenance are unchanged in this pass.

### Aerial city readability — 11 November 2026

The atlas gateway and city view reuse the existing Wilson survey-derived roof
masses, wall/Temple Mount traces and bundled elevation. Roof colors and wall
contrast are original cartographic styling. Wider muted valley tints follow the
existing Kidron and Hinnom traces; they are visual orientation aids, not a land
cover survey. No trees, buildings, historical boundaries or new geographic data
were added. The same illustrative heights and historical limitations apply.

City framing and bounded artifact presentation offsets changed; stored evidence
coordinates, source notes and selection pinpoints did not. No external asset or
runtime request was introduced. Cardo remains the locally bundled OFL font.

### Individually selectable Jerusalem accounts — 11 November 2026

Four account illustrations now reuse each node's existing locally bundled image
and artifact treatment. Their two-column arrangement is a reading/presentation
layout around the unchanged representative Jerusalem anchor, not four claimed
event locations or archaeological findspots. Ground-plane projection keeps the
illustrations above the terrain; it supplies no new geographic evidence. The
five finds' stored coordinates and all evidence/location notes are unchanged.
The short phone reader gives Bethesda extra bounded display clearance from the
accounts; selection still identifies the original site anchor. No new source,
image, city detail or external runtime request was introduced.

## Surface tone and sea depth — 13 November 2026

`sea-depth.webp` (1200 × 714, about 20 KB) is built by `scripts/build-atlas-bathymetry.cjs` from the same cached Tilezen / Mapzen Terrarium zoom 7 tiles as the land relief; below sea level those tiles carry ETOPO1 (NOAA NCEI). Depth is clamped at 4,500 m and square-root encoded. The water colour now follows this measured depth, with the earlier decorative pale shelf kept along the coast. No new data source is introduced.

The land shader keeps the Natural Earth II source colours and adjusts them at render time: dry, bright ground receives a warmer tone and a stronger hillshade multiplier (1.3 to 2.4, was 1.15); ground where green exceeds red has its saturation raised by 40%; high ground is paled slightly using mesh height. A thin ink line is drawn on the seaward side of the shore from the existing coast distance fields, and the sheet fades into the paper backdrop over its outer 4%. These are presentation choices and do not alter elevation, coastlines or source colour files.


### Jerusalem presentation polish — 14 November 2026

The existing elevation, Wilson survey, roof masses and valley traces are reused.
Olive valley tints, warm limestone tones, relief shading and a feathered sheet
edge are cartographic styling, not new land-cover or reconstruction evidence.
The overview now separates the full artifact/caption footprints in screen space;
additional presentation displacement tapers away below neighbourhood scale.
The earlier 1.2 km overview cap describes the initial fan only, before this
collision pass. Stored locations, selected site anchors and the close-up 25 m
fan cap remain unchanged. The four accounts remain a reading arrangement, not
four asserted findspots. No external assets or new geographic data were added.


### Jerusalem relief and account settings: 14 November 2026

This update supersedes the earlier 1.5× relief and shared-account-anchor descriptions.
The scene now uses 2.6× vertical relief with baked hill shading. Account medallions
are presentation markers, not archaeological findspots: the two crucifixion-related
entries use the traditional Holy Sepulchre setting; Apostles’ Testimony uses a
representative Temple Mount setting associated with Solomon’s Colonnade in Acts
3:11 and 5:12. These coordinates are not verified event locations. The appearance
to the five hundred (1 Corinthians 15:6) has no recorded location and is shown at
the map margin without a geographic leader.

The dashed blue tunnel curve is drawn schematically between the Gihon Spring and
the Siloam area. Its intermediate bends are not survey data. Map notes disclose
this directly. The survey-derived wall outline represents later Jerusalem,
not a newly sourced first-century city boundary.


### Grounded city markers: 14 November 2026

This correction supersedes the earlier screen-offset and collision-spread
descriptions. Artifact miniatures now have fixed world positions, dimensions,
and orientations at their existing source anchors. Small illustrative footings
meet the local terrain beneath each model; the Apostles' account rests on the
existing Temple Mount platform. These supports are display geometry, not claimed
archaeological foundations. Captions can still avoid nearby labels and controls.

The two crucifixion accounts share one compact marker labeled "Golgotha: 2
accounts" and "traditional location". Its explanation distinguishes the
traditional Holy Sepulchre setting from a verified findspot and the first-century
city from the displayed 16th-century walls. The appearance to the five hundred
remains accessible as a separate "place not recorded" control, with no invented
geographic anchor. The existing photographs and historical location limitations
are retained.

All finds fits the fixed marker set. City exploration uses an elevated camera
range to reduce terrain occlusion; the distinct Jaffa Gate close-up is named
explicitly. Markers can leave the viewport when zooming or panning. Neither the
terrain mesh, elevation data, lighting, nor 2.6× vertical relief changed in this
grounding correction. No external assets or new geographic data were introduced.

### Jerusalem geography corrections — 14 November 2026

This supersedes the earlier schematic tunnel, unchanged pool coordinates, and
valley-trace notes. The modern elevation grid and 2.6× vertical relief are
unchanged. The Temple Mount court is now a filled enclosure: its retaining faces
extend below the lowest sampled perimeter terrain rather than leaving a thin
slab suspended over the southeastern slope. The court remains an illustrative
level surface derived from the existing elevation, not a surveyed ancient level.

Hinnom's tint and captions now follow the southern valley trough on the Wilson
1865 plate and the bundled elevation grid. The shared junction and lower Kidron
were checked against [OSM way 844849843](https://www.openstreetmap.org/way/844849843).
The terrain heights themselves are not moved to match the tint.

The tunnel bends are generalized from Charles Warren and Claude Conder, *The
Survey of Western Palestine: Jerusalem* (1884), [public-domain plan](https://commons.wikimedia.org/wiki/File:Siloam_Tunnel_sketch_1884.png).
The original 362 × 588 pixel tracing and its two registration controls are retained
in `data/jerusalemGeography.json`. A similarity transform registers the plan to
the mapped spring and southern outlet areas; this is not a new excavation survey.
The generalized line is about 502 m long, compared with the approximately 533 m
physical passage, reflecting generalization and registration uncertainty. OSM's tunnel way is marked
as needing improved alignment and was not used for the course.

The dashed line is an underground overlay, at a nearly level water elevation
(635 m to 634.7 m), rather than a line climbing the modern ground. Elevation context:
[Shimron et al., 2022, Gihon waterworks](https://www.scirp.org/pdf/ad_2022042814140503.pdf).
The small end-to-end gradient is generalized, not surveyed intermediate tunnel
levels. It stops at the smaller Byzantine outlet pool and does not continue to
the later Second Temple pool. These locations were checked against the
[City of David 2022 visitor map](https://cityofdavid.org.il/wp-content/uploads/2023/03/Traveler_COD_Eng_2022_lowrez.pdf).
That copyrighted PDF was consulted for geographic facts, not reproduced or traced.

Modern registration controls: [Gihon spring-area node](https://www.openstreetmap.org/node/1608094098),
[tunnel outlet path](https://www.openstreetmap.org/way/755420509), and
[Second Temple pool site](https://www.openstreetmap.org/node/3309212388).
The tunnel evidence marker now identifies its southern outlet area; it does not
claim the exact inscription panel. Bethesda uses the pool-complex point in the
[University of Munich gazetteer](https://www.bedadls.gwi.uni-muenchen.de/topolist/).
These are site-area coordinates, not individual trench or artifact positions.

OSM-derived coordinates and linework are © OpenStreetMap contributors, ODbL,
retrieved 14 November 2026. Their source IDs and correction coordinates are
retained in `data/jerusalemGeography.json`; generation uses the same file. No new
third-party runtime request or image asset is introduced.
# Jerusalem around AD 30

The default city layer is an interpretive reconstruction for approximately AD 30.
The 1865 comparison retains the Wilson image, wall trace, roofscape and Ottoman
Jaffa Gate district. Switching dates preserves the camera. Evidence sculptures
are an optional layer covering several centuries, not ancient city buildings.

## Geometry and placement method

`data/jerusalemAD30.ts` records WGS84 control points, feature-specific evidence
notes, and source links. `lib/jerusalemAD30.ts` projects those points using the
same local kilometre coordinates as the survey. Coordinates for lost wall
connections and buildings are **generalized author placements**, constrained by
published topographic relationships; they are not a newly acquired excavation
GIS dataset or a metric digitization of a site plan. Decimal precision supports
stable rendering, not metre-level archaeological certainty.

- **First Wall:** generalized citadel–western hill–Mount Zion–Siloam–Ophel circuit,
  with its northern return toward the Temple causeway. Based on Josephus,
  *Jewish War* 5.4, Geva's archaeological synthesis, and the Mount Zion gateway
  excavation. Connecting segments, wall heights and tower spacing are illustrative.
- **Second Wall:** one proposed Gennath–Antonia course, rendered as ochre dashes
  instead of asserted masonry. Traditional Golgotha is outside this proposal.
  The Third Wall, begun under Agrippa I in AD 41–44, is excluded.
- **Temple:** the existing georeferenced enclosure is reused as a geographic
  anchor. The east-facing sanctuary near the rock, inner courts, porticoes and
  southern Royal Stoa are simplified interpretations of ancient descriptions
  and the conventional reconstruction. The rock identification, exact sanctuary
  position, reconstructed elevations and completion dates of individual portions
  are not asserted as excavated facts. No later mosque, church, Ottoman gate,
  Hadrianic arch or Byzantine cardo is rendered in the AD 30 layer.
- **Palace:** placed south of the citadel on the western hill. Two courtyard
  ranges and three northern towers express the literary description; their exact
  footprints and dimensions are illustrative. No trial room is identified.
- **Antonia:** a compact schematic fortress at the northwest Temple corner.
  Its extent and upper structures remain uncertain.
- **Residential areas:** broad western-hill, eastern-ridge and northern-suburb
  polygons. Courtyard architecture draws on the Upper City excavations. Generated
  plots and gaps do not reproduce an excavated street grid. They exclude the
  Temple, known site anchors and mapped wall/street corridors.
- **Bethesda:** shared gazetteer coordinate, generalized twin basins and five
  text-informed porticoes. Basin dimensions are illustrative. The subsequent
  church and healing-complex buildings are absent.
- **Siloam:** shared, distinct outlet and Birket al-Hamra coordinates. The latter
  uses a restrained basin with northern steps, not a completed four-sided stair
  reconstruction. The 2024 excavation report challenges both the identification
  and the function of the steps; the associated atlas entry now explains this.
- **Monumental street:** a generalized archaeological corridor appears only on
  request as blue dashes. The excavators date completed paving after AD 30/31 or
  31/32 and before AD 41/42, so a finished street is absent from the AD 30 fabric.
- **Ground:** the modern DEM and its uniform 2.6× relief remain shared between
  periods. This is not an ancient bedrock model. Ancient fill depths, quarries,
  street levels and retaining terraces are not recovered from it. The central
  Tyropoeon is indicated geographically without inventing an exact ancient depth.

## Reconstruction references

- [Josephus, *Jewish War*, Book 5, chapters 4–5](https://www.gutenberg.org/files/2850/2850-h/2850-h.htm): primary literary descriptions, read in their AD 70 narrative context and filtered for the earlier scene.
- [Hillel Geva (2022), Hasmonean Jerusalem through the Lens of Archaeology](https://publications.iaa.org.il/cornerstone/vol10/iss1/7/): First Wall, western-hill occupation, and the interpretive status of the Second Wall.
- [Upper City excavation overview](https://www.gov.il/en/pages/jerusalem-the-upper-city-during-the-second-temple-period): residential architecture and palace-tower remains.
- [Bargil Pixner, Mount Zion gateway excavations](https://www.zionsberg-jerusalem.de/wp-content/uploads/2017/12/Jerusalems-Essene-Gateway.pdf): southern wall context. The proposed Essene community identification is not adopted as a mapped neighborhood.
- [IAA, Wilson's Arch and the Great Causeway (2011)](https://hadashot.iaa.org.il/Report_Detail_Eng.aspx?id=1738&mag_id=118&print=all): archaeological phases of the western Temple approach.
- [Szanton, Hagbi, Uziel and Ariel (2019), Pontius Pilate in Jerusalem](https://www.academia.edu/40780377/Pontius_Pilate_in_Jerusalem_The_Monumental_Street_from_the_Siloam_Pool_to_the_Temple_Mount): street construction chronology.
- [Szanton, Vukosavović and Berko (2024), New archaeological discoveries at Birket al-Hamra](https://cityofdavid.org.il/wp-content/uploads/2025/12/New-archaeological-discoveries-in-the-excavations-of-Birket-al-Hamra-in-Jerusalem.pdf): basin remains, steps and revised interpretation. The upload date is later than the 2024 conference publication.
- [Avi-Yonah Jerusalem model guide](https://cojs.org/wp-content/uploads/Model_of_Ancient_Jerusalem.pdf): cross-check of broad landmark relationships and Third Wall chronology. This model depicts AD 66, not AD 30. Its speculative theater, hippodrome and named residences were not copied; its reconstructions are not treated as survey evidence.

Published illustrations were inspected as research references; no copyrighted
illustration is bundled in the public site. Meshes are original code-generated
interpretations. Existing survey, DEM and OSM credits below still apply.


## 1865 comparison rebuilt from the large-scale survey (15 November 2026)

Supersedes the earlier 24 m pink-cell roofscape and illustrative Jaffa district
entries above for the live 1865 scene. The city now uses hatched built-area
polygons and open courts extracted from Wilson’s 1:2,500 city sheet, with
separate Haram, Holy Sepulchre, pool and citadel architecture. Domestic roof
bays, heights and details remain interpretation, not surveyed house parcels.

- [Wilson 1865 city plan, Nottingham scan](https://commons.wikimedia.org/wiki/File:1865_Ordnance_Survey_of_Jerusalem_Old_City_full_map.jpg)
- [Wilson Haram Grounds, two sheets, BnF GE C-9410](https://gallica.bnf.fr/ark:/12148/btv1b53098730n)
- [McDonald, Pool of Hezekiah and city roofs, 1865](https://commons.wikimedia.org/wiki/File:James_McDonald._Views_within_the_city._Pool_of_Hezekiah._1865.jpg)
- [McDonald, city panorama, 1865, Library of Congress](https://www.loc.gov/pictures/item/90714422/)
- [Salzmann, Dome of the Rock from the west, 1854, Metropolitan Museum](https://www.metmuseum.org/art/collection/search/287019)

All these photographs and the two Haram sheets were inspected as references.
The full primary city scan is public domain. No remote imagery is fetched at
runtime. The 1865 ground locally incorporates the recorded 2,419-foot Haram
setting and 2,492-foot Holy Sepulchre setting, blended into the modern DEM.
The ground plane within the Haram is simplified; individual roof heights and
colours are not recoverable from a plan and monochrome photographs. Details
and reproducible pixel controls are in `docs/jerusalem-1865-reconstruction.md`.
AD 30 and the AD 30 regional-map gateway are unchanged.


### Jerusalem final scale and enclosure audit: 15 November 2026

This supersedes the earlier shared 2.6× relief and gateway-outline notes.
Both city periods use natural vertical scale, and the AD 30 regional miniature
uses that same city model. The regional atlas terrain is unchanged. Both city
periods share the full Wilson city-sheet enclosure tracing; the former AD 30
small-sheet trace differed by 14–32 metres at the enclosure corners.

The AD 30 outer court uses Wilson’s 2,419-foot level as a modelling reference,
with separate raised inner courts. It is not an excavated AD 30 floor level.
Modern surface returns inside the precinct are capped; local bridge and southern
gate approach surfaces are schematic corrections to meet the reconstructed
access geometry. They do not recover exact ancient fill or bedrock. Outside
these local adjustments the existing modern elevation samples are retained.
See `docs/jerusalem-1865-reconstruction.md` for method and verification.
