What is spatially real
The land surface streams a digital elevation model. The county boundary,
385 settlement points, arterial roads, rail geometry, four station points
and two school points are current public-map records clipped to Hailun.
How the 3D bars are built
Each column aggregates a 3 by 3 group of WorldPop 1 km population
cells. Boundary cells are fractionally clipped to Hailun, and each
2010 and 2020 grid is normalized so all bars sum to that year's
official census population. Height is linear: one resident equals
0.18 map meters. In 2020, the pale lower volume is the remaining
population; the coral volume between the 2020 and 2010 heights is
gross cell-level population loss. Small cell-level gains are excluded
from that loss volume and reported separately from the net decline.
What is measured through time
Population values are census benchmarks for 2000, 2010 and 2020.
The Population evidence control also compares countywide age structure
and average family-household size in 2010 and 2020. Only documented
census years appear. The 2010 age shares and household size are derived
from the seventh-census bulletin's published percentage-point changes.
These countywide values appear as 3D chart columns anchored near Hailun:
they are map-based diagrams, not estimates for the ground beneath them.
The school chapter compares the compiled ordinary-primary-school counts
for 2000 (390), 2010 (218) and 2014 (32). The 2022 value of 34 is derived
from the official report's 26 compliant primary schools at a 76.5% rate,
so the interface marks 2022 as a definition break rather than pretending
the series is perfectly identical. The official unique compulsory-school
total is 54. The 2D map removes the blue road network and ordinary
settlement dots. Historical years render each documented county count as
a schematic field distributed across mapped settlements; those comparison
dots are not archived school addresses. In 2022 the schematic field
disappears and exactly two current school sites with open coordinates
appear. The outflow
chapter moves to a national province-to-province view and compares
census-defined rural-to-urban flows in 2010 and 2020; it does not infer
Hailun destinations from the county's population decline. The road-grid
chapter measures boundary-clipped OpenStreetMap road length
in 2014, 2015, 2020 and 2025 using the ohsome history API.
How the 2D census view works
The Population chapter can lock the map to a north-up, zero-pitch view
and display Hailun's 2000, 2010 and 2020 census records as countywide
charts. Use the year buttons, timeline or play control to compare them.
The 2000 view omits age structure because no comparable local age table
was available; the 2010 age shares are reconstructed from changes published
in the seventh-census bulletin. The colored 3 km cells use the matching
WorldPop year normalized to each official census total. They provide
spatial context but are not township-level census observations.
How interaction works
The evidence timeline is a continuous drag control that snaps to
documented observations on release. Chart dots identify evidence
positions but are not navigation buttons. Orbit mode makes a normal
map drag rotate and tilt the terrain; switch it off to pan instead.
Road development observation
Four OpenStreetMap history snapshots show Hailun's mapped drivable-road
network rising fastest between 2015 and 2020, then continuing to expand
at a slower pace through 2025. The road lines become brighter and wider
as the recorded network length increases.
What the migration graphic means
The arrow field borrows the frequency principle of The New York Times
county movement map: more moving marks mean more people. The China view
is reconstructed from the sixth- and seventh-census long-form tables.
A rural origin is a hukou registration place recorded as a township or
a town village committee; an urban destination is a current residence
recorded as a city or town. The published 10% long-form counts are
scaled by ten, following the census research convention.
Public data still do not report a Hailun-to-destination-city matrix. The
national view therefore changes scale instead of inventing local routes.
Every drawn direction is a real province-to-province OD pair, but the line
ends at a province display anchor because the census table does not name a
city. Only routes of at least 500,000 people are drawn: 23 routes in 2010
and 31 in 2020. The headline totals include all 930 possible cross-province
pairs, including the smaller routes that are not drawn.
What is intentionally absent
No village-level closure dates are fabricated. The School chapter's
historical dots are explicitly labeled as a schematic comparison field
based on mapped settlements, not real former school addresses. Only the
two 2022 points claim current open-map coordinates. Age and household
evidence is countywide, so
those views replace the spatial population columns with explicitly labeled
countywide chart columns rather than inventing a village distribution.