Ray Apena — R3

ATLAS / PLATE 001

THE WALKING RIVER

Tracking the movement of The Yellow River Delta, 1984–2024 (it can't seem to stay put).

How do you draw a river that refuses to keep one shape?

contextplatenotes
  1. N.01

    THE IDEA

    I wanted to represent the Yellow River Delta as a temporal surface instead of choosing one river line. In a place where channels move and hydraulic intervention is continuous, a static trace hides the part I find interesting.

  2. N.02

    INPUTS

    The Plate uses the JRC Global Surface Water yearly-history record for 1984–2024. I grouped the validated yearly observations into four epochs, then used the same record to calculate full-period persistence and wet/dry transition counts.

  3. N.03

    HOW I BUILT IT

    I calculated the share of valid years containing water for 1984–1993, 1994–2003, 2004–2013 and 2014–2024. From the full sequence I also calculated persistence and the number of valid consecutive-year switches between water and non-water.

  4. N.04

    RESULT

    Across the full record, 3,310 km² reaches persistence of at least 0.75, while 1,259 km² sits between 0.25 and 0.75. Late-only water covers about 579 km², compared with about 70 km² early-only. The delta has a durable hydraulic core surrounded by a much less settled edge.

  5. N.05

    CHOICES

    I kept epochs, persistence and transitions as separate views because they answer different questions. I also kept the rendering severe. The work is about occupation and change, so scenic colour would add more atmosphere than information.

  6. N.06

    CAVEATS

    These are classified surface-water observations. They do not explain erosion, sediment movement, engineering decisions, water depth or what happened between annual observations. No-data years are excluded from each pixel’s ratio, so the number of observations varies spatially.

  7. N.07

    FILES + SOURCES

    SOURCE MATERIAL

    European Commission Joint Research Centre / Google, Global Surface Water / YearlyHistory, 1984-2024. Components: JRC/GSW1_0/YearlyHistory (1984-2015; version-of-record historical yearly classification); projects/global-surface-water/assets/GSW1_5/YearlyHistory_2016_2021 (2016-2021; v1.5 recomputed YearlyHistory); projects/global-surface-water/assets/GSW1_5/YearlyHistory_2022_2024 (2022-2024; Landsat Collection 2 extension). Classes: 0 = no data, 1 = not water, 2 = seasonal water, 3 = permanent water. Licence: Copernicus Programme; free of charge without restriction of use. Published-map attribution: Source: EC JRC/Google.

    METHOD

    Epoch occupancy = valid yearly classifications with waterClass>=2 / valid yearly classifications. Persistence uses the same ratio over 1984–2024. Transition count is the number of consecutive valid years whose binary water/non-water classification changes.

    TRANSFORMATIONS

    Exported from Earth Engine as one multiband GeoTIFF (sha256 27784ccd614a79e6f72bf521a636b15f7680cbdf1609987b1898ba3e34737046), reprojected to EPSG:32650 at 30 m pixels (7,496,622 valid pixels), processed by scripts/process_p001_export.py on 2026-10-04T01:19:58.501992+00:00. Scope bbox 118.45, 37.45, 119.45, 38.2 (WGS84). Contours: Threshold polygons are generated from a 4× block-mean downsample for tractable static display generation; quantitative statistics use full-resolution pixels.

    MEASURED VALUES

    Persistent water (persistence ≥ 0.75): 3310.3782 km². Intermittent (0.25–0.75): 1258.7859 km². Early-only: 70.3107 km²; late-only: 579.3903 km². Mean transition count over valid pixels: 1.9839370250701904; 95th percentile: 10. Area at occupancy ≥ 0.5 by epoch: 1984-1993 3325.5918 km², 1994-2003 3811.491 km², 2004-2013 4169.8332 km², 2014-2024 4253.8941 km².

    DATA LIMITS

    JRC documents residual Collection-1/Collection-2 co-registration offset, typically sub-pixel but potentially >= one 30 m pixel in some WRS-2 path/rows. Yearly classifications with no data are excluded from each ratio, so pixels differ in how many years inform them. Contour outlines are generalised; statistics use full-resolution pixels.

    SCOPE

    These are classified surface-water observations only. The map does not show erosion, sedimentation, causes, engineering effects or geomorphological process, and it says nothing about water depth or what happened within a year.

    DATA

    stats.json, annotations.json, provenance.json, source.json; rasters: four occupancy epochs, persistence, transitions; contours.geojson; scope.geojson. Attribution: Source: EC JRC/Google.

    CODE

    scripts/p001_export_earth_engine.js (authenticated Earth Engine export) → scripts/process_p001_export.py (all static P001 assets and statistics).

    SOURCES

METHOD
Temporal surface-water representation
SOURCE
Source: EC JRC/Google
PERIOD
1984–2024
GEOGRAPHY
Yellow River Delta
TYPE
TEMPORAL THEMATIC RASTER
NEXT PLATE002 — DISTORT A MAP ON PURPOSE →

§ CONTACT

I AM THE ONE WHO KNOCKS.

JK, feel free to reach out below!

ray@sleepingnook.com LinkedIn

© 2026 RAYMOND AKOLA-EREWUNMI APENA

PHILOSOPHER | DATA DAWG