FullBasin figures and tables for ESSD (essd-2026-402): changes
================================================================

This folder (v1.1/) was put online on 28 October 2026 and is linked
from the author comments in the public discussion. On 1 November 2026
five points were corrected here, in place, so that the page agrees with
the revised manuscript; the folder keeps its name and address. None of
these changes alters a basin, a region or a pixel of FullBasin or
FullTopo. In the input data of the scripts (fullbasin-figure-data.zip)
only Data/region_fine_global_v11.csv changes, in the two rows of the
antimeridian regions (see Fig. 3 below).

1 November 2026
- Fig. 7, image: Scripts/Fig07_BoundaryWalk_v11.py marked boundary
  pixels with a four-neighbor test (function boundary_4: a pixel of the
  catchment is on the boundary if its upper, lower, left or right
  neighbor is outside). It now uses the eight-neighbor test of the
  caption and of the production code
  (main_merit90m_flowmorph_step3_boundary_walk.c): the four diagonal
  neighbors are checked as well (function boundary_8). Nothing else in
  the script or its input data changes. ring(T1) grows from 592 to 842
  pixels and ring(T2) from 882 to 1268; the four panels look almost the
  same (0.26% of the image pixels differ). Replaced:
  Figures/Figure_07.png, Figures/Figure_07.pdf,
  Scripts/Fig07_BoundaryWalk_v11.py.
- Fig. 7, caption: "boundaries follow 4-connected pixel edges" is
  removed, and "the interval test on its neighbors, one comparison per
  candidate" now reads "the eight-neighbor interval test, two
  comparisons per candidate", as in Eq. (4) and Table 5 (R1-7).
- Fig. 5(f), caption: the largest-tributary chains were described as
  "the maximal runs of consecutive dfs_in values"; they are now "each
  occupying consecutive dfs_in values (Eq. (6))". A chain always
  occupies consecutive numbers, but a run of consecutive numbers need
  not be a chain (R1-7).
- Fig. 6, caption: the outlet is at 117.91 degrees E (117.9058 in
  ex8595_record.json), not 117.90 degrees E.
- Fig. 3(b), table and caption: the bounding boxes of the two
  antimeridian regions were written east across 180 degrees: region
  10001 from column 426,000 to 435,600 (175 to 183 degrees), above the
  grid width of 432,000, and region 10002 from 430,800 to 433,200 (179
  to 181 degrees). They are now written west of the antimeridian, as in
  the submitted manuscript, in Fig. 2 and in the FullBasin v1.1 tables
  (rewritten the same way on 1 November 2026): region 10001 from column
  -6,000 to 3,600 (-185 to -177 degrees) and region 10002 from -1,200
  to 1,200 (-181 to -179 degrees). The boxes and their pixels are the
  same; column c of the global grid is c mod 432,000. Only the two
  table rows of the figure change; the map does not. The caption now
  reads "the negative region_col_min (-6,000) reflects the bounding box
  extending west of the antimeridian". Replaced: Figures/Figure_03.png,
  Figures/Figure_03.pdf, and Data/region_fine_global_v11.csv inside
  fullbasin-figure-data.zip (its other 38 files are unchanged).
- index.html, README.txt and SHA256SUMS.txt are updated for these
  changes, and this file is added.

2 November 2026
- Table 3, the column of at least 0.05 km2: 2,569,927 now reads
  2,569,914 (all outlets) and 2,560,220 now reads 2,560,207 (exorheic).
  The old counts came from the areas of FullBasin v1.0, written to at
  most six significant digits and six decimals, which rounded 13 basins
  just below 0.05 km2 up to 0.05. FullBasin v1.1 keeps the areas at full
  precision, so these 13 basins are not counted. No basin changes.
  Replaced: Tables/Table_03.csv.
- Fig. 2, caption: for the same reason, 13,129,304 now reads 13,129,317.
- Table 2, note: "All basin-level tables share three linking keys" now
  reads "The BAT and its per-region subsets share three linking keys";
  the polygon layers and the group tables do not carry all three.
- index.html, README.txt and SHA256SUMS.txt are updated for these
  changes.
- Fig. 3, caption: added "The tiles of regions 10001 and 10002 are
  georeferenced 360 degrees east of their RAT longitudes."
- Fig. 7, caption: "ring(T1) and ring(T2)" now reads "Boundary(T1) and
  Boundary(T2)". The figure itself does not change.
- Table 2, note: added the conversion of seq_dfs to dfs_in
  (dfs_in = a + b - seq_dfs), as in the note of the manuscript.
- No figure, table or script changes in these three corrections.

Remaining differences from the revised manuscript
- The captions here were written separately and differ in wording;
  they agree in content with the manuscript.
- Table 1: the manuscript spells "labelled" as "labeled" and writes
  "Eq. 2" as "Eq. (2)".
- Table 2: the manuscript adds two columns, File Size and Compressed
  (R1-9). Some headings and variable names are shorter there; for
  example, the seq_dfs row reads "Reversed depth-first order". Both
  notes give the conversion of seq_dfs to dfs_in; Fig. A3 shows it.
- Tables 3 and 4: same content; in the manuscript the formulas of
  Table 4 are typeset as equations, and the perimeter formula no longer
  carries the pointer "(Appendix B)".
- Table 5: same rows and costs, in shorter wording; the cost of the
  upstream test adds "same region".
- Table 6: the manuscript gives the eight statistics of each attribute,
  rounded. Table_06.csv keeps them at full precision, with the pixel
  counts and the positions of the minimum and maximum. The rounded
  values agree with the CSV.

seq_dfs conversion
- This page and the Table 2 note and Fig. A3 of the manuscript read
  seq_dfs back to dfs_in within each basin (dfs_in = a + b - seq_dfs),
  as do the
  FullTopo v1.1 README (updated on 31 October 2026, same DOI) and
  FlowTopo 1.1.
