2021-11-05: I recently got the idea to show heat/value maps of map bytes. It will show 2d spatial shapes much better. Going through the TILEs looking for interesting patterns.
Offsets in this doc are from the first byte after the first TILE as 0x00.
0x04 - Rivers Definite clumping pattern on land, most likely terrain related.
0x05 - Owned tile / culture borders (player number)
0x06 - all 179s? Middle gray / middle value 0x07 - all 9s
Possibly an int32 here 0x08 - Resource graphics tile id, including volcanoes. Spotted differing numbers for volcanoes appearing differently 0x09 - Also associated with resources, but many zero out compared to 0x08. (Mayve all zeroes? Need to check.) 10 - Also resource-associated, but all zeroes? 11 - Also resource-associated, but all zeroes?
12 - Unit ID, I think. Almost certainly unit-correlated. Need to check to see if they are all unique 13 14 15 - ^ part of a unit ID
16 - Base terrain graphics tile index - This looks really funky in the value-as-texture view 17 - Base terrain graphics file ID
18 through 23 - All zeroes? (Possibly pre-Conquests data?)
24 - all 255 - From other notes, believed to be barb info, perhaps tribe ID 25 - all 255
26 - City ID 27 - All zeroes? Maybe part of city ID, but I think there may be 256 limit on cities
28 - all 255 - Believed to be Colony ID from other notes? 29 - all 255
30 - Continent ID 31 - All zeroes?
32 - All 6s
33 - Secret code? This seems to be zeroes except for horizontal data near the top of map. It's what I think it would look like if I stuffed data into a save with my idea for C3X. It doesn't really look like textual or structured data, or any reasonable collection of numbers, though. Not even particularly aligned in any way.
34 - 255s 35 - 255s
36 - Razed city! All 0s, but found a 1, and it's ruins in-game. Not seeing other values, so can rule out pollution. More 1s found for razed cities.
37 - 0s 38 - 0s 39 - 0s - possible msb of razed city int32? Doesn't really make sense
40 through 43 - "TILE" 44 through 47 = 12 (section length)
48 - Improvements flag (includes pollution which is 0x40 flag)
49 - 0s 50 - 0s 51 - 0s 52 - 0s
53 - Terrain (2 nybbles, ID referencdes to BIC)
54 - 0s 55 - 0s 56 - 0s 57 - 0s
58 - appears to be bit flag for start locations, bonus grasslands, forests chopped, etc. Bit flat 0x02 may be tile workable / available? See "Civ3C-SaveFileNotes" for more 59 - more bit flags; in fact I think the chopped one is 0x10 here.
60 through 63 - "TILE" 64 through 67 - 4 (32-bit length)
68 through 71 - 0s
72 through 75 - "TILE" 76 through 29 - 128 (32-bit length)
80 through 83 - Tile visible bitmask per other notes 84 through 87 - Tile visible to units bitmask 88 through 91 - colony visibility per other notes 92 through 95 - culture visibility
97 - All odd numbers! Interesting clusters, but can't find any real correlation yet. Maybe a packed terrain graphic indicator? 1 for base terrain? High numbers seem to be grouped with mountains and hills, but no tile-for-tile correlation. Hills can be 255 or 249, but not exclusively. Most sea and ocean is 255 but it varies. Try to find an even number; try to correlate with terrain.
98 - Terrain-related. Seems to vary where terrain types meet.
99 - 0 or 1. Speculated in other notes it may be part of previous bit flag.
100 - City ID of city working this tile (including city center tile) 101 - 0s - part of City ID?
(See other notes for Tile128 section from its 0x16 through 0x75) The trade network stuff must be close to right. This is fun to see the patterns in a large game with wars in progress. Parts of the world appear and disappear while scrolling through.
198 - This looks like serialized data in map tiles, again like I thought of trying. It looks a bit more like byte-aligned data. 199 - More serialized-data-lookin' stuff
208 - Start of next "TILE"s section