Peak ore distribution with triangular distributions is one or two blocks lower than intended

When ores generate, the ores with a triangular distribution have a peak that is one block lower than the distribution implied by the chart here: https://twitter.com/henrikkniberg/status/1364635258260455425/photo/1

I generated ore distributions in an Amplified world with seed 1 over an area of 262,144 chunks and plotted the distributions. I expected peaks of ore distributions to be at these Y levels:
Coal: 95
Iron: 31 and 255
Copper: 47
Gold: -17
Lapis: 0
(Redstone, and diamond excluded because their peaks are at the bottom of the world. Emerald excluded because its peak is high in the world.)

What I actually found:

Coal
91:1.7729%
92:1.7859%
93:1.8131% <Highest
94:1.8095%
95:1.7866% <Expected highest
96:1.7799%
97:1.7628%
98:1.7376%

Iron:
28:0.5401%
29:0.5502%
30:0.5535% <Highest
31:0.5434% <Expected highest
32:0.5300%
33:0.5174%
34:0.5055%
35:0.4953%

(The following is not reliable due to insufficient data)
251:0.8123%
252:0.8420% <Highest
253:0.8404%
254:0.8279%
255:0.8373% <Expected highest
256:0.8050%
257:0.8153%
258:0.7685%
259:0.7547%

Copper:
43:0.6514%
44:0.6691%
45:0.6794%
46:0.6798% <Highest
47:0.6686% <Expected highest
48:0.6511%
49:0.6359%
50:0.6214%
51:0.6130%

Gold:
-21:0.3102%
-20:0.3189%
-19:0.3240% <Highest
-18:0.3221%
-17:0.3182% <Expected highest
-16:0.3113%
-15:0.3030%
-14:0.2956%
-13:0.2909%

Lapis:
-5:0.0742%
-4:0.0757%
-3:0.0769%
-2:0.0773% <Highest
-1:0.0755% <Expected highest
0:0.0736%
1:0.0715%
2:0.0696%
3:0.0687%
4:0.0660%