Stop looking at the profile on the site. According to that the finish of the lap is lower than the start of the lap.
Apeanalysis:
3 could even be 4, with GPS analyzer and 2668 waypoints over 13,865 km (and now the missing 235 meters piss me off too) and srtm3 the start is at 223 highest point at 346, so 123 meters in 1,868 Km. Start line: Took a white line on the road that would fit where it should be and makes sense as a finish line from a bicycle race.
9 It's just a 9. Can't see anything that would make it a 10% or more
-9
-6
-2
0
-1
3 500 meters up at 7%, but then 5 meters flat, slightly down.
5 500 meters flat, 0,1, ok, first it goes slightly down, then slightly up. second 500 meters go up at 8,8%. Mainly the last 400 meters at almost 10%. Average for the km 4,5 (well, 4,454%) 5 makes sense.
-7
0 (really goes up 2% then down 2% for ca 500 meters each
2: But here make 1,4 km into 1 km. Since the next climb start almost in the middle of a km, but overall is a 6+% it should be in. Adjust now.
6 1,1 km long, 6,4% first up, then a flatter part, last 500 meters 8%. 7% possible too, but probably 6 is better.
-3 it's really 359 meters at 8,2% down..the other 506 meters were relocated earlier to get the 6 at the km before. then the mystery of the missing meters. Make km 1 4 and make this -4 possible too.
Not that the SRTM thingy here seems to ultra-precise, seemed to have problem getting good readings too, a bit to many saws in the hills for my taste actually. But not as rough as the profile on the Japan cup site definetly is.
In the end, don't care though, it's the Japan Cup. Just stop basing the whole thing on that profile.