I probably wouldn't call it fraud (others might, though) as much as I would call it scientific incompetence in general (it appears that many "scientists" these days, across a number of different fields, are actually quite bad at doing science; this probably isn't a new phenomenon, though, there's just a lot more of it these days), mixed with generally crap data, mixed with bad statistical methods (there's been a lot of talk about problems with this lately), mixed with an overwhelming need to get your papers published
by doing whatever it takes.
Having read scientific journals myself on and off for several decades now, one of the things I've noticed is that once a paper gets published in a reputable peer-reviewed journal, then that opens the floodgates for other related papers to get published using related data and related methods. And that's true even if the original paper used highly questionable data and highly questionable methods, as so many do these days.
The reality of scientific publishing today is that scientists need something - anything - of theirs to get published in order to build and maintain their professional reputations. And journals need something - anything - to publish; usually something provocative, too, otherwise nobody would bother paying the high fees that they charge. So a lot of what they publish is just their version of clickbait.
Funny you should mention Japan: Back when I did my deep dive, one of the cities in Japan (Tokyo, I think it was) was being held up as a worst case example for warming. And if you looked at the raw temperature data, the warming there was kind of scary enough as it was, but in the adjusted data it was just horrific. And I thought to myself that one of two things was going on here: Either the level of warming that was being claimed to have occurred there didn't actually occur, or that it puts lie to the notion that humans and flora and fauna can't readily adapt to such warming. After all, it's not like Tokyo is an apocalyptic dead zone or anything, now is it?
As to the raw data itself, the last time I checked the raw HadCRUT3 data that I used in my deep dive was no longer available (I kept running into broken links and such), and this data had only been provided under duress in the first place. I didn't really strain any muscles trying to look for it again, though. Nor did I originally have much luck trying to find similar data for HadCRUT4 (again I didn't look too hard), but I have seen passing references to it being out there somewhere.
And by "raw" I mean the data as it originally came in from the various temperature stations, without being manipulated in any way except maybe to get it all in a common format for easy processing. But apparently what I call "raw" and what some other folks call "raw" can be quite different things.
And unlike your situation, when I looked at the HadCRUT3 data (at least the version that existed at the time; I know they made some changes to it afterwards) what I found there was appalling. So either massive fraud was going on at the time, or (more likely) they had just allowed their computer algorithms to run amok on it without really quality checking the final results.
As to stations "not changing during that time period", I forget the details but you should be aware that some folks (not me) have gone so far as to track down a few such stations (those which were well-documented and well-maintained, but with no documented moves or changes), only to find that the algorithms had made adjustments to them anyway! As for the local station that I used, the adjustments were such that it made it look like this station had shifted from condition A to condition B (that it had moved or whatever), then shifted back to condition A, then back to condition B, and so on, and that it shifted by exactly the same amount every time, too. Then rinse and repeat, every few years, which is hardly a realistic scenario.
BTW, all of those changes were warming changes, too. There was never a cooling change that I saw, at least not for my local station nor the handful of other stations that I also checked. The most that I saw was those periodic shifts back from the "adjusted" (warmer) data to the "raw" (cooler) data.
I did that deep dive about ten years ago, BTW, and the local data set I used went back to about 1870 or so - a solid 130+ years of data. I know a lot has changed in those ten years, but I don't know that much has really changed for the better concerning the data or the algorithms being used to process it.