My understanding, and correct me if I'm wrong, is that ATAR is a reflection of academic achievement*, while UMAT is an aptitude test mostly (though not completely, in my opinion) unrelated to academic achievement.
Of the two, and as someone who administers these on the regular (including twice today), UMAT is closer in construct to an IQ assessment than ATAR will ever be. S3, in particular, is directly ripped from all the well known IQ assessments and is generally understood to reflect fluid intelligence. Processing speed and working memory components are firmly embedded in all sections, while visuospatial reasoning and verbal comprehension are definitely included in sections 1 and 2.
Add in some luck (most people will bulk guess 10 or so questions at the end), some practice effects (for some), and some psychological fun such as anxiety, and you've got the closest thing to a mass-administered IQ assessment as you're likely to get.
But what's IQ without proof you can actually use it? Especially when it comes to the government first shelling out gazillions of dollars to educate you with a CSP or BMP, then to trust the health of the nation to you. So that's where ATAR comes in. Or GPA. ATAR and GPA rely on intelligence, yes, but also staying power, motivation, teachability (technical term

), long term memory, learning, executive functioning, amongst other things.
Combine the two with a face to face interview to make sure the individual can string two words together with some degree of composure and maturity, and you're on your way to a method for culling thousands of very smart, motivated people down to the sort of numbers that can actually be accommodated.
I'm absolutely not saying it's a perfect system**, but I definitely follow the rationale and logic of it for the Universities that choose to use it (in whatever combination).
* have already admitted elsewhere that I'm too old to have ever received an ATAR so, seriously, correct me if I'm wrong!
(** a perfect system doesn't, and can't, exist, btw)
Finally, noting as an aside, I think it's generally the people with predicted ATARs that are, what?, 90+?, who sit UMAT in the first place. Using words like 'high' and 'low' to describes the two sets of scores in the same sentence can be misleading, as the individuals who comprise the two cohorts are not equal. Even someone with, say 50th %ile UMAT, is at 50%ile of the top 10% of ATAR scorers. I feel like Universities recognise this, which is why a combination of the two is used to fine tune who gets an offer.
Obviously, rurality, ATSI, EAS etc is a whole, slightly different, ball game
/my 2 cents on why I don't mind the ATAR/UMAT combo and why just one or the other doesn't provide the same amount or breadth of information as the two do...