Smyth says -οῦ, but LSJ says -ᾶ except in the case of κατωφαγᾶς where it lists -οῦ alongside -ᾶ. A Perseus collections search yields neither Βορρᾶ nor Βορροῦ. Does the genitive in -οῦ then occur in any other words?
If you try this link
http://www.lexigram.gr/lex/arch/
and look up βορέας you will find both forms listed for the genitive.
Bore/ou is Attic; Bore/a is Doric; and Bore/w (nom. Bore/hs) is Ionic.
From the LSJ:
Βορέας , ου, ὁ: Aeol. Βορίαις Alc.Supp.7.13; Ion. Βορέης Hom., or Βορῆς , έω, Hdt.7.189; Att. Βορρᾶς , ᾶ, Cratin.207, Th.6.2, al., Pl.Criti.112b, PPetr.1.21.11 (iii B. C.), Ev.Luc.13.29, etc.; but gen. Βορέου IG1^2.373.29
I read this as saying that the Attic version was generally nominative βορρᾶς, genitive βορρᾶ. I looked up the Luke reference, and that is what Westcott and Hort give. I don’t know what went wrong with your Perseus search.
However, from LSJ Supplement (not online):
Βορέας, after ‘Βορρᾶς, ᾶ’ insert ‘(also Βορροῦ Aristonym 7 K.-A.; cf. °Βορρόθεν)’; add ‘see also βορεύς’
I only have the translation of that fragment – Loeb gives it as “of the North” – so the fragment couldn’t have been much than more that one word.
The last part of the LSJ entry above, “but gen. Βορέου IG1^2.373.29” means that there is inscriptional evidence for this other version. The first number is region (I believe). So an Attic inscription but not one I could find online. Here are some recommended Google searches for inscriptions:
site:epigraphy.packhum.org βορρᾶ
site:epigraphy.packhum.org βορέου
βορέου is equivalent to βορροῦ by Smyth 117. Another example of a masculine εας noun would be Ανδρέας, but that’s fairly late. Do you have some others?
The accusative Βορρᾶν occurs at VI.2.5 in Thucydides, but the genitive Βορέου at III.23.5 is generally deleted by modern editors as a gloss. Thucydides also has uncontracted forms of Βορέας.
The word at Critias 112b is the neuter plural adjective πρόσβορρα, cited in LSJ for the double rho spelling, not for the genitive form.
I couldn’t find the Cratinus citation. In the Loeb edition of the Comic Fragments, 207 doesn’t have any form of Βορέας or Βορρᾶς.
I found the Aristonymus fragment here: https://books.google.com/books?id=3sMUAAAAIAAJ&pg=PA488#v=onepage&q&f=false
It’s cited as fragment 8:
Hesych. ‘βορροῦ’· τοῦ βορέου ἡ γενικὴ παρ’ Ἀριστωνύμῳ.
The Loeb that I was looking at (digitally) would be volume 513, pg 140. I couldn’t find Cratinus either. It would be in the same book that I link above, pg. 100 (if anyone here has access to decent library).
It turns out that LSJ’s Cratinus 207 is fragment 222 in the Loeb:
{A.} ἐς Συρίαν δ᾿ ἐνθένδ᾿ ἀφικνῇ μετέωρος ὑπ᾿ αὔρας.
{B.} ἱμάτιον μοχθηρόν, ὅταν βορρᾶς καταπνεύσῃ.
Not quite sure what the question was, but names in –ας (long alpha, 1st decl. masc., ionic –ης) do tend to form their genitives in –α (long alpha),originating as a contraction of the doric (and aeolic) form –αο (long alpha, ionic –ηο becoming–εω by quantitative metathesis). This applies equally to names in –έας contracted to -ᾶς. The –ου forms show leveling to the ordinary attic 1st-decl. masc. gen. form “borrowed” from the 2nd decl.
As to the internal contractions, βορέης > βορῆς (allegedly Herodotean) or βορέας > βορρᾶς (Attic), we can’t put much faith in either manuscripts or editors; they will vary and/or standardize. We need examples guaranteed by the meter. There’s nothing exceptional about βορρᾶς in the Cratinus (presumably LSJ cites it as the earliest extant occurrence), but it’s not metrically guaranteed either; uncontracted βορέας would scan just as well.