Etymology

Most likely continues the relatively widespread OE compound uppon, but it has been argued that it arose under influence from ON upp á (OED followed by ODEE) due to the date it first appears and the locality of the texts in which it is first prominent. ON influence might also account for the position of the stress on the preposition (as in PDE), though it is impossible to demonstrate where the accent fell in OE or ME forms with second element spelt <-on>, and the distribution of such forms is not straightforwardly N and E. 

PGmc Ancestor

*upp + *ana

Proposed ON Etymon (OIcel representative)

upp á 'upon, on'
(ONP upp (adv.) (Cl-V upp (sense IV)); á (3) (præp.))

Other Scandinavian Reflexes

Norw paa, Dan paa, MSw up a, op a, uppa, oppa, etc.; Sw. på

OE Cognate

uppon, uppan 'on, upon, up to, against

Phonological and morphological markers

Summary category

CCC4c

Attestation

The earliest texts in MED to prefer <-on> forms exclusively (as opposed to <-en> forms showing the accent on the first syllable and reduction of the vowel in original on  or <-on> spellings in the same mss which could potentially represent forms with the same accent) are a1121 Peterb.Chron.(LdMisc 636), and Orrm, but these are followed by a texts in a range of dial in 13c. and subsequent attestations.

Occurrences in the Gersum Corpus

Gaw 9, 37, 47, etc.; Cl 141, 268, 416, etc.; Pat 12, 134, 138, etc.; Erk 76, 92, 125, etc.; WA 31, 42, 158, etc.

Bibliography

MED upon (prep.) , upon (adv.) , OED upon (prep.) , HTOED , Dance vpon, de Vries upp; á (4), Mag. upp (1); á (3), Bj-L. på, Torp NnEO paa, Torp-Falk paa, Hellquist på, Nielsen på, Orel *uppa; *ana(i), Kroonen *eup ~ *upp; *ana, AEW upp; an, on (1)