There are known instances of outdoor transmission, and even some fairly large spread, but every study I've been able to find that attempted to differentiate between indoor and outdoor transmission found that indoor transmission accounted for the overwhelming majority of analyzed cases.
An early preprint in April by Qian et al found that "among our 7,324 identified cases in China with sufficient descriptions, only one outdoor outbreak involving two cases occurred". This refers to a single incidence of transmission (i.e., the two cases were one person who had COVID-19 and transmitted it to the other person).
A peer-reviewed article from June by Leclerc et al reviewed data from various published reports. Their data comprises 6123 cases, of which 5838 (95%) occurred in settings they categorized as "Indoor". 150 occurred in "outdoor" settings and 135 in "indoor/outdoor" settings (about 2% each).
Some of the authors of the latter study have created a project to collect information about transmission events. In July they released a brief note (not a peer-review publication) based on updated information about 616 known clusters, stating (among other things): "Over 96% of the identified clusters (and of all cases associated with clusters) were associated with indoor environments".
Their database appears to have grown since then, containing information about 192,000 cases in about 1500 clusters, of which about 176,000 (91%) were categorized as indoor, with an additional 15,000 or so indoor/outdoor, leaving less than 1,000 (<1%) categorized as outdoor or with no information recorded on this dimension. (Glancing at the data, it looks like they have recategorized some events from the earlier publication, such as some at construction sites, from "outdoor" to "indoor/outdoor".)
As regards "superspreading", the database mentioned above contains only three clusters explicitly identified as outdoors, consisting of 26, 11, and 2 cases. The 26-case event is listed as occurring at "Mountain King Potatoes" in Colorado. It appears to be referring to this event, but that report refers to a "potato warehouse" so the spread may not in fact have occurred outdoors. The 11-case event is listed as "Greenways Landscape Services" but I wasn't able to find specific information about that event. The status of the "indoor/outdoor" events is unclear since some may have been predominantly indoor with a bit of outdoor interaction, while others may have been the reverse, so it's possible that some larger superspreading events were driven, at least in part, by outdoor interaction.
As with many COVID-19 matters it remains somewhat difficult to find straightforward, peer-reviewed research on specific dimensions of transmission, presumably because there has been relatively little time to gather such information. Also, all the existing publications (and their associated datasets) come with many provisos about how they are not comprehensive and are necessarily convenience-based in that they rely on a hodgepodge of sources of varying reliability (e.g., news reports). That said, a group of researchers explicitly trying to create the "superspreading database" mentioned above, to track large-scale transmission events, and having so far accumulated data on nearly 200,000 such cases, have so far identified a grand total of 39 cases which they have categorized as being transmitted in events with no indoor component. Even including all events with any outdoor component (or an unknown indoor/outdoor status) brings the total to less than 10% of cases in the database. This is clearly suggestive of a very limited role for outdoor transmission in COVID-19 spread.
