When I look at the graphs of URL errors for a particular site, is the number in a given day the actual number of errors at that point in time, e.g. is it a snapshot of the number of reported errors or is it the number of errors found on that given day?
As @Simon-Hayter noted in his comment, the number of errors reported is a running aggregate/cumulative tally for the site.
It may sometimes spike/jump if the crawlers try to check the site during webserver downtime/maintenance. Likewise, if you see it trend downwards, it may have detected certain errors as self-corrected.
Possible self-correction measures include:
- Crawling an updated xml sitemap and realizing some urls can be delisted.
- 301 redirects having been setup/added for defunct url aliases/retired pages.
Check out the Google Search Console Crawl Errors Report (websites) page for more info about the types of errors reported and manual steps to flag/notify Google that errors have been corrected.