Great reply,
thanks for adding insight to it. I had in my mind the conception that, by virtue of the metadata or geo-location data, that maybe the SPC might segregate that data into "type of report" columns as opposed to just a report itself. I think when you're asking statisticians to do the work, I assume it's a lot easier to not factor in report type, and just focus on the report itself.
I am curious though and maybe you can fill me in. In terms of total reports, say 1 tornado has 50 observers simultaneously sending separate reports, so thats 50 reports for 1 tornado x how many additional reports are made as time continues and the storm path continues, over say 30min. so those same 50 make an addition 5 reports over 30 min. so now we are up to 300 reports for 1 tornado. Does that skew the annual data at all in some way? or does the SPC just say, reports are reports that's it.
I will preface this by saying that I am not, nor have I ever, been employed by NWS or any non-military government agency (12 years enlisted Army is the extent of my time working for the government), and I'm also not a degreed meteorologist or statistician. So I'm going off of what I believe to be an educated guess as to how this works, but have reason to believe that I am in fact correct.
So, when tornado reports come in, I believe they are logged as they come in (this could also vary by office) by type of hazard (hail, wind, tor, flood, etc). While some may make a comment RE means of reporting, many do not. That's why, when the SPC has their "preliminary" reports at the end of a given day, the numbers almost always go down over the course of the next few days. Those are often referred to as "unfiltered" reports as well, because there's been no consolidation of duplicate reports of the same event. Once it's determined that, for example, 50 reports are in reference to tornado A2 (just giving a storm and tornado designator for the sake of discussion), those reports are all consolidated into one tornado "report" because it was the same tornado. This number is what goes in the numbers that I believe
@adlyons is drawing from. I've noticed in the past, notations of "multiple spotters reported a tornado at this location" in the comment section, but then in the geographical data, it gives the starting time and ending time based on a number of factors. Also gives starting and ending location, which is based on the survey. All that to say that, those 50 reports all count as one tornado report, as do the 250 additional reports over the remainder of the tornado's life in your example Jason. So as far as overall tornado statistics for a year, once it's determined to be a single tornado, that should not skew the overall numbers for a year.
To the point that
@Pierre Ducray made, I'm sure there are some small tornadoes that get caught now that weren't caught years ago. But, for the most part, especially after 1970, I don't think it's a significant enough number that have not been reported and/or have not been surveyed to seriously skew the statistics. Like with the rest of this, I could be mistaken, but I don't believe I am.