Identify clusters of same dates in table

04-13-2024 02:17 PM
Labels (2)
New Contributor III

Hi: I've got a table of several thousand records over ~100 years (California fire history). I'd like to identify clusters of fires that occur on the same date (ALARM_DATE), hoping to eventually identify major wind events driving fires in different locations on the same day.

I tried Find Identical but that seems to look for identical records. I need to find clusters within the attribute table with the same event dates but the records will be different.



0 Kudos
4 Replies
MVP Esteemed Contributor

Select Layer By Attribute (Data Management)—ArcGIS Pro | Documentation

will allow for the selection by a date or a date range,


Split By Attributes (Analysis)—ArcGIS Pro | Documentation

would enable you to do the same, but separate features in to separate featureclasses

alternates  Space-time cluster analysis—ArcGIS Pro | Documentation


... sort of retired...
New Contributor III

Dan: Thanks. I think the cluster analysis is the most promising. It'll be a couple of days before I can start tinkering with it.

Unless I'm missing something (and that's always a possibility!), I think selecting by attribute and creating a layer would give an overlap of the same date across any year where I want a unique date & year.

Anyway, I'll post if your solution is right later this week (I hope..).

Thanks again!

also sort of retired

0 Kudos
Occasional Contributor III

1) Are you still searching for workflows?

2) Once you have clustered your fire events, how do you plan on analyzing it?

- Zach

0 Kudos
New Contributor III


Hi. I just replied to Dan that I've not yet tried the cluster analysis but will later this week so, short answer, I'm not sure if I have a solution or workflow.

So, probably TMI about stuff you might already know but:
The historical fire record has fire polygons since 1900 or so. What we're looking for are clusters of fires driven by Foehn type winds in the Sierra Nevada mountains of California. I'm thinking the clusters might be a proxy for Foehn events.That information isn't captured in the historical data set (though there are a few papers on recent wind events).

From that subset it's then a little easier to narrow a search of newspapers and the USFS archive on those dates or years to see if there were high winds or the fires were just coincidentally on the same day (not uncommon, of course) but just not driven by high winds.

I'm also looking at the shape of the fires. Long and narrow perimeters might indicate a fast moving wind-driven fire. There's a bunch of those around the Tahoe area but, as you go south, they're more amorphous blobs. This is in keeping with, apparently, fewer Foehn wind events occurring south as the Sierra rises in height. Also, many of the pre ~1970s records often don't have a start date only a year, so shape could be a proxy for wind.

One of our crew of retired GIS geeks has written a paper looking at fire insurance cancellation by county, forcing people to use the Fair Plan. The greatest threats of catastrophic loss (for both insurance companies and, of course, residents) is from these wind driven fires but cancellations don't seem proportional to that risk by county. As above, risk seems to decrease as you head farther south so we're trying to find other events in the historical record to confirm this. For instance, there was a major set of fires in 1936, most of which aren't mapped. Newspaper newspaper accounts describe a number of fires on the same day from Tahoe south to Mariposa as well as a couple in the coast range.

Anyway, I'm kinda flailing here but  always open to thoughts & suggestions!



0 Kudos