Using LiDAR and Fable 5.2 to identify ancestral homesites

LiDAR ancestral homesite map generated
 with Fable 5.2 and USGS

Mark Thompson's was the first post I saw in the Genealogy and Artificial Intelligence Facebook Group announcing the drop of Claude 5.2 The next post I saw was Professor Ethan Mollick's Claude 5.2 test drive and WOW! Read Mollick's whole blog post and watch the mind-blowing animations.

I tried out Fable 5.2 yesterday, but just for some small things. The heavy lifting came today! For the past two years, I have been learning LiDAR (Light Detection and Ranging) which uses a laser to map geographical features. There are many different occupations that utilize LiDAR but where a genealogist could use it is to locate graves and ruined structures. A year ago, my cousin, Brian Evans, mapped the plot of our ancestor's lease from Thomas Lord Fairfax in 1762 Fauquier County, Virginia. Thanks to Brian's map, we were able to view the land plat from the ground at several locations. Today, I fed the map and plat information into Fable 5.2 and asked it to locate potential areas for the homesite based on the USGS LiDAR. Only an on-the-ground survey could verify the findings, but helpful and interesting nonetheless. The USGS site has a notice that Topo Base Map will be retired on November 16, 2026 which may be unfortunate depending on whether it is needed; but it will hopefully be improved upon. So how I did this - there is no magic prompt. If you don't already know where your ancestor's land was, then you need to figure that out first. Fable 5.2 will walk you through every step of the way. When I got stuck on something, I took a photo of the screen and submitted it to Fable and it walked me through the fix. I should also mention that while Fable 5.2 came with a reset, and I was growing nervous I would tokenmax on this project, especially as the minutes ticked by (it took Fable over five minutes to generate three maps). Fortunately, it barely used anything for this project! As I write this, I am 1% of usage. The "A, B, C" near the circles on the map are the anomalies that LiDAR and Fable 5.2 found for a possible homesite. Lastly, the USGS LiDAR is useful for identifying large, man-made structures, but not so much for graves. It doesn't zoom close enough to the ground to pick up graves. More to come on that though. Below is Fable 5.2's comments: 

"I cut out the area around Lot 52A, and drew Brian Evans's four corners on it. Three working maps are attached: lot52A_annotated_hillshade.png, lot52A_contours.png and lot52A_local_relief.png. These are analysis images, not a chapter document.

**The plot checks out against the ground.** Hume Road shows as a clean line crossing the north end, and its junction with Route 730 falls at Evans's NE corner, as his poster says. His placement is sound.

**Water.** Two stream channels cross the northern third of the lot and run west. The upper one drops from about 659 feet near the modern pond to 631 feet where it leaves the northwest line. The lower one drops from about 667 to 634 feet. They join just west of the lot. So the lot had running water in two places, both near the road end. The LiDAR does not name the streams. The Dearing map's "West Br. Thumb Run" label is still the only naming evidence.

**Relief.** Ground inside the lot runs from about 603 to 777 feet. The low ground is along the western side and the streams. A knoll rising to about 800 feet sits on the southeast line, and the lot takes in its western flank. The southern half is open, plowed ground with straight fence lines.

**Features flagged for a ground check.** None of these is identified as a 1762 structure.

A, at 38.83259, -77.97197, elevation 651 feet. Two small sharp features a few meters across, on the bank just north of the upper stream and about 100 meters west of the pond.

B, at 38.83056, -77.97519, elevation 634 feet. A small irregular cluster roughly 10 meters across, on the flat between the two streams, about 60 meters inside the northwest line.

C, at 38.83053, -77.97559. A straight raised bank about 100 meters long beside the lower stream, next to B.

A and B could be stone piles, a spring head, tree throws or ruins. C could be a spoil bank or old lane edge. LiDAR cannot say which. If Brian or anyone else can walk the lot with permission, those coordinates will take a phone straight to each spot."