 Hello, welcome to another episode of the Art Practical Workshop series. My name is Nico Tripsovich. In this episode, we're going to georeference a CM-born historical fire insurance map and then connect it to census data in the recently released 1950 census. So let's go ahead and get started. So I have done a project here in QGIS where I'm going to georeference these sandborn maps and I started to create a geo package polygon layer here and I'm going to leave it in the EPSG 4326 Actually, let's use pseudomercator 3857, which is the reference system of these web-based satellite imagery services and I created a series of fields here as you can see street name, street address, house ID, which is going to be a combination of street name and street address, a unique ID for the house and first name, last name, gender, race, occupation, born. So these are basically mirroring the fields that are found in the 1950 census. I'm just reproducing these and typing in here Simply add them like that and get another field. So I've created these fields and let's go ahead and create that and now we have this polygon layer and I'm going to show you how I do reference these from sandborn. So now the sandborn maps prior to 19 25 or so are available on the Library of Congress in full color. These are fire insurance maps So the best way to find your study area is to use this index map which occurs at the beginning of each group at the Library of Congress and you find each block has a number and then there's a track or region that has a bolder number. So I would first find your I believe this is the map sheet, the page number in the map sheet and then in bold there and then you would find the actual block, city block in smaller number and I've gone ahead and found it already to save time. Here we are and this is the Peralta Acienda area, Peralta Creek and these these houses have been removed to create a park but they still exist here on the sandborn maps from 1928 so I'll go ahead and digitize the footprints off of these and then we can connect them to the census data from that area. These houses were removed in the 1980s. So the first thing I'll do is clip out the this segment of houses. One thing I would note is that you can see a curve here in the line. So there's some distortion introduced by the scanning process. These parcel lines are sometimes available from the county and one could georeference to the county parcel lines. However, sandborn, well first of all you may have error in the parcel lines in the projection as compared with the satellite imagery that's very high resolution and I've noticed that here in Alameda County and also when you're using a historical map source like this, consider the intent or the focus of the map maker and in this case it's the homes. So the parcel lines are somewhat secondary. I would be cautious about using the parcel lines as referencing layer. Finally, it's recommended that you not include non-priority data in your referencing. So for example, I wouldn't georeference this entire area in one go. I would reference this area like the city blocks. If we were focused on here, I would reference this area as a single unit and this area as a separate unit because sometimes how streets are compressed in order to squeeze blocks onto a single map sheet in sandborn. So the street may not be faithfully represented in terms of its width. So if you include the street in your referencing, you may introduce error. So here we have this area that we're interested in. We could be zoomed quite far out and again, I'm going to skip this street, right? I'm going to reference these houses in one go using the clip tool and we can clip it. I would include the street names, this sort of ancillary stuff and maybe the opposite corner but it's okay to include this extra information on the edges. Just don't use it to reference. So then I'm going to click the clip image. There it is and look, I can actually see how it's barely legible. If I increase the image size, it actually goes to the original and now we have a really high-res image and I can download this image. I click that and I'm going to right click on Windows here and save it. I'll just call it sandborn. I would name it more specifically for my project. I might call it the street name and range of addresses shown. So the next step is to georeference the sandborn maps using either the county parcel boundary map or the imagery services such as here I've got quick map services loaded with high-resolution imagery. As I mentioned, the parcel maps are a possibility because they appear on sandborns. However, sometimes the parcels are not accurately shown on the sandborns and the parcels from the county may not georeference either. So for example, here I've added the Alameda County parcel layer. I can show you how that was added using this ArcGIS REST server layer. And if we look here, it's Alameda County parcels. Here's the address and please note I clicked this when I created that layer, only request overlapping features overlapping the current view extents. So I zoomed back and I had that box checked and the result is this orange layer and notice how it ends here because it I wasn't zoomed out that far. And if you don't check that box, it can take a long time to look because it's the entire county. The other issue is look how it doesn't line up very well with the imagery behind. Based on some testing, I've determined the imagery is much closer than the county parcel boundaries. So I won't be using that, but I wanted to show you that that is a possible option since they do appear on sandborn and it would be really nice to just click on these with the snapping turned on. You can conceivably just click right on those corners. So we will be using this imagery and we'll be georeferencing using the georeference for tool that this is a QGIS version 3.26 and it is moved now to the layer menu from the raster menu. And I downloaded a sandborn, the one slightly west of what I was showing you earlier, I think we'll make a better candidate for this example and bringing it in that block here is this block in south. We'll be bringing that in using this imagery. And so one question is which of these houses are still there? For example, this area has been these buildings were removed in the expansion of the park and these buildings look much bigger than the houses shown on the sandborn. However, these little four houses do appear. Another trick is to go to Google street maps and here we are in the park and we can just kind of look around and look at the structures. And you can see here are those four houses and they have the appearance of houses built in 1920s as does this one. But these look newer. So these are those large houses we can see in the imagery that are probably built in the 50s or 60s. I also quickly showed you this tab. This is where I got the services layer that I showed you for parcel boundaries. So let's use the services layer shown. That's not using services layer. Let's use the imagery layer that I showed you. And I checked street view and I determined that those little four houses are in fact the same houses. So we'll go ahead and start to your referencing. I'm just going to do a quick job here so that there isn't too much time spent on this. The basic idea is that we are going to put some points around the map that correspond and keep in mind these houses have eaves generally. So I would move in a little ways from the very corner. We're assuming that the sandborn doesn't include the eaves. So let's just come in a little bit from the edge there. Sometimes you can find roof examples that don't have the eaves that have smaller eaves. Scroll here. So I'm going to put two points up there. Perhaps a point in the middle. Let's see. This house looks like it was here. Let's count houses. One, two, three, four, five, six. One, two, three, four, five, six. That's this house. This is this last house here. And then everything south of that was replaced. There have been some changes in the designs. You can see many of these houses were expanded. So it's probably best to find a house that looks like it hasn't been expanded and find the original corner. Let's go ahead and maybe drop a point. That was the wrong spot. I'll demonstrate how to move point. You just use this structure here. You can adjust your placement with that tool. Continuing, these were removed here. But this house, from personal experience, is one of the older houses in the area. So I'm going to use that one as a georeferencing point. Now, this corner is covered in a tree. But this corner is pretty visible. There's the deck. I have a large eave, as I recall. You can check street view if you need to. This one, I'll just go ahead and move this. There's a new point here. And then a point here, tracking the eave. All right. And then this structure is the original Peralta Hacienda. I know that that's original. Now, this Paxton Street was closed when they created the park. There it is. Also, yeah. So here's the Peralta House. It's a very large eave. This deck is a good candidate. The corner looks a little different and tricky. So I'm going to move this southern end here. So you can see it pretty well. But there seems to be a little offset there. So maybe I'll use the northern one because there's a glitch in the sandborne there. I'll use this corner. Okay. So I've got four, five points here. Now, that's probably enough. Note that this sandborne goes further south. But our focus is on this area. So you should concentrate your ground control points in the area distributed around the area of interest and not try to do the entire original map. In fact, it probably would have been better to clip the, when I was downloading this, it probably would have been better to clip it here and not attempt to download all of it because it implies that this is all georeference. And while it'll probably be reasonably well-referenced, focus is in this area. So let's go ahead and check the parameters. We're going to do a polynomial one. We're going into the Web Mercator coordinate reference system. I'm going to shorten those. We don't want too long final names. And let's just check what it's been saved. Go into the field school folder. And uncheck these. If you have a black collar around your original, a good idea to use zero for transparency to remove that collar. Although you can do it at a later step as well. I always save the ground control points. It saves them out into a little text file. And you can, they travel with the tip that you're saving out. And you can load them and make adjustments like I did earlier here. If you find that some of your residuals aren't very good. Before clicking the start georeferencing, I'm going to sort by the residuals column and look for the biggest error. Let's see which one that is the number three. So one thing you can do is sort of iterative process. We can select it, find it. It's going to be number three would be this one. You can turn it on and off. Oh, it's this one. And you can adjust it. If you think it could do a better job, you might consider sorting by residual and actually right click and removing or adjusting with this tool, this tool, the worst residual. So like you can uncheck it and see how it quickly adjusts formula. It comes up with the correction georeferencing solution without using that 19 number three. So it's a compromise between all the enabled ground control points. All right, let's just go ahead and use this one. There it is. Now we did get the black collar because I didn't specify. We can change it and add the georeferencing and rerun it. I'll have to remove it here. It doesn't have a conflict overriding. There you can see by checking that box, we no longer have that black collar. So that's a good idea. The trade off is you look at all the black pixels. For example, here are now transparent. So it's really a trade off. You get this sort of funny shimmery look on the black pixels when you make them transparent. All right. So this step is now complete. You can go ahead and georeference. I'm just going to do one house in this case. You can see that this one 2511 is now this park. Let's go ahead and georeference it. Maybe I'll change the appearance. Well, first I'm going to call it something different. I'll just go ahead and start georeferencing. You decide how much detail put in here. Georeferencing. Now you can see all those fields from the census that I created earlier are now visible. We know the street is 34th in the fourth app and the street address is 2511. There's a garage as well, which may be very interesting for archaeologists where the outbuildings are. Sometimes the privies are visible in these maps. All right. So let's go ahead and save those edits and turn off sandborn. We have that structure. Now you can see today it's a park and this is sort of a pavilion space associated with the park. Now continuing with the census here. The 1950 census was recently released because it has been 72 years and they are available open access here at thearchives.gov. One issue with using these is that you need to know the, it's a big help to know the enumeration district of the census that you're working in. Another option for many people have access to the university libraries to Ancestry Library and the Ancestry Library, one of the biggest advantages I found to these is that they do have this explore maps tool. So you click that and it will take you to, they've actually digitized and georeferenced the original enumeration district map. You can see an orange back there. And if you, so you can basically do a spatial search instead of having to know numerically that this is 67-152. There's the, there's the creek. There's the current park. You can just click it, drop a place mark like that and then there you go. Boom. It takes you to the census. And you page through here and I've already found, I paged through and I found that some pages into it. Here is 34th street, number 2511 and it's Mike Lewandowski, and Gucci, head of household, white male, 15 years old, born in Italy. And that was about his employment. Looks like he's a grocery store employee, proprietor. So I'll go ahead and type that in. It's editable. So now one thing you might consider is giving individual house IDs to the, to the structure. If you're really paying attention to your structures and different sources of data, it might make sense to come up with your own ID numbers for each of these houses that may differ from 25, from the street address like 2511 because these, like for example, this house is no longer standing. So you can have your own archaeological records. So now, right, grocery, born, so 1950, he was 50. So let's just say he was 1,900. There you go. And you want to perhaps label by his name. Or you might do, you could do a choropleth map showing other fields like occupation, race, age, and generalize about the historical character of this neighborhood. That is an example I wanted to provide, incorporating sandborne and census data into your GIS.