Methodology and Sources

The Atlas of Drowned Towns is a digital public history project that brings together the humanities, social sciences, and natural sciences to understand the processes and experiences of displacement and disappearance caused by the construction of dams. The project draws on “traditional” archival research at a variety of levels (for example, in the United States, from the National Archives and Records Administration to local historical societies), oral histories and interviews, and community outreach and archiving events.

How do you define a “drowned town”?

“Towns” is a bit of a misnomer — we’re actually interested in any kind of home that was moved or eliminated as part of the process of building a dam. That’s anything from an incorporated city or defined village to individual farms and ranches or areas occupied, used, and valued by someone or some community — for example, an Indigenous community. Generally, we include only those places that were populated at the time of dam construction. For example, we would not (usually) include an old mining village that had already become a depopulated “ghost town” by the time a dam was built.

What about places destroyed by dam failures, floods, or other kinds of drowning?

These are fascinating, important, and tragic stories, certainly — but not the subject of the Atlas of Drowned Towns. We are interested in deliberate and planned (however poorly) acts of displacement, inundation, and disappearance to make way for large dams and their reservoirs.

How do you find these “drowned towns”?

Great question! Most importantly: you! We get great tips and suggestions through our Share Your Story page and through email at drownedtowns@boisestate.edu. Once we get a tip about a drowned town, we check whether it fits our definition (see above), try to find out as much as we can about the place, and add it to the database.

Other scholars and researchers into the topic — see our Learn More page (and let us know who’s missing)

Various websites, newspaper articles, and other sources of information, like this, this, and this.

We also query various search engines, Wikipedia, and various LLMs like ChatGPT — and, of course, we check all of those (sometimes illegitimate and even outright false) tips.

Where does the quantitative data come from?

Information about dams (such as acre-feet of storage, dates of construction, etc.): US Army Corps of Engineers’ National Inventory of Dams.

Information about communities (such as population-at-time-of-inundation, cost of displacement, etc.): usually a combination of information from local historical societies, real estate records from dam-building agencies, real estate transaction records at the county level, memories and stories from displaced people and their descendants, newspaper articles and other fragments in various archives, and census information (see below for that process).

Population and demographic information: we currently use a limited, imprecise, painstaking, and enormously time-consuming process: (1) Determine the geographical boundaries of census tracts, enumeration districts, etc. that encompass the geographic unit in question (such as river basin or a place) (2) Access and download the relevant tables and data from the (thankfully digitized) records of the US Census Bureau through the US National Archives and Records Administration (3) Transcribe, aggregate, and calculate that data, using spreadsheets with multiple tabs representing different components of the geographical unit. For example, to get data for the Willamette Basin in 1930, we compare the river basin outline to a political map from that year, which results in Benton, Clackamas, Lane, Linn, Marion, Multnomah, Polk, Washington, and Yamhill counties — obviously, these political/census outlines do not perfectly match the geographic outlines of the region. Then we downloaded census data for Oregon from 1930 and transcribed the relevant data. Then we get the total population for relevant states and counties and the population of cities within those states and counties. Then we aggregated and calculated: Total population is equal to the combined populations of all the relevant states and counties; Urban population is equal to the combined populations of all cities within the relevant states and counties; Rural population is equal to the total population minus the urban population. Whew! Again: imprecise, limited, painstaking, and time-consuming. We’re working to improve the process and the data, and if you have suggestions, please (please!) email us.

What software do you use?

The backend database and frontend platform are custom and proprietary systems developed by a computer scientist at Boise State University. The website also uses WordPress plugins to populate some pages, including the Share Your Story submission page, the Learn More page, and the page you’re reading now. For geolocating and georeferencing artifacts (and georectifying historical maps and imagery), we use ArcGIS Pro. If you want to know more about the technical details, email us.