2014 UPDATE: Here is a new graphic that AccuWeather produced about the Pittsburgh flood, which was occurring at the same day, from the same storm system:

2009 Blog: Today is the 73-year anniversary of the Johnstown, Pennsylvania Saint Patrick's Day Flood. In 2007 I published a lengthy blog entry on what caused the Johnstown Floods. I mainly concentrated on the most famous Flood of 1889 and it's always bothered me that I didn't quantify how much rain and how much snow pack led to the 1936 flood. Today, I bring you that information. To recap what I said before...

The Killer Flood of March 17, 1936
The cause of the 1936 flood (24 killed, $43m damage) seems fairly straightforward -- the Johnstown Flood Museum website says that "On March 17, 1936, Johnstown experienced a devastating flood caused by heavy runoff from melting snow and three days of rain.". Indeed, a look at the daily weather map shows that the 1936 flood was caused by a strong late-winter storm below 29.30" pressure over the Carolinas, spreading 50-degree temperatures and heavy rain behind a warm front into Central Pennsylvania, causing rapid melting of snow, while heavy rain fell.

SURFACE MAP 8 AM 3/17/36 |ENLARGE | WHOLE U.S.
- Commentary Video by Johnstown Resident BeckyMarie on 1977 & 1936 Floods
- Photos of the 1936 Johnstown Flood
- WikiPedia: 1936 Pittsburgh Flood
- Google Newspaper Archives on 1936 Flood

OK, So How Much Rain Fell?

After a lot of research pulling me in a lot of different directions, and nothing useful from Google, I finally found a reference to the name of a USGS report (the contents of which was unGoogle-able). The USGS maintains electronic copies of reports from a hundred years ago, so I finally found the actual paper on their site. Here's a screencap from one of their PDFs:

This map shows that over 7 inches of rain fell in the area over a 4-day period (admittedly two of those being after the Johnstown flood, but most of that was probably north and east as the storm moved up the coast), with a widespread region of more than 5 inches. You may recall that "the big one" in 1899 had slightly higher amounts - a maximum of 7.9 inches in the immediate area, with over 10 inches in north-central PA. But 1936 had two other things going for it, flood-wise. One was additional rainfall between the 9th and the 22nd which added up to close-to-1899 values (see this map).

The other thing was snow melt. According to this map, between 4 and 10 water-equivalent inches (which would have melted from roughly 40-100 inches of snow, if you assume a 10:1 ratio, but you probably shouldn't, due to the packing of the snow, see Aaron's Comment below). We can't be sure how much of this melted, of course, but temperatures in the 50s with heavy rain probably got rid of most or all of it -- clearly that would have greatly exacerbated the problem.

All in all, the USGS estimates: "The total quantity of water that had to be disposed of in these ways ranged between 10 and 30 inches in depth over much of the region."