On the evening of November 1, the San Juan River at Pagosa Springs, Colorado, rose to 16.81 feet on the federal stream gauge in town. Nine days earlier, a storm that washed out roads, damaged homes and set off water rescues across southwest Colorado had pushed the same gauge to 15.02 feet. The National Weather Service counts anything above 12.6 feet there as major flooding. The second flood beat the first by almost two feet.
And the second flood came from the smaller storm. Colorado's state climatologist, Russ Schumacher, wrote that the November rain generally amounted to "'only' a 10 to 25-year rainstorm for southwestern Colorado." The quotation marks around "only" are his. The October storm, by his earlier account, ranked among the most extreme rainstorms in state records.
The explanation comes from following the water downhill, from the sky to the stream gauge.
The sky
A "10-year storm" sounds like a schedule. It's an odds estimate: a rainfall amount at one location, over a set length of time, with roughly a one-in-ten chance of being reached or exceeded in any given year. Two can show up in the same month.
For a spot beside the Pagosa Springs gauge, NOAA's rainfall estimates set the 25-year mark for three days of rain at 3.5 inches. Two volunteer rain gauges, one in town and one about a mile and a half to the south, caught roughly 3.7 to 3.9 inches over the three days ending the morning of November 2. That sits right around the mark, which matches Schumacher's description. In October, the same two gauges collected roughly 5.5 to 5.8 inches over five days. Measured in town, November's shorter storm brought about a third less rain.
A rain gauge is a bucket at one spot, though. The river at Pagosa Springs drains 298 square miles of mountains upstream, and rainfall up there can change sharply from one ridge to the next. Two buckets in town can't tell you what fell across all of it.
The ground
Soil holds water the way a sponge does. After a dry spell, rain soaks in and stays a while, and the river gets only part of it, and later. After weeks of rain the sponge is full. New rain runs off the surface or moves quickly through the ground, and it reaches the creeks while the storm is still going.
November's storm was the fifth round of heavy rain to hit southwest Colorado since mid-September. Counting it, Schumacher tallied more than 20 inches since September 1 at several stations around Pagosa Springs, and more than 30 at a couple of sites near Wolf Creek Pass, upstream. In parts of Archuleta County, which includes Pagosa Springs, totals through November 4 beat the wettest three months in records going back to 1895 by more than 6 inches. "Wetter than any previous three-month period, period," he wrote.
Hydrologists have watched this happen before. After Hurricane Floyd in 1999, the U.S. Geological Survey cautioned against assuming a 100-year rain produces a 100-year flood. Floyd fell on North Carolina ground that Hurricane Dennis had already soaked. A storm's rarity label measures only the rain. It says nothing about the condition of the ground the rain lands on.
The river before the rain
A stream gauge reports two numbers. One is height, the level of the water surface, read directly; that's where the feet come from. The other is flow, the volume of water passing each second, in cubic feet per second, and it's calculated from the height. (A cubic foot is about seven and a half gallons.)
Before the October storm, the San Juan was carrying about 450 cubic feet per second. By late October 30, a week after the October crest, the river had long since dropped below the level where the Weather Service starts watching for flooding. On a chart, that looked like recovery. The river was still carrying about 780 cubic feet per second, roughly three-quarters more than before.
Next to the roughly 11,200 cubic feet per second the river carried at its November peak, by preliminary estimates, that head start is small. What it tells you about is the land upstream. A river still running that high after several nearly dry days is draining hillsides the last storm soaked.
The channel
The storms also moved the landscape into the streams. The USGS has mapped hundreds of landslides and debris flows across the region and reported that they fed mud and debris into downstream flooding. Debris flows are fast-moving slurries of mud, rock and wood that come off steep, waterlogged slopes.
That debris can change what a gauge's numbers mean. The USGS converts height to flow using a rating, built from field crews measuring the river at different levels: at this height, this much water goes by. The rating holds only as long as the channel keeps its shape. If mud and rock make the bed shallower, the same amount of water has to stand higher to get past. A rating built on the old channel reads that higher water as more water, until crews go back out, measure again and revise it.
By the preliminary figures, November beat October on flow as well as height, about 11,200 cubic feet per second against 9,230. If debris raised the bed at Pagosa Springs, some of that difference could be the channel rather than the water. Whether it did hasn't been determined.
So the puzzle has two parts. The river stood higher in November. That was measured directly, and height is what reaches a house. Whether more water actually went by is the uncertain part. And if debris did raise the bed and stays put, the next flood of the same size will stand higher too, until the river scours it back out.
The gauge itself
A stream gauge has to stand in the river it measures, on the nights that river is doing its worst. Both floods left holes in the record. In October, an equipment malfunction knocked out readings for more than four hours before dawn on October 23, and the first good reading afterward became the recorded crest. In November, the recorded peak came at 8 p.m. By 9:45, when readings stopped, the river had slipped to 16.03 feet. Readings came back ten and a half hours later, with the river at 13.81.
Schumacher writes that because of "issues with the stream gage," the crest may have been higher than recorded. He doesn't say what failed. A falling river can rise again. A fresh burst of rain, or water arriving late from farther up the basin, could have pushed a second, higher crest through town overnight. The readings on either side of the gap can't say whether one did.
For scale, neither flood came near the gauge's largest known flow, about 25,000 cubic feet per second in October 1911.
Reading the next forecast
No agency analysis published so far has sorted out how much of November's crest came from heavier rain upstream, if it fell there, how much from soaked ground and how much from debris in the channel. The river's head start belongs with the soaked ground: it added a little to the peak directly, and it was the clearest sign of how wet the hillsides still were. For anyone living downstream of a wet stretch, the more useful fact is that soaked hillsides and a river still draining them were in place before the November rain started, and that a season of storms had been shoving mud and debris toward the channels the whole time.
During a wet stretch, a few questions fill in what the rainfall number leaves out.
- How much rain has fallen across the area over the past several weeks, in the mountains as well as in town?
- Has the river actually gone back to its usual flow, or has it only slipped under a warning line?
- Has a recent flood dumped mud and wreckage into the channels?
Some of the answers are already published. Each Weather Service office posts a forecast discussion several times a day, a short technical note where forecasters lay out their reasoning. Local flood warnings often say the same things in plainer language. On the afternoon of November 1, the Grand Junction office's discussion pointed to saturated soils as a reason to expect more flooding and more debris flows, hours before the San Juan crested that evening.
- Revisions to the crest: Every height and flow from the Pagosa Springs gauge is still provisional, and the USGS provisional data statement explains that such numbers have not received final approval and can change, so the official November peak may yet move.
- Where the mud went: The USGS landslide mapping for southwest Colorado is preliminary, and later updates may help show how much of the fall's slide debris reached the channels feeding the San Juan.
- Your own rainfall odds: NOAA's precipitation frequency estimates are looked up by latitude and longitude, so the 10- and 25-year rainfall depths shown here for Pagosa Springs can be checked for other locations and storm lengths from a day to ten days.
- What reporting damage means: The town's flood recovery page warns that submitting damage information is not an application for FEMA assistance and that the November 10–12 damage assessments do not mean federal aid has been approved.

