Farmington Flood: San Juan River’s Biggest Flow Since 1957

Crews in San Juan County, New Mexico, made at least 60 rescues by Friday, October 23, 2026, Source New Mexico reported. Searchlight New Mexico and the Santa Fe New Mexican also carried the story. A county official later estimated 200 to 300 rescues, some from rooftops, though no exact count was kept, the Albuquerque Journal reported. About 600 households in Farmington were told to evacuate, Hoodline reported, citing the Los Angeles Times.

On October 23, the San Juan River at Farmington carried at least 29,500 cubic feet of water per second, its most since 1957, by the gauge’s measured readings. The San Juan and Animas rivers spilled over their banks in Farmington, the National Weather Service (NWS) said. Rain began on October 21, with more than 5 inches in places, the Associated Press reported.

Downstream in Shiprock, on the Navajo Nation, residents waited on rooftops and the Shiprock bridge closed, the Navajo Times and Cedar News reported. Late Saturday, the Navajo Nation ordered Bluff, Montezuma Creek and Aneth in Utah to evacuate, Fox 13 and KSL reported.

How big was this flood next to the floods these towns have lived through? We checked the San Juan at Farmington, and 108 other stream gauges, against yearly records that go back as far as 1884. A gauge is a station that measures the water. Five of the 109 sit on canals or flood channels, not rivers.

What we found: The San Juan River at Farmington carried its most water since 1957. At 10:45 a.m. Mountain time on October 23, the gauge measured at least 29,500 cubic feet of water per second. The figure is provisional data from the United States Geological Survey (USGS). Cubic feet per second (cfs) measures how much water passes a point each second. That was 16 times its median flow of 1,830 cfs from October 13 to 19, the week before the rain.

The gauge stopped measuring 15 minutes later, while the river was still rising, so the true peak was higher. The 1957 flood carried 30,000 cfs, only 500 more. USGS estimates for the next five hours reach 32,600 cfs by 4 p.m., which would be the most since 1935. The NWS said the river reached a record height of 14.1 feet at the gauge before it stopped, above 12.3 feet in 1909. Height and flow are different measures: a record height does not mean a record amount of water.

News coverage gave heights and flows at single gauges. The Kiowa County Press reported a flow record on the Piedra River. We found no report that ranked every gauge’s flow against its full record.

How Wherobots makes this possible

USGS gauges report the river every 15 minutes. Each gauge also keeps a list of its biggest flow in each year, some going back more than a century.

With Wherobots, we joined the two for every gauge that reported flow in the Four Corners area. We found each gauge’s highest measured reading since October 20, ranked it against all of that gauge’s yearly peaks, and flagged gauges that stopped reporting. Then we placed every result on the map.

What we found about the Four Corners flood

1. How big was the flood where people were rescued?

Farmington and Shiprock sit on the San Juan below the point where the Animas River joins it.

• San Juan at Farmington: at least 29,500 cfs on October 23, the most since 1957. The record is 68,000 cfs, in 1927.
• San Juan at Shiprock: at least 26,600 cfs on October 24, the most since 1957. The record is 80,000 cfs, in 1929.
• Animas at Farmington: 12,000 cfs on October 24, the most since 1941.
• San Juan at Pagosa Springs: 9,150 cfs on October 23, the most since 1927.
• San Juan near Bluff, Utah: 38,000 cfs at 10:30 a.m. on October 26, the most since 1972, in data through 1:30 p.m. on October 26. The record is 70,000 cfs, in 1927.

At Farmington, five yearly peaks carried more: 1927, 1929, 1935, 1941 and 1957. The Farmington and Shiprock gauges both stopped while the river rose, so both peaks are minimums. At Pagosa Springs, an earlier reading of 9,420 cfs, at 3:30 a.m. on October 23, is now marked as a USGS estimate.

Bluff is downstream of Shiprock. Its gauge measured its highest flow at 10:30 a.m. on Monday, October 26: 38,000 cfs, with the water 26.2 feet high. USGS did not mark either reading as an estimate. Five of the gauge’s 102 yearly peaks were larger, the latest in 1972. That is below the 70,000 cfs flow record (1927) and the 32.0 ft height record.

The NWS had forecast the crest at Bluff for early Monday, October 26. From October 24 on, the gauge reported every 15 minutes until our data end, at 1:30 p.m. on October 26. At that time it read 37,300 cfs. The flow had stayed between 36,900 and 38,000 cfs for three hours, so the crest may not have passed.

Flow is not depth. A gauge records how much water passes, not how deep the water got in a given street.

2. Where did the rivers set records?

We counted a record when this week’s measured peak flow topped every yearly peak in a gauge’s list. We counted only gauges with at least 30 years of peaks: 79 of them reported flow this week. The 79 include two flood channels near Albuquerque, and neither set a record.

• 8 of 79 gauges set flow records. Six did so by 10% or more.
• Piedra River near Arboles, Colorado: 17,000 cfs, twice its 1970 record of 8,370.
• Moenkopi Wash, Arizona: 21,400 cfs, twice its 1983 record of 10,100.
• Animas at Silverton and below Silverton: 2,240 cfs against 1,470 in 1997, and 3,190 cfs against 2,970 in 1995.

On the main San Juan, Carracas, upstream, set a flow record. The town gauges at Farmington, Shiprock, Pagosa Springs and Bluff did not, in our data. Their record floods came between 1911 and 1929.

At Four Corners, below Shiprock, the provisional flow is 54,100 cfs, 3.7 times its 2003 record of 14,600. That figure is doubtful. It is about twice the measured peak upstream at Shiprock, 26,600 cfs, and that peak is a minimum, because the Shiprock gauge stopped while the river rose.

The height at Four Corners, 22.5 feet, is almost twice its 11.52-foot record. That points to a flow figure far outside the range USGS has measured there, so we do not count it as a flow record. The rule we set in advance is unchanged.

The water’s height, 22.5 feet against 11.52 feet, did set a record.

3. Did the gauges keep working?

Not all of them. Five gauges stopped reporting at their highest measured reading and had not restarted by the afternoon of October 26. Among them are the San Juan at Shiprock and Carracas. Their peaks are minimums.

At Farmington, the height sensor stopped at 11 a.m. on October 23, the NWS said. USGS then filled the flow record with hourly estimates. We left those out, because no gauge measured them. Our last height reading came at 10:45 a.m. on October 23. Measured readings restarted at noon on October 25, 49 hours and 15 minutes later, with 8,470 cfs.

We checked our readings against outside figures. At 10:45 a.m., our last height at Farmington was 13.94 feet. That is 0.16 feet below the 14.1 feet the NWS reported at 11 a.m. At Pagosa Springs, the gauge read 15.02 feet, the figure Colorado Public Radio reported.

The Shiprock gauge stopped reporting at 4 a.m. on Saturday, October 24. Its last height was 21.22 feet. The NWS gave 21.2 feet at 4 a.m. as the gauge’s last reading. Navajo Nation President Buu Nygren wrote on X that the river reached about 23 feet overnight, Rolling Out reported.

From river gauges to a flood record: the technology behind the analysis

• River readings: USGS 15-minute flow and height data, October 13 to 1:50 p.m. on October 26, 2026, provisional.
• Flood history: the USGS yearly peak flow of each gauge, the oldest from the 1880s.
• Gauge locations: the USGS site list for 35 to 38 degrees north and 105.5 to 112.5 degrees west.

In Wherobots, we compared 151,432 flow readings from 110 gauges with 6,351 yearly peaks from 113 gauges. The first job downloaded and ranked the data in under 5 minutes. A second job, with USGS estimates left out, produced the published numbers in 1 minute. A third counted the readings in 2 minutes. The same steps work for any river with a gauge record.

Why this matters beyond the Four Corners flood

Emergency managers can list the gauges that stopped at their highest reading, such as Shiprock at 26,600 cfs, and treat those peaks as minimums. Insurers and lenders can check whether the river near a property topped its record, as 8 of 79 long-record gauges did, before they price its flood risk. Federal flood estimates, which underlie many flood maps, are fitted to each gauge’s yearly peaks. A new record adds a large peak to that list and can raise the estimated size of a rare flood. Water and power utilities can check the rivers that feed their dams and intakes.

We also used Wherobots for Super Typhoon Champi, where we placed buildings in the storm’s path. Our guide to flood risk data explains the sources behind flood maps.

For the families rescued from rooftops in San Juan County, the measured flow at Farmington, at least 29,500 cfs, was the largest in 69 years. The river has run higher before, most of all in 1927.

Data sources and methodology

Sources. River readings, yearly peaks and gauge locations come from the USGS National Water Information System. We downloaded them on October 26, 2026, at 20:07 Coordinated Universal Time (2:07 p.m. Mountain time). The 15-minute readings are provisional.

River heights at Farmington and Shiprock and the record-height statements come from NWS flood warnings of October 23 and 24. Rescue and evacuation reports come from Source New Mexico, the Albuquerque Journal, Hoodline (citing the Los Angeles Times), the Navajo Times, Cedar News, Fox 13 and KSL.

The rain total is from the Associated Press, and President Nygren’s post is as reported by Rolling Out. The Pagosa Springs height report is from Colorado Public Radio. The earlier Piedra River flow-record report is from the Kiowa County Press. The Bluff crest forecast is from the NWS office in Grand Junction, Colorado.

Method. USGS documents a flood by comparing its peak flow at each gauge with the gauge’s yearly peaks, the series that federal flood estimates use (Bulletin 17C, USGS, 2019). We followed that comparison with measured readings only. We report each gauge’s rank as the last year with a bigger flood. At Farmington, 5 of its 95 yearly peaks, from 1924 on, were bigger.

We did not estimate how rare the flood was, such as a “100-year flood”. A gauge that stopped reporting at its highest reading gives a minimum.

What the numbers do not show. Flow at a gauge is not the depth of water in a neighborhood, and it does not count damaged homes. Dams upstream, such as Navajo Dam on the San Juan, change how much water reaches a gauge. USGS staff measure rivers after big floods and may revise the peak flows.

A change after review. Our first draft gave Farmington 32,600 cfs, the most since 1935. A check of the USGS data flags showed that this figure, and every flow after 10:45 a.m., is a USGS estimate, so we now use the last measured flow.

Updates with later data. Our first download ended at 8 a.m. on October 25, and our second at 9:35 p.m. that day. The second showed Bluff at 27,500 cfs, after a USGS revision. Both ended before the Bluff crest the NWS forecast for early October 26. So we downloaded the data a third time, through 1:50 p.m. on October 26.

In the third download, Bluff’s peak is 38,000 cfs, and Shiprock’s is 26,600 cfs, with readings USGS added. At Pagosa Springs, USGS now marks an earlier reading as an estimate, and the measured peak changed from 9,230 to 9,150 cfs. Costilla Creek near Garcia, Colorado, outside the San Juan’s basin, has no flow readings from October 20 on, so 79 long-record gauges reported flow, not 80. No record and no test verdict changed.

A correction after review. An earlier draft counted Four Corners as a record by its height. Our rule counts flow, and its flow reading is doubtful, so the count became 8 of 80, not 9. In the latest data it is 8 of 79.

How we checked this

Claim tested. News reports called the levels historic. So before looking at the data, we wrote down a claim to test. It says this week’s flood pushes the San Juan and its side rivers above the highest flows in their gauge records.

We set the rules below at the same time, before any run, to say how many records count.

Rules set in advance. The whole claim would be wrong if every gauge on the main San Juan with 30 or more years of peaks stayed below its record.

For tests 1 and 2, records at half or more of the gauges support the claim. If no gauge sets a record and none comes within 10% of one, the test contradicts the claim. Anything between is marked can’t determine.

Test 1, the main San Juan. One of 7 gauges set a flow record, Carracas. Four Corners set a record height, but its flow reading is doubtful. Farmington, Shiprock, Pagosa Springs, Bluff and Archuleta, below Navajo Dam, did not, in data through 1:50 p.m. on October 26. One record is fewer than half the gauges and more than none, so the verdict is can’t determine.

NWS record heights at Farmington and Shiprock do not change this result, which is about flow.

Test 2, the Animas. Two of 6 gauges set records, at Silverton and below Silverton, the second by less than 10%. Two more came within 10% of their record. Verdict: can’t determine.

Test 3, nearby basins. We set test 3 as a verdict test before the first run. Its rule, word for word: “Supports: records at 3 or more gauges outside the San Juan basin. Contradicts: none. Otherwise can’t determine.”

In the latest data, 3 of 51 gauges set records: Moenkopi Wash and two gauges on the Rio Chama. One of the three, Rio Chama near La Puente, beat its record by 3%.

Under its own pre-set rule, test 3 reads as support. Review then reclassified test 3 as a comparison, because its gauges are outside the San Juan’s basin and so outside the claim. It shows that the flood also set records in nearby basins. It does not count toward the verdict, which comes from tests 1 and 2.

Gauges in no test. The other 15 of the 79 gauges sit on smaller rivers in the San Juan’s basin. Two of them set records, the Piedra River near Arboles and Cement Creek at Silverton. That makes 8 records: 1, 2 and 3 in the tests, plus these 2.

Test 4, an outside check. Our heights are within 0.5 feet of the NWS figures: 13.94 against 14.1 feet at Farmington, and 21.22 against 21.2 feet at Shiprock. At Pagosa Springs, our 15.02 feet matches the figure Colorado Public Radio reported. This test checks the data, not the claim.

Conclusion. Overall verdict, from tests 1 and 2: can’t determine. Each found records at fewer than half of its gauges and at more than none. The rule that would disprove the claim was not met, because Carracas, on the main San Juan, set a flow record. Records came at 8 of 79 gauges, not at the town gauges.

Through 1:50 p.m. on October 26, the data rule out a measured flow record at Farmington, Shiprock, Pagosa Springs and Durango. At Bluff, the river reached 38,000 cfs at 10:30 a.m. on October 26, below its 1927 record of 70,000 cfs. It was still at 37,300 cfs at 1:30 p.m., so a later download would show whether it rose further. Reviewed USGS peaks would settle whether Farmington topped its 1957 flood of 30,000 cfs. The 32,600 cfs estimate would make it the most since 1935.