In July and again in October 2026, the South Skunk River at Ames crested within three inches of the same mark. The rain behind the two floods could hardly have been more different.
| Jul 4 | Oct 2 |
|---|
The gauge sits just below the confluence on the southeast side of Ames where Ioway Creek joins the South Skunk River. Rain falling anywhere across these 556 sq mi of Story, Hamilton and Boone counties ends up passing it. That land is the river's watershed, outlined in black on the map.
Two smaller watersheds feed it, shaded on the map. The South Skunk River watershed (green) begins near Blairsburg, then runs south through Story City and into Ames; the Ioway Creek watershed (pink) drains the west side of the basin and meets the S. Skunk after running through campus and town. Each has its own watershed management authority working on water quality and flooding.
The watershed sits firmly in the Des Moines Lobe: young glacial ground, flat, poorly drained by nature but artificially drained by tile lines and ditches since farming began. When the soil is saturated, rain water moves quickly to the streams.
The largest flood on record came on August 11, 2010, when the river at this gauge carried 36,200 cfs and put water through Hilton Coliseum. Both 2026 floods peaked at less than a third of that.
The summer flood came out of two bursts of rain. The first crossed the basin before dawn on July 3. The second was violent: in the hour ending at 2 AM on July 4, the airport gauge caught 2.60 in and the wettest 4 km grid cell got 2.67 in. The river crested 10 hours later.
The fall flood followed several days of on-and-off light rain, then a steadier soaking through the night into October 1. Its heaviest hour was gentler and spread across the whole basin. The river took 32 hours to crest after it.
In July the heaviest rain fell on the south end of the watershed, over Ioway Creek and Ames itself. Before the crest, Ioway Creek's basin averaged 5.4 in and the South Skunk's 3.8 in; the small local area around the gauge got 6.4 in. The airport sat near the bullseye, so its 7.54 in overstates what the basin as a whole received.
In October every part of the watershed got between 3.0 and 3.7 in, and the basin average was higher than the airport's reading.
NWS Stage IV gridded rainfall (about 4 km cells) via the Iowa Environmental Mesonet. Days run 7 AM to 7 AM CDT. Both floods share one color scale. Dashed lines mark the South Skunk River and Ioway Creek watersheds above their gauges.
Intense rain on a small area made a sharp, quick flood: the river rose nearly twice as fast in July and fell away sooner. The October flood built slowly and stayed high longer.
The surprise is in the water budget. October's flood carried slightly more storm runoff than July's from about three-quarters as much rain. Close to half of the fall rain ran off into surface waters; in July, less than a third did.
Storm runoff is flow above the river's level at the start of each 11-day window, spread over the 556 sq mi basin. Rain is counted through the morning after each crest; runoff share uses the Stage IV basin average.
Both 2026 floods were ordinary for this river: about a 1-in-4 chance of a peak that size in any year. Ten years since 1944 have seen bigger ones.
The July rain was not ordinary. At the airport, the 2-day total of 7.54 in is about a 1-in-90-year rain by NOAA Atlas 14. October's rain stayed under the 2-year line at every duration. Rarer rain does not guarantee a rarer flood.
The fall flood started from a higher river: 1,140 cfs before the storm, twice July's 573 cfs, with the stage 1.7 ft higher. The ground that feeds the river was already full. A soaked sponge can't absorb water, just like saturated soils can't absorb rain.
The weather explains part of it. In the month before the July flood, the corn and soybeans were growing fast under long, hot days, drawing water out of the ground (evapotranspiration ET) at about 0.15 in a day. By late September the crops were maturing (i.e. low metabolism, low ET), the days were shorter and cooler, and the ET totals fell to about 0.09 in a day. Less water left the landscape, so more was there when the rain came.
Satellite-based estimates make the difference plain. The day before the July rain, subsoil moisture across the watershed was close to normal for the time of year, averaging −0.6 points. Before the October rain it was 19.5 points wetter than normal, and every 9 km cell in the basin sat at least 11 points above normal.
The two Iowa State stations near Ames read about the same in both months (a few points wetter in the fall at one station). An anomaly compares each date with its own normal, so similar readings can still mean very different conditions for the season.
Adding up rain minus crop water use over the month before each crest, the two seasons come out close. By four days before the storms, rain had outpaced ET by about 3.5 in in both, but they got there by opposite paths. In July the balance had been falling for two weeks after the mid-June rains as the crops pulled the ground dry. In September it climbed steadily into the storm, with frequent light rain and little ET. The balances at the crest, +7.0 in in July and +6.4 in in October, are mostly the storms themselves.
Each flood is studied from 7 days before its crest day through 3 days after it. Charts measure time in hours or days from the crest at USGS 05471000.
15-minute stage and discharge from USGS. All 2026 values are provisional and may change when USGS reviews them.
Basin averages come from NWS Stage IV, area-weighted over the watershed boundary. Hourly grids are scaled cell by cell to match each day's quality-controlled total. The airport gauge (AMW) is a single point.
Rain is counted through the morning after each crest. The summer window also holds about 1 in of basin rain (0.43 in at the airport) that fell late on July 8, four days after the crest. It could not have fed that flood, so it is left out here. Counting it would lower the summer runoff share from 31% to 25%.
Storm runoff is flow above the starting level of each window (a constant-baseflow separation), divided by basin area. It is a simple method; a recession-based separation would give somewhat different numbers.
Lag is measured from the heaviest hour of rain to the crest. Rise rate is the largest one-hour stage increase. "Time above half its rise" counts hours the river stayed above the midpoint between its starting level and its crest.
Flood frequency is a log-Pearson Type III fit by method of moments to annual peaks through water year 2025, without the Bulletin 17C EMA or low-outlier test. Rainfall frequency uses NOAA Atlas 14 annual-maximum depths at the airport.
Subsoil moisture anomalies come from Crop-CASMA (USDA NASS / NASA SMAP) on its 9 km grid, for the day before each flood's rain began. An anomaly is the difference from the long-term normal for that time of year, exported as a fraction of water by volume and shown here as percentage points. Basin and branch averages weight each cell by the share of it inside the boundary; cells with data cover about three-quarters of the basin. Station readings are daily values at 12 inches from the ISU Soil Moisture Network.
The running balance adds daily rain minus OpenET actual ET over the 30 days before each crest. Rain is the Stage IV basin average (24 hours ending 7 AM, assigned to the day it mostly covers) or the AMW gauge's calendar-day total. It leaves out runoff, tile drainage and deep percolation, so it shows the weather's net push toward wetter or drier ground, not the water stored in it. The ET comes from one cropland sample, not the whole basin.
Daily rain totals from the ISU soil moisture stations were excluded: several fall days report 10 to 14 inches that no other gauge recorded. Their ET and soil moisture readings are used.
All downloads, calculations and static figures were produced in R; this page draws from the exported data.