Ice Age FloodsIce Age Floods National Geologic Trail
Home·Field notes·P64-29 Dry Gulch Terrace
Bretz-era field site · 1929

P64-29 Dry Gulch Terrace

Two accessory features of the Columbia Valley land color to the second interpretations. One is the existence, on the west side of the river, just north of the mouth of Dry Gulch, of a terrace at 1,000, a terrace whose surface definitely decends up the Columbia. Bretz Field Notes...

Year documented
1929Bretz 1929
Category
Bretz Field Site
Coordinates
47.3176, -120.1039WGS84
Trail
Ice Age Floods NGTWA / OR / ID / MT

Original field notebook

Two accessory features of the Columbia Valley land color to the second interpretations. One is the existence, on the west side of the river, just north of the mouth of Dry Gulch, of a terrace at 1,000, a terrace whose surface definitely decends up the Columbia. Bretz Field Notes .— Bretz 1929

This is one of more than 1,800 field sites cataloged in the early scabland surveys — the bedrock of the Ice Age Floods scientific record. The Bretz-era researchers walked the ground first; modern cosmogenic dating, LiDAR, and remote sensing have since extended and refined what they mapped.

Northwest Columbia and the Okanogan lobe

Geographic scope

This region covers the northwestern leg of the Columbia River — the "big bend" reach that swings west out of Lake Roosevelt around Grand Coulee Dam, runs south past Bridgeport, Brewster, and Pateros, accepts the Methow and Chelan inflows, and continues south past Entiat, Orondo, and Wenatchee. It includes the upper Columbia from the head of Lake Roosevelt (at the international border) down to Rock Island Dam below Wenatchee; the Withrow Moraine and Jameson Lake drumlin field on the Waterville Plateau; the Chelan trough and Lake Chelan; the Okanogan, Methow, and Sanpoil tributary valleys; and the Wenatchee River valley with its boulder-bar terraces near the Columbia confluence. The state-park anchor points are Bridgeport, Lincoln Rock, Daroga, and the Wenatchee Confluence area, plus the Withrow Moraine National Natural Landmark on the plateau west of Mansfield. This is the single most important corridor for understanding when the largest Missoula floods happened and why later flood routing flipped south into the Channeled Scablands.

The Ice Age Floods story here

This was the original highway of the Missoula floods. Before the Okanogan lobe of the Cordilleran Ice Sheet advanced across the upper Columbia, floodwater released from glacial Lake Missoula could follow the Columbia's natural northwest course around the big bend — south from Lake Roosevelt, past what is now Bridgeport and Pateros, through the Wenatchee reach, and onward toward the lower Columbia. The earliest and largest Missoula floods went this way. The reach near Wenatchee preserves the strongest physical evidence anywhere in the flood system for the peak discharges: crystalline ice-rafted erratics scattered as high as 495 m elevation — roughly 320 m above the modern Columbia River — across the sandstone hills of lower Wenatchee valley and the adjacent Columbia valley. These boulders are foreign rock (leucocratic feldspar-megacrystic granite, granodiorite, meta-argillite, quartzite, gneiss), originating hundreds of kilometers to the northeast, and can only have arrived stranded on the tops of icebergs floating in flood water that filled the valley to nearly 1,000 feet deep. Pangborn Bar on the east side of the river at Wenatchee — a crescentic flood bar roughly 600 feet (180 m) high, capped by giant current ripples with crests up to 20 feet tall and spaced ~300 feet apart — is the single largest depositional feature this corridor produced. Waitt and others place the building of high Pangborn Bar around 19,000 cal yr BP. Balbas and others (2017) anchored the peak-flood chronology by dating three ice-rafted erratics on the Wenatchee valley walls to 18.2 ± 1.5 ka using cosmogenic 10Be. This is the single most-cited date in the modern flood chronology and the reason this region is the anchor for the peak floods.

Then the routing flipped. The Okanogan lobe of the Cordilleran Ice Sheet, flowing south down the Okanogan trough, advanced across the Columbia River at the latitude of Bridgeport–Brewster between roughly 19 and 17 ka. As ice thickened across the upper Columbia and pushed south onto the Waterville Plateau, it built the Withrow Moraine — its terminal position roughly along the line of the modern town of Withrow, about 35 km south of the Columbia. Balbas and others (2017) dated boulders on the Withrow Moraine to a mean of 15.4 ± 1.4 ka, interpreted as the time the lobe began to retreat from its maximum extent (older estimates from varves and radiocarbon had placed the maximum a few thousand years earlier; the 10Be ages refined and shifted the retreat date forward). While the lobe was grounded across the Columbia, it dammed the entire Columbia valley from roughly Bridgeport upstream, impounding glacial Lake Columbia — a long, branching lake that flooded back up the Columbia channel into what is now Lake Roosevelt and far up the Sanpoil arm. With the natural Columbia outlet sealed, every subsequent Missoula flood that reached the upper Columbia was forced to spill south out of the lake. The spill points migrated as the lobe thickened: floodwater drained first through Foster Coulee, then through Moses Coulee as the ice closed Foster Coulee, and finally through upper Grand Coulee. Headward cataract retreat in upper Grand Coulee — dated by Balbas et al. (2017) from 10Be exposure ages of flood-plucked granodiorite surfaces in Northrup Canyon to 15.6 ± 1.3 ka — eventually breached north into the Columbia valley, locking in Grand Coulee as the dominant route for the youngest large Missoula floods until the last flood from Lake Missoula at 14.7 ± 1.2 ka.

The chronology has a coda. Glacial Lake Columbia outlived glacial Lake Missoula by 200–400 years. After the Purcell Trench lobe disintegrated and Lake Missoula could no longer refill, Lake Columbia continued to exist behind the thinning Okanogan ice dam. When the dam finally failed, the residual lake drained back down its original Columbia channel — and Balbas and others (2017) dated the youngest large flood deposits down the northwestern Columbia valley to 14.0 ± 1.4 to 14.4 ± 1.3 ka, interpreted as that final drainage of glacial Lake Columbia rather than a Missoula event. The Okanogan lobe was north of the Columbia by about 14.5 ± 0.3 ka and back on the Omak Plateau by 14.1 ± 0.3 ka. The full window from peak Missoula floods (18.2 ka) to the final Lake Columbia drainage (~14 ka) is therefore about 4,000 years of active megaflood activity through this single corridor, with the routing flipping from "Columbia direct" to "Scablands via Lake Columbia spillover" and then briefly back to "Columbia direct" as the lake terminated.

Bretz/Pardee-era fieldwork

J Harlen Bretz's argument for catastrophic flooding leaned hardest on the Channeled Scabland east of here, but the northwestern Columbia leg gave him essential supporting evidence — particularly the ice-rafted erratics at impossible elevations above the Columbia. Bretz's 1923 paper noted high erratics in the Wenatchee reach as proof that water depths had been hundreds of feet greater than any modern flow could explain. Joseph Pardee's 1942 paper "Unusual currents in glacial Lake Missoula, Montana" supplied Bretz with the missing water source by documenting giant current ripples on the floor of Lake Missoula, locking in the catastrophic-drainage model. Later workers including Richard Waitt, Bjornstad, and Atwater systematically mapped erratic distributions across the lower Wenatchee valley and adjacent Columbia walls, demonstrating that crystalline boulders foreign to the local sandstone bedrock occur up to 495 m elevation — the floor for any reconstructed peak flood water surface. Bill Long's altimeter-based field surveys in the 1960s–1970s established the systematic mapping framework. The Okanogan lobe's geometry was worked out by USGS and University of Washington researchers (Hanson, Easterbrook, and Booth among others) using glacial-geologic mapping of the Withrow Moraine, the Jameson Lake drumlin field, the Sims Corner eskers and kames, and the haystack basalt boulders scattered across the Waterville Plateau. By the 1980s the geometry was solid; what remained unsolved was the absolute chronology and, in particular, the relative timing of the lobe advance versus the largest floods.

Modern science — Balbas et al. 2017

Andrew Balbas, Anthony Barth, Paul Bierman, and Lee Corbett's 2017 paper, "10Be dating of late Pleistocene megafloods and Cordilleran Ice Sheet retreat in the northwestern United States" (Geology, vol. 45, no. 7, pages 583–586; DOI 10.1130/G38956.1) is the single most important paper for the chronology of this region. It is the first published study to apply direct cosmogenic-nuclide exposure dating to flood and ice-margin features across the entire northwestern Columbia and Channeled Scabland system, and it produced the absolute age framework still in use a decade later. Their sample sites in Figure 1B include sites 1 and 2 — ice-rafted granitic erratics on the valley walls near Wenatchee, perched 225–245 m above the Columbia River. Three boulders from these sites yielded 10Be exposure ages with a mean of 18.2 ± 1.5 ka. Because these erratics could only have been ice-rafted by water deep enough to float bergs against the valley wall at 225+ m above the river, the date directly records one of the largest Missoula floods routing down the northwestern Columbia before the Okanogan lobe closed the corridor. This 18.2 ka anchor is the central reference date for "peak Missoula floods" in essentially every subsequent paper.

The paper's second key result was dating the Withrow Moraine. Five boulders on the moraine surface yielded a mean age of 15.4 ± 1.4 ka, interpreted as the time when the Okanogan lobe began retreating from its terminal position on the Waterville Plateau. Together with cosmogenic ages from Purcell Trench lobe terminal moraines that returned similar dates, this established that the southern margin of the Cordilleran Ice Sheet was retreating across northern Washington and Idaho around 15.5 ka — likely driven by surface warming of the northeast Pacific Ocean. The third result was dating flood-plucked granodiorite surfaces in Northrup Canyon, a tributary of upper Grand Coulee at Steamboat Rock, to 15.6 ± 1.3 ka. Balbas and colleagues interpret this as the age of cataract incision into the Columbia valley — the moment upper Grand Coulee opened as a flood spillway and the dominant routing of the latest large Missoula floods shifted into the central Scablands. The fourth result was a 14.0–14.4 ka cluster of dates on younger flood deposits down the northwestern Columbia, interpreted as the final draining of glacial Lake Columbia after Lake Missoula had ceased to exist.

The implications for this region: peak floods (18.2 ka) routed through here; the Okanogan lobe was at maximum extent across the Columbia somewhere between roughly 19 and 17 ka and retreated by 15.4 ka; glacial Lake Columbia spilled south through Foster, Moses, and Grand coulees during the intervening interval; the lake itself terminated around 14 ka and produced the youngest large flood signal back down the Columbia. Post-2017 work has refined but not overturned this framework. The Waitt, O'Connor, Atwater et al. 2021 USGS review (Earth-Science Reviews) accepts the Balbas chronology as the working framework while noting persistent ±2,000-year uncertainty on the advance/retreat dates from different methods. Lehnigk and others' 2022 PNAS paper on glacial isostatic adjustment used Balbas dates to argue that crustal rebound progressively tilted divides and biased later flood routing toward the Telford–Crab Creek tract over the Cheney–Palouse. Recent 2024 GSA work on Okanogan-lobe glaciotectonics around the Withrow Moraine and upper Moses Coulee continues to refine the lobe's behavior at its terminus.

Visiting today

The Ice Age Floods Institute's Wenatchee Valley Erratics Chapter is the local organizing body and runs field trips, lectures, and audio tours throughout the region. Wenatchee Confluence State Park at the Wenatchee/Columbia confluence is the gateway destination; from there the Apple Capital Loop Trail in East Wenatchee passes giant erratics up to 40 feet across at marked stops near Mission Street Park, the Old Wenatchee Bridge, and along 2nd and 4th Streets east of Nile Avenue. Pangborn Bar itself is best seen on the Grant Road / 4th Street / Batterman Road grid in East Wenatchee, near Pangborn Memorial Airport — the megaripple field is most legible on LiDAR but visible from ground level on the bar surface. The Wenatchee Valley Museum and Cultural Center holds the East Wenatchee Clovis Site collection. Lincoln Rock State Park and Daroga State Park, both on US 97 north of Wenatchee, sit on Columbia River terraces inside the flood corridor and offer riverside access at the elevations the peak floods filled. Bridgeport State Park on Rufus Woods Lake (the Chief Joseph Dam reservoir) is the recommended stop for seeing the Okanogan-lobe story directly: the park's haystack rocks are local basalt boulders glacially transported and dumped by the lobe, and interpretive material describes the 1,000-foot ice thickness that covered the site. The Withrow Moraine and Jameson Lake Drumlin Field (National Natural Landmark) is accessible from Mansfield via county roads on the Waterville Plateau — a working-ranch landscape with the terminal moraine and drumlin field visible on LiDAR and from low ridge viewpoints. Lake Chelan State Park anchors the Chelan trough, the deepest glacially-carved valley in the region. The NPS Ice Age Floods National Geologic Trail brochure lists all of these as Washington-section sites.

Sources

About the researcher: J Harlen Bretz

Active: 1914-1979 (primary IAF fieldwork 1922-1932; Penrose Medal 1979) Affiliation: University of Chicago (Instructor 1914, Professor 1926, retired 1947, emeritus through 1979) Key paper: Bretz, J H. (1923). "The Channeled Scabland of the Columbia Plateau." Journal of Geology 31(8): 617-649.

Bretz spent seven summer field seasons beginning in 1922 walking the basalt coulees, dry cataracts, and giant gravel bars of eastern Washington and concluded that the landscape required a catastrophic flood at a scale no existing process could produce. He coined the term "Channeled Scabland" in his 1923 paper and over the next decade defended the hypothesis against a uniformitarian geology establishment that had no plausible source for the water and viewed catastrophism as scientifically illegitimate. Bretz refused to name a source until Pardee's 1942 paper supplied one: Glacial Lake Missoula. He continued publishing on the scablands as an emeritus professor through the 1950s and 1960s, including his synthesis "Washington's Channeled Scabland" (1959). The Geological Society of America awarded him the Penrose Medal in 1979 at age 96; he reportedly remarked to his son, "All my enemies are dead, so I have no one to gloat over."

Source: J Harlen Bretz - Wikipedia; HistoryLink.org; University of Chicago Magazine

Capture roadmap

What this site looks like once Phase 1 lands.

Every site along the trail will receive the full Terrain360 capture treatment: ground-level 360° panoramas, drone aerial imagery, and photogrammetry-based 3D models that visitors can spin in their browser. This page reserves the slots; the imagery flows in as field capture completes.

360° panoramic

Walk the site in your browser

Ground-level 360° panorama, every step along the feature, captured by Terrain360 field crews.

Phase 1 target · June–July 2026
Drone aerial

Read the landscape from above

Drone flyovers reveal the geometry of catastrophe — ripple marks, gravel bars, and scour patterns invisible from the ground.

Phase 1 target · June–July 2026
3D photogrammetry

Spin the geology in your browser

Photogrammetry and Gaussian-splat models let visitors rotate, measure, and inspect features in detail-page WebGL viewers.

Phase 1 target · June–July 2026
What we know now

How modern science extended the record

Satellite imagery and LiDAR have revealed that these coulees are among the largest flood-carved features on Earth. Grand Coulee alone is 50 miles long, up to 6 miles wide, and 900 feet deep — carved by floodwaters traveling at highway speeds.