Great Terrace of the Columbia #2 - The Geologists

The Great Terrace from Chelan butte. John Whitmer photo. WGS photo archive.


"I hold my conclusions lightly and am willing to change them if additional observations demand it." - I.C. Russell (1898)


Central Washington's Geology is Complicated

Two towering geologists of the American West, George Otis Smith and Bailey Willis, spent significant parts of their storied careers in Washington. In the introduction to their 1903 report on 'Geology and Physiography of Central Washington', they succinctly summarize their focus: Establish a framework for the study of and connections between the Cascade Range and the Columbia Basin,

Central Washington includes a part of two great topographic provinces; the great plain of the Columbia and the Cascade Range. The former, in its position and general desert-like character, suggests at once a resemblance to the Great Basin of Utah and Nevada; and the vastness of the desert plain is emphasized by the snowy peaks of the Cascades along its western border. These provinces are not to be regarded as unconnected in their geologic history, however great the contrast in their general features.

On the extensive basalt-covered plain monotony wearies the traveler, while on the rocky peaks of the Cascades the complexity taxes the powers of the observer.

Unlike the tortured and twisted bedrock of the North Cascades, river terraces demand little in the way of intellectual horsepower. Terraces are often simple landforms whose origins become evident upon first viewing.

The Great Terrace of the Columbia is different. Its more akin to multi-layered bedrock than to some run-of-the-mill surficial deposit. Fundamental questions regarding its origin, composition, age, and continuity with terrace flights in valleys to the north remain unresolved. In this article, I compile what geologists have written about the Great Terrace and include a few maps and photos.

Flint's stratified drift. Flint extended the Great Terrace 75 km up the Okanogan Valley, beyond Tonasket, and implied the landform exists in other valleys to the east.

Bretz's corner. Early on, Bretz recognized "discontinuous terraces on the margins of the scabland tracts, suggesting remnants of former valley fills" (Bretz (1923). Bretz's 1932 map, above, shows the Okanogan Valley and Great Terrace in a marginal position to the Channeled Scablands. His 1923 map showed less of the valley. His 1928 and 1959 maps don't show the Okanogan at all.

Sketch map. Place names, drainages, and features of the formerly glaciated southern Okanogan Valley and Waterville Plateau (Flint, 1935, Fig. 1). Shaded area is Columbia River Basalt. The triangle of basalt north of the Columbia River - the 'Okanogan Plateau' of Flint, - is today called the Karatar Plateau. The Columbia River used to flow north through Kartar Valley (Omak Trench) and join the Okanogan River at Omak, where it again resumed its path south. The river but was diverted by glacial ice and carved a new, shorter canyon through the basalt, its current route (Timm Ranch to Brewster).

Extent of the Great Terrace

The earliest mention of prominent river terraces in the northern Washington Territory comes from Bailey Willis (1887), who reports the observations of Professor T.C. Chamberlain,

...evidences of glaciation are found on that part of the Columbia...A broad gravel plateau, the analogue of the terraces along the Okinakane, lies between the right banks of the Co­lumbia and the Methow River, and is continued down the former wherever the canon walls are not too abrupt.

The "Great Terrace" of the Columbia River, named by Russell (1898), begins at the mouth of the Methow River and terminates a few miles south of the mouth of the Chelan River,

...on following the Columbia southward from the mouth of Chelan river, the valley soon broadens and one looks in vain for the great gravel terrace so conspicuous farther up stream. This termination of the great terrace of the Columbia when traced down stream is one of the reasons for believing in the former existence of a lake...

Smith and Calkins (1904) locate the Great Terrace somewhat differently, between Entiat and Brewster,

The Great Terrace is first clearly recognized by the traveler going north from Wenatchee about 3 miles beyond Entiat. For several miles northeast of that point it is well developed along the left bank of the river, and is shown by the contours on the Chelan topographic sheet of the United States Geological Survey. It is also in general well marked all along the west side of the valley from Chelan nearly to the Okanogan. The Great Terrace naturally extends up the tributaries of the Columbia, and in the lower portion of Okanogan Valley attains a breadth greater than the maximum observed along the master stream, amounting perhaps to as much as 3 miles just northeast of Brewster and about the same a few miles north of Clover [near Okanogan, WA].

Keyes (1917) sees continuity in terraces of the Columbia and Okanogan,

Above the mouth of Chelan river, the outlet of the lake of the same name, the bench about 300 feet above the water level appears to be the most important one. At least it is the largest and most persistent one...[and]...formed during the time when the course of the Columbia river was dammed by the great glacier which came out of the Lake Chelan valley. It seems to represent the level of the waters in the vast lake that was thus generated in the Columbia gorge above, the outlet of which was through the Grand Coulee or some other coulee near by. With this high level stage it seems most reasonable to associate the principal terraces of Okanogan valley.

Flint (1935) extends the terrace some 75 km up the Okanogan Valley beyond Tonasket to the mouth of Spectacle Lake Coulee,

This feature, termed by Russell the “Great Terrace,” consists of masses of silt (rock flour), sand, and gravel, forming benches along the Columbia canyon as far upstream as Bridgeport, and along the Okanogan trench as far north as Tonasket. In many places it is discontinuous, as a result here of lack of initial deposition, there of later erosional trimming; but single stretches many miles in length are common.

The terrace extends continuously up the Okanogan for several miles...Between Malott and Okanogan, the terrace expands into Tarheel Flat...Tarheel Flat merges north into the great Pogue Flat, twenty square miles in area, its surface rising from 1230 feet on the south, to 1330 feet near the point where the local tributary, Johnson Creek, enters the Okanogan trench. The “Great Terrace” also partly fills Wagonroad Coulee...Five miles north of Tonasket the “Great Terrace” re-appears as a narrow bench west of the river...This is its northern limit.

Hunt (1977). Map of the Columbia River and adjoining Fraser River drainage basins. Black blobs are modern lakes. Red rectangle is the Okanogan Valley and a portion of the Columbia Valley (Penticton, BC to Chelan, WA).

Relief map of WA and BC. Several conspicuous drainages extend across the US-Canada border from southern British Columbia into northern Washington. Most are hundreds of kilometers in length and have enormous watersheds. The valleys conveyed meltwater from the Cordilleran Ice Sheet south during the Ice Ages. Montana's Glacial Lake Missoula was filled by water from the Rocky Mountain Trench. The Okanagan-Okanogan Valley of BC and WA also carried meltwater south and contained Ice Age lakes. Water from Glacial Lake Missoula spilled suddenly and repeatedly into Eastern Washington, carving the Channeled Scabland. It is currently unclear whether large floods were similarly spilled from ice-dammed lakes in the Okanagan-Okanogan Valley to the Scablands. There are hints in the Columbia and Okanogan Valleys - landforms and deposits - that seem to suggest some outburst-type flooding did occur, but flood volumes appear, at this point anyway, to have been somewhat smaller and less frequent than the Missoula floods. We do not know their effects, if any, on the Scablands. Map modified from Marshall Richmond/USGS (2017).

Northern connections. The topographic and structural grain of northern Washington is oriented north-south. The Okanogan Highlands are drained by south-flowing modern rivers that feed the Columbia. All of their valleys were widened by Pleistocene ice, but the north-south valleys - a bedrock pattern - have existed for millions of years. They are pre-glacial and largely pre-Columbia River Basalt, as Willis (1887) noted more than a century ago, "...the Wenatchie, Methow, Okinakane, and San Puel strike the eye as members of a possibly consequent and unmodified system [by flood basalts]…carved upon the older surface of granite and crystalline sedimentary rocks". Base image modified from shaded relief image by Ralph Haugerud.

Modern contributions. Red arrows highlight the routes of modern streams draining to the Columbia. Glacial meltwater in various amounts followed these same channels in the past. Maybe some floods, too. Map was stolen from a spectacular GSA field guide by Waitt et al. (2021).

Campbell (1962) summarizes,

From its abrupt termination near Chelan outlet, northward far up the Okanogan River into Canada, it is a nearly flat-surfaced sheet of loose sediment into which the Columbia is entrenched several hundred feet. Early geologists thought it to be the delta of the Columbia, formed where it entered "Lake Lewis," a presumed vast lake that flooded most of central Washington during maximum glaciation farther north. The slanted bedding exposed in some of the road cuts would suggest this to be correct; but many complications have since been found to disrupt this simple picture…

Waitt (2009) agrees with Flint and Keyes,

...the 'great terrace' truly extensive compared to others...broad and prominent in lower Chelan valley as in adjacent Columbia valley...[and]...below the ice limit in Columbia valley, where it becomes a prominent outwash terrace traceable far downvalley...and up Okanogan valley 45 mi to beyond Tonasket. It suggests a level 1150-1200 ft here controlled the whole surface. That control is Knapp Coulee. As glacier ice retreated, mainstem Columbia valley must have remained blocked awhile by terminal moraine and outwash, all water upstream draining through Knapp Coulee.

According to Pine (1985), the Great Terrace is not,

...a single, regionally correlative, upper surface. Rather, a number of terraces...define the upper surface...[The terrace] between Chelan Butte and the mouth of the Okanogan...generally lies between 365 and 380 m above sea level...[and]...Upstream from Bridgeport...[it]...apparently merges with a set of prominent terraces (380 to 420 m a.s.l.) cut into the Nespelem silt...Along the Okanogan River...elevations range from 375 to 410 m (Flint, 1935b). In general...a progression of increasing...altitudes as one proceeds upstream...The extreme upstream and downstream segments of the "Great Terrace" are incised surfaces. The convex-upward terrace surfaces between Riverside and Okanogan are outwash fans built out from tributary coulees.

Jim O'Connor of USGS holds a more nuanced view (written communication, May 2022). Loosely paraphrasing his thoughts: From a point south of the Chelan River, a more or less distinct gravel train descends to the south from a morainal ridge nearly to Wenatchee. Terraces upstream of Chelan along the Columbia and Okanogan form a series of recessional outwash surfaces, laterally traceable over short distances. He adds,

I think the "great terrace" descriptor obscures the situation.

Historic photo. A flight of glacial terraces at Omak, WA. Pogue Flat in foreground. Mouth of Omak Creek in the hazy distance at left. View to the SE. Frank Matsura photo (#6887) c. 1910. Okanogan County Historical Society archives.

Synthetic photo. Oblique Google Earth image from roughly the same location. Vantage point near the intersection of Hwy 97 and Dayton Ave.

Giant Current Ripples

Giant current ripples, formed by glacial outburst floods, are found with some regularity in the Channeled Scabland to the south. In a few places, they are also found on the Great Terrace.

Waitt (2017, Stop 2.2) identifies giant current ripples on two terrace surfaces near Chelan,

Below [the Great Terrace surface] two gravel bars near the river. And two bars lie higher, including where we stand. These higher bars have giant current dunes across both levels - dunes truncated by water flow a the 'great terrace' level. The high dunes are spaced 160 m apart, their amplitude 10-12 m. This high surface can be traced only a few miles farther upvalley. So these floods preceded the last-glacial maximum at Chelan Falls.

The low bars near the river once displayed giant current dunes but all that is fubar since the housing development.

In 2019, I discovered giant current ripples in a lidar image of the Omak area and reported it here:

Spectacle Lake near Loomis. Terraces along the former course of the Similkameen River. WGS photo archive.

Clark Fork, MT. All water from Glacial Lake Missoula drained west through the Clark Fork Valley. WGS photo archive.

Lacustrine Sediment

Sediments that comprise the terrace in the Columbia and Okanogan Valleys are sandy, gravelly, and sometimes quite bouldery. But lacustrine deposits (lake deposits) are found there, too. It is unclear at this point whether ice-marginal lakes were large and few or small and many. It is not known how bulked the high terraces are by fine grained lacustrine sediment. Do similarly thick bodies of Ice Age till, sand, and gravel also contain lake beds (i.e., Rathdrum Prairie)? Early geologists had their own questions,

Willis (1887) speculates the valley during deglaciation did not host a sediment-dammed lake,

The valley of the [Okanogan River] belonged no doubt to the great system of which all these now abandoned channels were part. Like them, it was drift buried, but, unlike them, not dammed, because of its great width.

Flint (1935) speculates,

Evidently the great bulk of the silt and the sand is of lake origin, and was augmented by minor quantities of coarse débris introduced, not as deltaic beds but by fan-building streams of highly variable flow and frequently changing points of discharge. Some sections include minor unconformities, none of which, however, definitely indicates emergence or even sudden shoaling of the lake water. Fluctuation of the water surface may be admitted as possible, but is not proved.

Irrigation map. Irrigation project map for the Omak-Okanogan area with water dammed at Conconully. Green areas are broad terraces flats ("practicably irrigable land").

Waters (1933) weighs in,

Several factors indicate that deposition took place largely in ponded bodies of water. The finer-grained materials are invariably bedded horizontally and show the paper-thin lamination characteristic of lacustrine deposits. The gravels are crossbedded in deltaic fashion, and delta foresets are locally seen, extending for more than 100 feet along the dip. At many places the foresets dip up or else directly across the tributary valley, indicating that the delta was built by a stream flowing in the opposite direction or else at right angles to the present drainage. The majority of the terraces have horizontal tops, indicating that they were controlled by a lake surface, but a few terraces have sloping or irregular tops.

The field relations of the terraces indicate that they were built in lakes whose valleyward margins were dammed by ice that still occupied the Columbia Valley.

Waitt (2017) surmises,

…some evidence such as accordant levels of broad terraces far up the Okanogan, lacustrine silt composing some of it, suggests the lake (or some lake) outlasted the glacial Lake Columbia flood(s).

Pale laminated silt, clearly lacustrine, lies extensively about the slopes on both sides of lower Chelan basin up to at least 360 m, seemingly controlled by Knapp Coulee spillway (1980b, 1987). Upvalley of Chelan Falls the lower 3/4 of the 'great' terrace also consists of laminated lacustrine silty. Lower Chelan valley held an arm of a lake that grew larger and larger in the Columbia in front of ice receding up Columbia and Chelan valleys. I've called this "Lake Brewster" (Waitt and Thorson, 1983; Waitt, 1987).

Kettles on Terrace Flats

Kettles are irregularly-shaped pits in terrace flats. They are relict features, formed when large, half-buried chunks of ice (grounded icebergs) melted out and the surrounding sand/gravel collapsed inward. All who have worked on the glacial geology of the Okanogan Valley have mentioned kettles on terrace surfaces.

Near Chelan and in the Okanogan Valley, some of the broad terrace flats are ornamented with kettles and kettle fields. Other terraces contain no kettles. Where kettles are and are not is an important question to answer through detailed mapping (which I am currently doing). The presence of kettles speaks to ice stagnation and periods when chunks of calved ice were being transported down-valley by flowing water.


The Great Plain

The 'great plain' of Willis (1887) refers to the vast basalt-covered region in Eastern Washington that we today call the Columbia Plateau or Columbia Basin. Willis uses 'great plain', 'basalt-covered plain', and 'the desert plain' interchangeably. The term was utilized by others for a short time, but usage faded prior to WWII. Today, the term has all but disappeared from the literature despite its elegance and descriptive value. Richard Waitt is resurrecting it.

Return of 'Plain'. Maps by Willis (1887) and Waitt et al. (2021).


Downwasting of Okanogan Lobe Ice

Nasmith (162) provides a clear summary of how many suspect deglaciation proceeded,

No well-defined moraines are reported north of the terminal moraine near Coulee City, and it appears that retreat of the main ice front was accomplished largely by down-melting and stagnation of the ice mass as a whole, with no clearly defined halts or readvances. The down-melting of the ice surface ultimately left the plateau and highland areas ice-free, and when the remaining glacial ice was confined to the valleys the surface area of the melting ice was greatly reduced. It seems likely that the effective withdrawal of the ice may have proceeded much more rapidly in the early stages of retreat than in the later stages because the sharp reduction of the ablation area may have been sufficient to bring down-melting into temporary equilibrium with ice accumulation. In other words, although the change to a milder climate may have been continuous, the rate of disappearance of the ice may not.

Surficial geology matters. Surficial geology mapped in the Spectacle Lake area, Okanogan Valley by Pine (1985).

The Great Terrace south of Okanogan. R.W. Galster photo in 1976. WGS photo archive.

Okanogan Valley Terraces: Recessional Stream or Ice-marginal deposits?

The nature of the Great Terrace of the Columbia and the numerous, more voluminous, but seemingly subordinate terraces in the Okanogan Valley remains a central question.

Russell (1898) and Dawson (1898) both believed the Great Terrace to be of fluvial origin. They envisioned the Columbia emptying to great lake that covered much of south-central Washington - the Lake Lewis of Symons (1882). This idea has not stood the test of time,

As correctly stated by Mr. Dawson, the great terrace of the Columbia is a true stream terrace. It is one of a large series of similar terraces which occur along many of the rivers of eastern Washington...Snake...Spokane...Okanogan...Methow... [and] Yakima...

R.D. Salisbury and B. Willis reportedly examined terraces at Chelan c. 1900, interpreting them to be in part stream terraces and in part moraines of small glaciers, but I've not located that article. In 1901, G.H. Garrey and E. Blackwelder traced the Okanogan Lobe's moraine from Chelan to Colville (Salisbury, 1901).

Smith and Calkins (1904) recognized flights of terraces along the Columbia and Okanogan,

Upon the sides of the canyon of the Columbia are a succession of terraces, in part benches carved from bed rock, but mainly fill terraces built of stream-laid rounded of these gravel benches is markedly distinguished from all the others by its extent and perfection...the "Great Terrace of the Columbia"...

And interpreted some of their sediments as,

...aqueous deposits laid down by streams flowing along the sides of glaciers or in ponds held back by ice dams...

Daly (1912), working along the US-Canada border, was first to interpret the Okanogan terraces as ice-marginal features (kame terraces),

These are almost without question deposits of rock-debris which were washed into the valley and lodged between the valley wall and the Okanagan glacier. As the ice-sheet diminished these lateral terraces were formed at lower and lower levels. The resulting step-like forms are not, therefore, stream-cut terraces but are little-altered constructional reliefs formed in late Glacial time. Closely allied to these linear terracelets are truncated fans which at various levels were washed from the principal branch valleys into the trough and against the ice, the gravel-sand deposit in each case backing up int the branch valley. A deltoid form was thus produced, with the base of the delta marking the ice-wall which retained the detritus on the side of the main valley. These high-level fans are themselves sometimes terraced as if the ice-wall had lowered by successive stages.

Flint (1935) throws in with the ice-marginal crowd,

The writer believes that the “Great Terrace” in the Okanogan trench differs in no significant respects from the [Columbia River] terrace farther south, and agrees in the main with Waters’ interpretation. Having in mind the ice-marginal character of the higher, earlier-formed terraces, one may conceive of the much more bulky “Great Terrace” as recording a time when the ice in the Okanogan trench and the Columbia canyon had wasted enough to permit water to drain from one tributary around spurs to the tributary next down-valley, thus building sediment more or less continuously along the valley sides...

...extensive kettle complexes indicate the presence of wasting ice here during the accumulation of the “Great Terrace”...

Long profile. Generalized long profile of the Great Terrace in the Columbia and Okanogan Valleys (Oroville to Chelan) by K.A. Pine (1985, Fig. 27). From his caption, "The extreme upstream and downstream segments of the 'Great Terrace' are incised surfaces. The convex-upward terrace surfaces between Riverside and Okanogan are outwash fans built out from tributary coulees. Profile developed from 7 1/2-minute and 15-minute topographic map quadrangles and point elevations reported by Waters (1933)."

According to Pine (1985), a masters student at Western Washington University working in the Spectacle Lake area near Tonasket,

The most prominent set of kame terraces occurs along the Okanogan Valley and is collectively known as the "Great Terrace". In the study area, the "Great Terrace" is composed of highly deformed ice-contact deltaic and glaciolacustrine sediment overlain by a cap of outwash gravel. Sediments of the "Great Terrace" were deposited along and over stagnating ice in the Okanogan Valley, predominantly as deltas built by streams issuing from tributary coulees. The presence of numerous kettles on the upper surface of the "Great Terrace" and on younger terraces cut into it indicates that buried ice was present within the sediments of the "Great Terrace" for some time after its construction.

Jack C. Holden (2007), an amateur geologist and long time resident of the Okanogan Valley, agrees with Daly and Flint,

In modern parlance these “ice marginal deposits” are called kame terraces being lake and stream sediments deposited along the flanks of a shrinking glacier and the adjacent newly exposed highlands...the lake delta theory cannot account for the steps.

The kame terrace concept accounts for the sometimes multiplicity of terraces, one above the other like stair steps. With each successive shrinking of the glacier, lower terraces were formed until, just before the ice entirely disappeared, the last and lowest terrace was created. These last terraces are usually more extensive than the higher older ones because the valley floor became wider and accommodated extensive lakes and wandering stream channels as the ice shrunk - in addition to having an abundant source of ready-made sediments from the higher older terraces above.

In the Omak area the most extensive terrace is the four-mile-plus wide Pogue Flat/Sand Flat terrace complex, which has several levels, just north of town. The Sand Flat level is the lowest and to the east bordering the Okanogan River...The Pogue Flat terrace sediments were most likely derived from the Conconully drainage system to the west and probably consists, at least in greater or lesser part, of glacial lake deposits.

Ralph Dawes (2017), long time geology faculty at Wenatchee Valley College, summarizes the lay of the land this way,

A set of kame terraces deposited between ice and valley walls forms a composite "Great Terrace" 200 km along the sides of the Okanogan and Columbia River valleys, pocked by kettles, with local ice-contact-disturbed bedding. Ice-marginal, side-stream channels were eroded into bedrock adjacent to the Okanogan, Methow, and Columbia River valleys while the main valleys remained choked with last-stage glacial ice. Lacustrine beds in the Great Terrace, deposited in short-lived proglacial lakes, are interbedded with outwash and alluvial fans. A particularly thick, extensive sequence of lacustrine beds, including probable varves, extends for over 100 km along the British Columbia Okanagan Valley and was deposited in a large proglacial lake.

Ice-plugged valleys, ice-buttressed sedimentation, ice-marginal lakes, high level delta fans, and flow over ice during deglaciation are described in articles on the Okanagan Valley of south-central BC (i.e., Fulton, 1969; Nasmith, 1962; Eyles and Clague, 1991). Kiver and Stradling's work in the Upper Columbia and Pend Orielle Rivers deserves mention (i.e., Cararra et al., 1996), but I've not had time to read those articles through yet. I will.


Who were the FOOLs?

Jack Holden (entrepreneur-artist), Don Hruska (mining geologist), Gary Mundinger (bookstore owner), and John Whitecar (real estate appraiser) founded Friends of the Okanogan Lobe (FOOL) sometime around 2004. Together the FOOLs made many original and insightful observations on the glacial geology of the Okanogan Valley, publishing short papers in Okanogan County Heritage magazine. Holden was an illustrator with work in numerous publications including 'The Handbook of Unusual and Natural Phenomena' and 'Id of the Squid'. Hruska published a book 'Okanogan Ice Skating: Three million years of glacial impact and bedrock rumbling'.


Sketch map. Excerpt of Aaron C. Waters' study area map by (Waters, 1933).

Confluence. The Columbia Valley between the Okanogan River mouth and the Methow River mouth. Some important landforms are labeled, including Poverty Flat, Brewster Flat, Paslays Bench, and Alta Coulee. Several bedrock notches ("spillways") are labeled "SP". Excerpt of Waters (1933, Fig. 10).

High Level Fans

Remnants of glacial sediment perched high on valley walls, plastered to scoured bedrock, are conspicuous in the Okanogan Valley. They hint at an early, high level surface built against ice that has mostly been mostly removed by erosion.

Dawson (1898) observed high level deposits well above the level of the Great Terrace that,

...evidently accumulated at a time when the occupation of the Columbia gorge by the still moving Okanogan Glacier prevented the free escape of drainage waters.

Russell (1898) also noted them,

There are other terraces about 100 feet higher, but these belong in another category and have a different history than the great terrace...The terraces are conspicuous on the upper course of the rivers where their valleys or canyons are narrow, and disappear lower down the river where their valleys become broad.

As did Daly (1912),

...truncated various levels...The highest observed deposit of the kind was found at the 3,700-foot contour on the east side of Osoyoos lake. Others at a dozen or more different levels occur on both slopes...These high-level fans are themselves sometimes terraced as if the ice-wall had lowered by successive stages...

Smith and Calkins (1904) speculate as to the nature of the high level deposits,

The Great Terrace itself and those below it are of post-Glacial age, while those above it are believed to be contemporaneous with the glacial occupation of the valley.

These high terraces seem to have been either ignored or passed over briefly by previous writers, but they should be considered in any attempt to construct a complete history of Columbia River. They are attributed to the period when the valley was occupied by ice, the hypothesis being based upon the position of the terraces, their form, and the character of the materials composing them. The highest terrace is apparent at the highest level which the surface of the ice attained, according to the evidence of the terminal moraine, and the fact that many of them are on that part of the slope strewn with glacial bowlders precludes the supposition that these features are pre-Glacial. It appears, on the other hand, utterly improbable that they are of the nature of the Great Terrace and prove a filling of the valley with stream deposits to a depth of over 2,000 feet. As regards their form, they are less regular than the lower benches, and characterized by more numerous deep pittings than on the Great Terrace. The pebbles and bowlders entering into the construction of the terraces are generally subangular, suggesting glacial debris only slightly worked over by water. The facts of observation, therefore, are taken to indicate that the high terraces were built by streams flowing along the sides of the Okanogan Glacier during the period of its decline.

Kame terrace concept. Jack Holden's sketch explaining how ice-marginal deposits left behind by downwasting glacial ice formed the stair-stepped flights of kame terraces we see today in the Okanogan Valley (Holden, 2007).

Waters (1933) suggests the high terraces are intra-valley deposits,

The higher terraces are almost invariably [of and] confined to a single canyon. Some of the lower ones may extend around a spur and coalesce with terraces in a neighboring tributary, but, with the exception of an extensive compound terrace that lies at an elevation of approximately 1250 feet above tide and that is known in the literature as the “Great Terrace of the Columbia,” none of them is very extensive outside its own valley.

Nasmith (1962) takes an unconventional approach to "raised" alluvial fans and deltas, speculating they may be related to rainfall,

These are constructional features... built by streams graded to a base level higher than at present. Lowering of the base level, whether or tributary streams, resulted in down cutting and partial erosion by the streams that built them. Although similar in form to...glacial outwash, the raised alluvial fans and deltas are distinguished by the fact that the streams which built them were fed from normal precipitation and not by meltwater flowing from the ice lobe in the Okanagan Valley. Probably the climate and the intensity of erosion were both different when these alluvial fans and deltas were built, but no direct comparison with the present has been made.

Campbell (1962), in flowery prose of days gone by, notes,

As the glaciers melted, small lakes formed along their edges in the higher parts of the side-valleys; and the sediments, which settled in these lakes, can be seen clearly now as small terraces high up in the valleys. Their flat tops are a great boon to roadbuilders and they furnish many pleasant camp sites (the huckleberries are wonderful), though their gravelly flanks present problems in road maintenance. These ice-edge terraces form a sort of staircase down any particular valley, recording the ice wastage, and the elevations of their "treads" do not necessarily accord with those of "treads" in neighboring valleys. It is only in the main valleys, at lower elevations, that terraces matched in elevation can be found; and these have a different history, being the remains of the blanket of sediment washed out over the valley floor, during the final ice retreat.

According to Pine (1985),

Kame terraces at higher elevations are predominantly erosional in origin and probably signify relatively rapid lowering of the ice-sheet surface during their formation.

The higher terraces along Spectacle Lake, Horse Springs, and Whitestone Coulees are predominantly erosional in origin. They are underlain chiefly by till and typically have thin (approximately 1-3 m) caps of poorly sorted gravel.

At Loomis, Rinehart and Fox (1972) mapped terrace deposits in a generalized way as Qd (drift), but observed,

The upper surface of some of the terrace-like deposits...slope l0-20 degrees toward the axis of the valley. Where bedding is observed, it generally is grossly parallel to the upper surface. These features suggest that the deposits are fans rather than terraces and probably are composed of drift washed down from the higher slopes, perhaps both before and after recession of the glacier.

Old Okanogan. Panorama of Okanogan, WA with terraces c. 1909. View looking west. The square, white school building is on one terrace above town and a higher terrace surface is seen beyond. Matsura photo. Okanogan County Historical Society archive.

Boulder Fields

Boulder fields occur with frequency in the Okanogan Valley, though a pattern has not been worked out. Some appear directly related to high energy flows through bedrock canyons (i.e., boulder dumps). Others are clearly fan deposits, but very coarse grained ones. Yet others are erratics, some stranded in strange places.

Smith and Calkins (1904) describe a large, conspicuous accumulation of boulders at Alta Coulee-Methow River mouth just south of Pateros. The boulders have an obvious relationship to ice - likely they were quarried by a glacier - but may also be flood-transported - perhaps liberated from a nearby moraine by one or more great floods down the Columbia.

The authors of this report, in their course up Columbia and Okanogan valleys, had some opportunity to observe the evidences of their former occupation by ice. The most striking and unmistakable of these evidences are the erratic bowlders, the remarkable accumulation of which opposite the mouth of the Methow was first described by W. L. Dawson. The western slope of the Columbia Canyon at this locality is strewn with great numbers of huge fragments of basalt and granite, many of them larger than the ordinary settler's cabin. They continue to occur along the left bank for some miles below this point, and a few are found on the Great Terrace above the right bank. North of the mouth of the Methow the slope of the valley has many erratics of basalt, although that rock does not occur in place on that side of the Columbia. The highest of these glacial bowlders for which a barometric observation is recorded is 1,100 feet above the river, but others in the vicinity lie, by estimate, at least 300 feet higher.

Waitt (2017, Stop 1.6) visited this same accumulation, attributing the deposit to a megaflood,

Evidence of large post-ice sheet flood down Columbia's great bend include these huge boulders and high gravel bars just below the Methow's mouth. Alta and Antoine Coulees choked by bars and chaotically hummocky gravel (apparently) ice-jam deposits) and flood-scoured and boulder-strewn 'great terrace' outwash below Chelan Falls as high as 130m above the river's natural grade (Waitt, 1985, Waitt, 1994, 2016; Waitt et al., 2009). This flood seems the demise of glacial Lake Columbia - as Waters (1933) had postulated, but Flint (1935) dismissed.

Balbas et al. (2017) obtained 10Be ages on several Pateros boulders averaging 14,000 years old. PhD student Joel Gombiner and Dr. John Stone of the University of Washington will soon publish their own exposure-age dates for boulders in the floodway region.


Fryxell (1973) identified two major components of the Great Terrace, the "Azwell silt" and the "Brewster gravel", terms that have not held up over time.

At Chelan, Waitt (2009) describes the composition of four terrace levels.

The first, an arm of the Great Terrace, contains boulders in a mixed-lithology gravel "overlain by lenses of lake sediment but on the valley side overlies till containing striated boulders". On the next higher terrace, a complex mix of till, deltaic sand, outwash, and lake sediment are graded to the level of an ice-marginal channel nearby. The third terrace is not described, but the fourth contains boulders to 1m.

According to Pine (1985),

The sediments found high in the section along Spectacle Lake Coulee appear similar to the more proximal facies of the upper fan and upper midfan described by [New Englanders] Boothroyd and Ashley (1975). Both consist mainly of massive to crudely bedded gravel with scattered interbeds of stratified sand.

The large quantities of till exposed along the north wall of Spectacle Lake Coulee bear surfaces modified by fluvial processes. Much of the terraced area just north of Spectacle Lake is underlain by till which is widely exposed along the terrace fronts. In some places, the till is overlain by thick sequences of ice-contact glaciofluvial and glaciolacustrine sediments. Other portions of the kame terrace surface appear to have been only beveled and channeled by streams; on these, till is exposed at the surface or is overlain only by a very thin gravel cap. A similar situation exists along the east flank of Aeneas Mountain and on a small terrace on the north side of Whitestone Mountain. All of these areas where mapped as terraced till locally overlain by undifferentiated recessional stratified drift (Qtsd).

[The Great Terrace] is a very large kame terrace, or, more appropriately, a series of kame terraces. Although deformation partly obscures the facies relationships among the sediments comprising the "Great Terrace" in the study area, interpretation of available evidence suggests that the majority of the sediments were deposited in a proximal glaciolacustrine and prodelta slope environment.

Does this look simple? A diverse mix of sediments comprise the Great Terrace at Paradise Hill Rd at Brewster, WA.

Terraces in Adjacent Valleys

To a first order, the larger terraces in the Chelan, Methow, Okanogan, Sanpoil, Upper Columbia, and possibly those in the Wenatchee Valley and Pangborn Bar grade to a common base level, or base levels. Regional controls on sedimentation by ice (potential dams) and changes in accommodation space along the transient ice margin changed over time. I wonder, was more sediment available to early Pleistocene glaciers, outwash streams, and floods and less of it available later as it was removed, reworked, and transported downstream?

Smith and Calkins (1904) note,

Extensive gravel terraces border both Okanogan and Similkameen rivers as well as many of the dry channels.

Willis (1887) describes terraces in the Sinlahekin Valley, one of the West's hidden gems,

[In the Sinlahekin Valley, from Miner's Bend [Shanker's Bend] to Wagon Road pass, is a comparatively broad valley, a strip of marsh and lake be­tween steep mountain slopes. The terraces on its sides are continuous with others, which cling to the walls of the Similkameen Canon ...and they extend southward to the Three Pools [Blue Lakes], where they merge into the highest portion of the drift filling.

...between Wagon Road pass and Fish Lake, is a drift clogged canon, with abrupt granite walls of considerable height; its northern half presents a very gentle northward slope. From the Three Pools southward the drift surface is dotted with kettle-holes. Ter­races again appear on the slope of the mountain northeast of Fish Lake...held by a gravel dam...The descent from Fish Lake is very rapid and is strewn with bowlders of large size. After passing a very deep and narrow pool three-fourths of a mile long, into which the brook sinks, the valley opens out between limestone bluffs [lime belt between Wagonroad Coulee] and drift terraces, and ends abruptly in a cul de sac at the head of Johnson Creek [north of Omak]...

The valley of the [Okanogan River] belonged no doubt to the great system of which all these now abandoned channels were part. Like them, it was drift buried, but, unlike them, not dammed, because of its great width. The present river is a quiet stream, having a fall of about 3 feet per mile from Lake Osoyoos to the Columbia River, and usually flowing between terraces 400 feet high. The lake [Lake Osoyoos] is but a shallow expansion of the river, retained by a drift deposit.

High gravel terrace above the Pend Oreille River. O.P. Jenkins photo 1924. WGS photo archive.

Antoine Coulee. K. Stoffel photo in 1987. WGS photo archive.

Bretz (1923) at Moses Coulee notes,

The rock floor of Moses Coulee is fully as low as that of Columbia Valley at the junction of the two. Both contain a great gravel fill here. Columbia River has cut through it, a depth of more than 300 feet. There is no such trenching in Lower Moses Coulee, but a well at Appledale penetrates 300 feet of this gravel without encountering bedrock.

Waters (1933),

Down the Columbia from [Chelan Butte] the walls of the tributary valleys are free and unencumbered save in their lowermost parts. Upstream they are choked through­ out their entire length by dozens of terraces, which convert each tributary valley into a gigantic stairway...

Muessig (1967), though focused mainly on the bedrock in the Curlew area, noted,

The [Sanpoil-Curlew] valley contains many terraces, most of which do not seem to match across it, but several prominent ones at around 2,500 feet do appear to match, and, as far as I could determine with a hand level, they slope south.

All the valleys tributary to the Sanpoil-Curlew drainage have outwash terraces along them. For example…South Fork of O'Brien Creek…Most of the terraces cannot be traced very far and many cannot be matched with others across the valley. Some, with knolls and kettles, are obviously kame terraces; others are just outwash terraces…The great number of terraces, their random distribution, and their generally short extent indicate a complex history whose unraveling has not been attempted. The glaciofluvial deposits in the Sanpoil-Curlew valley are not much different from those in the tributaries, except that they are more extensive and…more striking...[and] ...composed of a heterogeneous accumulation of outwash material ranging in size from gravel to silt and locally to clay. Most of it is poorly sorted and does not reflect the lithologies of the nearby rocks.

The topography imposed on the area by glaciation has been virtually unmodified in postglacial time.

Terraces near Chelan. R.W. Galster photo in 1976. WGS photo archive.

Hanson (1970) recognized a connection between terraces in Foster Valley and those in Columbia Valley,

The torrential sediment fill in this embayment is preserved as a series of gravel terraces (dissected kame terraces) at and north of the junction of East and West Foster Creek Valleys. This terrace may relate to "The Great Terrace" which so prominently borders the Columbia River downstream...

Russell and Eddy (1971) report,

The topography of the [Aeneas Lake] area indicates kettle plain geomorphology. Kame terraces flank both sides of the coulee and Aeneas Lake [SW of Tonasket] is an excellent example of a kettle lake. The uplands to the southeast also show the classic [kame] mounds of sand and gravel. The valley floor consists of ground morainal and outwash material. This material is overlain by alluvial fans which were generated from the terrace material flanking the valley floor.

Under the sub-heading 'problematical landforms', Hunt (1977) notes,

Besides the "Scablands" channels, there are several landforms in the Columbia system that require explanations. The "Lake Missoula" beaches present an enigma as they occur not only on Clark Fork, but...the Pend Oreille lake basin. Had they developed behind an ice dam in Clark Fork, they should stop there.

Fieldwork by horseback. Index map of the region reconnaissance geologist Israel Cook Russell (once called "Gilbert's boy Russell") was responsible for surveying. His perceptive, wide ranging, and often poetic findings are documented in USGS Bulletin 108, "Geological Reconnoissance in Central Washington" (Russell, 1893).

Landscape sketch. Terraces at Okanagan Falls near Skaha in south-central BC (Nasmith, 1962, Plate II, p. 22). There are obvious landform, sedimentation, and process connections between the Okanagan (BC) and Okanogan (WA).

Age of the Great Terrace

This summary by Pine (1985) provides a jumping-off point for future geochronometric work,

The apparent absence of Glacier Peak tephra layer G from all but the lowermost 15 to 20 km of the Okanogan Valley suggested to Porter (1978) that the valley was still covered by ice at the time of the eruption about 11,200 yr BP (Mehringer and others, 1984).

Historic photo. Flights of glacial terraces in the Foster Creek Valley near Bridgeport, WA (Russell, 1893). Russell joined the USGS survey party as the photographer, a skill he had to pick up quickly, which he did.

Bedrock Spillover Channels and Subglacial Tunnel Valleys

The origin of deep, narrow bedrock gorges that parallel the main Okanogan Valley (the 'spillways' of Waters, 1933) is controversial. The valleys are often curvy, scoured free of sediment, and coursed today by woefully underfit streams. In other places, they are completely filled and make no modern topographic expression (i.e., Kettle Falls area). Either they were carved directly by glacial ice (conventional view), by pressurized subglacial jets of water (Lesemann and Brennand, 2009), or by ice-marginal streams.

According to Smith and Calkins (1904),

Throughout northern Washington it is apparent that stream erosion may have accomplished tasks that seem incredible, and nowhere is its efficiency more forcibly shown than in Okanogan Valley...

Along the sides of Columbia and Okanogan valleys there are numberless side channels of peculiar history...a perfect labyrinth of deserted stream channels...which commonly take the name of "coulees"...are so complex that a careful and detailed survey would be necessary before a correct interpretation of the drainage changes could be made.

On the east wall of Okanogan Valley there are several parallel gaps steep-sided gashes...striking features...

Thirteen miles due west, on the opposite side of the valley...a similar series of gaps, which plainly represent abandoned channels of Salmon Creek...successively occupied and deserted by this stream...There are four of them, the floors of which are successively higher...

Antoine Coulee near the mouth of the Methow is a similar feature on a much larger scale...10 miles in length...its floor is approximately at the level of the Great Terrace of the Columbia. This steep-walled defile, which cuts across the spurs sloping down to Columbia River, is believed to be also an abandoned stream channel.

Nasmith (1962) noted a lack of glacial scour at both both Antwine and Knapp Coulees near Chelan,

I traversed the entire length of Antwine's coulee [southern continuation of Alta Coulee] last September, but failed to find any evidence to suggest that it had been occupied by a glacier....entirely without glacial scoring, moraines or other evidence of ice work.

I traversed the entire length of Knapp's coulee last September, and, as in the similar instances cited above, failed to find any evidence of the work of ice...

Not so amateur. Wooten (2008) mapped "fluvioglacial bedrock channels" (pink areas), finding they occur in clusters and parallel trunk valleys.

Tonasket fan. Deeply-incised glacial terraces at Tonasket (Keyes, 1917, p. 3). Photo taken near the Tonasket Airport looking SW across the Okanogan River and Hwy 97. The town of Tonasket sits on a broad alluvial fan inset into the set of higher terraces and formed after ice retreated north. No one in Tonasket has ever been addicted to nightlife.

Feel free to contact me with comments, corrections, or additional information on the Great Terrace:


Barksdale, J.D., 1975, Geology of the Methow Valley, Okanogan County, Washington, Washington Division of Geology and Earth Resources Bulletin, v. 68, 72 pgs. + 1 plate

Bretz, J. 1923, Glacial drainage on the Columbia Plateau, Geological Society of America Bulletin, v. 34, p. 573-608

Bretz, J. 1932, The Grand Coulee, American Geographical Society Special Publication No. 15, 89 pgs.

Cox, R., 1831, Adventures on the Columbia River...

Dawson, W.L., 1898, Glacial phenomena in Okanogan County, Washington, American Geologist; v. 22, p. 203-217

Flint, R., 1935, Glacial features of the southern Okanogan region, Geological Society of America Bulletin, v. 46, p. 169-194

Flint, R., 1936, Stratified drift and deglaciation in eastern Washington, Geological Society of America Bulletin, v. 47, p. 1849-1884

Flint, R.; Irwin, W., 1939, Glacial geology of Grand Coulee Dam, Washington, Geological Society of America Bulletin, v. 50, p. 661-680

Freeman, O. W., 1933, Stagnation of the Okanogan lobe of the Cordilleran ice sheet and the resulting physiographic effects, Northwest Science, v. 7, p. 61-66

Fryxell (1973)...??

Keyes, C., 1917, High level terraces of Okanogan Valley, Washington, Iowa Academy of Sciences, v. 24, p. 47-51

Lesemann, J.; Brennand, T.A., 2009, Regional reconstruction of subglacial hydrology and glaciodynamic behavior along the southern margin of the Cordilleran Ice Sheet in British Columbia, Canada and northern Washington State, USA, Quaternary Science Reviews, v. 28, p. 2420-2444

Pine, K.A., 1985, Glacial geology of the Tonasket-Spectacle Lake area, Okanogan County, Washington, MS thesis, Western Washington University

Rinehart, C.D., and Fox, K.F., Jr. 1972. Geology and mineral deposits of the Loomis quadrangle, Okanogan County, Washington: Washington Division of Geology and Earth Resources Bulletin 64, 124 p. + 3 plates

Russell, I.S., 1893, A geological reconnaissance in central Washington: U.S.Geological Survey Bulletin 108, 108 pgs.

Russell, R.H.; Eddy, P.A., 1971, Geohydrologic evaluation of Aeneas Lake - Horse Springs Coulee, Okanogan County, Washington, Washington Department Ecology Technical Report 72-2, 31 pgs.

Salisbury, R.D., 1901, Glacial work in the western mountains in 1901, Journal of Geology, v. 9, p.718-731

Smith, G.O.; Calkins, F.C., 1904, A geological reconnaissance across the Cascade Range near the 49th Parallel, USGS Bulletin 235

Waitt, R.B.; Atwater, B.F.; Lehnigk, K.; Larsen, I.J.; Bjornstad, B.N.; Hanson, M.A.; O'Connor, J.E., 2021, Upper Grand Coulee: New views of a channeled scabland megafloods enigma, GSA Field Guide 62

Waitt, R.B., Denlinger, R.P.; O'Connor, J.E., 2009, Many monstrous Missoula floods down Channeled Scabland and Columbia Valley, in O'Connor, Dorsey, and Madin (editors), Volcanoes to Vineyards: Geologic field trips througt dynamic landscape of the Pacific Northwest, GSA Field Guide, v. 15, p. 775-844

Waitt, R.B., 2017, Pleistocene glaciers, lakes, and floods in north-central Washington State, GSA Field Guide 49, p. 175-205

Waitt, R.B., 1972 Geomorphology and glacial geology of the Methow drainage basin, eastern North Cascade Range, Washington, PhD Dissertation, University of Washington, 154 pgs.

Waters, A.C., 1933, Terraces and coulees along the Columbia River near Lake Chelan, Washington, Geological Society of America Bulletin, v. 44, p. 783-820

Willis, B., 1887, Changes in river courses in Washington Territory due to glaciation, USGS Bulletin 40, 10 pgs.

Terraces at Brewster Flats. Southern Okanogan Valley. R.W. Galster photo in 1976. WGS photo archive.


Last 50 Posts
All Posts by Month