The source of this uncorrected OCR text may be viewed in the DjVu format at: http://fax.libs.uga.edu/J84xNASx1x18xCx36/ or http://purl.galileo.usg.edu/ugafax/J84xNASx1x18xCx36/ The Channeled Scabland A Guide to the Geomorphology of the Columbia Basin, Washington Prepared for the Comparative Planetary Geology Field Conference held in the Columbia Basin June 5-8, 1978 edited by VICTOR R. BAKER Department of Geological Sciences The University of Texas at Austin Austin, Texas 78712 and DAG NUMMEDAL Department of Geology The University of South Carolina Columbia, S. C. 29208 Sponsored by Planetary Geology Program Office of Space Science National Aeronautics and Space Administration Washington, D. C. 20546 Cover: Dry Falls Cataract, Grand Coulee, Washington. Photo 5599 by John S. Shelton. Preface As the Mariner and Viking spacecrafts photo graphed the large channels on Mars, they gen erated a renewed interest in the erosional effects of large quantities of running water. As the debate continues whether these channels were caused by catastrophic floods, wind, lava or mudflows, it is imperative to examine, in the field, the morpho logic criteria which appears to be diagnostic of catastrophic flooding. Hence this field conference to the Channeled Scabland. Although we now propose to use the Channeled Scabland as a testing area for the Martian flood hypotheses, the flood origin of the scabland itself has a history of intense controversy. Ironically, many of the arguments against catastrophic floods on Mars are the same as those advanced against Bretz' "outrageous hypothesis" for the scabland origin. These include the source of such large quantities of water, the mechanism of its sudden release and the magnitude and nature of sub- fluvial erosion. Also, some of the alternative fluids proposed for channel generation on Mars were advanced in the debate about the origin of the Channeled Scabland. THE CHANNELED SCABLAND intends to provide a thorough description of the morphology of the region and a discussion of its inferred cause, including the hydrodynamics of high-veloc ity flood erosion. It is designed as a guide, written specifically for the Comparative Planetology Field Conference in the Columbia Basin in June, 1978. The editors hope, however, that the book will provide a useful guide also for future field trips and an introduction to the area for geologists who intend to launch their own studies on spe cific attributes of this fascinating landscape. The book is dedicated to J Harlen Bretz in recognition of his pioneering advocacy of the flood origin of the scablands. It is also an expres sion of our appreciation of the many other scien tists who directly or indirectly have contributed to our understanding of the scabland origin. The ideas on the Channeled Scabland set forth in this book have matured through discussions with nu merous colleagues, including Jon C. Boothroyd, William C. Bradley, J Harlen Bretz, Robert K. Fahnestock, Roald Fryxell, Harold E. Malde, George E. Neff, Peter C. Patton, Russell G. Shepherd, Richard B. Waitt, Jr. and Paul L. Weis. The field studies and analytical work on the problem involved the assistance of Pauline M. Baker, Frances A. Heaton, R. Craig Kochel, and John M. MacGregor. Field work by V. Baker was initially supported by National Science Foun dation Grant GA-21478. Subsequent studies were supported by NASA Grants NGR 44-012194 and NSG-7326, Planetology Program Office, and by The Geology Foundation, The University of Texas at Austin. Field studies by D. Nummedal have been supported by NASA Grant NSG-7076, Planetology Program Office, and the University of South Carolina. P. Jeffrey Brown and Duane T. Eppler, the University of South Carolina, provided great help in guidebook editing, logistics planning and co ordination of the field trip. We extend our sincere appreciation to John S. Shelton for providing numerous illustrations for the book and to Henry D. Neumann for assistance in obtaining the photos for the color plates. Neither the book nor the field trip would have been possible without the enthusiastic support by Stephen E. Dwornik, translated into financial assistance through grants to the participating prin cipal investigators in NASA's planetary geology program. We also thank Priscilla Ridgell and Carleen Sexton for the typing, Burk Scheper for the photo- technical work and Nanette Muzzy for the graph ics. D.N. V.R.B. in \ Biographical Sketch of J Harlen Bretz .1 J HARLEN BRETZ This book is dedicated to Dr. J Harlen Bretz, who, by painstaking field work, scientific insight and firm conviction, proved the catastrophic flood origin of the Channeled Scabland. J Harlen Bretz was born on September 2, 1882, in Saranac, Michigan. He attended Albion Col lege, where he studied biology and met classmate Fanny Challis, whom he married in 1906. As a high school teacher in Flint, Michigan and later Seattle, Washington, he developed an avid interest in geology. His spare time was consumed with studies of glacial geology in the Puget Sound region. His hobby became his profession when the Puget Sound study became the basis for a Ph.D. summa cum laude at the University of Chicago. The entire dissertation was published (Bretz, 1913) and is distinguished for his revisions of the area's glacial history which had been de scribed earlier by Bailey Willis and George Otis Smith. Bretz was also the first to recognize the significance of the great Osceola mudflow from Mount Rainier. Bretz next spent a year on the faculty of the University of Washington. Unfortunately, his col leagues there did not share his zeal for field work, and Bretz accepted an invitation from R. D. Salis bury to return to the University of Chicago. He began teaching a field course at Devil's Lake, Baraboo, Wisconsin. His love for field work soon brought him back to Washington for summer field courses. Starting in the Columbia Gorge be tween Oregon and Washington, his field course soon moved into the Channeled Scabland as Bretz formulated his flood hypothesis (Baker, Ch. 1, this volume). While a faculty member at the University of Chicago for over 30 years, 300 students com pleted their field geology training with "Doc" Bretz. The graduates include E. Dorf, M. K. Hub bert, W. C. Krumbein, J. L. Hough, A. N. Strahler, and H. S. Yoder, to name just a few. He wrote 20 major articles and monographs on the Channeled Scabland, served as an associate editor of the Journal of Geology and was asso ciated with the Washington, Illinois and Missouri Geological Surveys. In the 1930's, his research turned to physiographic studies in Greenland and the geology of the Chicago region. His two mono graphs on Chicago contain an ingenious analysis of the draining of Glacial Lake Chicago (a prede cessor of modern Lake Michigan). From 1938 to 1961, Bretz established one of the most important American schools of thought on the origin of limestone caverns. His studies of caves in 17 states, Mexico, and Bermuda placed speleology on a firm scientific base. His insights and energy are a major reason for the modern resurgence of karst geomorphic and hydrologie studies in the United States. Dr. Bretz officially retired from the University of Chicago in 1947. However, his years as Pro fessor Emeritus were nearly as productive scien tifically as those on the active faculty. With Leland Horberg, he published the first modern geological analysis of petrocalcic soil formation (caliche). Perhaps the most extensive survey of caves in one state was his book "Caves of Mis souri" (1956). In the 1960's, Bretz' monumental analysis of geomorphic history in the Ozarks of Missouri had him questioning the new geomor phic paradigm of "dynamic equilibrium." He is a member of the American Association for the Advancement of Science and a fellow of the Geological Society of America. Today, at age 95, Dr. Bretz still maintains a vigorous correspondence with former students and colleagues from his home at 2114 Cedar Road, Homewood, Illinois 60430. IV Contents Preface ni Biographical Sketch of J Karlen Bretz v Introduction 1 / Harlen Bretz Chapter 1. The Spokane Flood Controversy 3 Victor R. Baker Chapter 2. Quaternary Geology of the Channeled Scabland and Adjacent Areas 17 Victor R. Baker Chapter 3. Bedrock Geology of the Northern Columbia Plateau and Adjacent Areas 37 Donald A. Swanson and Thomas L. Wright Chapter 4. Paleohydraulics and Hydrodynamics of Scabland Floods 59 Victor R. Baker Chapter 5. Large-Scale Erosional and Deposi- tional Features of the Channeled Scabland 81 Victor R. Baker Chapter 6. Origin of the Cheney-Palouse Scabland Tract 117 Peter C. Patton and Victor R. Baker Color Plates Chapter 7. Field Trip Stop Descriptions 131 / Dag Nummedal Chapter 8. Aerial Field Guide 169 Dag Nummedal Chapter 9. The Touchet beds of the Walla \ Walla Valley 173 """ Robert J. Carson, Charles F. McKhann and Mark H. Pizey References 179 \ , I Vll Introduction J HARLEN BRETZ Pro/es.vor Emeritus of Geology The University of Chicago Chicago, Illinois 60637 The publication in 1923 of a geomorphic study of "The Channeled Scabland of the Columbia Plateau" in southeastern Washington state launched a controversy that lasted for decades. A map accompanying the paper depicted a pat tern of abandoned erosional waterways, many of them streamless canyons (coulees) with former cataract cliffs and plunge basins, potholes and deep rock basins, all eroded in the underlying basalt of the gently southwestward dipping slope of that part of the Columbia Plateau. The pattern of dry stream ways was described as at network, a plexus, an anastomosis; totally unlike any other drainage pattern on earth. A debacle was asked for, the volume of which filled existing normal stream valleys to overflowing. This great flood spilled over former divides, eroding their sum mits to complete the network. Associated with the enormously enlarged drainage ways in favor able places were similarly huge mounds of stream gravel which the writer called great river bars. Huge stream-rolled boulders occurred in these bars. The boulders were obviously plucked from the columnar basalt bedrock by the postulated high-velocity currents. The term valley would no longer suffice. The abandoned rock-bound former waterways were called channels, and the entire composite was named "Channeled Scabland." The total area in volved was 18 townships wide by 22 townships long (approximately 40,000 km2). The Cordilleran ice sheet had advanced to make contact with the heads of the dominant channel ways. Obviously the plexus of channels was in some way genetically related to the north ern ice sheet. But where on our planet had glacial wastage produced a record of flooding and the total remaking of existing topography? The writer called for a catastrophic event. Catastrophism had virtually vanished from geologic thinking when Button's concept of "the Present is the key to the Past" was accepted and Uniformitarianism was born. Was not this debacle that had been deduced from the Channeled Scabland simply a return, a retreat to catastroph- ism, to the dark ages of geology? It could not, it must not be tolerated. This, the writer of the 1923 article learned when, in 1927, he was invited to lecture on his finding and thinkings before the Geological So ciety of Washington, D.C. an organization heavily manned by the staff of the United States Geo logical Survey. A discussion followed the lecture, and six elders spoke their prepared rebuttals. They demanded, in effect, a return to sanity and Uniformitarianism. The upstart theorist was not upset nor silenced. Despite his knowledge that the country was full of other dissenters to his flood theory, he pro ceeded to publish more papers on his favorite topic, now named the Spokane Flood. He de scribed other features of the afflicted plateau which he claimed were inexplicable without his flood of glacially derived meltwater. His apostacy would not be corrected as advised by the elders. The one-man rebellion was still alive. 1924 saw two new papers, 1925 saw one more and 1927, three more. One of the three con tained the lecture and discussion already noted. Another paper in 1928 was the writer's reply to all alternative hypotheses thus far suggested. Also in 1928, he traced his flood down the Columbia as far as Portland, Oregon, adding a 200 square mile delta in the Willamette Valley. By 1930, he had found a source for that im mense discharge across the plateau. Clark Fork of the Columbia River, draining a large moun tainous region of western Montana, had been dammed by the Cordilleran ice sheet at its traverse of the Idaho panhandle. This formed an immense glacial lake with an estimated volume of 500 cubic miles. The lake had been named some years before as Glacial Lake Missoula. The first geologist to describe the lake ironically was one of the six challenging elders in Washington in 1927 and the author of a short papefl on problematical features, perhaps glacial in origin, in what came to be known as the Channeled Scabland. If Lake Missoula had a properly located place for its ice dam and a clear route thence to the Channeled Scabland, then presto, we would have the big problem solved. Missoula's depth at the dam was known from its shorelines to have been 2000 feet, and there was a clear route to Spokane and the Scabland. A catastrophic failure of the dam would release 500 cubic miles of glacially derived water with adequate gradient to Spokane. Late in the field study, a criterion of unde niable validity for the occurrence of a flood, or several such, came to light. Hidden largely by sagebrush were numerous occurrences of current ripple marks. They were discovered because the U.S. Bureau of Reclamation had taken aerial photographs of the area to be irrigated with Grand Coulee water. Then it became clear that some gravel surfaces, curiously humpy, were cov ered with giant current ripples. An investigator, standing between two humps, could not see over either one. Indeed, the size of these ripple ridges made them really small hills. Finally came the discovery of giant current ripples in parts of Lake Missoula where, in a catastrophic emptying, strong currents were formed. When investigated by a specialist in fluvial hydraulics and hydro dynamics there could be no question whatever of their origin, no other explanation for their rhythmic patterns than that of bedform develop ment by amazingly deep, swift flood water. Measurements of records for depths of water and gradients of water surfaces in channels with proper cross sections also yielded amazing figures for velocities. Plucking of channel bedrock to yield and transport huge boulders became under standable when this 1973 paleohydraulic and hydrodynamic study was made. Back in 1927 geologic opinion still wavered. One man accepted flooding but did it with local iceberg dams in different channels. Another was a die-hard Uniformitarian who in 1938 still argued that no flood had occurred; that the erosional complex on the plateau was made by "leisurely streams no larger than the Snake in flood today." (But that was before the giant ripple marks were discovered.) "Meticulously detailed" criticisms of the gross errors and assumptions in the 1938 non-flood interpretation appeared when three geologists spent much of the summer of 1952 (published in 1956) in studying chiefly the excavations made by the U.S. Bureau of Reclamation in diverting Columbia River water behind Grand Coulee dam. This water irrigates from a million acres of the Quincy Basin, an extensive fill made by scabland floods. There were repeated burstings of Lake Missoula dams and refillings when later advances of the ice front made other episodes in the lake's history. How many is not yet definitely deter mined, despite recent and continuing research on this problem. When the Geological Society of America held its annual meeting in Seattle in November, 1977, and field trips were made to the flood-ravaged Columbia Plateau and valley, guidebooks made clear that the catastrophic history of plateau and master river valley is being pursued with enthus iasm. Chapter 1 The Spokane Flood Controversy VICTOR R. BAKER Department of Geological Sciences The University of Texas at Austin Austin, Texas 78712 ABSTRACT The Spokane Flood controversy is both a story of ironies and a marvelous exposition of the scientific method. In a brilliant series of papers between 1923 and 1932, J Harlen Bretz shocked the geological community with his studies of an enormous plexus of proglacial channels eroded into the loess and basalt of the Columbia Plateau, eastern Washington. This region, which he named the "Channeled Scabland," contained erosional and depositional features that were unique among fluvial phenomena. With painstaking field work, before the advent of aeria] photographs and modern topograhic maps, Bretz documented the field relationships of the region. He argued that the landforms could only be explained as the product of a relatively brief, but enormous flood, which he called the "Spokane Flood." Consider ing the nature and vehemence of the opposition to this outrageous hypothesis, the eventual tri umph of that idea constitutes one of the most fascinating episodes in the history of modern geomorphology. INTRODUCTION The inimitable words of J Harlen Bretz ( 1928c, p. 446) describes the scene in eastern Wash ington: "No one with an eye for landforms can cross eastern Washington in daylight without en countering and being impressed by the "scab- land". Like great scars marring the otherwise fair face of the plateau are these elongated tracts of bare, or nearly bare, black rock carved into mazes of buttes and canyons. Everybody on the plateau knows scabland. It interrupts the wheat lands, parceling them out into hill tracts less than 40 acres to more than 40 square miles in extent. One can neither reach them nor depart from them without cross ing some part of the ramifying scabland. Aside from affording a scanty pasturage, scabland is almost without value. The popular name is an expressive metaphor. The scablands are wounds only partially healed—great wounds in the epi dermis of soil with which Nature protects the underlying rock. With eyes only a few feet above the ground the observer today must travel back and forth repeatedly and must record his observations mentally, photographically, by sketch and by map before he can form anything approaching a complete picture. Yet long before the paper bearing these words has yellowed, the average observer, looking down from the air as he crosses the region, will see almost at a glance the picture here drawn by piecing together the ground-level observations of months of work. The region is unique: let the observer take tho wings of the morning to the uttermost parts of the earth: he will nowhere find its likeness. Conceive of a roughly rectangular area of about 12,000 square miles, which has been tilted up along its northern side and eastern end to produce a regional slope approximately 20 feet to the mile. Consider this slope as the warped surface of a thick, resistant formation, over which lies a cover of unconsolidated ma terials a few feet to 250 feet thick. A slightly irregular dendritic drainage pattern in maturity has been developed in the weaker materials, but only the major stream ways have been eroded into the resistant underlying bed rock. Deep canyons bound the rectangle on the north, west, and south, the two master streams which oc cupy them converging and joining near the southwestern corner where the downwarping of the region is greatest. Conceive now that this drainage system of the gently tilted region is entered by glacial waters along more than a hundred miles of its northern high border. The volume of the in vading water much exceeds the capacity of the existing stream ways. The valleys entered be come river channels, they brim over into neigh boring ones, and minor divides within the sys tem are crossed in hundreds of places. Many of these divides are trenched to the level of the preexisting valley floors, others have the weaker superjacent formations entirely swept off for many miles. All told, 2800 square miles of the region are scoured clean onto the basalt bed rock, and 900 square miles are buried in the debris deposited by these great rivers. The topographic features produced during this epi sode are wholly river-bottom forms or are com- t. pounded of river-bottom modifications of the invaded and over-swept drainage network of hills and valleys. Hundreds of cataract ledges, of basins and canyons eroded into bed rock, of isolated buttes of the bed rock, of gravel bars piled high above valley floors, and of island hills of the weaker overlying formations are left at the cessation of this episode. No fluvia- tile plains are formed, no lacustrine flats are deposited, almost no debris is brought into the region with the invading waters. Everywhere the record is of extraordinarily vigorous sub- fluvial action. The physiographic expression of the region is without parallel; it is unique, this channeled scabland of the Columbia Plateau." A mere glance at a modern LANDSAT photograph of the Channeled Scabland (Fig. 1.1) *.* - ' • . '" I . . • t ' • ' j v » t~ « 4 r T \ * S. l. r t r ft V » » * . l- -*. V-// , » / Figure 1.1. LANDSAT photograph of the northern photograph. The far left scabland complex is the Grand part of the Channeled Scabland. Scabland channels form Coulee-Hartline Basin-Lenore Canyon tract. At the the dark-toned anastomosis that contrasts with the wheat center is the Telford-Crab Creek Scabland complex. At farms on the light-toned Palouse loess. The Columbia the right (east) is the Cheney-Palouse scabland tract and Spokane Rivers occur at the top (north) of the (LANDSAT E-1003-18150 composite, 26 July 1972). will show the features that Bretz, studying from the ground, developed as the basis of his flood hypothesis. The extensive wheat cultivation on the loess presents a vivid contrast to the flood- scared basalt exposed in the channel ways. The unique character of the dry river courses ("coulees") of the Channeled Scabland was ap preciated by the first scientific observers of the region. Rev. Samuel Parker (1838) provided the first published statement on the Grand Coulee: "[it] was indubitably the former channel of the river [Columbia]." Lieutenant T. W. Symons (1882) of the U.S. Army traversed the Grand Coulee, stating that he, "went north through the coulee, its perpendicular walls forming a vista like some grand old ruined roofless hall, down which we traveled hour after hour." Symons (1882) initiated the widely held notion that dur ing glacial episodes of the Pleistocene the Colum bia had simply been diverted across the Colum bia Plateau. Variations on this general theme were standard in the early literature (Russell, 1893; Dawson, 1898; Salisbury, 1901; Calkins, 1905). The Grand Coulee gained international fame in 1912 when it was traversed by the American Geographical Society's Transcontinental Excur sion. Karl Oestreich (1915) of the University of Utrecht described the coulee as "eines mächtigen Flusses Bett . . . ohne jede Spur von Zerfall der frischen Form." He provided an excellent de scription of significant features that required a special origin: exhumed granite hills, perpendic ular walls, and the hanging valleys marginal to the upper Coulee. He ascribed these hanging valleys to glacial erosion and to deepening of the coulee by the glacial Columbia River. Moreover, he recognized that the upper Grand Coulee was carved through a preglacial divide, which he cor rectly located just north of Coulee City. Another foreign observer on the American Geographical Society excursion was H. ßaulig, University of Rennes. Baulig (1913) described the loess, coulees, dry falls ("cataracte desséchée de la Columbia"), rock basins, and plunge pools. The origin of these features was ascribed to a glacial diversion of the Columbia. Nevertheless, he marveled at the, scale of erosion (Baulig, 1913, p. 159): "peutêtre unique du relief ter restre,—unique par ses dimensions, sinon par son origine." Dr. O. E. Meinzer, the eminent hydrologist of the U.S. Geological Survey, took an early interest in the western part of the Channeled Scabland. He observed (Meinzer, 1918) that the glacially diverted Columbia at Grand Coulee "cut precipitous gorges several hundred feet deep, de veloped three cataracts, at least one of which was higher than Niagara, . . . and performed an al most incredible amount of work in carrying boulders many miles and gouging out holes as much as two hundred feet deep." He implies that the great erosion occurred because the Columbia River was diverted across the steeply dipping basalt surface of the northern Columbia Plateau. It was not until the studies of J Harlen Bretz (1923-1932) that the scientific study of this re gion began in earnest. Brotz interpreted the ero- sional and depositional features of the region as the product of a brief but enormous flood, which he called the Spokane Flood. For geology in the 1920's this was clearly an outrageous hypothesis. Oison (1969) has described the reception of the idea. "During its not always calm history, the story of the development of the Channeled Scab- land was thought by some to have brushed be yond the dividing line in flaunting catastrophe too vividly in the face of the uniformity that had lent scientific dignity to interpretation of the his tory of the earth." The reaction of the scientific community was predictable, "this heresy must be gently but firmly stamped out" (Bretz and others, 1956, p. 961). AN OUTRAGEOUS HYPOTHESIS Because the Spokane Flood controversy is so tied to Bretz as its central figure, this review will consider part of his professional career during the years of his formulation of the flood hypothe sis. The ensuing debates were not always marked by scientific objectivity, but their recounting is a fascinating example of the triumph of an out rageous hypothesis. Only in the last two decades has the flood hypothesis gained general ac ceptance. It is a measure of scientific maturity that in current studies of the Channeled Scabland, "the idea, but not the man has become central" (Olson, 1969). While teaching at the University of Chicago, Bretz began conducting a summer field course in the wilds of the Columbia Gorge between Washington and Oregon. The idea for a study of the Channeled Scabland came during the summer of 1922. As he relates the story, "One summer I was out in Spokane. I saw a section of a topo graphic map of what is now called the Channeled Scabland, and from that I got the idea" (Quota tion from Seattle Times, Sunday Magazine, July 11, 1971, p. 13). Without the benefit of modern aerial photo graphs or even adequate topographic map cov erage, Bretz began to take parties of advanced students into the region for month-long field studies. The work continued over the next 7 years. He soon revised an earlier notion that a marine submergence had occurred just down stream from the Channeled Scabland (Bretz, 1919). Nevertheless, the erratic granite boulders, which he had used as evidence for the submerg- • ence, were scattered about the basalt plateau far beyond the limits reached by Pleistocene glacia tion. Bretz (1923a) named the glaciation re sponsible for these erratics the "Spokane Glacia tion." Although his first paper on the Channeled Scabland (actually the text of an oral presenta tion to the Geological Society of America) took care not to call upon cataclysmic origins, Bretz (1923a) provided a detailed description of physiographic relationships in the region. An example in his description (Bretz, 1923a, p. 601) of the pre-flood drainage line that was later en larged to form the lower part of Moses Coulee: "The cliffs here are deeply notched by wide-open V-shaped tributary valleys. . . . These notches give the cliffs a striking resemblance to a series of great rounded gables in alignment. . . . Both widening and deepening in the basalt occurred and the tributaries were left hanging. They have since attained topograhpic adjustment by build ing large alluvial fans out on the canyon floor." He further noted that prodigious quantities of water were involved in the erosion. Referring to three outlets at the south end of the Hartline Basin (Dry Coulee, Lenore Canyon, and Long Lake Canyon), Bretz (1923a, p. 593-594) states, ". . . these are truly distributary canyons. They mark a distributive or braided course of the Spokane glacial flood over a basalt surface which possessed no adequate pre-Spokane valleys." Bretz (1923a, p. 603) originally thought that the scabland gravels were organized into terrace remnants. However, after noting that they lacked a "sharp terrace form," this interpretation was quickly modified (Bretz, 1923b, p. 643): ". . . the evidence seems conclusive that all gravel de posits of the scablands are bars, built in favorable situations in the great streams which eroded the channels." With this conclusion he was forced to call upon catastrophic quantities of water. If the bars were over 100 feet in height, even greater water depths were required to form them. The second paper (Bretz, 1923b) also included the first detailed geomorphic map of the entire Channeled Scabland, showing the overall anas tomosing pattern assumed by a great flood of water. Bretz (1923b, p. 624-626) was the first to recognize the streamlined loess hills of the Cheney-Palouse scabland. He described them as follows: "A very striking and significant feature of the steepened slopes is their convergence at the northern ends of the groups to form great prows, pointing up the scabland's gradient. . . . The nose of a prow may extend as a sharp ridge from the scabland to the very summit of the hill. It is im possible to study these prow-pointed loessial hills, surrounded by the scarred and channeled basalt scablands, without seeing in them the result of a powerful eroding agent which attacked them about their bases and most effectively from the scabland's up-gradient direction." Bretz knew that his interpretation would be controversial. He argued (Bretz, 1923b, p. 621), "All other hypotheses meet fatal objections. Yet the reader of the following more detailed de scriptions, if now accepting the writer's interpre tation, is likely to pause repeatedly and question that interpretation. The magnitude of the erosive changes wrought by these glacial streams is noth ing short of amazing." Bretz subsequently argued that the rugged scabland of anastomosing channels and rock basins cut into the basalt was the product of sub- fluvial quarrying. He described this process for the modern Columbia River near The Dalles, Oregon (Bretz, 1924). Moreover, he asserted that only large vigorous streams could produce such forms.1 The eventual conclusion from these varying lines of evidence was that so much glacial meltwater occupied the pre-existing valleys on the Columbia Plateau that it must have constituted a vast but short-lived flood, the "Spokane flood" (Bretz, 1925, p. 98). The flood spilled across pre-glacial stream divides, eroding the maturely dissected loess topography to form linear chan nels, and leaving a legacy of scoured loess scarps, hanging distributary valleys, and high-level fluvial deposits. It also built the huge constructional bars of gravel and then subsided so quickly that these bedforms were left almost unmodified by the sub siding water (Bretz, 1925, p. 105). Bretz (1925) was able to trace the path of the great flood downstream through the Columbia Gorge to its debouchure into the Willamette low land, where it built the "Portland delta." On this great subfluvial fan he recognized the signif icance of macroturbulence in accounting for certain flood features: "The Rocky Butte fosse is but the unfilled locus of an eddy caused by downward deflection where the current impinged on the east face of the butte. . . . The dependent terrace to the west was deposited in the slack water below the obstruction" (Bretz, 1925, p. 256). Bretz (1925) even made the first estimate of the flood discharge. He chose Wallula Gap for this calculation because of the ponding effect of the constriction. His calculated maximum flow rate was 1.9 x 10« m3/s (66.1 x 10« cfs), but he noted that this erred toward the low side. Nevertheless, he stated, "it represents the melting of about 42 cubic miles of ice daily" (Bretz, 1925, p. 258). He then notes that the insolation properties of ice and the total available ice mass north of the Channeled Scabland brings the whole concept into doubt. "The writer," he says (Bretz, 1925, p. 259), "has repeatedly been driven to this position of doubt, only to be forced by reconsideration of the field evidence to use again the conception of enormous volume. . . . These remarkable records of running water on the Columbia Plateau and in the valleys of the Snake and Columbia Rivers cannot be interpreted in terms of ordinary river action and ordinary valley development. . . . Enormous volume, existing for a very short time, alone will account for their existence." Bretz (1925) then speculated on the somewhat obscure conditions that produced the Spokane Flood. He could only think of two possible ex planations: (1) a very rapid and short-lived climatic amelioration, and (2) a gigantic glacier burst produced by volcanic activity beneath an ice cap. He noted severe objections to either hy pothesis, but held that the great flood had oc curred in spite of the problems in accounting for its source. THE SPOKANE FLOOD DEBATE In 1927 the Geological Society of Washington, D.C., invited Bretz to give a lecture "Channeled Scabland and the Spokane Flood." It was a pur poseful invitation: a veritable phalanx of doubters had been assembled to debate the flood hypo thesis. Bretz (1927a) presented the basic out line of his theory to date, citing the detailed field evidence which he could not explain by any hy pothesis other than a great flood of water. The first discussant was W. C. Alden, who cautiously warned of the difficulties with the hypothesis. Lacking personal field experience in the region he suggested that the rock basins might be col lapsed lava caves, but he realized that the major features indicated stream erosion. "It seems to me impossible that such part of the great ice fields as would have drained across the Columbia Plateau could, under any probable conditions, have yielded so much water as is called for in so short a time. ... It appears that ice sheets of three distinct stages of glaciation invaded the borders of this region and may have afforded conditions of repeated floodings of much smaller volume" (Alden, 1927, p. 203). O. E. Meinzer voiced a commonly held view of the Channeled Scabland, ". . . the Columbia River is a very large stream, especially in its flood stages, and it was doubtless still larger in the Pleistocene epoch. Its erosive work in the Grand Coulee . . . appears to me about what would be expected from a stream of such size when diverted from its valley and poured for a long time over a surface of considerable relief that was wholly unadjusted to it" (Meinzer, 1927, p. 207). He argued that the glacially swollen Columbia could have easily cut the Dry Falls and deposited the great gravel fan of the northern Quincy Basin. He described the Quincy Basin as containing an extensive series of ter races. Moreover, the high-level channels were explained by progressive abandonment as the glacial Columbia progressively cut down to lower levels. One difficulty that Meinzer appreciated from his field work in the Quincy Basin (Schwennes- sen and Meinzer, 1918) was the fact that four great spillways led out of the region where water had obviously been ponded. Bretz (1923a) had shown that the upper limits of the torrents that poured through these spillways occurred at the same altitudes. Rather than ascribing this coinci dence to contemporaneous operation, Meinzer ac tually published the idea that the spillways had been cut one at a time, and subsequent minor earth movements had later brought them to an equivalent altitude. "This recent deformation may account to some extent for channels cut through ridges that can not otherwise be well explained except by assuming excessive depths of flood water" (Meinzer, 1927, p. 208). E. T. McKnight was also a participant in the Washington discussions. He subsequently sug gested (McKnight, 1927) that a glacially diverted Columbia River was a viable alternative to Bretz' hypothesis. In response Bretz (1927b) argued that the great flood channels and bars near Gable Mountain (in the Pasco Basin) were far too large to be ascribed to the Columbia River. He made his position quite clear (Bretz, 1927b, p. 468): "I think I am as eager as anyone to find an explanation for the Channeled Scabland of the Columbia Plateau which will fit all the facts and will satisfy geologists. I have put forth the flood hypothesis, only after much hesitation and only when accumulating data seemed to offer no alter native." Bretz continued to answer various criticisms of his flood hypothesis (Bretz, 1928a, 1928b), and he established some new lines of inquiry into the problem. He (Bretz, 1929) showed that each of the valleys entering the eastern margin of the scabland spillways contained flood deposits em- placed by phenomenally deep water flowing up the tributaries away from the scabland channels. Along the Snake River he traced these deposits to beyond Lewiston, Idaho, more than 85 miles upstream from the nearest scabland channel. The conclusion again defied conventional wisdom (Bretz, 1929, p. 509): "Upvalley currents of great depth and great vigor are essential. . . . No descending gradient of the valley floor can be held responsible. The gradient must have existed in the surface of that flood. The writer, forced by the field evidence to this hypothesis, though warned times without number that he will not be believed, must call for an unparalleled rapidity in the rise of the scabland rivers." Each subse quent study produced yet another affirmation of the flood theory. Bretz (1930b) writes: "The writer, at least normally sensitive to adverse criticism, has no desire to invite attention simply by advocating extremely novel views. Back of the repeated assertion of the verity of the Spokane Flood lies a unique assemblage of erosional forms and glacial water deposits; an assemblage which can be resolved into a genetic scheme only if time be very short, volume very large, velocity very high, and erosion chiefly by plucking of the jointed basalt." Among the spectators at the Washington lec ture was J. T. Pardee. Pardee (1922) also had written on the origin of the Channeled Scabland. W. C. Alden, who was Chief of Pleistocene Geology, U.S. Geological Survey, had sent Par- dee to study the scablands. He published a brief article (Pardee, 1922) proposing that the Cheney- Palouse scablands tract had been created by glaciation of rather unusual character. Bretz later visited Pardee's field locations and found that his "glacial" deposits were flood bars (Bretz, 1974). Correspondence between Alden, Bretz, and Par- dee suggests that Pardee was really considering a hypothesis that the scablands might be related to drainage from a large Pleistocene lake that he had studied in the western part of Montana (Fig. 1.2) (Pardee, 1910). It appears that Alden dissuaded him from that idea (Bretz, 1974). In his memorandum of September 25, 1922, to Figure 1.2. Late Pleistocene strandlines of Lake Mis- soula at Missoula, Montana. The highest strandlines reach 1280 m (4200 feet). David White, Chief Geologist of the U.S.G.S., Alden notes of Pardee's work: "... very signifi cant phenomena were discovered in the region southwest of Spokane. . . . The results so far ... require caution in their interpretation. The condi tions warn against premature publication." David White later asked Bretz if he knew what Alden's middle name was. When Bretz replied in the negative, White said, "It's Cautious, Bretz, Cau tious." It seems clear that the source of the great scabland floods was known even as Bretz was struggling to defend his hypothesis to doubters at the Washington meeting. One story has it that during the discussion Pardee leaned over to Kirk Bryan and said, "I know where Bretz' flood came from." Bretz finally solved the source problem for the Spokane Flood in 1928. Although Harding (1929) without consultation or acknowledgement made the first announcement of Bretz' idea, Bretz (1930a) later published the discovery that scab- land flooding resulted from an abrupt failure of the ice dam that retained Glacial Lake Missoula. Bretz (1932a) clearly illustrated the relationship of Lake Missoula to the Channeled Scabland. James Gilluly was another of those at the Washington meeting who was upset with Bretz' hypothesis. Although he had not studied the Channeled Scabland in the field, he presented an imaginative and persuasive argument for the crea tion of the unusual landforms by the long-con tinued erosion of present-sized streams (Gilluly, 1927, p. 203-205). He took exception to a minor point concerning the use of talus heights as time indicators and then attacked the major weak point in the flood hypothesis. At that time the only two explanations offered for achieving the great volumes of flood water were (1) a very sudden climatic amelioration, and (2) subglacial volcanism and a resulting glacier burst. Some simple calculations demonstrate the inadequacy of either explanation in producing the required volumes of water in so short a time. He con cluded, in essence, that Occam's razor did not apply to the Channeled Scabland and called for a more complex sequence of adjustments by rivers or floods not much larger than the Columbia. In reply Bretz (1927a) asked whether the lack of a documented source for the flood was proof that the flood had not occurred. He argued that the scabland phenomena themselves required the existence of a great flood. Aaron Waters (in Bretz, 1972) relates that Gilluly was later to change his mind in this matter. Many years after the incident at the Washington Academy of Science Gilluly visited the Channeled Scabland on a field excursion. As he observed the Palouse-Snake divide crossing, a major scabland stream channel, his astonishment changed to a smiling comment, "How could anyone have been so wrong?" Nevertheless, the emotion of those days is evinced by the geologists who continued to deny the flood hypothesis and apparently never changed their minds on the mat ter: W. C. Alden, K. Bryan, W. H. Hobbs, F. Leverett, C. R. Mansfield, J. C. Merrian, O. E. Meinzer, and G. O. Smith. The published record of the Spokane Flood debate is clear on one major point. Bretz re peatedly asked only that his flood hypothesis be considered not by emotion or intuition, but by the established principles of the scientific method. His detailed paper on the scabland bars contains the most eloquent expression of this plea (Bretz, 1928b, p. 701): "Ideas without precedent are generally looked on with disfavor and men are shocked if their conceptions of an orderly world are challenged. A hypothesis earnestly defended begets emo tional reaction which may cloud the protagon ist's view, but if such hypotheses outrage pre vailing modes of thought the view of antagon ists may also become fogged. On the other hand, geology is plagued with extravagant ideas which spring from faulty ob servation and misinterpretation. They are worse than "outrageous hypotheses," for they lead no where. The writer's Spokane Flood hypothesis may belong to the latter class, but it can not be placed there unless errors of observation and direct inference are demonstrated. The writer insists that until then it should not be judged by the principles applicable to valley forma tion, for the scabland phenomena are the prod uct of river channel mechanics. If this is in error, inherent disharmonies should establish the fact, and without adequate acquaintance with the region, this is the logical field for critics." THE REVISIONISTS By the early 1930's 4he Channeled Scabland problem had become something of a sensation 8 for American geology. Bretz (1932a, 1932b) had published the last of his field results, and he had embarked on new problems in Greenland and Alberta and ground-water studies in the U.S. His monumental but controversial field study was now open to the kind of attack that he himself had so strongly urged—new field studies. Ira S. Allison (1933) was the first to enter the new foray. His view was not a denial of the Spokane Flood, but a modification. He argued that it was ice, rather than mere volume, that was the critical factor in the flood. He presented detailed evidence for the ponding of flood water all the way from the Columbia River gorge to the Wallula Gap. This ponding was produced ("in spite of the obvious difficulties of such an explanation") by a blockade of ice in the Colum bia gorge. The blockade grew gradually hoadward until it extended into eastern Washington. As • water was dammed to higher levels it spilled across secondary drainage divides creating the enigmatic hanging valleys, high-level gravels and widely distributed erratics. One of the key in sights of Allison's motivation was in his last sen tence, "perhap this revision will make the idea of such a flood more generally acceptable" (Alli son, 1933, p. 722). Hodge (1934) published a brief interpretation of the Channeled Scabland involving mainly glacial processes. He hypothesized a complicated alternation of ice advances and drainage changes. The basalt was quarried by glacial erosion, and channel complexes in the basalt were produced by the diversion of meltwater streams around blocks of stagnant glacier ice and jams of berg ice. The theory was never adequately supported by published field evidence. Perhaps the most serious alternative to the Spokane Flood hypothesis was posed by Richard Foster Flint (1938b). In many ways Flint's study is one of the most ironic in the annals of geology. He presented a carefully worded argument that cited a considerable amount of field data. He stated that the scabland gravel was relatively fine: "Gravel coarser than pebble size is common only in the northern part of the tract" (Flint, 1938b, p. 472). This description was combined with the observation of relatively good size sorting and fair to good rounding to suggest, "a picture of leisurely streams with normal discharge" (Flint, 1938b, p. 472). It is obvious from Flint's sedi- mentological descriptions that he was giving most of his attention to the slackwater faciès of the Missoula flood deposits in the various scabland channels. One of Flint's most important arguments was that the surface form of the scabland deposits was that of "non-paired, stream-cut terraces in various states of dissection" (Flint, 1938b, p. 475). It was an idea that Bretz had introduced (Bretz, 1923a) and subsequently rejected after closer field study. Flint thought that Bretz' re vised interpretation of the deposits as construc tional bar forms could explain some, but not all of the field relationships. He suggested that a sequence of channel aggradation by normal pro- glacial outwash was followed by dissection to leave remnants of fill that occasionally resembled bar forms. Flint (1938b) accepted Bretz's (1928b) argu ments that the flood gravel often ( 1 ) occurred in the lee of island-like areas, (2) had rounded upper surfaces, and (3) exhibited a parallelism of surface slopes with the dip of underlying foresets. He argued that "terraces" had been extensively dissected by a downstream base level reduction. The "terraces" were prefentially pre served in the lee of island-like areas. In addi tion, the low precipitation plus the high perme ability of the gravel prevented gullying, so the gravel deposits developed rounded slopes by dry creep. Finally, he showed that many of the gravel slopes did indeed truncate the underlying bedding. As specific cases, he argued that Bretz' Willow Creek bar, Staircase Rapids bar, Palouse Canyon bar, Midcanyon bar, and Shoulder bar were all simply terrace remnants. Subsequent studies have shown that three of these bars have prominent giant current ripples on their upper surfaces (Fig. 1.3). Flint also described multiple scarps and benches on the Palouse loess. Instead of record ing the high-water mark of the Spokane Flood (Bretz, 'l928b, p. 701), he interpreted these scarps as evidence of lateral planation by progla- cial streams. Subsequent studies in the Cheney- Palouse scabland by Patton and Baker (Chap. 6, this volume) reveal that these scarps resulted from differential erosion of Palouse Formation paleosols and from the exposure of calichified gravel underlying local areas of Palouse loess. Flint traced the coarse scabland deposits down stream into the Pasco Basin. There he found that the deposits changed from sand and gravel to silt and fine sand containing erratic stones. He named the fine-grained facies the "Touchot beds." The deposits had already been described by Bretz (1928a, p. 325-328; 1929, p. 516-536; 1930b, p. 414), who ascribed them to ponded flood water; and by Allison ( 1933), who ascribed them to water ponded by ice jams. The silts are recognized only to a uniform elevation of about 350 m. The stratification ranges from rhythmic parallel bedding to cut-and-fill. The included erratic stones are granite, basalt, and other crystalline lithologies. Intense folding, fracturing, and clastic dikes imply slumping and sliding of the water-saturated silt on gentle subaqueous slopes. Flint thought that these relationships were most consistent with a large lake, which he pro posed was ponded by a landslide dam or glacier ice in the Columbia gorge. Following Symons (1882) he named this water body Lake Lewis. At this point Flint had the necessary tools to erect his hypothesis. The proglacial meltwater streams of normal discharge overran the northern margin of the Cheney-Palouse tract. This flow was derived from lobes of ice at the heads of the Cheney-Palouse and Telford-Crab Creek scabland tracts. Flint thought water from Lake Missoula (Bretz, 1930a) need not be involved. Instead, he observed that the discharge "was less than that -t v Figure 1.3. Oblique aerial photograph of Staircase Rapids bar. The bar is approximately 50 m high and composed of coarse flood gravel. The giant current rip ples on the upper bar surface (left foreground) were actually first described by Flint (1938b) who did not recognize their origin. Bretz and others (1956, p. 1000- 1002) later used these and other giant current ripple sets to demonstrate Flint's "faulty reasoning." of the Snake River today" (Flint, 1938b, p. 515). As Lake Lewis rose, the "leisurely" streams that Flint envisioned aggraded, forming a thick fill. This fill blocked preglacial tributaries to the Channeled Scabland, such as the Snake River, and formed marginal lakes which accumulated fine-grained sediments. The steep scarps on the Palouse loess were then cut by lateral planation of the streams flowing on this fill. When Lake Lewis finally drained, the streams gradually in cised the fill to form terraces. Moreover, Flint was able to explain the enigmatic notched spurs and slotlike hanging canyons as the result of super position of streams from the widespread fill rather than a consequence of divide crossing by cata strophic flood water. Flint argued that the complex of anastomosing channel ways cut into basalt was a consequence of erosion by relatively small streams operating on various profiles. He stated that scabland-type erosion should occur wherever rock material with vertical planes of weakness is subjected to stream flow. As examples of such erosion he cited Red Rock Pass, Idaho, an outlet of pluvial Lake Bonneville (Gilbert, 1890). He also noted the scabland erosion at Twin Falls, Idaho, where the Snake River flows in a canyon nearly as spec tacular as the scabland channels. He noted, "the . . . [basalt] flows yielded to the hydraulic force of the Snake River as similar flows on the Columbia Plateau yielded to the hydraulic force of proglacial streams, yet I am not aware that unusual floods have been held to have affected the upper Snake River" (Flint, 1938b, p. 492). These words were written 30 years too soon! Malde (1968) described the catastrophic out burst of Lake Bonneville that eroded the scab- land forms at Red Rock Pass and Twin Falls. In yet another ironic passage, Flint (1938b, p. 504-505) calculated the probable rate of filling for Lake Lewis at the modern discharge of the Columbia River. He stated, "the calculated time, 13 years 1 month, seems grossly inadequate for the deposition of the fill in the scabland tracts." He rationalized his interpretation, however, by referring back to the interpreted filling episode. Bretz' flood theory was so despicable that even circular reasoning could be employed to erect an alternative hypothesis. A careful examination of Flint's (1938b) 10 11 paper reveals that he observed and described the morphological feature which, more than any other, was absolutely incompatible with his ele gant theory. On the surfaces of the scabland "terraces" he described an intricate microtopog- raphy of anastomosing channels, small depres sions, and crescentic channels (Flint, 1938b, p. 475). In other areas he observed "mamillary un- dulatory topography." As an example he gives the precise location of the train of giant current ripples on the upper surface of Staircase Rapids Bar, 3 km north of Washtucna (Flint, 1938b, p. 486). Although the ripples that he describes are somewhat masked by overlying slackwater sedi ments, Flint (1938b, p. 499-500) even states the characteristic ripple magnitude: "The undula tions are 20 to 100 feet long, and have ampli tudes up to 10 feet. Their axes are generally transverse to the Snake River." How ironic that Flint was the first to accurately describe (without knowing what they were) the very feature that Bretz and others (1956) later presented as in controvertible evidence for catastrophic flood flows (Fig. 1.3)! It was Allison (1941) who published the first criticism of Flint's fill hypothesis for the origin of the Channeled Scabland. The first shortcoming noted was that the anastomosing channel patterns and deep rock basins could not have been eroded by "normal" streams. Second, Allison disputed Flint's correlation of the scabland gravels to the Touchet beds, suggesting that the Touchet se quence was younger than the gravels. Third, he agreed with Bretz that the peculiar shapes of the scabland deposits required extraordinary proc esses. The conclusion was that the complex jam ming of various channels with ice was the only reasonable explanation for the unusual drainage patterns and depositional features. Another example of the strong emotions evoked by the Spokane Flood controversy in volves W. H. Hobbs, an eminent glacial geologist from the University of Michigan. He spent sev eral weeks studying the terrain in southeastern Washington and prepared a paper explaining the landforms as the product of a "Scabland Glacial Lobe." Both Bretz and Flint reviewed the paper for the Geological Society of America, and both recommended rejection. The paper was then sub mitted to the American Philosophical Society, which had supplied part of the funds for the study. Bretz again reviewed the paper, and again it was rejected. Although a brief statement of the hypothesis was published (Hobbs, 1943), the main manuscript had to be published privately (Hobbs, 1947). The author expressed his feelings in the "Foreword" to his paper: "In the winter of 1942-43 I was listening with much interest to a lecture on the late geological history of the so-called Scabland area which is southwest of Spokane and close to the supposed southern front of the Pleistocene Cordilleran continental glacier. A map projected on the screen dozens of lakes, none of which trans gressed its border, an almost sure indication that this lobate area had once been actually covered by a Pleistocene glacier lobe. Surrounding this lobe on the lecturer's map could be seen a broad apron of gravels, and enveloping the gravels were heavy deposits of silt. These relationships of glacier lobe to out- wash and loess duplicated what I had observed in west Greenland. The lecturer explained, how ever, that the deposits represented upon his map had been laid down by a great flood of water of unknown origin, the "Spokane Flood." In the belief that my Greenland observations had given me an advantage in interpreting the evidence within the Scabland region, I then and there decided to make a personal study of it on the ground. Although two other very extended studies had already been made of it by Fellows of the Geological Society of America, and their conclusions had been published in extenso in its Bulletin, the Society provided me with a grant of money which made possible a new study of the area. This field investigation was carried out during two seasons, and the lesults and con clusions met with unusually enthusiastic general approval when they were presented to the Soci ety in 1945 at its Pittsburgh meeting. Following tumultuous applause in the crowded section the discussion was throughout approving." The Hobbs paper contains so many funda mental errors that one marvels at the absurd limits that were being stretched to find an alterna tive to catastrophic flooding as the cause of the Channeled Scabland. Hobbs (1947) argued that the scabland was a product of glacial scour and that the Palouse loess was deposited contempo raneous to this glaciation by anticyclonic winds off the ice that lay in the various "channels." He interpreted many scabland gravel deposits as moraine remnants modified by glacier-border drainage. VINDICATION At long last Pardee (1942) shared his observa tions of Glacial Lake Missoula that firmly indi cated its role as the source of catastrophic floods through the Channeled Scabland. He noted that about 500 cubic miles of water were impounded behind a glacial lobe which occupied the basin of modern Lake Pend Oreille in northern Idaho. Pardee believed that this glacial dam had failed suddenly with a resultant rapid draining of the lake. Evidence for this failure included severely scoured constrictions in the lake basin, huge bars of current-transported debris (Fig. 1.4), and giant current ripple marks with heights of 50 feet and spacings of 500 feet (Fig. 1.5). Lake Missoula was the obvious source for the catastrophic flood flows required by Bretz' hypothetical origin of the Channeled Scabland (Fig. 1.6). Pardee did not state the connection, perhaps leaving that point generously to Bretz. Even Alden remained cautious to the end. His last published report on Lake Missoula observed (Alden, 1953, p. 155): "Abrupt release of water from lowering of the ice dam . . . might result in floods of great magni tude. . . . Each may, perhaps, have been the origin of many violent floods that are supposed to have swept over the scablands." In the summer of 1952, Bretz, then nearly 70 years old, returned for his last summer of field t " <-. .* Figure 1.4. Large "gulch fill" formed at the mouth of a tributary canyon along the Flathead River, Perma, Montana. The deposit is an eddy bar (Baker, 1973a) lormed during the rapid draining of glacial Lake Mis soula. First recognized by Pardee (1942) this gravel de posit was later breached by a small stream to form the V-shaped notch visible at right. The low terrace in the foreground is composed of lacustrine silt. work in the Channeled Scabland. The purpose was to investigate new data that had been ob tained in surveys for the Bureau of Reclamation's Columbia Basin project. Professor H. T. U. Smith accompanied him, acting in the field as "skeptic for all identifications and interpretations" (Bretz and others, 1956). With the aid of Mr. George E. Neff of the Bureau of Reclamation that study (Bretz and others, 1956) answered with meticu lous detail all previous criticisms of the flood hy pothesis. Central to the 1956 investigation was the study of the scabland depositional features. Extensive excavations for the irrigation project and new topographic maps proved that the gravel hills called bars by Bretz (1928b) were indeed that, subfluvial depositional bedforms. Most convincing of all was the presence of giant current ripples on the upper bar surfaces. These showed clearly that bars 30 m high were completely inundated by phenomenal flows of water. Numerous exam ples of giant current ripples were found on the same bars which Flint had interpreted as ter races. Such features could only have been pro duced by the flow velocities associated with truly catastrophic discharges. Bretz and others (1956) and Bretz (1959) modified Bretz' earlier inter pretations to allow for several episodes of flood ing. The central theme of their study, however, was that only a hypothesis involving flooding could account for all the features of the Chan neled Scabland. More recent studies of the Figure 1.5. Giant current ripples at Camas Prairie, north of Plains, Montana. The ripples are composed of gravel and consist of ridges up to 15 m high and spaced as much as 200 m apart. The ripples cover approximately 10 km2 of the northern Camas Prairie. Faint strandlines of Lake Missoula are visible in the background. 12 13 WASHl N G T 0 N ! I MONTANA LAKE MISSOULA 2O 40 60 i—i SCALE (MILES) MODERN LAKES GLACIAL LOBE (EARLY PINEDALE) EARLY PINEDALE FLOODING Figure 1.6. Relationship of glacial Lake Missoula-'to the Channeled Scabland of eastern Washington (Baker, 1973a). Quaternary geology of eastern Washington have accepted this reasoning (Trimble, 1963; Fryxell and Cook, 1964; Richmond and others, 1965; Baker, 1973a). Perhaps the final words on the Channeled Scabland controversy were delivered following a field trip, Field Conference E of the 7th Congress, International Association for Quaternary Re search. During August, 1965, an international party of geologists observed the evidence in Mon tana for Lake Missoula's catastrophic outbursts. They then traveled through the Channeled Scab- land studying the giant current ripples, flood gravel bars, and scabland erosion forms. Dr. Bretz was unable to attend the trip because of health. When the field party reached Pullman, they sent a long telegram to him at Homewood, Illinois. The telegram opened with "greetings and salutations" and closed with the sentence, "We are now all catastrophists" (Bretz, 1969, 1973). MODERN RIVERS / , GLACIAL LAKE MISSOULA AXIS OF MAJOR ANTICLINAL RIDGE DISCUSSION When Bretz published his work on the Chan neled Scabland, the paradigm of Geology was uni formity. The Spokane Flood hypothesis appeared to contradict the uniformitarian tradition that made geology a science in the nineteenth century. Indeed it was not until after 1840 that the flood theory fell into serious decline. The catastrophist idea of the Noachian debacle was finally laid to rest when Louis Agassiz showed that his glacial theory could explain erratics, striations, till, flu- vioglacial activity, etc. Old ideas die hard, how ever, and catastrophist absurdities still appeared in the literature of the early 1900's (as they do even today). Little wonder then that Bretz' Spo kane flood hypothesis appeared as an anathema to many of his contemporaries. Simultaneously the Spokane Flood hypothesis established a conflict between two important cor- nerstones of geological philosophy: (1) the tri umph of the glacial theory over diluvian myth, and (2) the scientific tolerance of outrageous hy potheses. It is a classic dilemma for the scientist to distinguish absurdity from outrage. A foolish idea is always self-evident, but not so with the rare, creative insight that happens to pass all rea sonable bounds in the consensus of knowledge. The remarks of a former president of our society: "How narrowly limited is the special field, either in subject or locality, upon which a member of the Geological Society of America now ventures to address his colleagues. ... I wonder sometimes if younger men do not find our mecting rather demure, not to say a trifle dull; and whether they would not enjoy a return to the livelier manners of earlier times . . . (Their) feeling of discour agement must often be shared by the chairman of a meeting when, after his encouraging invitation, 'This interesting paper is now open for discus sion,' only silence follows. . . . We shall be indeed fortunate if geology is so marvelously enlarged in the next thirty years as physics has been in the last thirty. But to make such progress violence must be done to many of our accepted principles." After speaking these words in 1926, William Morris Davis made a case for the value of out rageous geological hypotheses, even suggesting that geologists seriously consider "the Wegener outrage of wandering continents." He concluded by saying that the valuable outrage was that which encouraged the contemplation of other possible behaviors. Such outrages deserve contemplation followed not, he states, "by an off-hand verdict of 'impossible' or 'absurd', but a contemplation delibrate enough to seek out just what conditions would make the outrage seem permissible and reasonable." Needless to say, W. M. Davis was one of the first to accept Bretz' interpretation in the 1920's. It is a commentary on those years that others were not so tolerant. "During all those years, I was fighting for my professional career." (Quota tion of Dr. Bretz by the Seattle Times, July 11, 1971.) Bretz himself explored the consequences of his "outrage." His 1956 paper resoundingly confirmed the catastrophic flood theory by answering in meticulous detail all the previous ob jections to his grand hypothesis. It took over 30 years and the coming of a new generation of geologists for his theory to gain general accept ance. The Spokane Flood controversy is both a story of ironies and a marvelous exposition of the scientific method. One cannot but be amazed at the spectacle of otherwise objective scientists twisting hypotheses to give a uniformitarian ex planation to the Channeled Scabland. Un doubtedly these men thought they were upholding the very framework of geology as it had been established in the writings of Hutton, Lyell, and Agassiz. The final irony may be that Bretz' critics never really appreciated the scientific implications of Agassiz' famous dictum, "study nature, not books." Perhaps no geologist has understood and lived the spirit of those words more enthusiast ically than J Harlen Bretz. As the Viking spacecrafts were orbiting Mars in the summer of 1976, the cameras were trained on the great Martian channel systems. They re vealed uplands streamlined by fluid flow, eroded scabland on the channel floor, and many other features that we now know to be diagnostic of bedrock erosion by catastrophic flooding. Fifty years after J Harlen Bretz' theory of scabland erosion on the Columbia Plateau was being de nounced at an infamous meeting of the Washing ton Academy of Science, Viking scientists were using Bretz' well-documented studies of the Chan neled Scabland as the major earth-analog to Martian channel erosion. Few geological con cepts, born amid bitter controversy over a half century ago, have continued to have such rele vance to our science. 14 15 Chapter 2 Quaternary Geology of the Channeled Scabland and Adjacent Areas VICTOR R. BAKER Department of Geological Sciences The University of Texas at Austin Austin, Texas 78712 ABSTRACT The Quaternary history of the Channeled Scabland is characterized by discrete episodes of catastrophic flooding and prolonged periods of loess accumulation and soil formation. The loess sequence is correlated with Richmond's Rocky Mountain glacial chronology. Two pre-Bull Lake (pre-Illinoian) loess units are characterized by siliceous petrocalcic horizons. The Bull Lake loess (Palouse Formation) apparently accumu lated episodically be downwind accretion, fol lowed by periods of relative stability and soil formation. The Palouse paleosols have platy calcic horizons but do not show petrocalcic hori zons. Pinedale (= Fraser Glaciation) loess is paler in color than the older units; its paleosol calcic horizons lack the platy structure of the older loess paleosols. At least five major catastrophic flood events oc curred in the general vicinity of the Channeled Scabland. The earliest episode occurred prior to the extensive deposition of the Palouse Forma tion. Its surviving records are but fragmentary. Probable Missoula flood deposits in the Cheney- Palouse scabland tract are overlain by the younger pre-Palouse loess and a petrocalcic soil profile. Flood deposits in the Quincy Basin came from an unknown western source across Babcock Ridge. The Quincy Basin flood 'deposits are over lain by other flood deposits of probable Bull Lake age (Illinoian) also derived from a western source. Catastrophic flooding from Lake Bonne- ville affected the southern margin of the Chan neled Scabland about 30,000 years B.P. The last major episode of flooding occurred between about 18,000 and 13,000 years ago. It probably consisted of two outbursts from Glacial Lake Missoula. The earlier outburst predates the Vashon maximum (= Withrow Moraine of the Okanogan ice lobe). This flood affected Moses Coulee, the Grand Coulee (prior to its present configuration), and the eastern Channeled Scab- land (Telford-Crab Creek and Cheney-Palouse scabland tracts). A second flood, probably in volving less volume than the first, coincides with the deglacial phase of the Okanogan ice lobe. It mainly affected the Columbia River northwest of the Channeled Scabland downstream from a hy draulic constriction of the canyon at the site of Coulee Dam. The last phase of that flood prob ably also involved catastrophic flow down the Grand Coulee. Slackwater faciès of this flood contain the Mount St. Helens set "S" ash erupted about 13,000 years B.P. according to D. R. Mul- lineaux and co-workers. INTRODUCTION The Channeled Scabland of eastern Wash ington (Fig. 2.1) consists of a spectacular com plex of anastomosing channels, cataracts, loess "islands," and immense gravel bars created by the catastrophic fluvial erosion of the loess and basalt 17 of the Columbia Plateau (Bretz and others, 1956). The erosion and deposition that produced the scabland topography resulted from the failure of the ice dam impounding glacial Lake Mis- soula. At its maximum outflow, near the end of the last major Pleistocene glaciation, the lake discharged as much as 21.3 x 10° m3/sec into the vicinity of Spokane, Washington (Baker, 1973a). Recently the origin and history of the Channeled Scabland has assumed new significance because of morphologic similarities to outflow channels of probable flood origin on Mars (Baker and Milton, 1974; Sharp and Malin, 1975; Masursky and others, 1977). Despite detailed study since the 1920's (Bretz, 1923a, 1928c, 1932b, 1959, 1969; Richmond and others 1965), the exact number and timing of major floods which resulted from the outpourings of Lake Missoula remains a major unresolved problem 120° in the Quaternary history of eastern Washington. This paper will summarize the general Quaternary geology of the Channeled Scabland and present some new data on the number of catastrophic floods. PHYSIOGRAPHY, CLIMATE, AND SOILS Freeman and others (1945) have formalized the physiographic divisions of the regions covered in this report. The entire study area lies in the Columbia Basin Subprovince of the Colum bia Intermontane Physiographic Province. The Columbia Basin is a regional lowland surrounded by the Blue Mountains to the south, Cascade Mountains to the west, the Okanogan Highlands to the north, and the mountains of northern Idaho in the east. The region is also informally called 48° -5 47° I20 Figure 2.1. Map of the Channeled Scabland in eastern Washington showing the distribution of channels and the general extent of loess (Palouse Formation) that was not stripped away by the last major episode of flooding. the "Columbia Plateau," but intense folding in the western part produces a series of basins with intervening ridges. The basalt bedrock of the basin has the regional aspect of a structural basin. The western part of the Columbia Basin, termed the "Yakima Folds," consists of a series of anticlinal ridges extending eastward from the Cascade Mountains. Several of these ridges are transected by the Columbia River. From north to south these are the Frenchman Hills, Saddle Mountains, and Horse Heaven Hills. The northwest portion of the Columbia Basin, the Waterville Plateau, was not appreciably dis turbed by the Neogene folding. A major canyon, Moses Coulee, is deeply excavated into the Water ville Plateau and extends southwestward from its center to the Columbia River Valley. The far eastern part of the Columbia Basin is characterized by relatively undeformed basalt overlain by as much as 75 m of Pleistocene loess. The loess has been dissected to form a rolling topography known as the Palouse Hills. Eleva tion of the Palouse Hills declines from about 750 m on the northeastern margin to 100-120 m in the southwest. This gradient reflects the regional dip of the basalt toward the center of the basin. Between the Palouse Hills and the Yakima Folds, extensive stripping occurred of the loess mantle by the catastrophic flooding of glacial Lake Missoula. The maturely dissected loess presents an abrupt contrast with the black cliffs and ragged appearance of the flood-eroded scab- land tracts. Steep scarps on the eroded channel margins resemble wave-cut headlands rising above a desiccated sea. Bretz (1923b) named this re gion the Channeled Scabland. The Columbia Plateau lies in the rain shadow of the Cascade Mountains. Precipitation in the western basins (elevation 180-300 m) is less than 20 cm per year, but increases with elevation northeastward, reaching nearly 50 cm per year on the updip margins of the plateau (elevation 850 m). The low precipitation results in a lack of perennial drainage through the huge ancient channels, called "coulees." The Columbia and Spokane Rivers, which are deeply entrenched along the northern and western margins of the plateau, intercept mos't of the drainage from the areas of higher precipitation in the bordering mountains. The precipitation pattern is paralleled by a change in Great Soil Groups to the northeast. Sierozem soils form in the dry, southwestern portions of the plateau. Toward the northeast, Brown, Chestnut, and Chernozem soils appear in successively wetter portions of the plateau. These soils occur only on the loess and other fine grained sediments. The eroded scabland channels are generally devoid of these parent materials for soil formation. ANTEDILUVIAN EVENTS ON THE COLUMBIA PLATEAU The Yakima Basalt comprises the bedrock in all but a few parts of the Channeled Scabland. This basalt unit is part of the extensive Neogene eruptions of plateau basalts that cover over 250,- 000 km2 in parts of Washington, Oregon, and Idaho. Most of the lava was erupted during the Miocene. The lava flows are exceptionally thick, and several can be traced over 150 km. Consid erable structural and lithologie variation can be found in the basalt sequence, including joint ing patterns, pillow-palagonite complexes, sedi mentary interbeds (from lakes on the Miocene land surface) and geochemical variation. On the north and east margins of the plateau, the basalt is interbedded with extensive deposits of siltstone and shale of the Latah Formation, deposited as drainages were blocked by the basalt outpour ings. Deformation of the basalt sequence was most extensive during the Pliocene. The entire Colum bia Plateau was regionally tilted from an eleva tion of about 760 m in the northeast to about 120 m in the southwest near Pasco, Washington. Superimposed on the regional structure are the east-west fold ridges described earlier. The up raised northern rim of the plateau is especially significant for the flood history of the Channeled Scabland. Only a truly phenomenal quantity of water could fill the great canyon of the Columbia River between Spokane and Coulee Dam. That filling would be necessary to have water spill over the northern rim of the plateau and flow south- westward, carving the great scabland channels. During the Pliocene, the great structural basins of the western scablands accumulated a sequence 18 19 of partly consolidated silt, gravel, and clay known as the Ringold Formation (Flint, 1938a; New- comb, 1958). The age of this formation was poorly interpreted until Eric Gustafson studied the extensive upper Ringold vertebrate fauna from the White Bluffs area. Gustafson's White Bluffs fauna is early Blancan (Pliocene) in age. The faunal assemblage correlates best with the Hag- erman fauna of Idaho, dated at about 3.5 x 10° years B.P. Gustafson (1973) interprets the .Ringold For mation as a sequence of stream-channel conglom erate, point-bar sandstone, and overbank deposits within a major fluvial depositional system. The predominance of browsing forms among the large mammals (interpreted from tooth form) in cluding especially Bretzia, Platygonwt, and Mega- lonyx, suggests that the Ringold flood plain sup ported considerable riparian forest. Circumstan tial evidence suggests a strongly seasonal climate with yearly rainfall between 25 and 50 cm. Today the rainfall averages 20 cm, and the vegetation isxerophytic (sagebrush). The Ringold Formation does not occur in the eastern Columbia Plateau, where late Pliocene and Pleistocene sedimentation was largely eplian, as expressed in a complex blanket of loess sheets (Fig. 2.2). The loess units provide a fairly com plete Pleistocene chronology that was correlated by Richmond and others (1965) to Richmond's (1965) glacial chronology of the Rocky Moun tains. The oldest loess units are considered to be pre- Bull Lake in age. By the revised glacial chron- \ Figure 2.2. Oblique aerial photograph of a road cut through "Palouse Hills" topography about 5 km west of Washtucna, Washington. The white layers are calcic paleosols in the loess sequence. ology of Pierce and others (1976), Bull Lake time correlates to Illinoian, about 125,000 to 200,000 years ago. Two pre-Bull Lake loess units can be recognized in the field by the well-in durated, siliceous petrocalcic horizons that formed on them. The older of the two petrocalcic hori zons is colored pinkish by an associated oxi dized tuff. These petrocalcic horizons consist of as much as 0.6 m of roughly horizontal carbonate laminae over a calcic horizon of carbonate- plugged loess. This profile is an extremely strong soil horizon that qualifies for the designation "K horizon" (Gile and others, 1966), defined by the presence of 50% or more CaCO3. Fryxell and Cook (1964) describe as much as 3 m of B hori zon associated with the pre-Bull Lake loess units. Most of the loess on the Columbia Plateau is correlated to Bull Lake Glaciation (Richmond and others, 1965). This loess, which has a thick ness of up to 75 m (Ringe, 1970), is formally designated the Palouse Formation (Newcomb, 1961). Richmond and others (1965) recognize three cycles of soil formation in the Palouse For mation, but unpublished observations by the au thor indicate that more cycles may have occurred. Each soil shows a mature profile with a well- developed textural B horizon. The underlying Cca horizon is strongly calcareous and has a well- developed platy structure. Unlike soils on older loess units, however, this calcic horizon is less thoroughly cemented and does not qualify as a K horizon. Boulders derived from the Cca or B horizons of Palouse Formation soils are com monly found in deposits of the late Wisconsin flooding on the Columbia Plateau. The Bull Lake age of the Palouse Formation was established by its relationship to glacial de posits in the vicinity of Spokane. It is obvious, however, that the detailed stratigraphie informa tion in this loess sequence is a better record of mid-Pleistocene events than is the glacial sequence to the north. The Palouse Formation should be given ' detailed study, employing modern tech niques as described by Kukla (1975) for the loess of Czechoslovakia. Paleomagnetic and tephrachronologic studies would probably yield an important new stratigraphie interpretation for the Palouse Formation and older loess units on the Columbia Plateau. A major unconformity separates the Palouse Formation from younger loess, correlated to the Pinedale Glaciation by Richmond and others (1965). The Pinedale loess is much paler in color than the Palouse Formation. Associated soil pro files have only weakly developed textural B hori zons and lack the structural Cca or K horizons of older soils. CaCO3 occurs in veins or nodules but not in continuous plates. In contrast, soils of Holocene age show A-C profiles of minimal de velopment. The soils within a typical loess hill show pe riods of stability during the progressive north westerly accumulation of loess. A section through a loess hill near La Crosse (SE Î4, Sec. l, T. 15N., R. 39E.) is shown in Figure 2.3. The three Bull Lake loess units have been truncated by a loess unit of a younger age (Pinedale?) which mantles the surface of the whole hill. The entire body of Bull Lake loess has a dull brown color (7.5YR 6/3). Color or textural B horizons were not apparent in this section. The younger loess units are dull yellow orange (10YR 7/3). PRE-WISCONSIN FLOODING IN THE CHANNELED SCABLAND The eastern portion of the Channeled Scabland (Fig. 2.4) contains several exposures of pre- Wisconsin flood gravel. The most complete sec tion is exposed in a railroad cut through a gravel bar on the downstream end of a residual loess island about one kilometer west of Marengo, Washington (location 1, Fig. 2.4). The cut ex poses a succession of two flood gravel units sep arated by three layers of loess (Fig. 2.5). These 2O- IO- O W three loess units are each capped by a pedogenic calcic horizon. The lower flood deposit is a poorly sorted mixture of basalt and loess pebbles and cobbles in a matrix of granule-sized basalt grains. The texture is similar to that of the extensive deposits of the last major episode of scabland flooding. Typical of many scabland gravel de posits is the presence of loess cobbles, implying fluvial transport in suspension which prevented destruction of the loess. Cobbles in this deposit are dull orange (7.5 YR 7/4) in color and have black manganese dioxide staining on their sur faces. The lower flood gravel is overlain by l m of dark yellowish brown (10 YR 6/6) loess which is capped by a petrocalcic horizon 60 cm thick. There is also a well-developed argillic horizon on this loess characterized by a pronounced pris matic structure. Grain-size analysis indicates that the prismatic structure is associated with a rela tively strong illuvial textural B horizon. Root casts infilled with caliche from the overlying petrocalcic horizon also occur in this argillic zone. The overlying petrocalcic horizon has a platy structure and completely plugs the uppermost part of this loess unit. Although no carbonate analysis was made on this horizon, its morphology is nearly pure carbonate (caliche) with little loess matrix. The horizon therefore qualifies for desig nation as a K horizon (> 50% carbonate) as described by Gile and others (1966). This pedocal soil appears to have been superimposed on the older, pedalfer soil (textural B horizon) and then eroded to its resistant carbonate layer prior to new loess deposition. -10 to ce r5 Cca SOIL HORIZON aROWN LOESS (BULL LAKE) 20 I YELLOW-ORANGE LOESS (PINEDALE-RECENT) 40 FEET 10 METERS Figure 2.3. Loess stratigraphy exposed by a road cut 2.5 km east of La Crosse, Washington. 21 20 to to \ \ o-i CO to UJ 2- 3- 4- 5-1 OLD MAID COULEE MARENGO MACALL REVERE SOIL HORIZONS e'-'e •V: °o„o UK mcca ncco m Cco ECca ZK EXPLANATION STRUCTURELESS LOESS LOESS WITH ORGANICS (SOIL A HORIZON) LOESS WITH PRISMATIC STRUCTURE CALCIFIED LOESS WITH PLATY STRUCTURE DESCRIPTION ALTITHERMAL SOIL" LATE PINEDALE LOESS FLOOD GRAVEL CALCCPALEOSOL LOESS :AUX PALEOSOL LOESS SUPERIMPOSED PETROCALCIC PALEOSCL ROOT CASTS TEXTURAL B HORIZON LOESS WITH FROST- HEAVED COBBLES PALOUSE FORMATION YOUNGER PRE -PALOUSE LOESS FLOOD GRAVEL (SECTION ERODED BY EARLY PINEDALE FLOOD) JLjtJ£ J »X* O o "^•i-^X CALCIFIED LOESS WITH NODULAR CALICHE FROST-HEAVED COBBLES IN LOESS MASSIVE ~) CALICHE SOIL r K PLATY HORIZON CALICHE ) ^% o o O Gt •3 o FLOOD GRAVEL WEATHERED BASALT COBBLES COBBLES WITH CARBONATE COATINGS COBBLES OF REDDISH PRE-PALOUSE LÛESS SOIL HORIZONS TflîÏÏÏÏ ._•>•.<) 1 ^7: r^.~iJ- frjjf o o ,o '•° • » O & G? ^ A B Cca JTCca M Cca EK JSBica E Bat ZCca JP__3> b 3 Figure 2.5. Stratigraphie section at Marengo, Washington and its probable correlation to other sections in the Channeled Scabland. to u> latter profile corresponds to the late Wisconsin- early Holocene soil that is common throughout the region on post-scabland loess deposits (Fryxell, 1965; Baker, 1973a). The uppermost gravel unit threfore correlates with the extensive deposits of the last major episode of scabland flooding. The lower flood gravel at Marengo has also been located at Revere (Locality 2, Fig. 2.4) and Macall (Locality 3, Fig. 2.4). These three new sections probably also correlate with a flood gravel exposed on a divide near the upper reaches of Old Maid Coulee (Locality 4, Fig. 2.4), a section previously interpreted by Bryan (1927, p. 27), by Flint (1938b, p. 518), and by Bretz and others (1956, p. 1006). It consists of foreset- bedded gravel overlain by calcified loess, which is, in turn, overlain by pale, water-transported silt (Fig. 2.6). About 20% of the basalt cobbles in the upper parts of the gravel are completely rotten, indicating prolonged weathering. The loess unit is capped by 60 cm of platy calcium car bonate, a soil petrocalcic or K horizon. The ero sion surface at the top of the caliche is overlain by silt that is dull yellow orange color (10 YR 6/3). It lacks the darker coloration and well-de veloped buried soils of the Palouse Formation. Moreover, its position on a divide and its uniform texture suggest that it may be suspended load from the last major flood that was deposited in a slackwater area. Flint (1938b) believed that the unit was a lake silt. He did suggest that this silt was approximately contemporaneous with the gravel of the main scabland channels, which are now considered to be approximately 13,000 years B.P. in age (Mullineaux and others, 1977). The loess unit and petrocalcic paleosol capping the old flood gravels at all four sections probably correlates with the younger pre-Palouse Forma tion loess and soil described earlier. Unlike other pre-Bull Lake flood deposits recognized in the western scablands at George and Winchester (Baker, 1973a), there was a significant pre- Palouse loess accumulation on these early flood gravels prior to a major period of soil formation (an interglaciation?). Moreover, the reddish loess cobbles in the flood gravel probably were eroded from the older of the two pre-Palouse loess units. The first episode of scabland flooding appears to have occurred during a Pleistocene glacial maxi mum prior to the maximum that produced the Bull Lake Glaciation. Subsequent loess accumula tion probably included a relatively humid interval that produced the pronounced textural B horizon, and it was followed by an arid interval that super imposed a petrocalcic horizon on the loess. An erosional episode, probably deflation, then stripped the surface horizons down to the resistant petrocalcic layer. These latter events all charac terized the latest pre-Bull Lake glacial and inter- glacial record on the Columbia Plateau. Bull Lake time is represented at the Marengo profile (Fig. 2.5) by two of the three known Palouse Formation loess units. This circumstance is typical of the cycles of soil formation, deflation, and lateral accretion that characterize the Palouse Formation (Lewis, 1960; Fryxell, 1966). During the Bull Lake Glaciation loess was ac creting behind pre-existing obstacles in a gen eral northeastward direction. Typical exposures through loess hills show that accretion was fol lowed by stability and soil formation. Subsequent - • r ti ' * - » >'* \ » r . J *" J " * .-• -vr ., > deflation would remove surficial soil horizons down to the resistant calcic layers. The cycle was then repeated as more deposition occurred up wind of the previous deposit. This pattern of lateral accretion plus the opportunity for stream dissection of the topography make it unlikely to find all three Palouse Formation loess units in one vertical section. An unresolved problem is the correlation of the pre-Wisconsin Cheney-Palouse flood to the pre-Wisconsin flood deposits of the western Quincy Basin (Baker, 1973a, p. 8). The section there (NW %, Sec. 31, T. 19N. R.24E) shows a typical foreset flood gravel containing boulders as large as l m in diameter. The uppermost 60 cm of the gravel is capped by a horizontally laminated petrocalcic horizon (Fig. 2.7). This is underlain by 30-60 cm of carbonate-cemented gravel. Local carbonate cementation in the coarser foresets occurs to a depth of 3 m. Carbonate coat ings on the underside of cobbles occur to a depth of 4 m. Weathering rinds on the basalt cobbles in the upper 1.5 m of the gravel exceed 7.5 cm in thickness. Many cobbles are completely rotten. The gravel below a depth of 3 m shows no evi dence of weathering. This weathering profile is Figure 2.6. Stratigraphie section at Old Maid Coulee. Figure 2.7. Oblique aerial photograph of a sanitary land fill at George, Washington. The white layer is a prom inent petrocalcic horizon developed at the top of coarse flood gravel. The gravel was laid down by a deep cata strophic flood that crossed the western rim of the Quincy Basin and flowed southeastward into the Quincy Basin. much more intense than that noted on the pre- Wisconsin Cheney-Palouse flood deposits. How ever, the definitive sequence of loess units is ab sent in the western Quincy Basin, so a precise correlation remains speculative. Richmond and others (1965) also recognize a Bull Lake episode of scabland flooding from widely scattered deposits on the Columbia Plateau. The best evidence occurs at Winchester Wasteway in the Quincy Basin, where the Bull Lake flood deposits overlie the older pre-Wiscon sin flood deposits seen at George (Baker, 1973a, p. 8). THE BONNEVILLE FLOOD Malde (1968) studied the catastrophic flood produced by the overflow and rapid lowering of Pleistocene Lake Bonneville (Fig. 2.8). He traced the course of this flood through the Snake River Plain of southern Idaho to Hells Canyon. Malde interpreted the age of this event to be about 30,- 000 years B.P., based on a radiocarbon date for molluscan fossils associated with flood debris and on the relict soil profile developed on the flood gravel (Melon Gravel). The soil has a thick calcic fc. •^* 's. ^ ^ ^. * •- • ,%* , : ' t ' -'' Figure 2.8. View of the Snake River Canyon down stream of Perrine Memorial Bridge in Twin Falls, Idaho. Scabland erosion of volcanic rocks, here produced by the Bonneville flood, is very similar to that produced by Missoula flooding of the Channeled Scabland. 24 25 horizon extending to depths greater than 2 m (Fig. 2.9). The soil on the Bonneville Flood de posits is believed to have formed during and since the mid-Wisconsin (Bull Lake-Pinedale) in- terglaciation or within the last 30,000 years. Downstream from Hells Canyon at Lewiston, Idaho, probable Bonneville flood deposits are overlain by slackwater surge deposits from the last major episode of scabland flooding (Fig. 2.10). Because Bonnevilfe flooding was confined to the Snake River Canyon, it skirted to the south of the Channeled Scabland. Nevertheless, studies in the Pasco Basin should eventually recognize Bonneville Flood deposits in association with Mis- soula flood deposits. THE LATE PLEISTOCENE DILUVIAN EVENTS The late Pleistocene glacial record of north western Washington and southwestern British Columbia is very well documented through de tailed radiocarbon dating (Armstrong and others, 1965; Fulton, 1971; Easterbrook, 1976). The last major glaciation, called the Fraser Glacia tion, extended from about 20,000 to 10,000 years Figure 2.9. Exposure of Melon Gravel in large boulder bar at mile 161 in the Melon Valley area of the Snake River Canyon, Idaho (Malde, 1968, p. 33). The relict paleosol here is moderately strong, lacking the petro- calcic horizons of pre-Wisconsin flood gravels in the Channeled Scabland. 26 ago. This is also the time interval that produced the last major episode of scabland flooding. At least one and probably two major outbursts of Lake Missoula occurred during this interval. Al though the evidence for the floods is overwhelm ing, the precise dating of the events in this interval is a matter of current controversy. This section will attempt to summarize the diverse arguments. Glacial Stratigraphy The Fraser Glaciation is precisely dated on the western side of the Cascade Mountains. During the Evans Creek Stade large alpine glaciers ad vanced to their maxium extent into the Puget lowland. This was followed by an advance of Cordilleran ice into the lowland from the north sometime after 19,000 years B.P. (Easterbrook, 1976). If the Pend Oreille Lobe of the Cordil leran Ice Sheet correlated with the Puget Lobe, then Lake Missoula could not have formed until after 19,000 years B.P. The Puget Lobe reached its maximum extent 14,000 to 15,000 years ago, during the Vashon Stage. Figure 2.10. Slackwater faciès of the last episode of scabland flooding overlying probable flood gravel of the Bonneville Flood at a gravel pit 5.5 km south of Lewis- ton, Idaho. The foresets and lithologies indicate that the gravel was deposited by flows coming down the Snake River through Hells Canyon. The slackwater deposits were deposited by backwater flooding up the Snake River from the mouth of the Palouse River, a distance of over 120 km. This deposit was described as Tammany Bar by Bretz (1969, p. 531-532). Between 14,000 and 13,000 years B.P. a major recession occurred in the Vashon glacier of the Puget lowland. The interval from 13,500 to 11,- 500 years ago is characterized by relative stability of Cordilleran ice during the Everson Glacio- marine Interstade, which was followed by a rather minor readvance perhaps between 11,500 and 11,000 years B.P. The glacial chronology on the Columbia Plateau is severely hampered by a lack of radio carbon dating. Work on the Waterville Plateau by Hanson (1970) and Easterbrook (1976) shows that an episode of major scabland flooding oc curred prior to the major advance of the Okano- gan Lobe on to the Columbia Plateau. That ad vance produced a spectacular moraine, the With- row Moraine (Waters, 1933; Flint, 1935). This is estimated to have formed about 14,000 to 15,000 years B.P. (Fryxell, 1962; Waitt, 1972a; Easter brook 1976). Relationships near Jamison Lake show the moraine overlying a Moses Coulee flood gravel bar (Fig. 2.11). Subsequent outwash (Fig. 2.12) and dramatic ice stagnation features record a rapid retreat of the Okanogan Lobe (Hanson, 1970; Waitt 1972a), perhaps contemporaneous to the analogous phase of the Puget Lobe. WITHROW MORAIN Figure 2.11. Topographic map of upper Moses Coulee, Withrow Moraine, a pendant bar of flood gravel, and an showing the physiographic relationships between the outwash terrace (from Easterbrook, 1976). 27 Late-Quaternary Vegetation A pollen sequence in a mire on the Telford- Crab Creek scabland tract near Creston, Wash ington records the vegetation changes on the Columbia Plateau that followed the scabland flooding of that region (Hansen, 1947). Mack and others (1976) have reinterpreted the pollen sequence at this locality using improved palyno- logical techniques. The postflood vegetation dur ing the Glacier Peak ash fall (12,000-13,000 B.P.) and slightly earlier consisted of steppe vegetation dominated by sagebrush (Artemisia) occupying extensive areas of stony patterned ground. Nearby loess hills were occupied by pine forest. A warm ing trend is indicated to have begun about 9,000 years ago. The Creston fen accumulated 60 cm of sedi ment between 9,390 ± 480 and 6,700 years B.P. and 70-100 cm between 12,000-13,000 and 9,- 390 ± 480 years B.P. (Mack and others, 1976; their Fig. 2). These accumulation rates range from 1.75 to 2.6 cm per 100 years. Since the mire accumulated 30-50 cm of sediment after scour by the scabland flooding and prior to Glacier Peak ash accumulation (12,000-13,000 years B.P.), that flooding must have occurred at least 13,500-16,000 years B.P. Of course, the depositional rates could have been different in this interval, so this very tenuous age estimate must be compared to other data. Figure 2.12. Oblique aerial photograph of the outwash terrace that extends downstream from the Withrow Moraine in Moses Coulee. Weathering and Soils Near Vantage, Washington, wood, dated by radiocarbon at 32,700 ± 900 years B.P., was found in deposits of the last major flood (Fryxell, 1962). The wood was probably derived from a preflood, interstadial bog on a scabland surface (Bretz, 1969). Soil-stratigraphic evidence (Baker, 1973a) implies that the flooding occurred much later than this lower limiting date, but when? On the Columbia Plateau, one indication of the age of the flooding is provided by the weathering of the basalt boulders on the surfaces of scab- land flood bars. Weathering rinds on these bould ers never exceed 3 mm in thickness. By contrast, rinds on similar basalt boulders deposited by the Bonneville Flood on the Snake River Plain of Idaho may be greater than 1.2 cm thick (Malde, 1968). Since the Snake River Plain has a similar climate to that of the Columbia Plateau, the Mis- soula Flood deposits probably considerably post date the Bonneville Flood, which Malde (1968, p. 10) believes occurred 30,000 years ago. The relict soils on the scabland floor gravels are poorly developed. There are only a few centi meters of near-surface oxidation. Buried cobbles and boulders lack weathering rinds. The Cca hori zon dominates the soil profile, but shows only weak development. Calcium carbonate occurs as coatings on the undersides of pebbles and cobbles to a depth of about 1 m. There is no platy caliche as found on the Bonneville Flood deposits (Fig. 2.9), which were weathered in a similar soil- forming environment. In the eastern Channeled Scabland the last ma jor flood scoured channel ways through the en tire sequence of brown loess units that comprise the Palouse Formation (Fig. 2.13). Small chan nels eroded into the flood deposits by postflood streams contain pale loess and loess-derived alluvium. The oldest soil profiles on postflood loess deposits show textural B horizons, with the <2u size fraction increasing from 8% to 11% (Baker, 1973a). Bretz' early studies concluded that the eastern Channeled Scabland (Cheney-Palouse tract) was somewhat older ' than the western Channeled Scabland (Grand Coulee region). The interpre tation was developed by physiographic evidence (Bretz and others, 1956; Bretz, 1969). Every scabland channel way entering from the east is clearly blocked by bars that were deposited by later flooding down the Grand Coulee system. Because the weathering of the flood gravel in these bars differs little from that of bars in the Cheney-Palouse scabland, either these floods are consequences of the same outburst event or they are only separated by a few thousand years in time. Unpublished observations of post-flood eolian silt accumulation on scabland surfaces may be relevant to this question. Bars and giant current ripples in the west scablands typically have 0.15 to 0.30 m of post-flood eolian accumulation on exposed areas and no more than 0.5 in protected swales. In contrast, the eastern scabland bars have 0.5 to 1.2 m of silt in exposed areas and up to 2.5 m in swales. This would appear to indicate greater age for the eastern scablands, but a prob lem remains. Silt thickness on scabland surface follows a northeasterly gradient such that y - 0.46e°-03* where y is the thickness of postflood eolian silt (ft.) and x is the linear distance (miles) in a northeastward direction from Hanford in the Pasco Basin. Thus, silt thickness may reflect source area and wind-direction controls rather than age. Post-flood silt is so thick at the TSCR ripple field, in the eastern scablands near Tokio Station, that it completely buries the giant current ripples (Fig. 2.14). The TSCR ripples apparent on aerial photographs are actually the silt accumulations over the former ripple troughs. The textural data indicate that this silt contains two weakly devel- 1500 _ NW 1400- o i- I3oo^ Lü Lu 1200 H 0 L 2 MILES RECENT LOESS LOESS AND ALLUVIUM (PINEOALE) :] PALOUSE FORMATION PRE-BULL LAKE LOESS "•".*:| FLOOD GRAVEL X'.l (E. PINEDALE) YAKIMA BASALT Figure 2.13. Diagramatic cross section showing strati- graphic relationships at Staircase Rapids Bar. Thickness of the loess is slightly exaggerated. 28 29 Figure 2.14. Exposure of post-flood eolian silt overly ing flood gravel just north of Tokio Station. The gravel pit occurs at SW V* sec. 15 T. 20 N., R. 36R oped paleosols (Fig. 2.15). This section presents a marked contrast to flood surfaces in the west ern Channeled Scabland. Tephrachronology ^/ Eruptions of volcanoes in the Cascade Moun tains showered ash over wide areas of the north western U. S. during the late Quaternary. Where these ashes can be distinguished, they provide a valuable series of marker beds for dating the asso ciated sedimentary deposits and for tying together the geologic history of widely separated provinces. Two of the most useful and widespread ash falls for work in the Channeled Scabland are those from Glacier Peak and Mt. Mazama (Fry- xell, 1965). Extensive radiocarbon work dates the Mazama ash approximately 7,000 years B.P. (Kittleman, 1973). The Glacier Peak tephra is older than 12,000 and younger than 13,000 years B.P. Lemke and others (1975) designate it as younger than 12,500 years. These'two ashes are easily distinguished by their refractive indices (Mazama > 1.502; G.P. < 1.502), but electron microproble analysis of the shards is necessary to avoid confusion of Glacier Peak and certain Mount St. Helens ashes. In the field the Mazama ash is usually granular, while the Glacier Peak ash forms lentils or pods with the texture of tapioca. Most prolific of the ash sources in the North west was Mount St. Helens. The individual erup tions of this volcano have been dated by the de tailed work of Mullineaux, Crandell, and Rubin, using the carbonaceous material in the coarse airfall material on the flanks of the volcano. The tephra occur in distinct groups, called "sets." The oldest is set S, erupted between 18,000 and 12,000 years B.P. Mullineaux and others (1975) present arguments for the most widespread mem bers of the set S eruption (layers Sg and So) being dated at about 13,000 years B.P. Mount St. Helens tephra set J was erupted from slightly less than 12,000 to slightly more than 8,000 I- e £ TOKIO PIT .' ,' ' •' ' •_' • ' .' Aeolian Si'lt with Paleosols ' ' , • • - <2//CLAY(%) 0 5 10 — White Platy Petrocalcic Horizon HZZ_ —- 4J Figure 2.15. Stratigraphie relationships at the Tokio section. years ago. Tephra set Y was extensively dis tributed in the Northwest between 4,000 and 3,400 years ago. Set W is the most recent, dated at about 450 years B.P. All these tephra units were carried east of Mount St. Helens and all have now been recognized in Quaternary sediments of eastern Washington (Mullineaux and others, 1975). The Mount St. Helens set S ash occurs as "couplets" and "triplets" in fine-grained clastic sediments, often designated as "slackwater faciès." Moody (1977) finds the triplet north of the Saddle Mountains at sites such as Sentinel Gap, Lind Coulee (Table 2.1 ), and Lynch Coulee (Fig. 2.16). South of this range only the upper and middle ash are recognized, forming a "couplet." The couplet is well exposed in "slack- water" sediments on the northern slopes of the Horse Heaven Hills, just south of Kiona, Wash ington. Mullineaux and others (1977) have con cluded that tephra set S dates the last major scabland flood. They support the age estimate of 13,000 years B.P. for the flood with a radiocar bon date of 13,080 ± 350 on peat directly over lying the "Portland delta," a Missoula Flood de posit at Portland, Oregon. Slackwater Sediments The dating of the last major scabland flood is complicated by the fact that the catastrophic flood deposits occur in two facies. The main-channel faciès can always be recognized as unequivocal flood deposition by its coarseness, sedimentary structures, angular boulders, broken rounds, er ratics, etc. However, some of the slackwater facies are easily confused with lacustrine deposits. An unequivocal flood origin can be established by continuous tracing between main-channel areas and slackwater areas, as in the Tucannon Valley sequence (Baker, 1973a). At the Tucannon valley one can follow a complete transition from chaotically deposited boulder and cobble gravel in the main channel to rhythmically bedded silt in slackwater areas 15 km up a preflood tributary of the main channel (Baker, 1973a; p. 42-47). Carl Gustafson (1976) divides "slackwater deposits" into 2 groups, one contemporaneous to scabland flooding and the other postflood. The postflood slackwater sediments contain articu lated bivalve mollusks that lived in a lacustrine environment. At Lind Coulee, Gustafson (in Webster and others, 1976, p. 15) interprets de posits of the major scabland flood (18,000- 20,000 years B.P.), consisting of main-channel and slackwater facies, to be unconformably over lain by younger "slackwater sediments" of lacus trine origin. This lacustrine unit contains the dis tinctive "triplet" of volcanic ash layers from Mt. St. Helens, which is interpreted as the Mt. St. Helens "S" set by Moody (1977). Working in the Lower Snake River Canyon, Hammatt and others (1976) interpreted two flood events and one lacustrine phase for the . l --*">•.. .-»'*• •»•^•«^ir • K ' **" "-,.>/**• - ' * .*-' *A Figure 2.16. FJcposure of the Mount St. Helens ash "triplet" in fine-grained "slackwater" sediment that im mediately overlies scabland gravel at the mouth of Lynch Coulee, near its junction with the Columbia River at West Bar. A coarse white ash layer occurs at "B" and a fine grained thinner layer occurs at "A." The lowest of the three layers occurs discontinuously in the granule gravel "C." The sediment containing the ash (Mt. St. Helens set S) is either time equivalent to or slightly younger than the last major catastrophic flood to affect the Columbia River Canyon northwest of the Channeled Scabland. 30 31 Table. 2.1. Correlation of some late-Quaternary deposits and events on tbe Columbia Plateau. Age Approximate years B.P.) 6700— _ annn _ ^^y\j\j\j — 12,500— 13,000— ±500 14,000— 18,000? Lind Coulee (various sources) (Artemisia on patterned ground; haploxylon pine on loess) and -v—— — ^—— s— -,? .^^^-^-^ i "" ^ ' Marmes Rocksheiter (Fryxell) ROCKFALL — transition — LOESS ASH XXXXXXXXX ^•^ Shell ^•v Midden Loess ^-\ Rockfall ^v Human Burial ROCKFALL ^-^- disconf ormity — ^~ X ASH XXXXXX .^^^,?^^^ ' Okanogan Lobe (Easterbrook, 1976) Holocene Sumas Stade Everson Interstade Recession •i a 9 o > Advance early post-glacial period. Flood gravel, which they interpret as 20,000 years B.P., is uncon- formably overlain by thin bedded silt containing Mt. St. Helens tephra. The silt, interpreted as lacustrine, is unconformably overlain by sandy faciès interpreted as a flood slackwater deposit laid down approximately 14,000 years B.P. This upper unit forms an undulating mantle over earlier deposits and is characterized by distinctly graded bedding. All three units are cut by clastic dikes. Sedimentological criteria must be established to distinguish slackwater faciès of flood origin (Baker, 1973a) from lacustrine deposits. Work on this problem is currently in progress. Flood Deposits Northwest of the Channeled Scabland Waitt (1977b) presents detailed evidence for late Pleistocene catastrophic flooding coming down the Columbia River, through the region that was blocked by the Okanogan lobe until about 13,500 years B.P. Ice-rafted erratics and upvalley-dipping crossbeds in gravel show that this down-Columbia flood was as deep as 400 m at the junction with the. Methow River (Waitt, 1977a). These depths require water at Coulee Dam to have had a surface elevation of 760 m. Waitt (1972b, 1977a) suggests that these rela tionships require hydraulic ponding of a Lake Missoula outburst at the Columbia gorge down stream from Coulee Dam. Because of the chrono logy of the Okanogan lobe, this flood could only have occurred about 13,500 to 13,000 years ago (Waitt, 1977b). Flood sediment relationships at Lynch Coulee (Waitt, 1977b, p. 15) suggest that the Columbia flood was approximately contemporaneous to flooding in the Quincy Basin (presumably de rived from the Grand Coulee). The Columbia flood deposits indicate transport up Lynch Coulee. These are overlain by flood deposits dip ping downcoulee (having flowed from Crater Coulee). The contact shows no weathering. At Moses Coulee, however, the downcoulee deposits came first and even surged up the Colombia (Waitt, 1977b, p. 17). The Moses Coulee flood deposits are overlain by Columbia flood deposits that surged up Moses Coulee near its mouth. XXXX Volcanic Ash Discussion The emerging stratigraphie evidence sum marized above has not yet completely resolved the number and timing of floods in the last epi sode of scabland flooding (19,000-13,000 years ago). The tephrachronology suggests that the last major flood affected the Columbia River, north west of the scablands, and probably the western Channeled Scabland itself. This flood occurred just prior to or nearly coincident with the erup tion of Mt. St. Helens tephra set S (approximately 13,000 years B.P.). This event probably coin cided with the wastage and breakup of the Okanogan ice sheet on the Waterville Plateau. It is also probable that another major episode of scabland flooding preceded the Okanogan breakout by several thousands years. This flood probably affected Moses Coulee, the Grand Cou lee, the Telford-Crab Creek scabland, and the Cheney-Palouse. The dating of this event is less precise, but includes the following: (1) the Withrow Moranie overlying flood gravel in Moses Coulee (Fig. 2.11), (2) physiographic relation ships in the Grand Coulee (Bretz, 1932a, 1969), (3) the amount of sedimentation at the Creston mire prior to the Glacier Peak ashfall, (4) the more extensive silt deposition on the Cheney- Palouse bars, (5) the younger event (13,000 years B.P.?) has bars blocking the mouths of eastern scabland distributaries (Bretz and others, 1956), and (6) stratigraphie relationships in the Snake River Canyon (Hammatt and others, 1976). A tenative scenario that has yet to be fully tested in the field is as follows. The largest out burst of Lake Missoula occurred just prior to the Vashon glacial maximum, but while the Okanogan lobe was advancing. This flood flowed around the Okanogan lobe through the "Mans field channels" (Hanson, 1970) and down Moses Coulee. This route was possible because ( 1 ) the advancing Okanogan lobe had just recently blocked the Columbia gorge, and (2) the upper Grand Coulee cataract had not yet receded to Coulee Dam (Bretz, 1932a). It is likely that this flood initiated the 250 m cataract in the upper Grand Coulee and that it receded to the Steam boat Rock position during the course of the "early Vashon" flooding. The same flood would also put water into the Telford-Crab Creek and 33 Cheney-Palouse scabland tracks. Slackwater de posits from this flood would be the lower se quence recognized by Hammat and others (1976) along the Snake River. The Okanogan lobe advanced to its maximum position, the Withrow Moraine, very shortly af ter the above flood. Wastage of the Okanogan lobe then led to a second flood that also included a burst from Lake Missoula. Richmond and others (1965) refer to this as the "middle Pine- dale" final flood on the Columbia River. How ever, that flood probably also put water over the Grand Coulee cataract head. During the course of this flood, which was mainly influencing the Columbia River in its early phases, the upper Grand Coulee cataract broke through to the Columbia gorge at Coulee Dam. This sent a final surge of water down the Grand Coulee and into the Quincy Basin. Thus, the Lynch Coulee rela tionships are explained by the dynamics of the last major scabland flood. That flood affected both the Columbia River and the western Channeled Scabland, but not the eastern scablands. Moses Coulee did not operate because the cataract re cession of the Grand Coulee in the earlier flood had pirated the channels that fed its upstream end. One variation of the above scenario is possible by using Waitt's (1972b, 1977a) hypothesis of hydraulic ponding at Coulee Dam. This would allow completion of the upper Grand Coulee in the earlier flood. Water would then be simul taneously flowing in the Columbia Canyon (free of the Okanogan lobe) and in the Grand Coulee. POSTDILUVIAN STRATIGRAPHY OF THE COLUMBIA PLATEAU Post-flood conditions in the Channeled Scab- land region have been studied extensively by Roald Fryxell, Carl E. Gustafson, and other workers at the Department of Anthropology, Washington State University. Current data sug gest that man entered the region as early as 12,000 years B.P. He encountered a cold dry steppe with considerable lacustrine areas. De ranged drainage conditions and high water tables were the legacy of recently wasted glaciers and catastrophic flooding. The most intensive study of a post-flood lake in the Channeled Scabland was made by Landye 34 (1973). Lake Bretz, whose shoreline altitude was 350 m (1160 ft.) occupied the closed de pression in the lower Grand Coulee upstream from the bar at Soap Lake. The lake probably lasted a few hundred years, long enough to ac cumulate an abundant population of mollusks and fish. At the maximum extent, the lake ex tended northward about 30 km from Soap Lake, Washington, to the head of Dry Falls, near Coulee City. The lake was probably maintained by high ground-water levels in the basalt associated with the recent déglaciation of the Waterville Plateau. The decline in water tables eventually resulted in the lake's demise. Its former basin is now oc cupied by a chain of somewhat saline lakes, in cluding Deep Lake, Fall Lake, Park Lake, Blue Lake, Lake Lenore, and Soap Lake. Fryxell (1965) recovered coarse volcanic ash from the sediments of Lake Bretz. This ash was identified as Glacier Peak ash. A radiocarbon date of 12,000 ± 30 years B. P. was obtained on the Lake Bretz mollusk shells. This analysis provides an upper limit to the age of the ash, and it also dates the lacustrine phase. However, new data on the Glacier Peak ash show that this eruption may have been the oldest of several. Mehringer and others (1977) describe two Glacier Peak ashfalls in Montana dated at about 11,250 years B.P. and separated by 10 to 25 years. Marmes Rocksheiter, located near the con fluence of the Snake and Palouse rivers, pro duced a long history of occupation (Fryxell and others, 1968; Rice, 1972). Artifacts discovered include bone needles, bone points, stemmed and lanceolate projectile points, and bola stones. A complex burial sequence was found beneath a shell midden dated by radiocarbon at 7,550 ± 100 years B.P. The midden, in turn, is overlain by Mazama ash (6,700 years B.P.). Carl Gustafson (1972) investigated faunal remains from the site and found bones including those of Arctic fox, large elk, pronghorn antelope, deer, bison, ro dents and salmonid fish. These vertebrates be longed mainly to an early postglacial steppe fauna that characterized the region prior to 7,500 years B.P. Marshall (1971) analyzed nearby flood- plain sediments and interpreted precipitation and steam run-off to have been greater than now prior to 7,500 years ago. Frost polygons were found in the overbank silt deposits in which the early Marmes cultural material was discovered, but do not form in the area at present. Rockfall frequen cies also showed that a cool moist climate was present during the time of the early occupation in the site (Fryxell and others, 1968). Because lacustrine sediment containing Glacier Peak ash was found immediately beneath the cultural zone, it is likely that occupation occurred over a long phase of cool, wet conditions between 12,000 and 7,500 years ago. The Lind Coulee archeological site near War den, Washington was the first locality in eastern Washington to yield evidence of paleo-Indian hunters. The early excavation work by Dougherty (1956) was one of the first to use the radiocar bon method, dating the earliest occupation at 8,700 years B.P. A new series of excavations at the site began in 1972. Although those studies are not yet complete, the preliminary results and correlations (Table 2.1) give an excellent late- glacial and post-glacial record. C. E. Gustafson (in Webster and others, 1976) recognizes a catastrophic flood phase unconformably overlain by lacustrine(?) silt containing the St. Helens "triplet" (set S). Inset into this slackwater silt is another silt unit. An overbank unit is set into the silts and contains two discrete ash layers in cluding ashes from St. Helens set J and Glacier Peak (Moody, 1977). Another younger overbank sequence contains another St. Helens J ash dated at 8,700 years B.P. Mazama ash occurs in the eolian silt that blankets the overbank sequence. CONCLUSION Based primarily on physiographic evidence, Bretz (1969) has proposed as many as eight separate scabland floods, of which at least four encountered an ice-blocked Columbia River and spilled over the northern margin of the Columbia Plateau. Bretz' other four floods were not diverted by ice and flowed down the Columbia River valley. Bretz (1969, p. 513) believed that the earliest plateau-crossing flood occurred during the Bull Lake Glaciation. Subsequent floods then enlarged and altered segments of this established drainage. Baker (1973a) noted that some of the physio graphic relationships described by Bretz (1969) could be produced during the dynamic progres sion of a single flood. Only flood deposits recog nized in a firm stratigraphie sequence can be con sidered unequivocal evidence for multiple flood ing. The best stratigraphie information to date suggests that there certainly were five major floods in the general vicincity of the Channeled Scab- land and perhaps six or seven. One, possibly two, floods were pre-Bull Lake and are overlain either by thick caliche or by Palouse Formation. Another flood occurred during the Palouse de position (Bull Lake Glaciation). The third flood was the Bonneville event, restricted to the Snake River. The final flood phase, 18,000-13,000 years ago, seems to include two floods. One of these precedes the Vashon maximum, and the other occurred during or just prior to the Mount St. Helens set "S" eruptions. 35 Il Chapter 3 Bedrock Geology of The Northern Columbia Plateau and Adjacent Areas DONALD A. SWANSON U.S. Geological Survey Menlo Park, California 94025 and ABSTRACT The Columbia Plateau is surrounded by a com plex assemblage of highly deformed Precambrian to lower Tertiary continental and oceanic rocks that reflects numerous episodes of continental ac cretion. The plateau itself is comprised of the Columbia River Basalt Group, a tholeiitic flood- basalt province of moderate size covering an area of about 2 x 105 km2 with an estimated volume of 2 x 105 km3. The Columbia River basalt, formed between about 16.5 x 10" years B.P. and 6 x 109 years B.P., is the youngest known flood basalt. More than 99 percent of the basalt was erupted during a 2.5-3 x 10C years interval centered about 15 x 10° years B.P., building a featureless plateau that sloped toward its center, reflecting concurrent subsidence and volcanism. Eruptions were infre quent between about 14 and 6 x 10G years B.P., allowing time for erosion and deformation be tween successive outpourings. The present-day courses of much of the Snake River, and parts of the Columbia River, across the plateau date from this time. Basalt produced during this waning activity is more heterogeneous chemically and isotopically than older flows, reflecting its pro longed period of volcanism. Most of the flows are thick and ponded behind natural levees. They were erupted from north-northwest-trending linear fissure systems tens of kilometers long, revealed today by dikes and relic vent areas. Eurption rates are estimated for various flows as between 1 km3/ day and ICH km3/day per linear kilometer of ac tive fissure, with flow rates of 5 to 15 km/hr down slopes of 1:1,000 considerd typical. Current mag ma production rates in Hawaii could have pro duced the basalt in the allotted time. No available THOMAS L. WRIGHT U.S. Geological Survey Reston, Virginia 22092 models adequately account for the tectonic setting of the province and its relation to coeval calc- alkaline volcanism in the Cascade Range. INTRODUCTION The Columbia Plateau is perhaps the best area on Earth to view evidence of catastrophic flood ing of truly impressive magnitude. The youngest series of floods produced the famed Channeled Scabland, which we will examine during this field conference. But an earlier episode of repeated flooding left telltale marks of a totally different but equally impressive kind; this flooding, involv ing lava, not water, deposited the flood basalt that built the Columbia Plateau in Miocene time. It is a remarkable coincidence that two such large yet unrelated inundations occurred in the same area and that evidence for both of them can be examined in many of the same outcrops. This paper attempts to summarize briefly the bedrock geology of the northern Columbia Plateau and its surroundings, emphasizing the deposits of the lava floods, the Columbia River Basalt Group. ROCKS BORDERING THE NORTHERN COLUMBIA PLATEAU The Miocene basalt overlies and encroaches upon a diverse assemblage of Precambrian to lower Tertiary rocks (Figs. 3.1 and 3.2). The de- positional, intrusive, and structural histories of these rocks are very complex and poorly known. 37 KOOIENAY- FLATHEAO SU8POOVINCE KOOTENAY ARC SUBPROVINCE WESTERN PROVINCE / • Pend Oreille Lake COEUR D'ALENE UBPROVINCE KEttOGO D'Alene Lake SPOKANE •CHENEY ~" COLUMBIA ra Bonn» Lake^ ^ ^Rock Lake *STEPTOE BUTTE Potholes voir LIND C FRENCHMAN HILLS HOLLOW CLEARWATER EMBAYMENT LEWISTON KAMIAH Figure 3.I. Index map showing localities and boun daries of provinces and subprovinces mentioned in text. Boundaries north of Columbia Plateau from Yates and They form highlands that rise abruptly from the margins of the plateau. No glacial floods coursed across them, except between the ice dam across the Clark Fork River and the plateau near Spo- kane. Glaciers descending from the highlands blocked major drainages and played an important role in scabland development by diverting water onto the plateau. Yates and others (1966) provide an excellent summary of the pre-Miocene rocks north of the plateau and east of the Okanogan Valley. They have subdivided the region into the geologically 38 others (1966). Other province boundaries based on New- comb (1970). / distinct Eastern und Western Provinces, separated by the valley of the Columbia River (Fig. 3.1). The Eastern Province is structurally divisible into three subprovinces, which Yates and others (1966) designated the Kootenay-Flathead, Coeur d'Alêne, and Kootenay Arc subprovinces (Fig. 3.1). The Kootenay-Flathead subprovince is characterized by subparallel, broad, open folds trending northwest to north-northwest; normal faults with displacements of hundreds of meters or more parallel the folds. The major folding is of Laramide age, but the faults are somewhat — o 2 e J- » FI3S 2 - O k O 2 3 r Lis g- o o- s- s n^ > U: ô "2 — o V CO /1 .S 1 è S o s al e£ n .._ *"A *w ^^ 3 O "Ö Ip^l 2 « M <^ g es B ri ? w nf d CO O U d A . g M .a ü 2 2 H .S u ? ri —• o 73 V3 O a, •a s g z s. h i*1 M