The following essay briefly outlines alternative hypotheses considered and evidence investigated prior to undertaking the Missouri River drainage basin landform origins research project, which is based entirely on previously unstudied topographic map evidence. Each phase of my study described here is documented by communications submitted to the North Dakota Academy of Science and/or by abstracts of papers or papers presented at regional and/or national meetings of the Geological Society of America and/or to a variety of other professional organizations. The essay illustrates how investigation of what for the geological research community was and still is unexplained and anomalous evidence kept leading to additional unexplained and anomalous evidence. Perhaps the most disturbing aspect of this study was the discovery that ruling theories have caused geologists to consciously or unconsciously see and interpret evidence that does not exist and to be blind to and/or ignore evidence that does exist. While many published reports are probably excellent and even some information in flawed reports is probably good, there is no way to know what information can be trusted and what information is flawed without personally checking all of the reported evidence. It was this latter discovery that finally led to my decision to base the Missouri River drainage basin landform origins research project study solely on previously unstudied topographic map evidence and by making a conscious effort to follow good scientific procedures. Dates indicate when communications presenting the information were submitted to the North Dakota Academy of Science. Academy reviewers rejected the 1999 and 2002 communications and I submitted that information directly to Academy members.   

1980: The Missouri Escarpment problem and hypothesis:

The Missouri Escarpment is a pronounced step of approximately 100 meters between high plains to the west and south and lower plains to the east and north and in North Dakota extends in a northwest-to-southeast direction from near the North Dakota northwest corner to near the center of North Dakota and then in a south direction into South Dakota. The Missouri Escarpment can be traced in a northwest direction from North Dakota across southern Saskatchewan and then into eastern Alberta and from the North Dakota-South Dakota border in a south direction to south central South Dakota. In all the Missouri Escarpment is approximately 1000 kilometers in length, all which is located somewhat east and north of the commonly recognized continental ice sheet southwest margin. Immediately south and west of the Missouri Escarpment and beginning at the Missouri Escarpment crest is the Missouri Coteau, which is zone of thick glacial sediments ranging from 20 to 50 kilometers in width, and which since the 1960s has been described as a dead ice moraine, although some older literature considered it to be a terminal moraine. The North Dakota Missouri River valley approximately follows the Missouri Coteau south and west margin. Glacial sediments on the Missouri Escarpment slope are thin. Escarpments similar to the Missouri Escarpment bound the Prairie Coteau upland in South Dakota, the North Dakota-Manitoba Turtle Mountains, and a number of southeast Saskatchewan and southwest Manitoba uplands. Uplands surrounded by these other escarpments are also capped with thick glacial sediments similar in nature to the Missouri Coteau glacial material. Lowlands between the various escarpments, including the Missouri Escarpment, can be viewed as broad valleys ranging from 100 to 200 kilometers in width. Glacial sediments on the lowlands are generally thin, but large ice thrust blocks of bedrock are commonly found suggesting the presence of a wet based ice sheet.

I began my study in 1979 by wanting to know how and when the Missouri Escarpment and similar escarpments had been formed. While numerous geologic reports briefly mention and even describe the Missouri Escarpment and/or one of the other regional escarpments as erosional escarpments most reports provide no other information. What attracted my attention were a handful of reports that claimed the Missouri Escarpment was a pre-ice age erosional feature because valleys of pre-ice age rivers had been eroded across it. Bedrock in the region where the escarpments are found is generally easily eroded claystone that has little or no strength when wet. Having seen evidence for what I had been told was deep erosion of hard rocks by Rocky Mountain valley glaciers I could not understand how the Missouri Escarpment and the so called pre-ice age river valleys could have survived continental ice sheet erosive activities. I then proposed at a North Dakota Academy of Science meeting that thin ice lobes widened valleys that had been eroded as diversion channels eroded by water from north-flowing rivers blocked by an advancing ice sheet. The model I proposed showed how multiple such valleys would be formed as the ice sheet advanced with each valley providing a route for further ice sheet advance. Other North Dakota geologists immediately opposed my thin ice lobe model claiming it was impossible because pre-ice age river valleys crossed the Missouri Escarpment.

1981: The western North Dakota regional slope and drainage network hypothesis

By 1981 the strong negative reaction of other North Dakota geologists to my erosion by thin ice lobes hypothesis for the Missouri Escarpment origin had convinced me that the Missouri Escarpment origin could not be determined until the so-called pre-ice age river valleys were better understood. I began by looking at North Dakota’s regional slope, which continues into eastern and central Montana, northern Wyoming, and western South Dakota. This regional slope is an erosion surface cutting across stratigraphic units and is usually regarded to have formed during Tertiary time (prior to any recognized continental ice sheets). The so-called pre-ice age river valleys that cross the Missouri Escarpment are the key evidence supporting a pre-ice age for the North Dakota regional slope. While the Tertiary age of this regional slope is assumed in numerous published reports I could find no literature describing how and why the regional slope was eroded or how and why the pre-ice age river system had developed. The Missouri Escarpment is an important step in this regional slope and I decided to better understand the Missouri Escarpment I needed to also better understand the regional slope and evolution of drainage routes on it.

I began my investigation by looking for an erosion agent that might lower base level to the northeast of North Dakota so as to create a northeast oriented drainage system. The Canadian Shield is located to the northeast of North Dakota and was a large region where all sedimentary strata had been completely removed. I also knew of abundant literature descriptions and from my own observations that continental ice sheets, which at their furthest advances had extended into North Dakota, had once covered the entire Canadian Shield area. Having visited several Rocky Mountain ranges where valley glaciers were interpreted to have eroded deep valleys in granitic and other types of hard bedrock I decided to develop and test a hypothesis that continental ice sheet erosion had deeply eroded the region it had occupied, with the most intense erosion being in the Canadian Shield area. This hypothesis assumed that prior to the continental ice sheet erosion stratigraphic units now truncated by the northeast oriented regional slope once extended much further to the northeast and probably covered much or all of the present day Canadian Shield area. The hypothesis further suggested glacial erosion had lowered base level by several hundred meters, which resulted in a generally lowering of the bedrock surface throughout the region. 

To test my newly developed continental ice sheet erosion hypothesis I started looking at drainage routes along and near the mapped continental ice sheet margin, which in North Dakota is located south and west of the Missouri River. In southwest North Dakota the Little Missouri River flows in a deep north oriented valley before turning to flow in more of an east direction to join the Missouri River. East of the north oriented Little Missouri River valley is an asymmetric drainage divide where one can stand and look west down steep slopes into the Little Missouri River valley or turn and look east along a gently sloping plain drained by southeast oriented headwaters of the Heart, Cannonball, and Grand Rivers. Abundant coarse-grained alluvium, including small boulders, consisting of rock types that were derived from distant bedrock sources are easily found as lag deposits scattered across the gently sloping plain. Standing 100 to 300 meters above this gently sloping plain are isolated buttes. Some of the these buttes contain coarse-grained alluvium (including small boulders) from distant source areas in stratigraphic units that have been mapped as being “Oligocene”, “Miocene”, and Pliocene” in age.

Based on the asymmetric drainage divide and evidence the coarse-grained alluvium had been transported great distances from somewhere other than North Dakota I determined the deep Little Missouri River valley and Yellowstone River valley had been eroded headward as they captured flow from a series of large southeast oriented rivers that had stripped approximately 300 meters of bedrock material from all of southwest North Dakota. I considered the most likely mechanism responsible for the erosion were large southeast oriented ice-marginal melt water floods flowing along a continental ice sheet margin. Projecting the ice-marginal floods headward along the mapped continental ice sheet margin I was led to the Flaxville, Wood Mountain, and Cypress Hills alluvial deposits, which cap isolated and successively higher upland regions. The Flaxville, Wood Mountain, and the Cypress Hills alluvium, like some of the North Dakota alluvium, contain “Pliocene”, “Miocene”, and “Oligocene” fossils. I suggested the isolated Cypress Hills, Wood Mountain, and Flaxville uplands were evidence that southeast oriented ice-marginal melt water rivers had stripped the surrounding bedrock.

I further proposed that alluvium capping the Cypress Hills, Wood Mountain, and the Flaxville upland, and included in the mapped North Dakota “Oligocene”, “Miocene”, and “Pliocene” sediments had been transported by the ice-marginal melt water floods and that fossils contained in that alluvium had been reworked from eroded bedrock units. Needless to say vertebrate paleontologists immediately jumped on my hypothesis and said it was impossible because there was good evidence the fossils had not been reworked.  I revised the hypothesis to say melt water floods had carried animal carcasses intact for great distances, but vertebrate paleontologists still insisted there was a 30 million year gap between the time the fossils had been deposited and the time continental ice sheets had formed.

1982: The southwest North Dakota “Oligocene” problem

By 1982 I was extremely puzzled because the fossil age dates appeared to be blocking excellent geomorphic evidence that supported deep erosion by continental ice sheets so I decided to look more closely at the mapped North Dakota “Oligocene” and “Miocene” sediments. A new USGS Geologic Map of North Dakota had been published and made available and on that map alluvium in southwest North Dakota “Oligocene” and other undated alluvial sediments, labeled “QTU” (Quaternary-Tertiary undivided), were described as having been derived from either the Black Hills or the Rocky Mountains. These were the melt water transported alluvial materials in my previous hypothesis. My immediate thought on seeing that newly published map was the “Oligocene” and “QTU” origin needed to be better determined and began a literature search to determine why the alluvium source area had not been better pinned down.

The alluvium in question consists of large quantities of gravel, cobbles, and even small boulders of igneous rocks, quartzites, and other rock types, none of which outcrop anywhere near southwest North Dakota. The igneous rocks include a wide variety of distinctive rock types some of which are easy to identify and trace. Alluvium in some of the mapped “Oligocene” deposits was located primarily in isolated buttes that stand up to 300 meters above the surrounding landscape and appeared to me to have been rapidly dumped in narrow valleys with topographic inversion occurring when the surrounding bedrock was removed. “Oligocene” alluvium deposit locations and cross bed directions suggested to me that those deposits might have been deposited by east oriented rivers, but I could not make a convincing case from the observed evidence and the map’s suggested possible Black Hills source argued for deposition by north-flowing rivers. As previously noted today the alluvium is found on a gently sloping erosional surface drained by southeast oriented Heart, Cannonball, and Grand River headwaters and is located just east of the deep north oriented Little Missouri River valley.  

The mapped “QTU” alluvial deposits are similar in composition to the “Oligocene” alluvial deposits, although appear to be reworked material possibly derived from earlier “Oligocene” alluvium, and were much more extensive than shown on the newly published geologic map. Many of the “QTU” deposits are linear and extend for several kilometers in west to east and northwest to southeast directions suggesting transport and deposition by multiple parallel and large east and/or southeast oriented rivers, which were flowing a short distance south and west of the mapped continental ice sheet southwest margin location. ”QTU” deposits were located on both sides of the Little Missouri River valley, which I interpret to mean the Little Missouri River valley did not exist at the time large east and southeast oriented rivers had transported and deposited the “QTU” alluvium. I was also able to trace linear deposits of the alluvium in a southeast direction across the present day Cannonball River-Grand River drainage divide and across the present day Grand River-Moreau River drainage divide (into the southeast oriented Thunder Butte Creek valley, which drains to the Moreau River). 

I searched the literature to try to discover why the new Geologic Map of North Dakota was showing two alternate source areas for the southwest North Dakota alluvium. I found an extensive literature written primarily by vertebrate paleontologists that extended back to the early 1900s in which the North Dakota “Oligocene” alluvium had been briefly noted and described and in which the Black Hills was reported to be the alluvium source. Several of the reports even claimed the observed presence of Black Hills alluvium in the North Dakota “Oligocene” sediments was proof north oriented drainage along the present day Little Missouri River route dated back to at least Oligocene time. I found no literature published previous to 1965 that suggested a Rocky Mountain source for the southwest North Dakota alluvium. In 1965 Denson and Gill in USGS Professional Paper 463 reported that they could not find a match between Black Hills rock types and a distinctive igneous rock type found abundantly in the North Dakota “Oligocene” alluvium and suggested the source of that distinctive igneous rock type was in the Beartooth Mountains of Montana. The USGS report also includes a map showing a hypothesized route the alluvium had traveled, which started in the Absaroka Mountains and crossed the Powder River Basin before reaching southwest North Dakota.

What was most intriguing about my literature search was that even after publication of the Denson and Gill report vertebrate paleontologists and others kept publishing reports saying the North Dakota “Oligocene” alluvium source area was the Black Hills. Perhaps the most interesting report was a PhD dissertation, which subsequently was published as a state geological survey report in which the author claimed the North Dakota “Oligocene” alluvium could not have been derived from the USGS report’s suggested source area. He then presented an argument that the alluvium was not only from a Black Hills source that had been completely eroded away, but that North Dakota “Oligocene” alluvium could be used to help reconstruct Black Hills igneous history. Both the PhD dissertation and the state geological survey report quietly disappeared after I reported evidence described in the following two paragraphs, although even to this day North Dakota geologists and vertebrate paleontologists are still publishing reports in which the alluvium source is not clearly defined and that hint at a Black Hills source.    

Assuming the USGS report to be correct in saying the alluvium did not come from the Black Hills and the PhD student report to be correct that the alluvium did not come the USGS report suggested source, I looked for alternate source areas along my projected ice-marginal flood flow routes to check. After reading additional literature I decided the most likely alluvium source was in the Bear Paw Mountains, where based on literature descriptions there appeared to be similar rock types. Confident that I had solved the problem, at my first opportunity I drove from Minot to the Bear Paw Mountains and much to my disappointment was unable to match rock types. I started driving back from the Bear Paw Mountains in the direction of the North Dakota “Oligocene” deposits and checked every lag gravel and stream alluvial deposit I encountered to see where the North Dakota alluvium rock types first appeared. No rock types matching the North Dakota alluvium were encountered until I entered the Redwater River valley and then the distinctive North Dakota stuff was everywhere. Continuing further east the Redwater-Yellowstone River drainage divide was covered with thick deposits of alluvium matching the North Dakota “Oligocene” alluvium. The stuff was also common throughout the Yellowstone River valley and easily observed in lag deposits as well as in the mapped “QTU” deposits crossing the Yellowstone River-Little Missouri River valley.

I next proceeded up the Yellowstone River valley (on subsequent trips) checking every Yellowstone River tributary valley to determine where the alluvium first entered the Yellowstone River valley and found the distinctive igneous rock source in dikes and sills exposed along walls of Beartooth Mountain valleys, right where the USGS report had suggested. I also checked with negative results Black Hills rock types to see if matches were possible and looked to see if there was any evidence of a trail leading from the Black Hills to southwest North Dakota. Not only could I find no evidence of a trail there were mapped “Oligocene” deposits in northwest South Dakota containing coarse-grained alluvium that very definitely did not contain the distinctive igneous rock types found in the North Dakota “Oligocene” alluvium. I further checked, also with negative results, the USGS report suggested route across the Powder River Basin and found no evidence to support transport of the distinctive alluvium along that route.

The distinctive Beartooth Mountains dike and sill alluvium can also be found on floors of the so-called “pre-ice age” valleys north and east of the present day Missouri River valley both in North Dakota and in neighboring southwest Manitoba. This evidence is often used to demonstrate the “pre-ice age” origin of those valleys and certainly demonstrates those valleys were eroded prior to the North Dakota Missouri River formation. While this evidence could not be ignored I decided to place that evidence on the back burner so I could first better understand what was happening south and west of the mapped continental ice sheet margin. It took almost ten years of developing and testing alternate hypotheses before a hypothesis that explained the alluvium distribution pattern both south and west and north and east of the mapped continental ice sheet margin.

1982: The northwest South Dakota “Oligocene” problem

I next started to investigate literature describing northwest South Dakota “Oligocene” deposits and again found a fascinating story. The northwest South Dakota “Oligocene-Miocene” deposits are also usually located in isolated buttes, which appear to me to be erosional residuals left after up to 300 meters of bedrock had been removed from the surrounding landscape. Some butte shapes and orientations (e.g. Slim Buttes) suggest to me the “Oligocene-Miocene” and possibly later sediments were probably deposited in narrow valleys, which may have had a northwest to southeast orientation, and subsequent deep regional erosion then removed the surrounding bedrock material and inverted the topography. Most of the present day regional drainage is oriented in a southeast direction suggesting multiple southeast oriented drainage routes were responsible for removing up to 300 meters of regional bedrock material. Reva Gap is today an obvious wind gap cut across the Slim Buttes upland drained to the east by a southeast oriented stream.

W.C. Toepelman, who was a University of Chicago PhD student working under the supervision of J Harlan Bretz, in 1925 completed the first detailed geological study of the Reva Gap “Oligocene” and “Miocene” sediments and found evidence for large landslide blocks along the walls of what he reported to have been a deep northwest to southeast oriented “Oligocene” or “Miocene” canyon. A 1959 South Dakota Academy of Science Proceeding paper by Malhotra and Teglund reported on a heavy mineral analysis of Reva Gap “Oligocene-Miocene” sediments that found no match with Black Hills heavy minerals and suggested a source to the northwest. In 1962 Gill published in the GSA Bulletin (vol. 73, p. 725-735) a report in which he included maps of landslide blocks along the walls of northwest-to-southeast oriented “Oligocene-Miocene” canyons at several Harding County, South Dakota and southeast Montana “Oligocene” and “Miocene” deposits. Gill’s maps include a map of the landslide block at the Reva Gap location that Toepelman’s thesis had described.

Yet in 1970 Jason Lillegraven published his vertebrate paleontology masters thesis (GSA Bull vol. 81, p. 831-850) in which he rejected the landslide block hypothesis because he could find no evidence for an “Oligocene-Miocene” River. On page 848 Lillegraven makes the following statement “No evidence was observed at Slim Buttes for the presence of a latest Oligocene or earliest Miocene major stream capable of cutting the deep valleys necessary to accommodate and to preserve 440 feet of Brule slump-block sediments. Certainly any stream of this magnitude would have left some indication of its existence.” Lillegraven did not specifically state the Slim Buttes “Oligocene” sediments were derived from the Black Hills although he correlated them with similar sediments in the South Dakota Big Badlands and implied the Black Hills were the source area.

Shortly after doing that literature research I visited the Reva Gap location with a group of freshman students. Immediately after we stepped out of the vans, almost as a joke, I suggested there might be evidence the locality (which is more than 100 kilometers south and west of the mapped continental ice sheet margin) had been glaciated. Within five minutes of making that comment every one of those freshmen students had found rounded cobbles and small boulders of quartzites, granite, and other rock types and was asking if those rocks suggested the area had been glaciated. I had to tell the students the rounded cobbles and small boulders they had found were probably evidence a major river had once crossed the region and that the river had transported the rounded cobbles and small boulders from a distant mountain range, but probably the cobbles and boulders were not evidence the Reva Gap area had been glaciated.

I subsequently visited every locality Gill had mapped and at each locality found cobbles and small boulders of rock types similar to those at the Reva Gap. Cobbles and boulders and other alluvium found in and associated with the Harding County, South Dakota “Oligocene-Miocene” deposits do not include any of the distinctive igneous rock types found in the North Dakota “Oligocene-Miocene” and “QTU” deposits. I was unable to trace any of the Harding County alluvium to source areas, although one conglomeratic quartzite boulder found at Reva Gap appeared to match a conglomeratic quartzite I had seen in the Clarks Fork of the Yellowstone drainage basin. I could not rule out other source areas for that boulder and also felt a much better match was needed. It is possible the alluvium came from the Black Hills, but I was unable to find any evidence of a coarse-grained alluvium trail reaching from the Black Hills to the isolated Harding County “Oligocene-Miocene” deposits. Critics and reviewers have several times told me the Lillegraven publication is the more recent work and that it disproved the Toepelman, Malhotra and Teglund, and the Gill interpretations of the Reva Gap evidence.

Before concluding my discussion of the northwest South Dakota “Oligocene” and “Miocene” I should state many published reports, including the Gill report, assume the present day small and isolated “Oligocene” and “Miocene” deposits to be erosional residuals that provide evidence of what were once continuous layers of “Oligocene”, “Miocene”, and even Pliocene” sediments that extended northward from the Black Hills into North Dakota. While the 1959 Malhotra and Teglund report provided heavy mineral analysis evidence that the Slim Buttes sediments had been derived from a source other than the Black Hills subsequent workers did attempt to use heavy minerals to establish a Black Hills source area. For example, Clark (1975 in GSA Memoir 144, p 95-117) discusses inconclusive heavy mineral data that might suggest a Black Hills source and then concludes by saying the Slim Buttes “Oligocene” sediments were derived from the Black Hills and shows a map (p. 100) with an arrow pointing from the Black Hills to the Slim Buttes and implying Black Hills sediments may have continued further north.

1982: The valley glacier melt water hypothesis

In the early 1980s I was ready to propose a new hypothesis to explain the evidence I had observed. There is an excellent trail of distinctive alluvium leading from Beartooth Mountain valleys to the southwest North Dakota “Oligocene” and “QTU” deposits. There is good evidence the Beartooth Mountain valleys from which the distinctive alluvium was eroded were occupied by large valley glaciers. The trail suggested distinctive coarse-grained alluvium had been eroded from the Beartooth Mountain valley walls and then transported by a large river flowing roughly parallel to the present day Yellowstone River route, although at an elevation equivalent to the present day Yellowstone-Redwater River drainage divide. Something in eastern Montana blocked, or at least partially blocked that high-level river causing the river to deposit large quantities of alluvium on its floor and also diverting water to flow in multiple east and southeast oriented channels into and across southwest North Dakota. Some of these channels eroded deep valleys into the North Dakota bedrock and these deep valleys were subsequently filled as the river continued to transport alluvium into the region. The eastern Montana blockage occurred at approximately the location of a mapped continental ice sheet southwest margin and the diverted water had flowed parallel to and near that mapped continental ice sheet margin.

Absent the “Oligocene” age determination for the southwest North Dakota “Oligocene” sediments and the presence of the north-oriented Little Missouri River anyone looking at this set of evidence would logically suggest the Beartooth Mountain valley glaciers had eroded the alluvium and then melt water from those glaciers had transported the alluvium to eastern Montana where a continental ice sheet margin had blocked the melt water river and diverted the water and alluvium in an east and southeast direction. Further there was excellent evidence the North Dakota “Oligocene” alluvium did not match the South Dakota “Oligocene” alluvium suggesting the two areas of isolated “Oligocene” deposits did not correlate with each other. With this information in hand I then developed and proposed my first hypothesis to explain the evidence.

My first hypothesis simply stated “Pleistocene” Beartooth Mountain valley glaciers had deeply eroded Beartooth Mountain valleys and large melt water rivers had then transported the eroded material in a northeast direction to eastern Montana where a “Pleistocene” continental ice sheet margin had blocked the melt water river and diverted at least some of the water and alluvium in an east and southeast direction along that continental ice sheet margin. I also suggested the mapped North Dakota “Oligocene” sediments did not correlate with the South Dakota “Oligocene” sediments and should be remapped as “Pleistocene” sediments. Vertebrate paleontologists objected strongly to my suggestion that the North Dakota “Oligocene” deposits might be “Pleistocene” in age and just as strongly to my suggestion that the North Dakota “Oligocene” sediments did not correlate with the South Dakota “Oligocene” sediments.

1983: Jokullhlaups from a Yellowstone Plateau ice cap hypothesis

After thinking about the vertebrate paleontologists’ insistence that the North Dakota “Oligocene” sediments correlated with the South Dakota “Oligocene” I decided in 1982 to trust the geomorphic and other evidence I had collected and to treat the vertebrate paleontologists’ insistence that the North Dakota “Oligocene” sediments and fossils correlate with the South Dakota “Oligocene” sediments and fossils and with “Oligocene” sediments and fossils throughout the Rocky Mountains and Great Plains also as evidence. In other words, I decided to test the hypothesis that all Rocky Mountain and Great Plains “Oligocene” sediments and fossils had been deposited at a time when a continental ice sheet was present. I started by looking for “Oligocene” sediments located near or along the mapped continental ice sheet margin. I had already suggested melt water rivers had transported and deposited the Cypress Hills alluvium along or near an ice sheet margin, although up until that time I had not tried to figure out any details. I also had found evidence reported by previous investigators and confirmed by freshman students that suggested southeast oriented melt water rivers flowing roughly parallel to a continental ice sheet margin had deposited and eroded northwest South Dakota and southeast Montana “Oligocene” sediments. I then noted vertebrate paleontologists’ had described “Oligocene” sediments in the South Dakota Big Badlands as having been deposited in an east oriented valley that led directly to the mapped continental ice sheet margin.

After studying the literature and making my own observations of these ice-marginal “Oligocene” sediments I determined age dates and depositional environments vertebrate paleontologists were assigning to the fossils and sediments were the only hard evidence suggesting these “ice-marginal” “Oligocene” sediments had not been deposited along or near a continental ice sheet margin. To further test my hypothesis I then looked for a water source that would create a large enough flood so as to transport coarse-grained alluvium to multiple locations along a continental ice sheet margin extending from near the present day Cypress Hills to near the present day Black Hills. After some thought and literature research I decided volcanic eruptions under an ice cap covering the Yellowstone Plateau could account for the various mapped “ice sheet marginal” “Oligocene” and “Miocene” deposits. I began testing this new hypothesis by taking a special course on Yellowstone volcanic history and then by hiking across mountain passes to inspect routes the floods would have had take to reach the critical Beartooth Mountain valleys. Considerable time was also spent exploring map evidence to see routes the floods would have taken to reach the other “ice sheet margin” “Oligocene” sediments.

The hypothesis proved highly productive. There were identifiable flood flow routes to each of the mapped “ice sheet marginal” “Oligocene” deposits and there was evidence that large volumes of water had once flowed along at least segments of those routes. Again I studied the literature looking for evidence that might negate the hypothesis and again the conflicting evidence consisted of the age dates assigned by vertebrate paleontologists and geochronology experts to those sediments and the fossils and the published depositional environment interpretations. But the depositional environment interpretations could be reinterpreted. For example, vertebrate paleontologists often interpreted concentrations of vertebrate fossils at many of the sites to mean the climate was dry and that animals were using the sites as watering holes. A reinterpretation that would explain those same fossil concentrations would be large outburst floods overwhelmed the animals at distant locations and transported the animal carcasses intact to near a continental ice sheet margin and then buried the intact carcasses in other flood transported debris.

An unexpected bonus the jokullhlaups from the Yellowstone Plateau ice cap hypothesis provided was that flood flow routes could also be identified that led to numerous mapped Wyoming and Montana “Oligocene” sediments that were not located near a mapped continental ice sheet margin. Published literature descriptions of these other “Oligocene” sediments often mentioned the presence of coarse-grained alluvium and other than the fossil age interpretation did not present what I considered evidence that prevented the possibility of deposition by large floods emanating from the Yellowstone Plateau area. I did not make special trips to see these additional Wyoming and Montana “Oligocene” sites in person, although I had seen some of the Wyoming “Oligocene” and “Miocene” sediments when doing thesis work in Wyoming some 15 years earlier.

The jokullhlaup hypothesis not only explained the alluvium transport route from the Beartooth Mountains to southwest North Dakota, but it also explained why relatively small and isolated “Oligocene” deposits containing coarse-grained alluvium would be located along or near a mapped continental ice sheet margin and also why “Oligocene” deposits might be found elsewhere in many Montana and Wyoming localities. The hypothesis also provided a possible explanation for the evolution of major drainage routes that today radiate in all directions from the Yellowstone Plateau. However, the hypothesis eventually had to be rejected because I could not rule out the possibility that the vertebrate paleontologists’ correlation of “Oligocene” fossils and sediments was correct and if so I needed a melt water source able to reach and explain all mapped western United States terrestrial “Oligocene” deposits. Other than the age conflict melt water floods generated by volcanic eruptions under a Yellowstone Plateau ice cap could explain numerous “Oligocene” deposits, but there are also numerous western United States “Oligocene” deposits that could not have been reached by such melt water floods.

Before abandoning the hypothesis I did test to see if adding outburst floods originating from volcanic eruptions under ice caps covering the Colorado San Juan Mountains and in the Cascade Mountain area could explain “Oligocene” and “Miocene” sediments that floods from the Yellowstone Plateau could not reach. While successful in being able to reach many of the previously unreachable “Oligocene” sediments I decided the addition of these new flood source locations unnecessarily complicated the hypothesis and Occam’s razor was being violated. The hypothesis did convince me that vertebrate paleontologists’ correlation of terrestrial “Oligocene” sediments might be valid. If so, I interpreted the correlation to mean the sediments may have been deposited under similar conditions, although perhaps not at identical times.

1985 and later years: Ice sheet margin erosion hypotheses

The decision to reject the jokullhlaup hypothesis forced me to take a close look at the North and South Dakota continental ice sheet margin evidence. West and south of the North and South Dakota Missouri River valley free-standing glacial erratics, with little or no inclosing matrix or fine-grained drift of any kind, had been used by previous workers to mark the maximum ice sheet advance position. North and east of the Missouri River valley fine-grained drift is abundant and surrounds the erratics. Some previous workers regarded the difference to be the result of two different ice sheets, with an earlier ice sheet advancing to a location south and west of the Missouri River valley, while the younger ice sheet only extended as far as the Missouri River valley location.

However, Flint (1955, USGS Professional Paper 262, p. 85-87) had compared weathering of erratics both south and west of the Missouri River with weathering of erratics north and east of the Missouri River and found no differences. Based on the weathering data Flint proposed the same ice sheet that had deposited the erratics north and east of the Missouri River valley had deposited the erratics south and west of the Missouri River trench. Flint then suggested the Missouri River valley and its accompanying breaks had served as a baffle that had trapped the flow of basal ice, while the upper part of the ice sheet, which contained little fine-grained drift material, was permitted to flow freely beyond the Missouri River valley. Flint’s conclusion about the age of the erratics south and west of the Missouri River valley intrigued me, because up until reading the Flint report I had considered the deposition by two or more ice sheets hypothesis to have been possible, although I had always thought the similarities between the two (or more) ice sheet maximum advances were rather remarkable.

Regardless of how many ice sheets were involved the hypothesis I wanted to test related to whether the lack of fine-grained drift materials south and west of the Missouri River was evidence that the drift had been deposited on a topographic surface at least as high as the highest mapped “Oligocene”, “Miocene”, and “Pliocene” sediments. My hypothesis was the mapped “Oligocene”, ”Miocene”, and “Pliocene” sediments had been deposited in valleys carved by ice-marginal melt water rivers and that continued ice-marginal melt water flow had then inverted the topography as bedrock surrounding the “Oligocene”, Miocene”, and “Pliocene” valley fills was removed.

To test my hypothesis I decided to compare deeply eroded glacial sediments north and east of the Missouri River valley with glacial deposits south and west of the Missouri River valley. The area north and east of the Missouri River valley that I selected was in the Ward County, North Dakota Souris River valley about ten kilometers northeast of Velva, North Dakota. The Souris River valley has been described by other workers as a glacial lake spillway carved by a large catastrophic flood. Evidence for glaciation in the catastrophic flood-eroded valley study area consisted primarily of erratic boulders with the largest being 3 by 4 meters in dimension, although on both sides of the flood eroded valley uplands were covered with several meters of fine-grained glacial materials. Erratic boulders on the valley floor and on the upland surfaces showed similar weathering characteristics, with little or no weathering observed in either location.

The size, concentration, and degree of weathering of erratic boulders in the Souris River valley study area were then compared with the same characteristics of erratic boulders between Mandan and New Salem in Morton County, North Dakota. The largest erratic boulder observed in Morton County was 6 by 4 meters in size, which suggests it was probably transported by an ice sheet and not ice rafted to the location. The Morton County erratic boulders appeared to be similar in size, concentration, and weathering characteristics to the Ward County erratic boulders. The only obvious difference was the Morton County erratic boulders were many kilometers from the nearest fine-grained drift, while fine-grained drift materials were abundantly present on either side of the Souris River valley. I saw no evidence that caused me to question Flint’s earlier conclusion that erratic boulders found south and west of the Missouri River valley were probably deposited by the same ice sheet that deposited the erratic boulders found north and east of the Missouri River valley.

This study convinced me that ice-marginal melt water floods had removed 300 or more meters of bedrock material from the southwest North Dakota region south and west of the Missouri River and that mapped southwest North Dakota  “Oligocene”, “Miocene”, and “Pliocene” sediments had been deposited in deep valleys carved into a topographic surface at least 300 meters higher than southwest North Dakota’s present day erosion surface. I concluded that ice-marginal melt water floods would have been the logical agent for eroding the deep valleys, for subsequently filling the valleys with “Oligocene”, “Miocene”, and “Pliocene” sediments, and for inverting the topography by stripping bedrock materials surrounding the “Oligocene”, “Miocene”, and “Pliocene” valley fills. 

1988: Investigating the age date problem

By the mid 1980s I was perplexed. The jokullhlaup hypothesis explained so much evidence that I was certain it was very close to what had actually happened, but as noted it did not explain all of the evidence and then there was the continuing age date problem, which if correct might mean all of the geomorphic and other data I had collected was meaningless. Unable to tweak the jokullhlaup hypothesis in a way that explained all western United States terrestrial “Oligocene” deposits I decided to look more closely at the age date problem. Before continuing I should say geologists use two types of dating. The first is relative dating where events are put in order. I am great believer in relative age dating and use it constantly in my geomorphology studies. Further, no evidence that I encountered as my research led up to the jokullhlaup hypothesis or found  during the jokullhlaup hypothesis testing was inconsistent with relative age dates previous investigators had established. In fact the jokullhlaup hypothesis supported vertebrate paleontologists when they correlate “Oligocene” sediments from region to region.

Absolute age dates represent the second type of dating geologists frequently use and is done by assigning specific number of years before present to fossils, sediments, volcanic rocks or ashes, or to just about anything else geologists can find a way date. A great variety of absolute age dating techniques exist, however they are all based on assumptions, which may or may not be correct. Also, every absolute dating technique requires calibration by comparing its results with an already known history. Techniques used to obtain absolute dates for Cenozoic history events frequently have been calibrated by use of vertebrate paleontology established Cenozoic history interpretations, which again may or may not be correct. In other words, unlike relative age dates where there is little room for error if a few relatively simple rules are followed, assumptions required by almost all absolute age techniques open up numerous error opportunities.

Evidence I had assembled suggested a continental ice sheet had been present when western United States terrestrial “Oligocene” sediments were deposited and even suggested large floods had deposited the sediments. Large floods able to reach all of the “Oligocene” sediment localities could best be explained by glacial melting. Yet according to geochronologists who have been assigning absolute age dates to various geologic events there was a time gap of approximately 30 million years between the time “Oligocene” sediments were deposited and the time continental ice sheets were present. I spent considerable time looking at absolute age dates being assigned by different authors to sediments I was investigating and while I did not have a simple way of challenging the published absolute age dates two lines of evidence suggested the published absolute age dates might be little more than pure gibberish.

The first line of evidence was in the isolated and mapped southwest North Dakota and northwest South Dakota “Oligocene”, “Miocene”, and “Pliocene” sediments previously described. At several sites, which are today isolated buttes, mapped “Oligocene” sediments are found on top of mapped “late Eocene “ sediments and the mapped “Miocene” sediments are on top of the “Oligocene” sediments with the mapped “Pliocene” sediments on top of the entire sequence. Geologists who first mapped these sediments assumed the buttes were erosional remnants of continuous layers of “Oligocene”, “Miocene”, and “Pliocene” sediments that once extended from the Black Hills into southwest North Dakota and perhaps much further. However, the sediments to me suggest that high-energy rivers flowing in relatively narrow valleys had deposited them. If my interpretation was correct the mapped  “Oligocene”, “Miocene”, and “Pliocene” sediments could have been deposited in a matter of just a few years. Yet according to geochronologists the “Oligocene” began 33.9 million years ago and the “Pliocene” occurred between 5.3 and 2.5 million years ago. The sediments themselves were telling me the assigned absolute age dates made no sense.

The second line of evidence was I noted several inconsistencies between absolute dating done by different methods and/or used in different studies. I then looked for a way to devise a hypothesis that would determine whether absolute age dates calculated by different techniques and being used to date sediments other than “Oligocene” sediments could be used to determine absolute age dates for a mapped “Oligocene” sedimentary deposit. As already described I had established a link between the southwest North Dakota “Oligocene” sediments and alluvium found along Yellowstone River terraces. I noted several different researchers using several different age dating techniques had published age dates for Yellowstone River terraces or for terraces along major Yellowstone River tributaries and were using those age dates to determine Yellowstone River valley incision rates. While evidence I had found suggested the Yellowstone River valley had probably been rapidly eroded headward during large melt water floods I decided for the purposes of my test to assume the Yellowstone River valley incision occurred gradually as most other researchers were suggesting.

For my test I used published age dates for a Yellowstone River terrace south of Circle, MT and a Missouri River terrace south of Flaxville, MT (Colton, Naseser, and Naeser, 1986, GSA Abs. with Prog. v. 18, p 347), a Yellowstone River terrace near Billings, MT and a Clarks Fork terrace near Red Lodge, MT (Reheis and Agard, 1984, GSA Abs with Prog., v. 16, p. 632), a Tongue River terrace at Ashland, MT (Coates, Naeser, and Heffern, 1987, 12^th Int. Cong., Inter. Union, Quaternary Res. Prog. With Abs., p. 145), a Bighorn River terrace in the Wyoming Bighorn Basin (Palmquist, 1983, in Geology of the Bighorn Basin: 34^th Field Conference Guidebook, Wyo. Geol. Assoc., p. 217-231), and a Greybull River terrace near Tatman, Wyoming (Ritter, 1975, in Geology and Mineral Resource of the Bighorn Basin: 27^th Annual Field Conf. Guidebook. Wyo. Geol. Assoc., p. 37-44). Using each of those published age dates I calculated Yellowstone Valley incision rates and used those Yellowstone River valley incision rates to determine an absolute age date for the Yellowstone River drainage basin alluvium found in the southwest North Dakota mapped “Oligocene” sediments. The calculated incision rates (as determined from the published terrace absolute dates) suggested southwest North Dakota “Oligocene” sediments were deposited between 1 Ma and 14.2 Ma (depending on which incision rate location and which southwest North Dakota “Oligocene” sediment location were used). In other words, even though I considered all of the Yellowstone River and tributary terrace age dates to be suggesting an incorrect Yellowstone River and tributary erosion model, every one of those published age dates suggests the southwest North Dakota mapped “Oligocene” sediments are much younger than the 33.0 to 28.0 Ma age vertebrate paleontologists and others were insisting on.   

1988: Melt water flow routes along an ice sheet’s southwest margin hypothesis

My absolute age date investigation convinced me that mapped “Oligocene” sediments are probably much younger than the 33.0 to 28.0 MA age being claimed by vertebrate paleontologists and others, but it did not solve the problem of a water source capable of producing large floods that could reach all mapped western United States “Oligocene” sediments. The only possible source I could think of was a continental ice sheet, but most published glacial models showed continental ice sheet melt water flowing in or near the present day Missouri-Mississippi River valleys and nowhere near the mapped “Oligocene” sedimentary deposits I needed the melt water floods to reach. Because I was getting nowhere in terms of my efforts to convince vertebrate paleontologists that a continental ice sheet had been present when the southwest North Dakota mapped “Oligocene” sediments had been deposited I decided to put the “Oligocene” evidence on the back burner and to focus my studies on using topographic map evidence and some limited field work to determine where continental ice sheet melt water had gone.

There is good evidence in North Dakota and Montana that one or more continental ice sheets extended across the present day Missouri River valley location. The next hypothesis that I developed and tested was that there should be topographic map evidence of routes used by melt water when the ice sheet margin was located south and west of the present day Missouri River valley location. The published literature provided a few suggestions, but those suggested melt water flow routes did not appear adequate to explain where the continental ice sheet melt water had gone. In an effort to find the melt water flow routes I started studying Montana and North Dakota drainage divides immediately south and west of the mapped continental ice sheet margin and noticed that many of those divides are crossed by multiple through valleys (or wind gaps) that could be aligned with through valleys (or wind gaps) on adjacent drainage divides and were remnants of a drainage system that preceded the present day drainage system.

Based on literature descriptions of anastomosing channel complexes and on personal inspection of an anastomosing channel complex near Minot, ND (identified by Kehew (1982, GSA Bull, v. 93, p. 1051-1058) I decided the through valleys (wind gaps) crossing the drainage divides had been eroded by a large complex of flood eroded anastomosing channels that had been systematically dismembered by headward erosion of the deeper present day trunk stream valleys. Using this hypothesis I then started to trace the anastomosing channel complexes headward to try to determine a flood source. The drainage divide crossings (through valleys and wind gaps) are best seen on detailed topographic maps and my efforts to trace the anastomosing channel complexes headward required the use of large topographic map mosaics, which had to be laid out on classroom floors and then did not cover large enough areas. I experimented with large mosaics of photographically reduced detailed topographic maps, which aided in visualizing the flood flow routes.

Flood landforms, which I was able to identify on the large mosaics of photographically reduced topographic maps, included systems of anastomosing channel complexes comparable in scale to those found in the Channeled Scablands of eastern Washington, streamlined erosion residuals, sequences of high-level drainage divide crossings, extensive flood-scoured surfaces, giant spillway trenches where floodwaters breached former drainage divides, large flood-eroded escarpments, and labyrinthine complexes where huge sheets of water poured off upland surfaces into spillway trenches. I identified the immense sheets of ice-marginal water to have been at least 100 kilometers wide and suggested the catastrophic flood landforms extended into western South Dakota, perhaps as far as the Pine Ridge Escarpment. But, in 1988 I did not know where the floods began, where the floods went, or how far to the southwest the flood evidence could be found.

The ice-marginal melt water floods had flowed in a southeast direction and in eastern Montana had crossed the northeast oriented route the distinctive Beartooth Mountain alluvium had traveled before turning in an east and southeast direction to reach the mapped southwest North Dakota “Oligocene” and “QTU” locations. While at first these conflicting flood flow directions appeared to be confusing, especially because the distinctive alluvium is present throughout the Redwater River valley, I decided the evidence could be explained if in addition to being blocked and diverted along an ice sheet margin floodwaters were also spilling onto a lowering ice sheet surface. This model explained the thick alluvium deposits along the high eastern Montana Redwater River-Yellowstone River drainage divide, the diversion of floodwaters along the ice sheet margin, and headward erosion of the Little Missouri, Yellowstone, and Redwater River valleys in sequence as southeast oriented flood flow merged with northeast oriented flood flow from the Beartooth Mountain area. At the time I had not thought much about where water spilling onto the ice sheet surface might be going, but a progressively lowering ice sheet surface not only appeared to explain the evidence, but also the headward erosion of the deep Yellowstone River valley toward the Beartooth Mountain region.

1989: Melt water flood origin for the Great Plains drainage network hypothesis

Use of topographic map evidence convinced me that large floods had crossed the Redwater-Yellowstone, Yellowstone-Little Missouri, Powder-Little Missouri, and Little Missouri-Missouri River drainage divides and had used at least some flood flow routes identified during studies of mapped “Oligocene” and QTU” distinctive alluvium. The logical time when this giant floodway would have been active was at the continental ice sheet maximum advance when all continental ice sheet melt water and drainage from what I in 1989 still considered to be pre-glacial north oriented Rocky Mountain drainage systems would have had to flow in a southeast direction between the continental ice sheet southwest margin and the higher Rocky Mountains to the southwest. I next decided to look for topographic map and additional literature and field evidence to determine if the floodway could be traced headward into southern Alberta and southward into Texas.

Previous researchers had described ice-dammed lakes in north central Montana and southern Alberta that might have been flood sources, although I immediately ran into a major problem. Topographic map evidence suggested the southeast oriented floodwaters had crossed the Powder River Basin and elevations of the upstream ice-dammed lakes, as described in the literature, were not high enough to account for the Powder River Basin evidence. One of my students then investigated the possibility of a much higher ice dam between the Montana Bear Paw and Highwood Mountains than previous geologists had suggested. Her study found inconclusive evidence that could be explained in other ways. I subsequently abandoned the glacial lake outburst flood hypothesis because I found evidence floodwaters had crossed Rocky Mountain regions that were much higher than any of the previously described ice-dammed lakes or than any logical ice-dammed lake I could reasonably construct based on present day topography.

I next studied literature reporting immense floods of what was probably glacial melt water that entered the Gulf of Mexico and are today recorded in deep sea cores and proceeded to follow the floodway in a south direction and identified a floodway that would have been the only open drainage route for much of western and central North America at times of ice sheet maxima (when the ice sheet extended beyond the present day Missouri River valley location). At such times floodwaters would have been forced west of the Ozarks and reached the Gulf of Mexico by crossing Texas. What intrigued me about that floodway was the fact that several river valleys now cross the floodway, which based on their orientations and other characteristics (e.g. asymmetric drainage divides and stream capture evidence) had been eroded in a progressive sequence from south to north. This evidence plus evidence from observing drainage networks in a small clay pit caused me to propose the present day river valleys had been eroded headward as deep trenches across large south oriented floods with headward erosion of each deep valley beheading south oriented flood flow to the newly eroded valley just to the south.

Evidence for this progressive sequence of headward erosion suggested the first trenches to erode headward were in the south and may have included segments of the Rio Grande and Pecos valleys. Subsequent trenches to erode headward across the identified floodway in order of erosion were the Colorado (Texas), Brazos, Canadian, Cimarron, Arkansas, Smoky Hill, Republican, Platte (Nebraska), Niobrara, White (South Dakota), Cheyenne, Belle Fourche, Moreau, Grand, Cannonball, Heart, Little Missouri, Yellowstone-Powder, Yellowstone-Tongue, Yellowstone (upstream from Tongue), Musselshell, Missouri (upper), and Milk. The sequence definitely led to southern Alberta and I spent time looking at southern Alberta evidence to see if large ice-dammed lakes could provide a source of water. I could see no way to create southern Alberta glacial lakes large enough and high enough to erode the river valleys that crossed the floodway. Further, some river valleys crossing the floodway extended headward into the high Rocky Mountains and needed to be investigated further.   

1990: The asymmetric drainage divide hypothesis

The principle of cross cutting relationships was applied using asymmetric drainage divides to determine the relationships between erosion surfaces associated with each Northern Great Plains Missouri River tributary. The principle of cross cutting relationships was used as follows: if erosion surface A is cut by erosion surface B, then erosion surface A must predate erosion surface B. Likewise, if river valley C cuts erosion surface B, then erosion surface B must predate river valley C. United States Geological Survey 1:250,000 scale maps were used and asymmetric drainage divides were found separating most Missouri River tributary drainage basins from each other. The asymmetric divide evidence confirmed that Missouri River tributary drainage basins had been eroded in a remarkable sequence beginning with the Niobrara River drainage basin (which was as far south as the study looked) and progressing upstream to north central Montana.

Asymmetric drainage divides confirming this remarkable sequence were found between the Niobrara and White Rivers (Pine Ridge Escarpment), the White and Cheyenne Rivers, the Cheyenne and Belle Fourche Rivers, the Moreau and Grand Rivers, the Heart and Knife Rivers, the Moreau, Grand, Cannonball, Heart, and Knife Rivers and the Little Missouri River, the Little Missouri and Yellowstone Rivers, the Little Missouri and Powder Rivers, the Yellowstone and Redwater Rivers, the Redwater River and Prairie Dog Creek, Prairie Dog and Big Dry Creeks, Big Dry Creek and the Musselshell River, the Musselshell River and Armells Creek, Armells and Dog Creek, Dog Creek and the Judith River, and the Judith River and Arrow Creek.

Based on this study I concluded Northern Great Plains drainage basins had formed in the following sequence (from oldest to youngest, although the sequence does not mean one drainage basin was completed before the next drainage basin began): Niobrara River, White River, Cheyenne River, Belle Fourche River, Moreau River, Grand River, Cannonball River, Heart River, Knife River, Little Missouri River, Powder River-Lower Yellowstone River, Redwater River, Prairie Dog Creek, Big Dry Creek, Musselshell River, Armells Creek, Dog Creek, Judith River, and Arrow Creek and that any explanation of Northern Great Plains drainage history must account for this identifiable sequence.

1991: The Scenic and Sage Creek Basin escarpment-surrounded basin hypothesis

When doing the asymmetric divide study I noted some anomalous drainage divides especially in the Scenic and Sage Creek Basin area of the South Dakota Big Badlands (east of the Black Hills). The Scenic and Sage Creek Basins are located between the northeast oriented Cheyenne River valley and the northeast and east oriented White River valley to the southeast and east (the White River valley turns from a northeast orientation to an east orientation a short distance east of the Scenic and Sage Creek Basins). The Scenic and Sage Creek Basins are escarpment-surrounded (or amphitheater-shaped) basins, which open to the east and the White River valley. Escarpment slopes surrounding the Scenic and Sage Creek Basins are steeper than slopes leading from the escarpment crests into the Cheyenne River valley. Also the Scenic and Sage Creek Basins are both drained by northwest oriented streams through narrow valleys (water gaps) cut across the escarpment ridge to the Cheyenne River rather than by east oriented streams flowing to the amphitheater basin openings and the adjacent White River valley.

The Scenic and Sage Creek Basin evidence presented two major problems for my study. First the asymmetric divide evidence was reversed from what it was elsewhere and throughout the upper Missouri River drainage basin. Second, and perhaps more serious for my hypothesis, was I could not at first figure out a way to move southeast oriented floodwaters so as to erode the northwest oriented water gaps now draining the Scenic and Sage Creek Basins. I probably spent more than three months looking at the Scenic and Sage Creek Basin evidence before I finally figured how those escarpment-surrounded basins and the northwest oriented water gaps draining them were eroded.

To understand Scenic and Sage Creek Basin puzzle solution it is necessary to understand the present regional topography and drainage routes. The Scenic and Sage Creek Basins are located east of the Black Hills. The Cheyenne River originates west of the Black Hills and as it approaches the Black Hills turns in a southeast direction to flow to the Black Hills south end where it turns to flow in a northeast direction (just west of Scenic and Sage Creek Basins) to eventually join the south oriented Missouri River as a barbed tributary. The Belle Fourche River also originates west of the Black Hills (north of southeast oriented Cheyenne River tributaries) and flows in a northeast direction to the Black Hills northern tip where the Belle Fourche River turns to flow in a southeast direction and eventually joins the northeast oriented Cheyenne River. In addition to the Belle Fourche River the northeast oriented Cheyenne River segment east of the Black Hills is joined by several southeast oriented tributaries, which originate in the Black Hills. The White River originates on the high plains and flows in a northeast direction across the Pine Ridge Escarpment to the lower plains between the Pine Ridge Escarpment and the Black Hills and then continues in a northeast direction before turning just east of the Scenic and Sage Creek Basins to flow in an east direction to join the Missouri River.  

The solution to the Scenic and Sage Creek drainage history puzzle requires a large southeast oriented flood to have flowed across the present day Powder River Basin and to have split into two floods that flowed around the Black Hills. One flood flowed in a southeast direction along the Black Hills’ northeast margin (probably floodwaters extended all the way from the Black Hills to the ice sheet margin). The other flood flowed around the Black Hills south end to the deep White River valley, which was eroding headward from what at that time was an actively eroding Missouri River valley (in south central South Dakota). The Pine Ridge Escarpment and the Wyoming Hat Creek Breaks Escarpment were carved as the deep White River valley eroded headward from the ice sheet margin in south central South Dakota. Headward erosion of the northeast oriented Cheyenne River valley then captured flood flow near the Black Hills south end that had been moving to the newly eroded White River valley. However, southeast oriented flood flow routes further to the southwest were not captured and eroded the present day White River valley across the Pine Ridge Escarpment.

With this background and recognition that the White River valley was eroded headward slightly in advance of headward erosion of the northeast oriented Cheyenne River valley (east of the Black Hills) it is possible to unravel the Scenic and Sage Creek Basin puzzle. The White River valley bend from a northeast oriented valley to an east oriented valley (just east of the southeast-facing escarpment-surrounded Scenic and Sage Creek Basins) is located where floodwaters flowing around the Black Hills south end and then northeast to the actively eroding White River valley converged with floodwaters flowing around the Black Hills north end and then southeast to the actively eroding White River valley.

The Scenic and Sage Creek Basin escarpment-surrounded basins were eroded as large headcuts by the southeast oriented flood flow, but were abandoned after headward erosion of the northeast oriented Cheyenne River valley captured the southeast oriented flood flow. Floodwaters flowing around the Black Hills south end and on the flood flow routes further to the southwest continued to erode the present day northeast oriented White River valley headward, although only for a brief time before Cheyenne River valley headward erosion captured all or most of the flood flow. However, during that brief time floodwaters filled the newly eroded White River valley and floodwaters spilled over the newly formed drainage divide into the newly eroded Cheyenne River valley. The spillage occurred where prior to Cheyenne River valley headward erosion southeast oriented flood flow had been moving to the actively eroding Scenic and Sage Creek Basin headcuts and eroded the northwest oriented valleys that now drain the Scenic and Sage Creek Basin floors. Apparently flood flow in the White River valley ended as the northwest oriented valleys drained the Scenic and Sage Basin areas. This interpretation not only solved the Scenic and Sage Creek Basin puzzle, but it was also consistent with all of the flow interpretations I had made up to that time and was further evidence immense floods had flowed in a southeast direction across the Powder River Basin.     

1993: Using escarpment-surrounded basin to reconstruct flood flow routes hypothesis

Recognizing the Scenic and Sage Creek Basins as flood-eroded headcuts provided an explanation for dozens and possibly hundreds of similar escarpment-surrounded basins found throughout the Great Plains and Rocky Mountain region. Geologists have not developed good explanations for the origin of these escarpment-surrounded basins, although a commonly used hypothesis is ground water seepage or spring sapping is causing the escarpments to gradually retreat. Believing the flood-eroded headcut hypothesis to be a superior explanation I decided to test the hypothesis that similar escarpment-surrounded basins could be used to quickly identify logical flood flow routes that were consistent with other topographic map evidence and that might have deposited “Oligocene” sediments identified during my previous studies.

I then spent considerable time scanning topographic maps for the entire Rocky Mountain and Great Plains region looking for similar flood-eroded headcuts. Some of the larger ones found were along the Russian Springs Escarpment in southwest North Dakota, at the Jumpoff near the Short Pine Hills in Harding County, South Dakota, Hoskins Basin east of Billings, Montana, the Goshen Hole Escarpment straddling the Wyoming-Nebraska border, Bates Hole escarpment in Wyoming, the Beaver Rim Escarpment in Wyoming, and the Red Hills Escarpment in Kansas in addition to many others. I then used those identified escarpment-surrounded basins to reconstruct the flood flow movements that would have eroded them. Immediately some unanticipated results showed up.

For example, escarpments surrounding Goshen Hole converge to form a “V” at the west end, which provides excellent evidence to support the flood origin hypothesis. However, the Goshen Hole escarpment-surrounded basin is located directly east of the Wyoming Laramie Range, which meant floodwaters that eroded the Goshen Hole escarpment-surrounded basin had to have crossed the Laramie Range. Further, directly west of Goshen Hole, an anastomosing complex of deep canyons crosses the Laramie Range. Geologists have never provided a good explanation for those canyons and interpreting the Goshen Hole escarpment-surrounded basin as a flood-eroded headcut also meant the canyons crossing the Laramie Range could be interpreted as a flood-formed anastomosing channel complex. The new questions that arose were how was it possible for melt water floods to cross a high Rocky Mountain range and did the Laramie Range rise as floodwaters crossed it or did floodwaters remove sedimentary rocks that had buried the Laramie Range. Evidence from other escarpment-surrounded basins raised similar questions for other Rocky Mountain ranges.

Another interesting observation was that some Rocky Mountain area escarpment-surrounded basins only made sense if south and southeast oriented floodwaters had reversed direction to flow in north and northwest directions. Erosion of the northwest-facing Bates Hole escarpment-surrounded basin, for example, could only have occurred if the southeast and east oriented floodwaters flowing across the northern Laramie Basin and then the Laramie Range to erode the east-facing Goshen Hole escarpment-surrounded basin had reversed direction. Similarly south-facing escarpment-surrounded basins in the Big Horn Basin suggested floodwaters had flowed in a south direction across that basin, while north-facing escarpment-surrounded basins in the Wind River Basin suggested the south-oriented flood flow had been reversed to flow north through the Big Horn Basin. I then used stream capture sequences to test all flood flow routes determined by the escarpment-surrounded basin orientations and the stream capture evidence supported the flood flow routes and reversals the escarpment-surrounded basin evidence suggested.

I next studied escarpment-surrounded basins found throughout the western United States and determined melt water floods had flowed south from central Montana, through the Big Horn Basin, to the Great Divide Basin (Wyoming) where major melt water flow routes diverged with one route continuing south to the Colorado Plateau and eventually to the Pacific Ocean and another route proceeding across the Laramie Basin and Laramie Range and eventually to the Gulf of Mexico. I also found evidence melt water floods had flowed west into Utah and Nevada and overflowed to erode an anastomosing canyon complex across the present day Sierras. By use of the escarpment-surrounded basin evidence I was able to plot melt water flood flow routes that explained the erosion of almost all western United States valleys and canyons.

1994: North American glaciation and deglaciation hypothesis

Melt water flood flow routes identified by the escarpment-surrounded basin hypothesis served as a Rosetta stone that enabled me to learn how to read topographic map evidence. Each of the major melt water flood flow routes had been dismembered in a sequential manner proceeding headward toward the melt water flood source. After studying western United States drainage route evolution I proceeded to check eastern United States drainage routes, which in many respects provided a mirror image of the western United States evidence. Using data from the entire United States I identified drainage divides between what I defined as outward-flowing drainage systems, which flowed directly to an ocean, and inward-flowing drainage that flowed to the Missouri and Ohio Rivers or to the Great Lakes and Saint Lawrence River.

I interpreted drainage divides separating the outward flowing drainage from the inward-flowing drainage as evidence of a large “hole” that was once occupied by the continental ice sheet that produced the immense melt water floods. Outward flowing drainage basins had eroded headward in identifiable sequences along melt water flood flow routes radiating in all directions from the melting ice sheet. Inward-flowing drainage systems were eroded in sequence as the ice sheet melted and opened up space in the large “hole”. This model was consistent with drainage route orientations, stream capture sequences, and orientations of major erosional escarpments in addition to being consistent with flood flow routes determined from the escarpment-surrounded basin evidence.

The model also suggested a multiple-step deglaciation hypothesis beginning with an Antarctic-sized ice sheet occupying the large “hole” location. The model does not say why the ice sheet formed, but a combination of erosive activity under the ice sheet and the ice sheet’s weight created the “hole”. The ice sheet’s weight probably also caused tectonic uplift to occur in non-glaciated continental regions and probably played a significant role in raising what are today high mountain ranges and plateau areas. Melt water from this thick ice sheet then flowed in all directions, but with much of the melt water ending up in the Gulf of Mexico, where it displaced warm water that then moved into the Atlantic Ocean causing Northern Hemisphere climates to ameliorate.

Ameliorating Northern Hemisphere climates then increased ice sheet melting rates, which resulted in greater melt water volumes reaching the Gulf of Mexico to further ameliorate North Hemisphere climates. In other words once this Antarctic-sized ice sheet started melting the melt down process kept accelerating until enough of the ice sheet had melted that space in the large “hole” started to open up. At first the space being opened up was in the south and was drained by the actively eroding Mississippi River valley and tributary valley system. However, in time ice sheet melting progressed to the point that melt water floods could flow in north directions to reach the North Atlantic Ocean.

As the north oriented drainage routes across the large “hole” to the North Atlantic Ocean captured melt water floods that had been flowing to the Gulf of Mexico Northern Hemisphere climates began to deteriorate, which not only reduced the rate of ice sheet melting, but which also caused melt water floods on the decaying ice sheet’s floor to freeze (from the top down) and to form a thin ice sheet with remnants of the former thick ice sheet embedded in it. This thin ice sheet blocked the newly formed north oriented drainage routes causing the Ohio and Missouri River drainage systems to develop. Subsequently the thin ice sheet melted, although without producing the immense melt water floods the thick ice sheet had. It is possible the south-north melt water flood flow process was repeated several times before the second ice sheet completely melted.

While many details still needed to be worked out, including integration of my “Oligocene” evidence, this multiple step deglaciation process explained most if not all large-scale United States drainage networks. Still trying to express the new model in manner consistent with published glacial history models and oceanographic evidence age date evidence in 1994 I suggested the Antarctic-sized ice sheet, which the geology research community has yet to recognize, was formed and melted during “Pliocene” or “early Pleistocene” time, while the second “thin” ice sheet, which geologists do recognize, formed and melted during “Pleistocene” time, although my “Oligocene” evidence suggested the thick ice sheet had existed at a much earlier time.

1994: Melt water origin for most recent ice sheets hypothesis

Also in 1994 I began the process of integrating the newly developed glacial history model, which was based on topographic map evidence, with my previously obtained “Oligocene” and other evidence. While a “Pliocene” or “early Pleistocene” time  for the thick ice sheet existence would have been consistent with published glacial history age dates, I decided a much better interpretation would have said the thick ice sheet existed during the “middle and late Tertiary” and the “thin” ice sheet existed during the “Pleistocene”, even though such a definition probably requires much shorter “Oligocene”, Miocene”, and “Pliocene” time periods than geochronologists presently assign. I did not and still do not know and have no way of determining how long it takes for an Antarctic-sized North American ice sheet to form, exist, and melt, especially when the melting eventually develops into a rapid melt down. However I have great trouble believing such a melt down process took more than 30 million years as presently assigned “Tertiary” absolute age dates imply.

Regardless of the age date situation the newly developed melt water flood flow routes identified from escarpment-surrounded basins and the multiple step deglaciation model did explain evidence I had put on the back burner several years previously. The identified melt water flood flow routes showed large melt water floods emerging from the thick ice sheet margin at a relatively small number of locations, while at other locations floodwaters appeared to be flowing back onto the ice sheet surface. I developed the hypothesis that floodwaters flowing back onto the ice sheet surface were flowing to large supra-glacial melt water rivers that carved deep ice-walled (and subsequently) bedrock-floored canyons into the ice sheet surface. Evidence for the giant ice-walled and bedrock-floored canyons that ultimately formed are today Northern Plains area escarpments such as the Missouri Escarpment and escarpments surrounding various uplands such as the Turtle Mountains, Moose Mountain, Last Mountain, the Prairie Coteau, Riding Mountain, and Duck Mountain. At last I had found a viable Missouri Escarpment origin, which was my goal when the study began 15 years earlier.

Evidence for melt water floods flowing back onto the ice sheet surface to a deep ice-walled canyon also explained north and northeast oriented valleys in the present day upper Missouri River valley drainage basin. The north and northeast oriented valleys eroded headward from a deep ice-walled canyon to the ice sheet’s southwest margin and then eroded headward across massive southeast oriented ice-marginal melt water floods. This model also explained how distinctive Beartooth Mountain alluvium was transported both along the ice sheet margin and across the ice sheet margin location onto the ice sheet floor and is presently found north and east of the former ice sheet margin.

In addition the model explained the difference between ice sheet deposited sediments on the Missouri Coteau and capping escarpment-surrounded uplands such as the Turtle Mountains, Moose Mountain, Last Mountain, the Prairie Coteau, Riding Mountain, and Duck Mountain and ice sheet deposited sediments found on lowlands between the escarpment-bounded uplands. Glacial sediments on the Missouri Coteau and capping the escarpment-bounded uplands are generally thick and characteristic of dead ice moraines, where an ice sheet containing large quantities of debris gradually melted. Glacial sediments found on lowlands between the escarpment-bounded uplands are generally thin and contain blocks of ice thrust material suggesting the presence of an actively moving ice sheet riding on a wet base.

The difference between the two types of glacial sediments made sense if the lowlands between the escarpment-bounded uplands had been floors of giant melt water carved ice-walled and bedrock-floored ice-walled canyons, which had been carved into the surface of the decaying ice sheet. The dead ice moraines covering the upland regions were deposited as decaying ice sheet remnants between those ice-walled and bedrock-floored canyons gradually melted.

The ice-walled and bedrock-floored canyons were the routes where the melt water floods were flowing. At first these canyons drained in south directions, but as the thick ice decayed new ice-walled and bedrock-floored canyons opened up and provided shorter northeast and north oriented routes to the North Atlantic Ocean. Diversion of the large melt water floods from the Gulf of Mexico to the North Atlantic changed climates and the deteriorating climate caused the melt water floods to freeze (from the top down) and to create a thin wet based ice based ice sheet with remnants of the decaying thick ice sheet embedded in it.   

1996: Further development of the ice-walled and bedrock-floored canyons hypothesis

Recognizing that present day Northern Great Plains escarpments were remnants of ice-walled and bedrock-floored canyons supra-glacial melt water rivers had carved into the decaying ice sheet surface enabled me to better understand how the thick ice sheet melted. The present day Missouri Escarpment is a remnant of the southwest and west wall of a southeast and south oriented ice-walled and bedrock-floored canyon that gradually detached the decaying ice sheet’s southwest margin and that played a major role in developing what are today north and northeast oriented Montana, northern Wyoming, and western North and South Dakota drainage routes as well as in eroding the Missouri River valley downstream from Yankton, South Dakota. I named that southeast and south oriented ice-walled and bedrock-floored canyon as the Midcontinent Trench and the melt water river that carved it as the Midcontinent River. The ice sheet’s detached southwest margin, which was located where the Missouri Coteau is located today, was named the Southwest Ice Sheet. Other Northern Great Plains escarpments defined a series of other ice-walled and bedrock-floored canyon locations.

Today the Midcontinent Trench floor is drained by a series of discontinuous rivers, At the south end the James River flows in a south direction from central North Dakota to join the Missouri River near Yankton, South Dakota. The Sheyenne River originates near the James River origin and then flows in an east and south direction on the Midcontinent Trench floor before making a U-turn and flowing to join the north oriented Red River, which flows on the floor of a different trench (Agassiz Trench), that also originated as an ice-walled and bedrock-floored canyon. North and west of the James and Sheyenne River headwaters area the Midcontinent Trench is drained by the Souris River, which originates near Weyburn, Saskatchewan and flows in a southeast direction on the Midcontinent Trench floor into north central North Dakota where it also makes a U-turn and flows in a north direction on the floor of a north-south oriented trench between the Moose and Turtle Mountains (Assiniboine Trench) to join the south and east oriented Assiniboine River. North of the Souris River headwaters a short segment of the Midcontinent Trench floor is drained by northwest and southeast oriented Qu’Appelle River headwaters. Further upstream (on the former Midcontinent River route) a north and northwest oriented South Saskatchewan River segment drains a Midcontinent Trench floor segment and still further to the northwest the Midcontinent Trench floor is drained by a southeast oriented North Saskatchewan River segment.

Present day routes of rivers draining the Midcontinent Trench floor tell not only the story of how the giant Midcontinent River was dismembered, but also the story of how the ice sheet decay systematically diverted south-flowing melt water floods in a northeast and north direction to the North Atlantic Ocean. Melt water floods draining much of the decaying ice sheet’s southwest margin area (both on the ice sheet surface and adjacent to the ice sheet) first flowed in the Midcontinent Trench in a southeast and south direction to eventually reach the Gulf of Mexico. At the same time melt water floods flowed in a south direction in the Agassiz Trench, which carved the Prairie Coteau northeast escarpment as its southwest wall to reach the actively eroding Mississippi River valley and the Gulf of Mexico. When ice sheet decay opened a route from the Agassiz Trench to the Great Lakes area flow in the Agassiz Trench along the present day North Dakota-Minnesota border was reversed. Reversal of flow in the Agassiz Trench captured the Midcontinent River east half, which accounts for the present day Sheyenne River U-turn (and the adjacent Maple River U-turn).

As new routes to the Great Lakes area and eventually the Saint Lawrence River route and the North Atlantic Ocean additional Agassiz Trench segments were reversed. Eventually ice sheet decay opened up routes to the present day Hudson Bay location and all Agassiz Trench segments north of the North Dakota south border were reversed to flow in a north direction across the Hudson Bay area to reach the North Atlantic Ocean. At some point during this Agassiz Trench reversal process south flowing Assiniboine Trench melt water floods were captured by the reversed Agassiz Trench and diverted in an east direction between Riding Mountain and the Turtle Mountains to reach the newly reversed Agassiz Trench route. Capture of flow from the south flowing Assiniboine Trench in southwest Manitoba caused a reversal of flow on the Assiniboine Trench south end (south of the elbow of capture and north of where the Assiniboine Trench joined the Midcontinent Trench in north central North Dakota). Reversal of flow in the Assiniboine Trench south end captured all Midcontinent Trench melt water flood flow creating the present day Souris River U-turn. Headward erosion of the east-oriented Qu’Appelle valley, probably as an ice-walled and bedrock-floored canyon, next captured the Midcontinent River melt water flood flow in southeast Saskatchewan. Next a northeast oriented (and possibly newly reversed) trench on the present day Saskatchewan River alignment captured the Midcontinent River melt water flood flow with North Saskatchewan River valley headward erosion next capturing the Midcontinent River melt water flood flow.     

Up until the time of the flood flow reversals large melt water floods from Midcontinent River emerged from the Midcontinent River mouth along the decaying ice sheet margin in south central South Dakota and first flowed in a south direction across eastern Nebraska before headward erosion of the deeper Missouri River valley captured the flood flow (note how the Missouri River valley downstream from Yankton, South Dakota is much wider than upstream). Also, the Midcontinent River was joined in south central South Dakota by southeast oriented ice marginal floods. Headward erosion of the deep Missouri River valley to the Midcontinent Trench location also permitted the southeast and east oriented ice marginal floods, which were flowing around the emerging Black Hills, to erode deep valleys headward around the Black Hills south end. The Nebraska-South Dakota Pine Ridge Escarpment is what remains of the south wall of the large valley that eroded headward from deep Missouri River valley (at the Midcontinent River mouth) around the Black Hills’ south end. Southeast oriented flood flow moving into this newly eroded and deep east oriented valley removed most of that valley’s north wall and generally lowered all Northern Great Plains surfaces north of the Pine Ridge Escarpment. I named the flood-eroded Northern Great Plains surface between the Pine Ridge Escarpment and the ice sheet’s southwest margin as the Southwest Trench. Further to the northwest the Southwest Trench’s southwest wall was the Rocky Mountains.       

1997: Cypress Hills, Wood Mountain, and Flaxville alluvium origin hypothesis

As early as 1981 I had suspected coarse-grained alluvial deposits capping the southwest Saskatchewan Cypress Hills and Wood Mountain uplands and the northeast Montana Flaxville upland had been deposited along an ice sheet margin, although at that time I did not understand any of the details. However, when looked at in the context of my newly developed deglaciation model the Cypress Hills, Wood Mountain, and Flaxville coarse-grained alluvial deposits were not only explained, but also provided valuable evidence that further helped explain the thick ice sheet melt down process.

Today the Cypress Hills, Wood Mountain, and Flaxville alluvial deposits are found capping isolated buttes or mesas with the Cypress Hills upland being the highest and being located near the Saskatchewan southwest corner (and Alberta southeast corner). Today the alluvium covered Cypress Hills upland surfaces stand several hundred meters above the surrounding landscape. Previous workers have used vertebrate fossil evidence to assign late Eocene, early Oligocene, and Miocene ages to the Cypress Hills alluvium. Also, previous workers have established the alluvium was transported significant distances by what might have been a north or northeast flowing river from somewhere in the Rocky Mountains prior to the erosion cycle that removed hundreds of meters of bedrock from most of north central Montana, southeast Alberta, and southwest Saskatchewan. Previous workers had not explained why the alluvium was deposited out in the middle of the plains or why or when several hundred meters of bedrock had been stripped from the surrounding region.

My new deglaciation model had identified a large southeast oriented melt water river flowing from the decaying ice sheet’s western margin in Alberta and British Columbia along routes of present day Canadian Rocky Mountain linear valleys, including the Rocky Mountain Trench, (although at that time the valleys were still being eroded) into Montana and then across the Yellowstone Plateau area and into Wyoming. The immense melt water river, which I named the Great Divide River, was systematically dismembered from south to north as mountain ranges emerged. Mountain range emergence was occurring as ice sheet weight was causing uplift of regions elsewhere on the continent and as immense melt water floods were deeply eroding surrounding bedrock materials.

Great Divide River dismemberment as the Yellowstone Plateau emerged probably occurred when ice-marginal melt water floods spilled across the decaying ice sheet surface in the Cypress Hills region to reach what at that time was the supra-glacial Midcontinent River. The spillage carved an ice-walled and ice-floored canyon headward from the developing Midcontinent Trench to the ice sheet margin and then headward across southeast oriented ice-marginal melt water floods to capture all of the southeast oriented Great Divide River flood flow. The resulting flood capture events began to develop what is today the north oriented southwest and central Montana Missouri drainage system, although in north central Montana the captured flood flow moved toward the ice sheet margin breach on a topographic surface equivalent to or higher than the Cypress Hills elevation. The captured melt water flood flow crossed the ice sheet margin to reach what at that time was the supra-glacial southeast oriented Midcontinent River and then flowed in a southeast and south direction to eventually emerge from the ice sheet margin in present day south central South Dakota.

The ice sheet margin breach through which this north oriented Midcontinent River tributary flowed apparently became blocked from time to time. Blockages may have been due to seasonal freezing, collapse of huge ice blocks from canyon walls into the ice-walled and ice-floored canyon, or even unusually large melt water floods overwhelming the developing Midcontinent Trench tributary ice-walled and ice-floored canyon. In any case the blockages caused temporary ponding the captured Great Divide River flow, which had been diverted north to the ice sheet margin, and resulted in the deposition of whatever coarse-grained alluvial material was being transported. Blockages also forced the captured Great Divide River melt water floods to flow in a southeast direction along the ice sheet margin and to erode the landscape surrounding the newly deposited Cypress Hills alluvium, which served as a protective cap rock.

Either through downstream migration of the Cypress Hills ice sheet margin breaches (along the Midcontinent River route) or through the opening up of new ice sheet margin breaches the north-flowing early Missouri River again spilled across the ice sheet margin in the Wood Mountain location at a somewhat lower elevation to reach what was by that time a deeper Midcontinent River ice-walled and ice-floored canyon. Again large volumes of coarse-grained alluvium were deposited, probably as the result of temporary blockages along the ice sheet margin. The process was repeated as the ice breaches shifted in a southeast direction (downstream along the Midcontinent River route) to the Flaxville region of northeast Montana and more deposits of coarse-grained alluvium were deposited. Shifting of the ice sheet breach location downstream along the Midcontinent River continued, but by this time the Midcontinent Trench had become an ice-walled and bedrock-floored canyon and was detaching the decaying ice sheet’s southwest margin. The new ice sheet breach was located in the Medicine Lake region of northeast Montana and crossed the Midcontinent Trench southwest valley wall near Crosby, North Dakota and is today a large dry valley partially filled with glacial deposits (and has been described by many previous workers as the pre-glacial Missouri River valley). 

1998: Glacial lakes explained as slack water features hypothesis

Previous researchers have identified a number of glacial lake locations where “pre-glacial” north-flowing rivers were dammed by continental ice sheet margins. There is little or no evidence at some of the identified locations that an ice sheet dammed lake ever existed and the lack of evidence for such glacial lakes supports the ice sheet model presented here. However, at some of the locations excellent evidence in the form of wave cut shoreline features, deltaic deposits, lacustrine sediments, outlet spillways, and other evidence strongly supports the ice sheet dammed lake hypothesis and an alternate explanation based on the newly developed ice sheet melt down model and able to explain the existing glacial lake evidence was needed. Addressing this problem I observed that the existing glacial lake evidence is located near points where identified primary melt water flood flow routes had converged, both south and west of the ice sheet margin and on the decaying ice sheet floor, and suggested large slack water lakes could have formed.

For example, Lake Agassiz, which is one of the best-documented glacial lakes, was formed when slack water associated with a progression of Midcontinent River dismemberment stages inundated the Agassiz Trench and adjacent lowlands. Prior to dismemberment the Midcontinent River had flowed to the Gulf of Mexico and major tributaries had included the south-flowing Agassiz River (on the present day north flowing Red River route) and a south-flowing proto-Assiniboine River (flowing on the present day north oriented Souris River route west of the Turtle Mountains). Midcontinent River dismemberment occurred because the Agassiz River reversed flow direction to flow in a north direction, which captured much of the Midcontinent River water and diverted it north through what is now the Red River valley. The second Midcontinent River dismemberment stage when ice sheet melting opened up a new west-to-east oriented flow route between the present day Turtle Mountains and Riding Mountain (the present day Assiniboine River route). This route opened up a shorter Midcontinent River flood flow route, with the Midcontinent River making a U-turn in north central North Dakota. Many of the well documented Lake Agassiz features developed during this second dismemberment stage as slack water inundated the Agassiz Trench to the south of the present day Assiniboine-Red River confluence. The process was repeated as reversed flow in the Agassiz Trench systematically dismembered the Midcontinent River.

Not only did the slack water hypothesis explain glacial lake evidence in the Northern Great Plains region it also explained evidence for what previous workers were claiming to be ice-dammed glacial lakes in other locations. Perhaps the most interesting example was how the slack water hypothesis explained Glacial Lake Missoula. Glacial Lake Missoula is usually explained by the presence of a series large ice dams (located in the Idaho Lake Pend Oreille area), which blocked the northwest oriented Clark Fork River. My ice sheet melt down hypothesis does not recognize an ice dam in that location, but does note how the Purcell Trench links the Kootenai River valley with the Clark Fork Valley. The Purcell Trench is hypothesized to have been eroded by immense south oriented floods flowing from the western margin of the decaying thick ice sheet and then flowing along the present day Rocky Mountain Trench alignment to reach the Kootenai River valley. Some of the floodwaters then crossed from the Kootenai River alignment to reach the Flathead River alignment and what was at that time the northwest oriented Clark Fork valley while other floodwaters continued on the Kootenai River alignment to the Purcell Trench and then flowed south along the Purcell Trench to join the Clark Fork flood flow in the Lake Pend Oreille area. The immense south oriented floods flowing through the Purcell Trench blocked northwest oriented floodwaters in the Clark Fork valley and those blocked floodwaters filled the Clark Fork valley to form Glacial Lake Missoula. The slack water lake hypothesis explains in a simple manner how Glacial Lake Missoula could be filled and empty numerous times. The ice dam hypothesis requires repeated ice dams and is unnecessarily complicated.

1999: Rocky Mountain Trench origin hypothesis

When studying how melt water floods from the decaying ice sheet’s western margin flowed across what are today the Canadian Rocky Mountains to western Montana and to the Yellowstone Plateau area to form what I called the Great Divide River (which roughly followed the present day east-west continental divide alignment) I noted similarities between drainage patterns on the Rocky Mountain Trench floor and those on the Midcontinent Trench floor. Both the Midcontinent Trench and the Rocky Mountain Trench are erosional features, which were cut by large rivers that do not now exist. Present-day drainage patterns permit reconstruction of the dismemberment histories. Midcontinent River dismemberment has already been described. This section addresses Great Divide River dismemberment, especially in the Rocky Mountain Trench region.

Great Divide River water was probably obtained from large southwest oriented ice-walled canyons carved into the decaying ice sheet surface that flowed in southwest directions roughly along alignments of the present day northeast oriented South Saskatchewan, North Saskatchewan, Athabasca, and Peace River valleys. The southwest oriented melt water floods flowed from the decaying ice sheet into what at that time were the emerging Canadian Rocky Mountains and then were diverted in south and southeast directions along the alignments of what are now deep valleys and trenches between the various Canadian Rocky Mountain ranges. These south and southeast oriented floodwaters formed what I have named the Great Divide River, which initially flowed roughly along the present day east-west continental divide alignment until being dismembered. Essays in the Missouri River drainage basin landforms origins research project essay series describe how Great Divide River dismemberment systematically proceeded from the southern Colorado Rocky Mountains into southwest Montana.

Great Divide River dismemberment north of southwest Montana began with a major reversal of flood flow on the present day Clark Fork River alignment. The flood flow reversal caused floodwaters, which had been moving in southeast directions to reach what at that time were headwaters valleys for the newly formed reversed central Montana Missouri River, to make a giant U-turn and to flow in a northwest direction to eventually reach the actively eroding Columbia River valley. Next a similar flood flow reversal, probably also related to headward erosion of the deep and south oriented Columbia River valley, enabled the Kootenai River valley to capture all south oriented flood flow to the Flathead River valley. Headward erosion of the deep south-oriented Columbia River valley next captured all south oriented flood flow in the Rocky Mountain Trench that had been moving to the Kootenai River valley and reversed flow on the Rocky Mountain Trench floor north of the present day Kootenai River head waters area. Next Fraser River valley headward erosion beheaded and reversed flood flow on the Rocky Mountain Trench floor that had been moving to the newly eroded Columbia River valley. Finally a flood flow reversal on the present day Peace River alignment completed the Great Divide River dismemberment process at the Rocky Mountain Trench north end.       

2002: Missouri River drainage basin first order stream valley hypothesis

My North Dakota Academy of Science communications were submitted in the hope of pointing out neglected evidence and of stimulating discussion as I used the method of multiple working hypotheses to test various ideas. By 1999 this testing of hypotheses was enabling me to zero in on a hypothesis that explained much of the evidence I had been working with. However, instead of stimulating discussion the North Dakota Academy of Science simply rejected my submitted communication discussing the 1999 Rocky Mountain Trench and Midcontinent Trench similarities. One reviewer objected because the submitted one page communication did not provide adequate data to support my hypothesis. A second reviewer objected because “I see no new information in this manuscript. The reasons for changes in directions of flow are important, but missing from the manuscript. There is no apparent science in this manuscript.”

After thinking about the reviewer comments I designed an experiment that would look at and also assemble large quantities of data, which could also be used to test whether melt water flood flow routes such as my proposed Great Divide River and Midcontinent River had existed as my 1999 communication described. The experiment I conducted was to systematically check every Missouri River drainage basin and adjacent drainage basin first order stream valley to determine if there was evidence the stream valley had been eroded by water that came from outside the stream’s present day drainage basin. United States Geological Survey 1:100,000 scale topographic maps were used both to identify the first order streams and to check for evidence that water from one or more adjacent drainage basins had once flowed into the stream’s valley. All named streams lacking tributaries and located in an identifiable valley were considered to be first order streams. An abandoned valley of some sort linking a first order stream’s headwaters area with an adjacent drainage basin was considered evidence the first order stream valley had been eroded by water coming from outside the stream’s present day drainage basin.

Not only would this experiment serve as a test of the flood flow routes my previous research had identified, but it would also test a commonly accepted assumption that unless there is good reason to believe otherwise most present day streams eroded the valleys in which they flow. The experiment was conducted by systematically using the 1:100,000 scale topographic maps (sometimes supplemented by 1:24,000 scale topographic maps and in Canadian areas 1:50,000 topographic maps were used) to follow every Missouri River tributary drainage route headward and then back so as to identify and test every first order stream. At each first order stream encountered I looked for three types of landforms. First, I checked for an identifiable abandoned valley that linked the first order stream valley with an adjacent drainage basin, Second I checked to see if the first order stream valley head could be considered a headcut that had been eroded by water coming from outside the stream’s present day drainage basin. Third I checked to see if there was evidence of a stream capture that had beheaded flow to the identified headcut and that had diverted the flow in a different direction.

Collected data was entered in a database where each record names, locates (map name, and area of the map on which the landform was identified, and state or province name). Each record also provides a brief forty (40) to eighty (80) word description of the landform and a forty (40) to eighty (80) word interpretation of how the landform originated. In addition to investigating the entire Missouri River drainage basin the study also checked much of the upper Colorado River drainage basin, many western Montana and adjacent Alberta and British Columbia drainage basins, and drainage basins of headwaters of rivers draining the Rocky Mountain Trench. The investigation took almost two and one half years of nearly full time work to complete. When completed the database had records for 42,361 different landforms. Database evidence showed almost every first order stream valley encountered had been eroded by water from outside the stream’s present day drainage basin and also showed flow routes consistent with my previously collected data.

Concluding comment

The 2009-2013 Missouri River drainage basin landform origins research project, which focuses on Missouri River drainage basin drainage divides, was in some respects a repeat of my 1999-2001 database construction project for the purpose of making sure the first database results were repeatable and of creating a website where the previously unstudied erosional landform (topographic map) evidence and my interpretations of that evidence would be freely available to any interested persons.