Every child likes rivers. The constantly changing surface, the movement, and of course the water are irresistible. From floating sticks to building dams, they offer entertainment and learning. But rivers are also important to geology. Rivers feel the lay of the land, accurately showing the tilt. They cut away the surface and reveal the rocks below. And their sediments build the land, as well as show what is upstream. A lot can be learned from flowing water.
The Snake River is a good example. It bends its way through volcanic history, traversing a rich landscape, from the alpine plateau of Yellowstone to the arid lands of Idaho, until its end at Burbank (near Kennewick), Washington where it meets its superior, the Columbia river. Everywhere it flows it reveals a deep geological history, of plumes, rifts, and even lost island arcs. Mountains obstruct the passage in places, but elsewhere the river flows through wide plains. Nowadays, dams have added their obstructions, more of a problem for the salmon than for the water. The river begins in some of America’s youngest mountains; as it flows towards the Pacific, it cuts through progressively older rock. But even the aged rock isn’t old. This is a dry, wild, young land, build by massive eruptions and shaped by biblical floods. The river is recent; it seems out of place in this semi-desert.
Although recent, in a way the Snake river began before the landscape. It follows a patchwork of pre-existing channels, formed by older rivers, which were slowly captured by the growing Snake. The river evolved with the land, with uplift and tilt even changing the direction of the flow. This is a river as could have been devised by Escher.
For those who like statistics, the Snake river is over 1700 km long, and typically discharges 1500 m3/s, the 6th most voluminous US river. If you like history, the many different names of the river come from the diverse native communities who depended on it. But for volcanoholics, this is the river that trailed the Yellowstone hotspot.
Colombia river basalts
The geology begins at the end. Between the river mouth at Burbank, Washington, and Ontario, Oregon, where the river defines the Idaho-Oregon border, the Snake river flows over the Colombia river basalt, the youngest flood basalt on Earth. Most of it was erupted within a 1-million year period, 16 million years ago although some residual volcanism continued until 6 million years ago. These were not a few isolated lava flows: the never-ending outpourings covered an area the size of the UK to a depth of a kilometer.
The Snake river follows the fissures which created the Colombia river basalts and which erupted huge, sheet-like flows. No hidden lava tubes here! The wide lava sheets reached the Pacific Ocean, hundreds of kilometers distant, and continued on the ocean floor. Some 300 separate flows have been identified, each on average having a lava volume of 500 km3. Each separate flow appears to have been short lived: the rocks show that there was very little cooling along each sheet. To avoid cooling, each flow could not have lasted for more than one week, and the flow must have laminar, not turbulent. To get to the ocean in that time, the lava would have moved at perhaps 20 km/hr: if people had been around, they would have seen the approaching lava front, 50 meters tall and coming twice as fast as they could run! But this model is disputed. Other scientists argue that the flows could have taken years, in spurts, where the real flow occurs underneath a solidified layer with break-outs at the front. The large depth of the flow could have kept the interior hot and flowing. In the fast model, each individual flow would have been as large as 20 Laki’s, and erupted 20 times faster, at a rate of 50km3 per day. Toba was worse, but there it happened only once. In the slow model, each flow would have been only a few times worse than Laki but continuing for far longer. As real eruptions tend to show decaying eruption rates, with much of the lava erupted early on, the Colombia flows may have combined aspects of both models. Toba has shown that the fast model should not be discounted.
300 flows over a million year amounts to one every 3000 years, and gives an average eruption volume of 0.1km3/year. This is not far from what Iceland produces. However, these numbers are only an average – at times the flows followed each other much more rapidly, perhaps less than 100 years apart. And just the sulfur in each eruption could have wiped out life hundreds of kilometers away. The devastation is hard to imagine.
The Colombia river basalts are among the first outpourings of the Yellowstone hotspot. The Yellowstone hotspot is currently near Yellowstone. It was stationary while the continental plate drift west-southwest. At the typical rate of 4 cm/yr, 16 million years ago the hotspot would have been underneath Oregon, 600 km southwest of Yellowstone. This puts it near the Oregon-Nevada border. The Snake river flows 200-400 km north and east of there. At such distances, was it really the hotspot triggering the local eruptions?
The oldest eruptions are from southeastern Oregon, but flooded a huge area, including that of the current Snake river. The eruptions which formed the Colombia plateau, further north, occured between half and one million year later, and covered the northern extend of the older flows. The flows became progressively younger towards the north. It seems volcanism traveled north, away from the hotspot.
The region show swarms of dykes, roughly aligned north-south, and many close to the Idaho-Oregon border. The swarms don’t quite point in the same direction: there is a bit of a rotation. Allowing for continental motion. it is possible to identify a centre from which the dykes radiated away. This central spot turns out to be Steens Mountain, southeast Oregon. This is also the location of the oldest dated flows, at 17Myr. It seems that Steens mountain is where the Yellowstone hotspot first erupted on the scene. Shortly after, 15 million year ago, a number of large caldera eruptions occured to the south, around McDermitt on the Nevada border.
So what happened? One model is that the mantle plume first came up underneath Steens Mountain, generating multiple magma chambers. The plume and the early eruptions build up a large bulge on the surface. The weight of the bulge pushed the magma outward: it followed a ‘downward’ trajectory (in so far as that makes sense underground), a path of least resistance. The Holohraun magma did the same: at each point the dyke followed the steepest contour of the land above. The Steens magma chambers thus migrated north and caused eruptions far from the original source. The Snake river still follows the path formed by these dykes. In this region, the rivers originally flowed towards the south. But the uplift from the hotspot reversed the incline of the land, and the rivers stopped, and began to flow the other way.
While the main volcanism migrated north, eruptions also started to move towards the south, forming the McDermitt calderas, and west, with the most recent outpourings as fast west as next to the Cascades. The latter is opposite to the track of the presumed hotspot! How is that possible?
Camp and Ross, in 2004, suggested that the thick lithosphere underneath Idaho stopped the plume head in its tracks. The plume itself is a fairly narrow channel; underneath the crust it stops and widens, forming a bulbous head much wider than the channel that fed it. This wide head was sheared off from the plume channel underneath by the deep Idaho crust. Tied to the crust, the head was forced to follow the continental drift. The plume and the head were now moving in opposite directions.
Along the Idaho-Orgeon border, the Snake river passes through a high plateau, which reaches far above the level of the river. The river does not go around it, as other rivers in the area do, but enters a deep canyon which disects the plateau. By some measures this is the deepest canyon of the US. The Snake river runs more than 1.5 kilometer below the level of the plateau, and compared to the mountains east of the canyon, the depth is almost 2.5 kilometer. 80 kilometer long, 15 kilometer wide, and 500 meter deeper than the Grand Canyon, this is Hells Canyon, inaccessible, and roadless.
The high plateau was formed in the flood basalt epoch. The canyon slices through this, down to the much older landscape which the flood basalt covered. Originally, this was the west coast of North America. About 120 million years ago, a volcanic island arc accreted on to the continent, extending it to the west. The Colombia flood basalt covered an ancient, volcanic landscape, now uncovered again by the erosion from the Snake river. Several dams have been build in the canyon, harvesting hydroelectricity from this monument to eroding lava.
Western Snake valley and Lake Idaho
In western Orgeon, the Snake river follows a low plain, an old rift valley or graben, from about 17 million years ago. The river flows close to the southwestern edge of the 50-kilometer wide valley. The valley diverges from the track of the Yellowstone hotspot, turning towards the northwest. Only at the eastern end of the valley does the Snake river join the central track of the Yellowstone hotspot. The graben formed when rising heat from below pushed the crust sideways, extending it. The central area dropped down in response, forming a rift valley. In a less resilient land, perhaps it would have split the continent. But in Idaho, no chance of that. Eruptions and a layer of sediment over a kilometer thick covered the valley and surrounding lands, but the valley is still here. The last eruption was less than 1 million year ago. The valley is still subsiding, in response to the loss of the heat from below. Why no flood basalt here? Perhaps the hotspot took too long to reform its magma chambers, after losing its head. Or perhaps the much thicker lithosphere here only allowed a little magma to travel through.
Volcanic eruption started explosively, rhyolitic, 11.5 million years ago, followed by basaltic eruptions 7-9 million years ago. A second phase of basaltic volcanism happened much more recent, 2 million year ago. Later, alternating layers of sediment and basalt formed. The vents include maars, cinder cones and shield volcanoes. The shield volcanoes are 200-300 meter tall and fairly steep, suggesting viscous lava. Most flows came from the edge of the rift and flowed towards the centre.
10 million years ago, Hells canyon wasn’t deep enough to drain this valley. The Hells canyon river, not the modern Snake, instead flowed towards the south and drained into the valley. Large lakes formed, eventually becoming one huge lake, extending to close to Twin Falls. This was Lake Idaho. Perhaps 2 million years ago, it first began to overflow into the deepening Hells canyon. The result was a biblical flood. It deepened Hells canyon and flash erosion widened it. This widening was completed by the Bonneville flood, 15,000 years ago.
The Bonneville flood came from Utah. During the ice age, a huge lake had build up, dwarfing the Great Salt Lake. When the dam gave way, a flood came, north, entering the Snake river valley, and created the 200-meter deep Snake river canyon, formed the Shoshone falls, and crashed through Hells canyon. At its peak, the flood flow was a million m3 per second. Sediment from this flood covers much of the Western Snake valley, all the way to the mouth of the river.
The Eastern Snake Valley, the great American rift and the Craters of the Moon
Beyond Twin Falls the land begins to rise, and the valley now closely follows the hotspot track. It is no longer a graben but became a valley because of the weight of volcanic rock, perhaps aided by the emptying and cooling of the magma chambers below. The volcanism is overall a bit younger than that of the western valley although it started at the same time or a little earlier. Basalt is everywhere. There are also five major rhyolitic domes: Big Southern Butte, Cedar Butte, Middle Butte, Unnamed Butte, and East Butte, but these are young, from within the past million year.
The eruptions started as massive rhyolitic explosions, forming five overlapping calderas. Rhyolitic magma is silica-rich, and form from melted crust. In the eastern valley, there is a deep layer of rhyolite (in places 4 kilometer thick), with a 1-2 kilometer layer of basaltic (silica-poor, mantle melt) magma on top, showing that both types of eruptions occured, in that order. As the conduit through the crust became wider, the mantle magma was able to reach the surface without requiring to melt more crust. Underneath the entire valley is a sill of intruded basalt, 10 kilometer thick and 90 kilometer wide.
East of Twin Falls, on the northern side of the Snake river plain, are three complex lava fields, together called (slightly misnamed) the Craters of the Moon. They were formed in a series of 8 eruptions, between 15,000 to 2,00 years ago. More than 25, often nested, cinder cones dot the desolate landscape, with over 60 individual lava flows. Each eruption started with gas-rich explosions, building the cinder cones, becoming effusive as the gas ran out. The southern-most flow, closest to the Snake river, is the Wapi field, a shield volcano that formed from a single eruption 2300 years ago. The volcanic activity here is dormant, but not dead: new eruptions are likely within the next 1,000 years.
The rifts of the Craters of the Moon run SE to NW, perpendicular to the Snake river plain, and clearly have a distinct origin. The orientation is the same as that of the largest geological feature of North America, the Great American Rift. Also called the Basin and Range Province, it runs from Mexico to Canada and covers much of the mountainous west with series of flat, low-lying basins and narrow, steep mountain ranges. Famous Death Valley is one of them. The basins are due to crustal extension. The volcanism along the Snake Valley plane is in part due to this extension. The Craters of the Moon are one expression: crustal extension in one direction, crossing a crust weakened by a passing hotspot are a dangerous combination. The western Snake valley, Hells canyon, and the Colombia flood basalts fissures all follow this orientation.
The Tetons are perhaps the most impressive mountain range in the US. One of the ranges of the great American rift, steeply rising more than 2 kilometer over the surrounding plains to over 4 kilometer height, they form an immense barrier, just south of Yellowstone. The Snake river runs along and around the mountain range, flowing into and out of Jackson Lake, one of the most scenic locations in the US. Here is paradise, of the overwhelming type. The Teton mountains formed along an active fault, with uplifting on one side. Although quiescent in recent times, the fault is subject to large earthquakes, but it is not volcanic.
The Teton mountains began to form 9 million years. Jackson Lake fills a basin next to it, making it a fairly typical structure for the western US.
Going up the river, we finally come to the end, its current beginning. Yellowstone is a wonder. Over half of all the world’s geysers are here, powered by the modern Yellowstone hotspot. The hotspot may not have found its last resting place, but it has lingered here, even in its current phase of quiet creating an unworldly landscape. The hotspot has uplifted the land, and captured the continental watershed which originally was much further west. On the east of the divide is the Yellowstone river which carries its water to the Gulf of Mexico. The Snake river, starting in the south of the park only 5 kilometer from the Yellowstone river (which starts far to the north), feeds the Pacific.
The hotspot has done immense damage here. Massive eruptions took out the heart of the region, leaving a broken ring of mountains surrounding the plateau. Three times the world exploded, 2.2, 1.3 and 0.64 million years ago, each eruption perhaps lasting only days or weeks. Numerous smaller eruptions added their mark, from rhyolitic eruption to basalt depositions. The eruptions followed a cyclical pattern: slow uplift as the magma chamber build, with minor eruptions, the explosion followed by caldera collapse, post caldera eruptions, and resurgent doming within the caldera. At the end of the cycle, volcanism ceases and the geysers become the main outlet for the magmatic heat underground. After a period of quiescence, the cycle begins anew. The cycles may have finished now. Perhaps in a few million years the hotspot will pop up further east.
Hotspot or not?
The Yellowstone hotspot, with its massive eruptions that repeatedly devastated a continent, is legendary. Still, its existence is not universally accepted. The evidence in favour of a hot spot is twofold. First, the large melt volumes and the fact the region has been pushed up by kilometers shows the effect of heat from below. Second, the linear progression from west to east shows the effect of the continent traversing a stationary spot. This is strengthened by the fact that the distance between this spot and Hawaii has remained the same – as the Hawaiian spot is known to be deep and stationary, it suggest Yellowstone may be too.
But this has not convinced everyone. There are three basic problems with the hotspot model. (1) Along the Snake river valley, the basaltic eruptions are about 2 million years after the passing of the hotspot; (2) Much of the volcanism is far to the north of the hotspot track; (3) The seismological evidence does not clearly show a deep plume. On this last point, there is agreement that there is a warm region here, but how deep does it go? Underneath the western US lies a subducted plate, the Farallon plate. Seismological mapping has traced the remnant of this cold plate, but little trace of a deep warm bubble penetrating this plate has been found underneath Yellowstone. The heat maps are shallow. It is possible that the plume is not in fact underneath Yellowstone, but is further west, deflected at shallow levels towards Yellowstone by the sinking plate. There are indications for a deeper plume 200 kilometer to the northwest of Yellowstone.
And why did the hotspot and the great American rift, which should be unconnected events, start or intensify at the same time, 17 million years ago? Where was the hotspot before this time? Edge convection related to the Farallon plate could explain both. But to me, the Ayes have it and the case for the hotspot is too strong to be dismissed. The hotspot was affected and perhaps energized by the great rift. In fact, the extension of the first flood basalt outpourings to the northwest follows the orientation of the rift. Elsewhere too, the northwest trending rifts provided pathways for the magma to travel long distances. The rift feeding Holohraun may provide a good model for this. But the heat came from below and the magma from the mantle. Still, the depth is disputed, the origin unclear, and immovable opinions make the arguments as heated as the ground below.
The Snake river, on the trail of the hotspot, follows some of the most impressive and geologically fascinating scenery in the US. It is a young river, which took its current shape only within the past million year, stitching together beds from many pre-existing rivers and flood beds. As the hotspot uplifted the land, the river beds tilted and the Snake, while following the hotspot, started to slither west. The river is not yet finished: when the hotspot moves again, so will the Snake. At that time, as the continental divide follows the hotspot to the east, it may capture the Yellowstone river. The Snake river is an ever evolving, living heritage. Entertaining, educational, and fascinating.