Abstract The Channeled Scabland of eastern Washington State, USA, brought megafloods to the scientific forefront. A 30,000-km2 landscape of coulees and cataracts carved into the region’s loess-covered basalt attests to overwhelming volumes of energetic water. The scarred landscape, garnished by huge boulder bars and far-travelled ice-rafted erratics, spurred J Harlen Bretz’s vigorously disputed flood hypothesis in the 1920s. First known as the Spokane flood, it was rebranded the Missoula flood once understood that the water came from glacial Lake Missoula, formed when the Purcell Trench lobe of the last-glacial Cordilleran ice sheet dammed the Clark Fork valley in northwestern Idaho with ice a kilometer thick. Bretz’s flood evidence in the once-remote Channeled Scabland, widely seen and elaborated by the 1950s, eventually swayed consensus for cataclysmic flooding. Missoula flood questions then turned to some that continue today: how many? when? how big? what routes? what processes? The Missoula floods passed through eastern Washington by a multitude of valleys, coulees and scabland tracts, some contemporaneously, some sequentially. Routings and their timing depended on the positions of various lobes of the multi-pronged Cordilleran ice sheet and the erosional development of the channels themselves. The first floods mostly followed the big bend of Columbia valley looping through north-central Washington. But the south-advancing Okanogan ice lobe soon blocked that path, forming long-lasting glacial Lake Columbia in the impounded Columbia valley. Missoula floods into this lake were diverted south out of the Columbia valley and into eastern Washington coulees and scabland tracts. At least four floods entered Moses Coulee, but then as the Okanogan lobe advanced over and blocked the head of that coulee, more eastern paths took the water, including Grand Coulee and the Telford-Crab-Creek and Cheney-Palouse scabland tracts. Flood routing also depended on the erosion of the coulees. At some point, headward erosion of upper Grand Coulee lowered the divide saddle between the west-running Columbia valley and the deep and wide Grand Coulee heading southwest. Still uncertain is when this happened and the consequences with respect to the stage and extent of glacial Lake Columbia and to flood access to the other, higher, flood routes. Downstream, all flood routes converged into Pasco Basin, flowed through Wallula Gap and the Columbia River Gorge into the Pacific Ocean, following submarine canyons and depositing sediment layers on abyssal plains. Stratigraphic studies indicate dozens—likely more than a hundred—separate Missoula floods during the last glacial period. Over the length of the flood route, backwater areas and depositional basins preserve multiple flood beds, many of which are separated by signs of time, including volcanic ash layers and soil development in subaerial environments; and varve-like beds and pelagic mud layers in lacustrine and marine settings. Evidence also comes from the glacial Lake Missoula basin, where stratigraphy indicates dozens of filling and emptying cycles. Varve counts in conjunction with radiocarbon dating and paleomagnetic secular variation show the repeated filling-and-release cycles of glacial Lake Missoula had intervals possibly as long as 100 years early in the lake’s history but diminished to just one or two years for the last few floods. This behavior accords with jokulhlaup-style floods released by subglacial drainage from a self-dumping ice-dammed lake. Not yet clear is whether such a mechanism applies to all the floods or if some emptied more cataclysmically as hypothesized by some. Radiocarbon dating of sparse organic materials remains key to defining flood chronology but has been lately bolstered by analyses of terrestrial cosmogenic nuclides and optically stimulated luminescence. Varve counts and paleomagnetic secular variation studies help to define durations and intervals represented by sequences of flood beds. The ~16 ka Mount St. Helens Set S tephra is commonly interbedded within flood deposits, enabling correlation of deposits among sites. Tephra from the 13.7–13.4 ka eruption of Glacier Peak overlies all glacial Lake Missoula and Missoula flood deposits, defining an end time. Overall conclusions are that glacial Lake Missoula was extant and producing floods for at least 3–4 ky during 20–14 ka. At least ~75 floods preceded Mount St Helens Set S, followed by 30 or more after the tephra fall. Most floods entered glacial Lake Columbia, impounded by the Okanogan lobe, for 2–5 ky between about 18.5 and 15 ka. Glacial Lake Columbia outlived Lake Missoula by >200–400 yr but may have been born later since at least one flood came down the Columbia valley before the Okanogan ice lobe blocked the Columbia valley at 18.5–18 ka. The maximum extent of the Okanogan and Purcell Trench lobes, many Missoula floods, substantial erosion of upper Grand Coulee, and the widespread tephra falls from Mount St. Helens eruptions all happened about 17–15 ka. People, in the area since 16.6–15.3 ka, almost certainly witnessed the last of the Missoula floods and later large floods from other ice-dammed lakes in the Columbia River basin. Quantitative flow analyses give peak discharge estimates and support understanding of erosional and depositional processes. The first flow assessments were simple cross-section calculations but recent assessments employ two-dimensional hydrodynamic models. The general finding is that emplacement of the maximum stage evidence requires about 20 million m3/s near the Lake Missoula outlet and about 5–15 million m3/s through Wallula Gap and downstream in the Columbia River Gorge. These hydraulic analyses raise still-unresolved questions regarding canyon erosion and possible additional water sources. The large Pleistocene Bonneville flood entered the Columbia River system from the southeast from pluvial Lake Bonneville, the Pleistocene predecessor to Great Salt Lake in the eastern Great Basin. During the last glacial, the lake basin filled, covering >50,000 km2 with 10,400 km3 of water before reaching its maximum possible stage governed by Red Rock Pass, the lowest divide separating the basin from the Snake River basin to the north. The overtopping lake rapidly incised 108–125 m into the Red Rock Pass outlet, spilling half of its total lake volume. G.K. Gilbert described the essential sequence in the 1870s, but the flood was mostly forgotten until the late 1950s when Harold Malde linked the spectacular scabland topography and bouldery “melon gravel” on the Snake River Plain to the Lake Bonneville overflow. The Bonneville flood appears to have been a singular event at about 18 ka. No evidence of multiple or pre-last-glacial spillovers has yet been found. Its total volume was about twice that of a maximum Lake Missoula flood yet its peak discharge was ~1 million m3/s, less than a tenth of the largest Missoula floods. Its comparatively simple flow path and much steadier flow make the Bonneville flood ideal for new studies of erosional and depositional processes. At least two floods seem to have passed down the Columbia valley after the last of the Missoula floods, including a large flood about ~14 ka likely from cataclysmic demise of the thinning Okanogan ice lobe dam impounding glacial Lake Columbia. Floods from earlier glacial ages left scant yet clear evidence in the Channeled Scabland and Columbia valley. But their source, timing, and magnitudes are little understood. Some deposits are paleomagnetically reversed, thus older than ~800 ka. Last-glacial floods and perhaps older ones affected the Snake River Plain, some likely sourced in lakes dammed by alpine glaciers in central Idaho.

中文翻译：

---

米苏拉和博纳维尔洪水——哥伦比亚河流域冰河时代特大洪水回顾

摘要 美国华盛顿州东部的 Channeled Scabland 将特大洪水带到了科学前沿。在该地区覆盖着黄土的玄武岩上雕刻着 30,000 平方公里的古力和白内障景观，证明了大量的能量水。巨大的巨石栏杆和远距离漂流的冰筏漂流物点缀着伤痕累累的景观，激发了 J Harlen Bretz 在 1920 年代备受争议的洪水假说。最初被称为斯波坎洪水，它被重新命名为米苏拉洪水，一旦人们了解到水来自冰川湖米苏拉，当末次冰川科迪勒拉冰盖的珀塞尔海沟叶用冰在爱达荷州西北部堵塞克拉克福克山谷时形成一公里厚的。Bretz 在曾经偏远的 Channeled Scabland 中的洪水证据，在 1950 年代被广泛看到和阐述，最终动摇了灾难性洪水的共识。米苏拉洪水问题然后转向了一些今天仍在继续的问题：有多少？什么时候？多大？什么路线？什么流程？米苏拉洪水通过许多山谷、古道和 scabland 区域穿过华盛顿东部，有些是同时发生的，有些是依次发生的。路线及其时间取决于多分支科迪勒拉冰盖的各个裂片的位置以及通道本身的侵蚀发展。第一次洪水主要是在哥伦比亚河谷的大弯流经华盛顿中北部。但向南推进的奥卡诺根冰瓣很快挡住了这条道路，在被蓄水的哥伦比亚山谷中形成了长期存在的冰川哥伦比亚湖。进入这个湖的米苏拉洪水被从哥伦比亚山谷向南转移，进入华盛顿东部的古力和 scabland 大片。至少有四次洪水进入了摩西古力，但随着奥卡诺根海瓣前进并挡住了古力的头部，更多的东部路径淹没了水，包括大古力和德福-蟹溪和切尼-帕卢斯 scabland 大片。洪水路线还取决于古力的侵蚀。在某个时候，大古力上游的上游侵蚀降低了向西延伸的哥伦比亚山谷和向西南方向延伸的深而宽的大古力之间的分水岭。仍然不确定的是何时发生这种情况以及与哥伦比亚冰川的阶段和范围以及洪水进入其他更高的洪水路线有关的后果。下游，所有洪水路线都汇入帕斯科盆地，流经瓦卢拉峡谷和哥伦比亚河峡谷进入太平洋，顺着海底峡谷和深海平原沉积沉积层。地层学研究表明，在最后一个冰河期，有数十次——可能超过 100 次——独立的米苏拉洪水。在洪水路线的整个长度上，回水区和沉积盆地保留了多个洪水床，其中许多被时间的痕迹隔开，包括火山灰层和地下环境中的土壤发育；以及湖泊和海洋环境中的类似海藻的床和远洋泥层。证据还来自冰川密苏拉湖盆地，那里的地层显示数十次充填和排空循环。Varve 计数与放射性碳测年和古地磁长期变化相结合，表明米苏拉冰川湖的重复充填和释放周期在该湖的历史早期可能长达 100 年，但在最近几次洪水中减少到只有一两年. 这种行为与自倾式冰坝湖的冰下排水所释放的 jokulhlaup 式洪水相一致。目前尚不清楚这种机制是否适用于所有洪水，或者是否如某些人所假设的那样以更灾难性的方式排空。稀有有机材料的放射性碳测年仍然是定义洪水年代学的关键，但最近得到了对陆地宇宙成因核素和光激发光的分析的支持。变流计数和古地磁长期变化研究有助于确定由洪水床序列代表的持续时间和间隔。~16 ka Mount St. Helens Set S tephra 通常夹在洪水沉积物中，从而能够将不同地点的沉积物关联起来。13.7-13.4 ka 冰川峰喷发的火山灰覆盖在所有冰川湖米苏拉和米苏拉洪水沉积物上，定义了结束时间。总体结论是冰川密苏拉湖存在并在 20-14 ka 期间产生了至少 3-4 ky 的洪水。至少约 75 次洪水在圣海伦斯山 S 组之前发生，随后在火山灰坠落之后发生了 30 次或更多。大多数洪水进入冰川哥伦比亚湖，由奥卡诺根叶蓄水，在大约 18.5 至 15 ka 之间持续 2 至 5 ky。哥伦比亚冰川湖比米苏拉湖更长寿 > 200-400 年，但可能出生较晚，因为在奥卡诺根冰叶在 18.5-18 ka 阻塞哥伦比亚山谷之前，至少有一次洪水从哥伦比亚山谷流过。奥卡诺根海沟和珀塞尔海沟裂片的最大范围、许多米苏拉洪水、大古力上游的大量侵蚀以及圣海伦火山喷发造成的广泛的火山灰都发生在大约 17-15 ka。自 16.6-15.3 ka 以来，该地区的人们几乎可以肯定目睹了最后一次米苏拉洪水，以及后来来自哥伦比亚河流域其他冰坝湖泊的大洪水。定量流量分析提供峰值流量估计并支持对侵蚀和沉积过程的理解。最初的流量评估是简单的横截面计算，但最近的评估采用二维流体动力学模型。一般发现，最大阶段证据的就位需要在密苏拉湖出口附近约 2000 万立方米/秒，通过 Wallula Gap 和哥伦比亚河峡谷下游约 5-1500 万立方米/秒。这些水力分析提出了关于峡谷侵蚀和可能的额外水源的尚未解决的问题。大更新世博纳维尔大洪水从东南部从大盆地东部大盐湖的更新世前身博纳维尔大洪水湖进入哥伦比亚河系统。在最后一次冰期，湖盆充满了 10,400 平方公里的水，覆盖了 > 50,000 平方公里，然后达到了由红岩通道控制的最大可能阶段，这是将盆地与北部的蛇河流域分隔开的最低分水岭。溢流湖迅速切入 Red Rock Pass 出口 108-125 m，溢出其总湖量的一半。GK Gilbert 描述了 1870 年代的基本序列，但直到 1950 年代后期，哈罗德·马尔德 (Harold Malde) 将 Snake River 平原上壮观的 scabland 地形和巨石“瓜砾石”与博纳维尔湖的溢流联系起来时，洪水大多被遗忘了。Bonneville 洪水似乎是发生在大约 18 ka 的单一事件。尚未发现多次或末次冰期前溢出的证据。其总体积约为米苏拉湖最大洪水的两倍，但其峰值流量约为 100 万立方米/秒，不到米苏拉最大洪水的十分之一。其相对简单的流动路径和更稳定的流动使邦纳维尔洪水成为侵蚀和沉积过程新研究的理想选择。在最后一次米苏拉洪水之后，至少有两次洪水似乎已经从哥伦比亚山谷流过，其中一次大约 14 ka 的大洪水可能是由于蓄水哥伦比亚冰川湖的奥卡诺根冰叶坝变薄造成的灾难性消亡。来自早期冰川时代的洪水在 Channeled Scabland 和哥伦比亚山谷留下了很少但清晰的证据。但它们的来源、时间和大小却鲜为人知。一些矿床古地磁反转，因此年龄超过~800 ka。末次冰期洪水和可能更老的洪水影响了蛇河平原，其中一些可能来自爱达荷州中部被高山冰川筑坝的湖泊。包括大约 14 ka 的大洪水，这可能是由于蓄水哥伦比亚冰川湖的奥卡诺根冰叶坝变薄造成的灾难性消亡。来自早期冰川时代的洪水在 Channeled Scabland 和哥伦比亚山谷留下了很少但清晰的证据。但它们的来源、时间和大小却鲜为人知。一些矿床古地磁反转，因此年龄超过~800 ka。末次冰期洪水和可能更老的洪水影响了蛇河平原，其中一些可能来自爱达荷州中部被高山冰川筑坝的湖泊。包括大约 14 ka 的大洪水，这可能是由于蓄水哥伦比亚冰川湖的奥卡诺根冰叶坝变薄造成的灾难性消亡。来自早期冰川时代的洪水在 Channeled Scabland 和哥伦比亚山谷留下了很少但清晰的证据。但它们的来源、时间和大小却鲜为人知。一些矿床古地磁反转，因此年龄超过~800 ka。末次冰期洪水和可能更老的洪水影响了蛇河平原，其中一些可能来自爱达荷州中部被高山冰川筑坝的湖泊。