Carbonate boulders transported down steep tributary channels by debris flow came to rest on Holocene debris fans beside the Colorado River in Grand Canyon National Park. Weakly acidic rainfall and the metabolic activity of blue-green algae have produced roughly hemispheric dissolution pits as much as 2-cm deep on the initially smooth surfaces of the boulders. The average depth of dissolution pits increases with relative age of fan surfaces. The deepening rate averages 2.4 mm/1000 yr (standard error = 0.2 mm/1000 yr), as calculated from several radiometrically dated surfaces and an archeological structure. This linear rate, which appears constant over at least the past 3000 yr, is consistent with field relations limiting the maximum age of the fans and with the physical chemistry of limestone dissolution. Dissolution-pit measurements ( n= 6973) were made on 617 boulders on 71 fan surfaces at the 26 largest debris fans in Grand Canyon. Among these fan surfaces, the average pit depth ranges from 1.2 to 17.4 mm, and the resulting pit dissolution ages range from 500 to 7300 cal yr B.P. Most (75%) surfaces are younger than 3000 yr, probably because of removal of older debris fans by the Colorado River. Many of the ages are close to 800, 1600, 2300, 3100, or 4300 cal yr B.P. If not the result of differential preservation of fan surfaces, this clustering implies periods of heightened debris-flow activity and increased precipitation.

One of the most intriguing episodes in the Quaternary evolution of the Grand Canyon of the Colorado River, Arizona, was the development of vast lakes that are thought to have backed up behind lava erupted into the gorge. Stratigraphic evidence for these deep lava-dammed lakes is expectedly sparse. Possible lacustrine deposits at six areas in the eastern canyon yielded no compelling evidence for sediment deposited in a deep lake. At two of the sites the sediment was associated with late Quaternary spring-fed pools and marshes. Water-lain silt and sand at lower Havasu Creek was deposited ∼3000 cal yr ago. The deposit contains an ostracode assemblage similar to that living in the modern travertine-dammed pools adjacent to the outcrop. The second deposit, at Lees Ferry, formed in a spring-fed marsh ∼43,000 cal yr ago, as determined by 14C and amino acid geochronology. It contains abundant ostracode and mollusk fossils, the richest assemblages reported from the Grand Canyon to date. Our interpretation of these sediments as spring-fed deposits, and their relative youth, provides an alternative to the conventional view that deposits like these were formed in deep lava-dammed lakes that filled the Grand Canyon.

In the western Grand Canyon, fluvial terraces and pediment surfaces, both associated with a Pleistocene basalt flow, document Quaternary aggradation and downcutting by the Colorado River, illuminate the river's response to overload and the end of overload, and allow calibration of soil-carbonate stages and determination of downcutting rates. Four downcutting–aggradation cycles are present. Each begins with erosion of older deposits to form a new river channel in which a characteristic suite of deposits is laid down. The current cycle (I) started ∼700 yr B.P. The oldest (IV) includes the 603,000 ± 8000 to 524,000 ± 7000 yr Black Ledge basalt flow, emplaced when the river channel was ∼30 m higher than it is now. The flow is overlain by basalt–cobble gravel and basalt sand. Soils reach the stage V level of carbonate development. Calibrated ages for soil stages are Stage V, ∼525,000 yr; stage IV, <525,000 yr, ≥250,000 yr; stage III, <250,000 yr, ≥100,000 yr. The monolithologic basalt sand beds represent overloading by volcanic ash produced by an eruption 30–50 km upstream. The basalt–cobble beds signal breaching and rapid destruction of lava dams and erosion of flows. These deposits show that the Colorado River responds to overload by aggrading vigorously during the overload and then downcutting equally vigorously when the overload ends. The overall downcutting rate for the interval studied is 1.6 cm/1000 yr, much lower than rates upstream. The current downcutting rate, 11–14 m/1000 yr, likely is a response both to the end of late Pleistocene and early Holocene overload and to the reduction of sediment supply caused by Glen Canyon Dam.

Holocene and modern travertine formed in spring-fed Havasu Creek of the Grand Canyon, Arizona, was studied to determine the factors governing its oxygen-isotope composition. Analysis of substrate-grown travertine indicates that calculated calcite-formation temperatures compare favorably with measured water temperatures, and include silt-rich laminae deposited by monsoon-driven floods. Ancient spring-pool travertine is dated by U-series at 7380 ± 110 yr and consists of 14 travertine-silt couplets of probable annual deposition. One hundred eighty high-resolution δ18O analyses of this mid-Holocene sample average −11.0‰ PDB. The average value for modern travertine is ∼0.5‰ lower, perhaps because mid-Holocene temperature was higher or there was proportionally greater summer recharge. δ18O cyclicity in the mid-Holocene travertine has average amplitude of 1.9 ± 0.5‰ PDB, slightly less than the inferred modern-day annual temperature range of Havasu Creek. The annual temperature range might have been reduced during the 14-yr interval compared to present, although other non-temperature factors could account for the muted annual variation. Silt-rich laminae within isotopically lower calcite in the modern and mid-Holocene travertine verifies the seasonal resolution of both samples, and suggests that similar temperature-precipitation conditions, as well as monsoon-generated summer floods, prevailed in the mid-Holocene as they do throughout the Grand Canyon region today.

The failure of a lava dam 165,000 yr ago produced the largest known flood on the Colorado River in Grand Canyon. The Hyaloclastite Dam was up to 366 m high, and geochemical evidence linked this structure to outburst-flood deposits that occurred for 32 km downstream. Using the Hyaloclastite outburst-flood deposits as paleostage indicators, we used dam-failure and unsteady flow modeling to estimate a peak discharge and flow hydrograph. Failure of the Hyaloclastite Dam released a maximum 11 × 109 m3 of water in 31 h. Peak discharges, estimated from uncertainty in channel geometry, dam height, and hydraulic characteristics, ranged from 2.3 to 5.3 × 105 m3 s−1 for the Hyaloclastite outburst flood. This discharge is an order of magnitude greater than the largest known discharge on the Colorado River (1.4 × 104 m3 s−1) and the largest peak discharge resulting from failure of a constructed dam in the USA (6.5 × 104 m3 s−1). Moreover, the Hyaloclastite outburst flood is the oldest documented Quaternary flood and one of the largest to have occurred in the continental USA. The peak discharge for this flood ranks in the top 30 floods (>105 m3 s−1) known worldwide and in the top ten largest floods in North America.