Lake of the Woods (LOTW) is a large, complex lake basin once occupied by glacial Lake Agassiz. High-resolution seismic-reflection profiles and cores in the shallow, open southern basin of LOTW reveal a sedimentary architecture comprising four lacustrine units separated by three low-stand unconformities. These units represent several phases of Lake Agassiz and its changing configuration. One unconformity marks the Moorhead low phase and another marks the separation of LOTW from Lake Agassiz, perhaps ~10 cal ka BP, as the level of the latter fell, but before final drainage of Agassiz. Initially, the separate Holocene lake in the southern basin was broad and shallow, sometimes marshy or dry. Shortly after 8 cal ka BP, the southern basin dried up completely, despite the progressive rise of the northern outlet of the lake due to differential isostatic uplift. The resulting hiatus is related to the well-documented mid-Holocene arid interval in central North America. A return to wetter conditions in the late Holocene caused the southern basin of LOTW to refill since about 3800 cal yr BP. Late Holocene sediments have accumulated slightly asymmetrically in the basin, possible due to continued southward transgression of the lake as a result of isostatic tilting.

Geomorphic analysis and optically stimulated luminescence (OSL) ages from undated Lake Agassiz beaches and adjacent fluvial sediments on Riding Mountain in Manitoba provide insight into their early history. New OSL ages of 14.5±2.4 and 13.4±0.7 ka on the oldest (Herman to Norcross) beaches of Lake Agassiz near the Canada-U.S. border indicate that the Laurentide Ice Sheet (LIS) retreated from that part of the Agassiz basin by ~14.5 ka. To the north along Riding Mountain, the Herman strandlines are absent, and OSL ages on the oldest beach there average 12.9 ka, which links it to the younger Norcross-Tintah strandlines. In adjacent Riding Mountain, OSL ages and geomorphological relationships of a large abandoned glacial spillway >200 m above the oldest beaches of Lake Agassiz indicate that this channel predates retreat of the LIS and formation of beaches in this part of the Agassiz basin, with ice remaining in this area until after 14.5 ka. OSL ages on the Gimli beach 170 km to the east are >3000 yr older than conventional assignments, suggesting that it formed during the Moorhead low-water phase 12.8–10.6 ka. Luminescence ages support the conclusion that the Campbell beach formed ~10.9 ka near the end of the Moorhead low-water phase.

Lake Agassiz was ponded on the northward-sloping surface of the Hudson Bay and Arctic Ocean basins, as the Laurentide Ice Sheet retreated. The level of Lake Agassiz abruptly fell ~ 12.9 cal (11 14C) ka BP, exposing the lake floor over a large region for > 1000 yr. The routing of overflow during this (Moorhead low-water) period is uncertain, and there is evidence on the continent and in ocean basins for both an easterly route through the Great Lakes–St. Lawrence to the North Atlantic and for a northwesterly route through the Clearwater–Athabasca–Mackenzie system to the Arctic Ocean. The Moorhead low water phase coincides with the Younger Dryas cooling, and a cause–effect relationship has been proposed by attributing a change in ocean thermohaline circulation to the re-routing of Lake Agassiz freshwaters from the Gulf of Mexico to more northern oceans. Paleoclimatic interpretations from ecosystems in lake sediments in the region, and a simple calculation of the paleohydrological budget of Lake Agassiz, indicate that the climate remained wet and cool throughout the YD in this region, and was not warm nor dry enough to allow evaporative loss to offset the influx of meltwater and precipitation; thus, the Moorhead phase resulted from changes in the outlet that carried overflow.

Sediment cores with new radiocarbon dates from the southern outlet of glacial Lake Agassiz indicate that meltwater delivery to the Mississippi valley was disrupted at 10,800 14C yr B.P. and the outlet was abandoned by 9400 14C yr B.P. These findings confirm the timing of generally accepted terminations of the Lockhart and Emerson Phases of Lake Agassiz. Additionally, the radiocarbon chronology indicates that the spillway was fully formed by 10,800 14C yr B.P. and that the occupancy in late-Emerson time was likely short-lived with minimal spillway erosion.

Computer reconstructions of the bathymetry of the lake were used to quantify variations in the size and form of Lake Agassiz during its final two phases (the Nipigon and Ojibway phases), between about 9200 and 7700 14C yr B.P. (ca. 10,300–8400 cal yr B.P.). New bathymetric models for four Nipigon Phase stages (corresponding to the McCauleyville, Hillsboro, Burnside, and The Pas strandlines) indicate that Lake Agassiz ranged between about 19,200 and 4600 km3 in volume and 254,000 and 151,000 km2 in areal extent at those times. A bathymetric model of the last (Ponton) stage of the lake, corresponding to the period in which Lake Agassiz was combined with glacial Lake Ojbway to the east, shows that Lake Agassiz–Ojibway was about 163,000 km3 in volume and 841,000 km2 in areal extent prior to the final release of lake waters into the Tyrrell Sea. During the Nipigon Phase, a number of catastrophic releases of water from Lake Agassiz occurred as more northerly (lower) outlets were made available by the retreating southern margin of the Laurentide Ice Sheet; we estimate that each of the four newly investigated Nipigon Phase releases involved water volumes of between 1600 and 2300 km3. The final release of Lake Agassiz waters into the Tyrrell Sea at about 7700 14C yr B.P. is estimated to have been about 163,000 km3 in volume.

The volume and surface area of glacial Lake Agassiz varied considerably during its 4000-year history. Computer models for seven stages of Lake Agassiz were used to quantify these variations over the lake's early history, between about 11,000 and 9300 14C yr B.P. (ca. 13,000 to 10,300 cal yr B.P.). Just after formation of the Herman strandlines (ca. 11,000 14C yr B.P.), the volume of Lake Agassiz appears to have decreased by >85% as a consequence of the abrupt rerouting of overflow to its eastern outlet from its southward routing into the Mississippi River basin. This drainage released about 9500 km3 of water into the North Atlantic Ocean via the Great Lakes and Gulf of St. Lawrence. Following closure of this eastern routing of overflow, the lake reached its maximum size at about 9400 14C yr B.P. with an area of >260,000 km2 and a volume of >22,700 km3. A second major reduction in volume occurred shortly after that, when its volume decreased >10% following the opening of the Kaiashk outlet to the east into the Great Lakes, and 2500–7000 km3 of water was released into the North Atlantic Ocean. These discharges may have affected ocean circulation and North Atlantic Deep Water production.

Between 10,500 and 9000 cal yr BP, δ18O values of benthic ostracodes within glaciolacustrine varves from Lake Superior range from − 18 to − 22‰ PDB. In contrast, coeval ostracode and bivalve records from the Lake Huron and Lake Michigan basins are characterized by extreme δ18O variations, ranging from values that reflect a source that is primarily glacial (∼ − 20‰ PDB) to much higher values characteristic of a regional meteoric source (∼ − 5‰ PDB). Re-evaluated age models for the Huron and Michigan records yield a more consistent δ18O stratigraphy. The striking feature of these records is a sharp drop in δ18O values between 9400 and 9000 cal yr BP. In the Huron basin, this low δ18O excursion was ascribed to the late Stanley lowstand, and in the Lake Michigan basin to Lake Agassiz flooding. Catastrophic flooding from Lake Agassiz is likely, but a second possibility is that the low δ18O excursion records the switching of overflow from the Lake Superior basin from an undocumented northern outlet back into the Great Lakes basin. Quantifying freshwater fluxes for this system remains difficult because the benthic ostracodes in the glaciolacustrine varves of Lake Superior and Lake Agassiz may not record the average δ18O value of surface water.

Four traditionally recognized strandline complexes in the southern basin of glacial Lake Agassiz are the Herman, Norcross, Tintah and Campbell, whose names correspond to towns in west-central Minnesota that lie on a linear transect defined by the Great Northern railroad grade; the active corridor for commerce at the time when Warren Upham was mapping and naming the shorelines of Lake Agassiz (ca.1880–1895). Because shorelines represent static water planes, their extension around the lake margin establishes time-synchronous lake levels. Transitions between shoreline positions represent significant water-level fluctuations. However, geologic ages have never been obtained from sites near the namesake towns in the vicinity of the southern outlet. Here we report the first geologic ages for Lake Agassiz shorelines obtained at field sites along the namesake transect, and evaluate the emerging chronology in light of other paleoclimate records. Our current work from 11 sampling sites has yielded 16 independent ages. These results combined with a growing OSL age data set for Lake Agassiz's southern basin provide robust age constraints for the Herman, Norcross and Campbell strandlines with averages and standard deviations of 14.1 ± 0.3 ka, 13.6 ± 0.2 ka, and 10.5 ± 0.3 ka, respectively.