Stratigraphy of individual basalt flows in the N 2 magnetostratigraphic unit of the Grande Ronde Basalt (GRB) within the central Columbia Plateau has been developed using data from seven surface sections and fifteen boreholes. Twenty-one individual flows have been identified and grouped into eight flow packages. The flow correlations were developed based on chemical composition, paleomagnetic vector direction, stratigraphic position, and thickness of individual flows. A multivariate statistical procedure, discriminant analysis, was used to test the validity of using chemical composition alone to define the flow packages. Results of the test show that samples can be correctly classified within one adjacent flow package in 94 percent of the cases. Application of discriminant analysis to chemical composition data indicates that within the Pasco Basin the upper two-thirds of the N 2 GRB contains 17 individual flows, of which only 9 to 15 may be present at any one location. Seven of these flows are present throughout the portion of the Pasco Basin studied. Correlation of flows between boreholes or surface sections means that units with the same stratigraphic position and diagnostic characteristics have been identified. In most cases this means that a single flow formed from one eruption. However, some flows may not be continuous, or some correlated flows may represent eruptions separated by as much as several thousand years but which gave rise to flows with identical stratigraphic positions and similar characteristics. The greatest number of flows occurs in the southeastern part of the basin, but the thickest total section occurs in the western and northwestern parts of the basin. The detailed flow correlations provide evidence of deformation during emplacement of the N 2 GRB. Variations of the thickness of individual flows and packages of flows disclose that subsidence was greatest in the western portion of the basin and that growth of the Yakima Ridges began at least by late GRB time. This timing of deformation of the Yakima Ridges is consistent with previous interpretations. We conclude that discriminant analysis applied to chemical composition of GRB flows provides a means of quantitatively defining correlation of flow packages. The method yields estimates of uncertainty in correlations but must be applied in the context of an appropriate stratigraphic framework. Discriminant analysis should be useful in a variety of volcanic terranes where subtle but consistent differences in composition occur between individual flows or flow packages. Successful application of the technique requires that variations in the chemistry of individual flows or packages of flows be generally less than chemical differences among flow packages and that a relatively large number of samples be available for classification.

Thick (>30 m) flows of Columbia River basalt contain internal vesicular zones within otherwise dense, sparsely vesicular basalt. These zones are continuous over distances ranging from 0.5 km to 30 km; most are characterized by an abrupt transition from vesicle-rich to vesicle-poor rock above and by a gradational lower margin. The zones were formed by post-emplacement migration, coalescence, and entrapment of aqueous vapor bubbles. Under appropriate physicochemical conditions, bubbles nucleate at the lower solidification front and rise buoyantly until retarded by the higher viscosities below the upper solidification front. In some cases, ponding of bubbles against this ceiling occurs before freezing-in of the vesicles by the downward passage of the upper front. We have developed a one-dimensional dynamic model of the process of vesicular zone emplacement by starting with the measured vesicle distribution in a lava flow and calculating movement of vesicles according to Stokes Law as the flow is melted, or “uncrystallized.” Data for the solidification history of the Cohassett flow are based on a previously developed cooling model. Melting of the flow was accomplished by reversing movement of the upper and lower solidification fronts; when a solidification front passes a vesicle, the vesicle is free to move in the reverse direction since time is moving in reverse. The model clearly illustrates the ponding effect of the upper solidification front and shows that nucleation of vesicles on the lower solidification front is required in trials where the model approximates a reasonable distribution of bubbles at T = 0. Internal vesicular zones consist of layers representing variations in number and size of vesicles. The regular layer spacing suggests a cyclic enrichment and depletion of bubble nuclei. If the production of nuclei is controlled by the state of the system at the lower solidification front, then the layers can be explained by oscillations between periods of crystallization, during which the system moves toward higher oversaturation pressure and periods of vapor exsolution tending to reestablish equilibrium. The layered distribution of vesicles in Columbia River basalt flows, for example, the McCoy Canyon and Rocky Coulee flows, can be explained by such a mechanism.