HR: 0800h
AN: V41A-1360    [Abstracts]
TI: Quantifying the distribution of bubble sizes in volcanic rocks: Generalized statistical formulation and application to vesicular lavas
AU: * Proussevitch, A
EM: alex.proussevitch@unh.edu
AF: University of New Hampshire, Morse Hall, EOS, Durham, NH 03824 United States
AU: Sahagian, D
EM: dork.sahagian@lehigh.edu
AF: Lehigh University, 31 Williams Dr., Bethlehem, PA 18015 United States
AB: We have developed a generalized analytical and computational approach to analyze the full spectrum of observed bubble size distributions in volcanic rocks. All natural bubble populations can be characterized within the logarithmic family of statistical distributions. Populations are typically described with best fit functions. We have generalized this approach to employ complimentary use of distribution density and exceedance forms for probability. This results in very accurate estimates of distribution moments and bubble number density for each mode in any distribution. We demonstrate that transformation linearization of logarithmic distributions can be accomplished by changing the scale of measurement units. This enables the investigator to visualize distribution characteristics and thus eliminates the problem of observational bias resulting from limitations in instrument resolution, sample size, etc. In an initial application of this analytical technique, we have studied bubble populations in volcanic rocks from a collection of vesicular basalts from the Colorado Plateau. The samples reveal a variety of mono- and polymodal distributions within the logarithmic family of statistical functions. Within this family, most populations have a bimodal log-normal distribution, but the others are represented by mono or bimodal log logistic, and Weibull distributions. We present eleven bubble population types based on distribution function, mode location and intensity. Trends within some of these population types can be interpreted as evolution of vesiculation processes. Larger bubble size modes can result from coalescence of smaller bubbles within the distribution. In the course of coalescence larger modes grow at the expense of smaller modes. Another type of trend leads to monomodal Weibull (or exponential) distributions as a result of superposition of multiple log normal distributions if modes are sufficiently close in size and intensity.
DE: 8429 Lava rheology and morphology
DE: 8439 Physics and chemistry of magma bodies
DE: 8494 Instruments and techniques
SC: Volcanology, Geochemistry, Petrology [V]
MN: 2004 AGU Fall Meeting