Previous page     Contents     Next page

Image analysis is a powerful tool used to solve microscopic problems but it also can be dangerous when used in an automated capacity. We have used the image analysis computer program Diana to describe crystal size and vesicle size distributions in Mt. St. Helens gray dacite (MSHD) and Merapi basaltic andesite (MRP9603), first using the length of the crystal to characterize its size. The program considers the longest dimension of the crystal as its length, irrespective of its crystallographic axes, causing both errors in true length (defined by a crystallographic axis) and crystal orientation (Fig. 3.8-7). An improved analysis is made using the area of the crystals as size compared to the short axis. We found the results of the later method consistent with the crystal accumulation in the particle-distribution of experimentally obtained pyroclasts of MRP9603. While using the method on pyroclast samples of Mt. St. Helens gray dacite, the distribution curve did not verify our image analytical results. In determining the crystal-distribution Diana did not recognize a fracture system in the phenocryst. Closer examination of the dacite using reflected light revealed that most of the phenocrysts were broken. In MSHD the mechanism that hinders fracturing from destroying the phenocrysts failed to do so in MRP9603. SEM images of MSHD experimental-pyroclastic particles show multiple pyroclasts (glass and crystal) in all size fractions, whereas MRP9603 reveals many unbroken crystals with thin glass rims.
 

Fig. 3.8-7: Schematic of crystal. Length of crystal in alignment with crystallographic axis (l) and computer processed length (l´ ). Errors up to 21% are observed.

Characterizing vesicle sizes in pyroclastic rocks is commonly carried out by the examination of binary images of analyses. Either scanned SEM or reflection-light-microscopy images are used for the investigations. Most primary (non-filtered) pictures are composed of a wide range of grayscales (256). Transferring the images into binary files, the program searches for a selected grayscale to calculate the black and white border. Diana allows some recalculation in the event of illumination effects from light microscopy. Materials like MSHD ground mass do not allow precise measurement by reflected light microscopy, and SEM pictures are required. By microscopic investigation we found two vesicle-systems in MSHD: a few large (mm) irregularly formed vesicles and a network of several small micron-wide tubes. A 2-D image analysis of these branched tubes shows that 34% of the MSHD-ground mass is built up of vesicles with a 3 µm diameter. BET (multi-point gas sorption) surface analysis reveals a surface area of 0.1347m²/g for MSHD, or 2.36 m² for the 17.5 g sample analyzed. We then calculate the open tube system of 3 µm diameter to have a length of approximately 24000 m. To describe this system on a 2-D basis is not possible. Mercury immersion tests have revealed a third locally connected vesicle system of micron size. In order to use image analysis to describe this type of vesicle system we will correlate permeability measurements and vesicle-form-characteristics as is practiced by sedimentologists.