V31D-01 INVITED 08:00h
Basaltic Lava Channels
In Hawaii, the mode of lava transport - through open channels or through insulating lava tubes - determines the thermal, rheological, and emplacement history of a lava flow. Most Hawaiian lavas are erupted at near-liquidus temperatures and are therefore crystal-poor; lava transport through open channels allows rapid cooling and consequent rapid increases in lava crystallinity. Solidified aa flows resulting from channelized flow are typically fine-grained throughout their thickness, indicating cooling of the entire flow thickness during transport. In contrast, transport of lava through insulating tubes permits flow over long distances with little cooling. Flows emerging from such tubes typically have pahoehoe flow surfaces with glassy crusts. Groundmass textures that coarsen from the flow rind to the interior reflect rates of post-emplacement, rather than syn-emplacement, cooling. To distinguish eruption conditions that result in lava channels from those that allow formation of lava tubes, we have performed a series of laboratory experiments involving injection of PEG 600 (a wax with a Newtonian rheology and freezing temperature of 19°C) into cold water through both uniform and non-uniform sloping channels. In uniform channels, tube formation can be distinguished from open channel flow using a dimensionless parameter based on a solidification time scale, an advection time scale, and a Rayleigh number that describes convection by heat loss from crust-free shear zones. Theoretical analysis predicts that in the open channel regime, the width of the crust (dc) will vary with the channel width (W) as dc = W$^{5/3}$. Crustal coverage of non-uniform channels in both laboratory experiments and field examples from Kilauea Volcano, Hawaii, is consistent with this prediction. However, experiments in non-uniform channels illustrate additional controls on the surface coverage of lava channels. Most important is crustal extension resulting from flow acceleration through constrictions or channel bends that exposes more core lava to cooling than simply that of the shear zones. Thus the channel geometry plays a major role in the thermal history of a flow. As lava flows rarely flow through pre-existing channels of prescribed geometry, we have performed an additional set of analog laboratory experiments to determine the relationship between flow rate, slope, and channel formation in solidifying flows. All flows develop stable uniform channels within solidified levees except when the flow rate is sufficiently low to permit flow front solidification, inflation, and tube formation. On constant slopes, increasing flow rates result in increases in both the rate of flow advance rate and the channel width, and a decrease in levee width. At constant flow rates, both channel width and levee width decrease with increasing slope while flow advance rate increases. Limited data on the geometry of basaltic lava channels indicate that experimental data are consistent with field observations, however, both additional field data and scaling relationships are required to fully utilize the laboratory experiments to predict channel development in basaltic lava flows.
V31D-02 08:12h
Channel Formation in Viscous Fluids Independent of Scale and Composition
Observations of active basalt, dacite, and polyethylene glycol (PEG) flows suggest that the channel-forming process is similar through various scales and is independent of composition. Specifically, analysis of subaerial and simulated flows shows the systematic formation of channel zones (stable, transitional, zone of dispersed flow), as described by Lipman and Banks [1987] to characterize a 27 km-long a'a channel from the 1984 Mauna Loa (MLV), Hawaii eruption, are observed in all viscous fluid channels. A channelized dacite flow, emplaced from 1999-2002 on Santiaguito volcano (SV), Guatemala, displays similar downstream transitions in structure and morphology along its 3.75 km flow length. We observed the formation of these zones in meter-scale pahoehoe breakouts in active flows from the Pu'u O'o Mother's Day flow in 2002-2003, and in solidified flows from Mauna Ulu, Kilauea volcano (KV), Hawaii. Controlled laboratory simulations produce the structural and morphologic zones on the centimeter-scale ($<$ 70 cm long) as warm PEG (22-24 $\deg$C) is emplaced into a cold sucrose solution (6-12 $\deg$C) on a range of underlying slopes (8-$30\deg$). As an initial comparison, we measured the fraction of the flow length characterized by the stable, transitional, and zone of dispersed flow for MLV, SV, KV, and PEG flows. The stable zone dominates kilometer-scale flows, covering the majority of the flow length for MLV (0.49) and SV (0.64). Pahoehoe breakouts on Mauna Ulu have a lower range for stable zones of 0.25-0.49, while in PEG flows $<$0.28 of the flow length is stable. The fraction of flow contained within the transitional zone varies, with meter-, and kilometer-long flows ranging from 0.13-0.31, whereas PEG flows are $>$0.30 of the flow. The zone of dispersed flow dominates the flow length on centimeter- ($>$0.40) and meter-scale (0.25-0.54) flows, but is much less (0.20) of the flow length for MLV and SV. The time-scales for the emplacement of the channels ranges from minutes for PEG and pahoehoe breakouts, weeks for the a'a channel on MLV, to years for the dacitic flow on SV, yet each channel displays the same systematic downstream variation in morphology. The fraction of flow-length characterized by each zone at various length-scales may be attributed to the timescale of the eruptions, whereas longer emplacement times allow for maturity of the stable and transition zones downstream. To better understand the fundamental similarities of channel formation in the solar system, we will continue to analyze lava channels on greater length-scales (100's km), and in various environments including Venus, Mars, and Io.
V31D-03 08:24h
Can Satellite-Based Sensors, Hand-Held Thermal Imagers and Thermal Infrared Radiometers Calculate Reliable Eruption Rates at Active Lava Flows, Domes and Lakes?
Thermal data provide viable means of extracting eruption rates at active lava flows, domes and lakes. The initial algorithm, developed since 1994, uses total heat flux (Q), extracted from satellite-sensor-derived thermal images of the active lava body to extract eruption rate (E) using E = Q / $\rho$ [c $\Delta$ T + f L]. Here $\rho$ and c are lava density and specific heat capacity, $\Delta$T is lava cooling, f is fractional crystallization and L is latent heat of crystallization. Later it was shown that this reduces to a linear relationship: E = a Q + b, where a and b are defined by values assumed for $\rho$, c, $\Delta$T, f, and L. We present three case studies that: (1) demonstrate the variety of thermal data and activity styles that can be used in this approach, and (2) validate the approach through cross-checks with independent, field-based data: {\it(I) Stromboli: Satellite and thermal imager-based lava flow eruption rates.} A safe, easy and rapid method to calculate lava effusion rates using hand-held thermal image data was developed in June 2003 at Stromboli (Italy). FLIR data were used as input to the thermal effusion rate model, previously applied to satellite data, allowing automated effusion rate extraction. A comparison between a thermally-derived (0.2 - 0.9 m$^{3}$/s) and dimensionally-derived (i.e. channel depth x width x velocity) effusion rate (0.6 m$^{3}$/s) showed excellent agreement. In addition, the comparison between FLIR-derived effusion rates and satellite (AVHRR) derived values showed a good correlation (R = 0.9). {\it(II) Santiaguito: Satellite-derived eruption rates for a lava dome} A time series of 21 Landsat ETM+ and TM images acquired during 1986-2003, were used to calculate eruption rates at Santiaguito dome (Guatemala) to yield a time-averaged effusion rate of 0.4 m$^{3}$/s. Field-based flow dimension and velocity measurements during 1987 and 2000-03 yielded a values of 0.6$\pm$0.3 and 0.5$\pm$0.2 m$^{3}$/s which compared with an ETM+ derived values of 0.7$\pm$0.1 and 0.5$\pm$0.1 m$^{3}$/s, respectively. {\it(III) Erta Ale: Thermal infrared thermometer derived eruption rates} Using the typical surface temperature recorded by a thermal infrared thermometer targeted at the center of the Erta Ale lava lake (Ethiopia), a flux of ~1-4 x 10$^{7}$ W, to give a mass flux of 30-110 kg/s. These compare with satellite-based (TM) estimates of 40-100 kg/s. At Etna (Italy) 657 effusion rate measurements were made from combined thermal and field-based sources during 1999-2004. These all show excellent cross-data-set agreement. We have achieved a calibration of a and b in E = a Q + b. Suitable values for $\rho$, c, f, and L show that $\Delta$T is consistently in the range of 150 to 250 $\deg$C.
V31D-04 08:36h
Heat Loss From Skylights in Lava Tube Systems
Skylights serve as windows into the lava distribution systems that comprise tube-fed lava flow fields. They can thus provide insights into the thermo-rheological dynamics and flow stability in such feeder systems. Skylights form when a portion of the roof of a lava tube collapses, exposing the lava flowing in the tube to the atmosphere. The radiative and convective heat loss through skylights is likely significant and has not been adequately quantified previously. The loss of radiant heat through a skylight is relatively simple to constrain and results in a transient drop in the surface temperature of the molten lava exposed at the skylight. Given a typical lava flow surface T of ~1080 $\deg$C, we calculate Q$_{rad}$ of 1.71 x 10$^{5}$ W/m$^{2}$ [Q$_{rad}$ = $\sigma$ $\epsilon$ T$^{4}$]. This gives total cooling of 0.02 - 0.07 $\deg$C over a 3 m long skylight-exposed length. This is equivalent to 5 - 22 $\deg$C/km which compares to a typical cooling rate of $\sim$1 $\deg$C/km for the tubed section. A thin crust of cooled lava forms on the surface beneath the skylight which is then entrained back into the flowing lava causing a further (entrainment-related) heat loss. Convective heat loss from skylights is more complicated [Q$_{conv}$ = h$_{c}$(T$_{lava}$ - T$_{air}$)]. Our measurements show that air can flow in or out of a skylight. This represents a balance whereby hot, buoyant air blows out of a lava tube to be replaced by an influx of cold air that is sucked into a tube. Variations in the flow of air in and out of skylights may be caused by: 1) atmospheric pressure variations (cf. breathing cave systems), 2) convective instabilities that form in the tube-contained air, 3) changes in lava mass flux in a tube, increasing or decreasing the size of the headspace of air in the tube, or 4) transient combustion of hydrogen gas. We present temperature measurements of lava and gas made at skylights on the active flow field of Kilauea volcano using thermal infrared thermometers, thermocouples, and a thermal imager. From these data we calculate the radiative and convective heat losses from lava tubes through skylights and show how heat losses and flow rates vary with time. Our aim is to fully quantify the effects of this heat loss on the rheology and cooling rate of lava flowing in tubes.
V31D-05 08:48h
An analysis of the development of surface textures and flow morphology on small lava flows on Kilauea using high frame rate (30 Hz) digital thermal imagery.
Understanding the way that lavas cool is an important step in the development and validation of models of lava flows, and thermal imaging cameras are playing an important role in this investigation. Recent technological advances have led to the development of uncooled thermal imaging cameras, increases in data-acquisition rates and a new generation of high frame rate cameras. During August 2004, a hand-held FLIR ThermaCAM S40 was used to collect several thousand 30 Hz thermal images from 4 active lava flows on the Pulama pali fault scarp, Kilauea volcano, Hawaii. Short observation distances (3 to 10 m) resulted in the collection of high spatial (4-13 mm) resolution thermal data. The images acquired cover the formation, cooling, inflation and budding of new pahoehoe lobes and the formation of a range of surface textures on a small (35 m long) open channel flow. The images reveal the temperatures at which these flows developed surface textures such as a ductile crust and ropes, together with the transitions to brittle deformation such as tension cracks and the formation of aa clinker. They also recorded the apparent temperatures at which flow fronts of different dimensions stopped advancing, inflation occurred and new breakouts emerged from the front or margins of flows. Surface temperatures during the transition from channel to tube flow through regressive crustal formation were also determined. Methods of converting apparent temperatures to surface temperatures using appropriate lab-based emissivities and atmospheric attenuation have been made, and these are compared with similar transitions on basaltic flows on Etna. These then allow surface heat loss, core cooling and down-flow (temperature-dependant) rheological changes to be calculated. These are compared with similar transitions, cooling rates and rheological models for basaltic flows on Etna. The advantage of high frame rate thermal images is that they allow complete thermal characterisation of entire flows allowing crossflow and downflow temperature distributions to be determined and compared with theoretical values. Natural tracers on the surface of the flows allows velocity profiles to be constructed, and this enables the range of temperatures and strain rates at which different surface textures develop on different lava flows to be established.
V31D-06 09:00h
Lava flow-field emplacement at Rock Corral Butte, Eastern Snake River Plains, Idaho: It's doesn't look like Hawaii from here
Rock Corral Butte (RCB) is a basaltic shield volcano located at $43\deg$ 22.25'N, $113\deg$ 1.20'W within the Eastern Snake River Plains. The summit region is characterized by complex topography associated with late-stage eruption of spatter. Rock Corral Butte is surrounded by a large flow field ($>$16 km2) that is remarkable for its rolling, 2 to 4-m-scale topography that is superposed on large, kilometer-scale topographic terraces. The topographic highs within the flow are marked by relatively steep margins 2 to 4 m high, and flat or depressed interiors; the mounds are on the order of 5 to 10 m long and 3 to 10 m wide. These mounds may be tumuli or large lava lobes that have drained out or deflated (similar to flow-lobe tumuli observed in Icelandic shields). The mounds within the RCB flow field are distributed in distinct topographic terraces around the volcano. Initial analysis suggests that there is no significant change in petrology in the flow field with distance from the vent (up to 6 km), although the size and spatial density of the mounds decreases with distance. Preliminary research suggests that the distribution, size and morphology of these mounds at RCB are distinct from tumuli studied at Mauna Ulu, Hawaii, and are perhaps more similar to the distribution of similar features observed around Icelandic shields. This suggests that there may be fundamental differences in emplacement, and the formation of flow-field topography, at Mauna Ulu and RCB. Alternatively, a similar mechanism may create the RCB mounds as the tumuli at Mauna Ulu, but at different scales (both in time and space): tumuli at Mauna Ulu have been interpreted to require pulsating emplacement of small lava batches during the eruption along preferred pathways. If this holds true at RCB as well, it has implications for the size of lava pathways within the flow and eruption duration. We are continuing to examine and test models of flow field emplacement to constrain the behavior of RCB and similar shields along the Snake River Plains.
V31D-07 09:12h
New Insights From Whole Rock and Mineral Data on the Magmatic and Tectonic Evolution of the Columbia River Basalt Group (USA)
The Miocene Columbia River Basalt Group (CRBG) of north-western USA was emplaced in a geologically dynamic setting characterized by a close association between magmatism and lithospheric thinning and rifting. We present and discuss electron probe microanalysis and XRFA data obtained from samples spanning the entire sequence of the CRBG. The examined basalts have near-aphyric textures. No glass is present, and plagioclase and augitic clinopyroxene are dominant matrix and groundmass phases. Plagioclase microcrysts are labradoritic to bytownitic. Whole rock compositions were taken as proxies of the liquid compositions. Application of plagioclase / melt and clinopyroxene / melt geothermobarometers indicated that during crustal ascent the magmas were dry, and that pre-eruptive pressures and temperatures ranged from 0 to 0.66 GPa and 1393 to 1495 K, respectively. In a P-T diagram most of the samples are distributed along a general CRBG trend, while some samples plot along a parallel higher temperature trend. The calculated P-T values, the positive correlation between calculated P and T, and no horizontal alignment of the data, exclude the presence of upper crustal solidification fronts, and indicate that magma aggregation zones were located deeper than 25 km, plausibly immediately below the Moho, that in this region is at a depth of approximately 35 km. Episodic stretching of the lithosphere best explains the observed parallel P-T trends. Whole rock major element abundances resulted from fractional crystallization of the magmas during ascent. To retrieve the compositions of the primitive melts we added to the bulk rock compositions variable amounts of magnesian olivine [Mg/(Mg+Fe) = 0.88], and derived the evolution of olivine fractionating magmas in equilibrium with mantle harzburgite. Two groups of samples were found, corresponding to the parallel P-T trends obtained from mineral / melt calculations. The highest temperature trend corresponds to samples whose calculated primitive compositions are in agreement with those obtained from peridotite melting experiments (as published in the relevant literature). Interpretation of results for rocks belonging to the general CRBG trend suggests, either: (a) that higher forsteritic content olivine should be used in the calculations; or, (b) that melt / ol / opx reactions occurred. Investigation of the CRBG primitive compositions has relevance with regard to the geodynamic evolution models of this region. We are currently undertaking melt inclusion studies of suitable CRBG samples.
V31D-08 09:24h
Basaltic Dike Propagation at Yucca Mountain, Nevada, USA
We describe simulations of the propagation of basaltic dikes using a 2-dimensional, incompressible hydrofracture code including the effects of the free surface with specific application to potential interactions of rising magma with a nuclear waste repository at Yucca Mountain, Nevada. As the leading edge of the dike approaches the free surface, confinement at the crack tip is reduced and the tip accelerates relative to the magma front. In the absence of either excess confining stress or excess gas pressure in the tip cavity, this leads to an increase of crack-tip velocity by more than an order of magnitude. By casting the results in nondimensional form, they can be applied to a wide variety of intrusive situations. When applied to an alkali basalt intrusion at the proposed high-level nuclear waste repository at Yucca Mountain, the results provide for a description of the subsurface phenomena. For magma rising at 1 m/s and dikes wider than about 0.5 m, the tip of the fissure would already have breached the surface by the time magma arrived at the nominal 300-m repository depth. An approximation of the effect of magma expansion on dike propagation is used to show that removing the restriction of an incompressible magma would result in even greater crack-tip acceleration as the dike approached the surface. A second analysis with a distinct element code indicates that a dike could penetrate the repository even during the first 2000 years after closure during which time heating from radioactive decay of waste would raise the minimum horizontal compressive stress above the vertical stress for about 80 m above and below the repository horizon. Rather than sill formation, the analysis indicates that increased pressure and dike width below the repository cause the crack tip to penetrate the horizon, but much more slowly than under in situ stress conditions. The analysis did not address the effects of either anisotropic joints or heat loss on this result.
V31D-09 09:36h
Drilling an Active Pahoehoe Lava Flow
Core-Drilling of an actively inflating pahoehoe lava flow on Kilauea Volcano, Hawaii has provided new insight into the timing and causes of widely recognized petrologic variations within individual basalt flows. Seven closely spaced and successively longer cores through the crust-melt interface, along with melt samples, were recovered from a single lava flow during inflation and then throughout final solidification. Petrologic studies of these melt and core samples have yielded time-constrained vertical profiles of whole-rock, glass and mineral chemistry; glass thermometry and vesicularity. Theodolite measurements document flow inflation to 1.96 m height over the first 9.8 hrs after emplacement, concurrent with sporadic lava breakouts in the immediate vicinity of the drill site. Down-hole depth-profiles obtained at 3.0, 4.4 and 6.5 hrs, reveal steadily increasing upper and lower crustal thickness to 25 cm, while a melt thickness of 1.1-1.2 m was sustained within the inflating flow lobe. The temperature, composition and vesicularity of the upper crust-melt interface and central melt zone remained fairly constant as the flow inflated. However, olivine accumulated toward the base of the flow during this interval of dynamic recharge. As activity waned, the flow deflated for 17 hrs to a final height of 1.56m. Depth-profiles obtained from drilling at total elapsed times of 10.3, 11.1, 27.5 and 72.8 hrs, reveal a steadily diminishing melt thickness during deflation which culminated with a 45 cm, highly viscous zone in the core of the flow. As the melt receded from the deflating lobe, the interior flow cooled at $0.5\deg$C/hr and subsequent closed-system fractionation ensued with slower cooling at $0.2\deg$C/hr. During cooling, vesicles coalesced and migrated, leaving a dense inner flow core. Static processes associated with solidification of the flow core include development of a differentiated plagioclase mush at the flow-core top, an olivine rich layer at the flow-core bottom, and upward infiltration of late-stage residual melt. The flow core was completely solidified within 300 hrs after emplacement, when two final cores were collected. These complete cores allow correlation of the final intra-flow stratigraphy with the progressive changes documented by drilling of the flow while it was active.
V31D-10 INVITED 09:48h
Field Experiments on Active Kilauea Lava Flows to Improve Cooling Models for Pahoehoe Lava Flows
Previous attempts to model the cooling of pahoehoe lava flows have shown that uncertainties in the cooling by the wind and non-equilibrium crystallization caused significant errors. New analyses of field data collected over the past decade from active pahoehoe flows on Kilauea Volcano, Hawaii, have helped reduce these uncertainties. Field measurements of the cooling of pahoehoe surfaces by the wind show that the heat transfer coefficient to the atmosphere is 45-50 W m$^{-2}$ K^{-1}$ at a wind speed of 10 m/s and a surface temperature of 500 $\deg$C. This is consistent with earlier field data, but is more than 5 times higher than predicted by theory. We also find that cooling by the wind is the most inefficient when there is a slight breeze, rather than in still air. This may be because the convective patterns that would be set up in still air are disrupted by a breeze. Another field experiment quenched small pahoehoe lobes at different points in their cooling histories by cutting them off with an ax and dumping them into a bucket of water. The lobes were instrumented with thermocouples and cooled naturally for 370-2500 s before quenching. From these lobes we have produced a series of thin sections for which we have measured cooling histories. We find that the observed crystallinity can be adequately modeled as a function of either cooling rate or undercooling, but the the abundance of pre-existing nucleation sites is a critical parameter. In general, glass is formed at cooling rates above 1-3 $\deg$C/s. Initial undercooling of the lava was approximately 2-8 $\deg$C. Finally, the drilling of a 1.5 m thick inflating pahoehoe flow provides direct views into a flow that is intermediate in size between small pahoehoe lobes and flood basalt lava flows. From this experiment, we can constrain the growth of the upper crust as a function of time, when bubbles grew and migrated in the lava, and the overpressure during inflation. These field observations are being incorporated into an improved numerical model for the cooling of pahoehoe lava flows of all thicknesses.