Introduction

Imaging of volcanic plumes permits measurement of parameters such as plume height, ascent velocity and spreading rate. These can be used in dispersal models and to constrain laboratory or theoretical studies. Previous studies have used cameras (e.g. Chouet et al. 1974; Wilson and Self 1980; Ripepe et al. 1993), standard video (e.g. Clarke et al. 2002; Formenti et al. 2003; Bluth and Rose 2004), satellite remote sensing (e.g. Holasek et al. 1996; Searcy et al. 1998) and, more recently, thermal video (Calvari et al. 2006; Harris and Ripepe 2007; Patrick et al. 2007). Thermal video has the advantage of allowing measurement of plume temperature during ascent, as well as capturing plume dynamics by day and night. Vertical ash emissions have persisted at Santiaguito’s Caliente vent (Guatemala) since 1976 (Rose 1987). Through 2004, these persisted at a rate of 0.5 to 2 events per hour sending plumes 1 to 4 km above the vent (Bluth and Rose 2004). On January 25, 2005, we imaged 25 of Caliente’s plumes using thermal video, our aim being to quantify plume dynamics and thermal properties during emission and ascent. In doing so, we use the terminology of Turner (1962; 1969) and Morton (1959) [and reviewed by Patrick (2007)] so that: (i) a jet is a high-velocity flow driven by initial momentum, (ii) a starting plume is a buoyant plume where the convecting plume front is fed by a steady feeder plume that develops below it, and (iii) a thermal is a detached vortex rising by buoyancy (Fig. 1).

Fig. 1
figure 1

Image sequences of the two plume types recorded at Santiaguito. a Type A plumes developing from individual jets to a starting plume and, finally, a rooted thermal. b Type B plume showing transition from a starting plume to a thermal

Full size image

The thermal camera and data set

We used a Forward-Looking Infrared systems ThermaCamTM (Model S40) to obtain plume data. This thermal video camera operates in the 7–13 μm range, producing 320 × 240 pixel calibrated temperature images at 30 frames per second. The camera was tripod mounted on a ridge 4.5 km south of, and ~1000 m in elevation below, Caliente. Corrections for atmospheric transmissivity were applied using field-input distance, ambient temperature and relative humidity. The pixel spatial resolution over this line-of-sight distance, and corrected for a look angle that is titled upwards by ~13°, is ~6 m, with the image covering a 1920 × 1440 m area. Given the vent location in the image, this allowed plumes to be tracked over their first ~1140 m of ascent (Fig. 1). For each emission, plume front height, width and plume apparent temperature were measured at 10 m intervals over the first 100 m of ascent, and every 20 m thereafter. Because the exact mix of gas, aerosol and ash in the plume is not known, assigning a correct emissivity is difficult, we thus use the term apparent temperature and leave our temperature uncorrected for emissivity. For comparative purposes, we divided the plumes into two types. Type A (n = 10) were relatively high and ascended beyond the top of the image; therefore having final ascent heights >1.1 km (Fig. 1a). Type B (n = 15) plumes were lower and remained within the image through the entire emission-dispersal process, so that final ascent heights were <1.1 km (Fig. 1b).

Results

In some cases emission began from vents that are arranged in a ring around the circumference of Caliente’s summit crater (Bluth and Rose 2004), to be followed by crater-wide emission. In others, crater-wide emission occurred from the start. Nine of the observed plumes were initiated by single or multiple jets from individual vents, each jet being 5 to 10 m in radius. The subsequent summit-wide emission overtook the initial jets to form a starting plume (Fig. 1). The largest plume displayed a rooted thermal morphology, defined by Patrick (2007) as having ring vortex at the base of the plume front (Fig. 1a). Primary plume front parameters were measured directly from the thermal images. These were plume front (i) vertical position at each time step, (ii) radius and (iii) apparent temperature (Fig. 2a–c). From these parameters we derived ascent velocity, spreading rate and cooling rate (Fig. 2d–f). Following Patrick et al. (2007) we used the shape of the height vs. time curve, as well as the derived acceleration values to detemine the transition from gas thrust to convective regimes (i.e., convective phases had insignificant decelerations).

Fig. 2
figure 2

Plume front parameters with ascent height measured for type A (black lines) and type B (gray lines) plumes; the two thick lines are the two plumes that generated pycroclastc flows. a Plume front position with time, b plume front radius, and c plume front apparent temperature. First order derivative of parameters given in (a) to (b) to obtain plume front ascent rate d, lateral spreading rate e and cooling rate f with height

Full size image

Plume ascent velocity and spreading rate

The velocity profiles show two distinct regions common to both plume types (Fig. 2a, d). The lower of these two regions is marked by high initial ascent velocities followed by significant deceleration within the first 20–50 m of ascent. Above this is a region marked by lower, but generally steady, ascent velocities. The change in ascent velocity can be attributed to a transition from momentum-dominated (i.e. gas thrust driven) ascent immediately above the vent, to buoyancy driven ascent thereafter. All but one of the 25 plumes imaged exhibited a gas thrust region, with maximum (at-vent) velocities ranging from 15 to 55 m s−1, with a mean of 25 m s−1 (Fig. 3a). Velocities in the buoyant region were lower and ranged from 3.5 to 15.5 m s−1, with a mean of 7.5 m s−1 (Fig. 3b). Type A plumes had the highest velocities in the buoyant region and very little deceleration. In contrast, type B plumes had lowest velocities in the buoyant region, with some deceleration (Fig. 3). Plume front radial spreading (Fig. 2b, e) also reveals two distinct zones. The first is characterized by rapid lateral spreading and extends to 50 m. Above this zone, lateral spreading rates are lower and plume front diameter increases steadily; that is plume front expansion proceeds linearly (Fig. 2b). These two zones approximately coincide with the gas thrust and buoyant regions defined by the ascent velocity (Fig. 2).

Fig. 3
figure 3

Parameters measured for type A (circles) and B (triangles) plumes. Filled circle and triangle denote pyroclastic-flow-generating plumes. Tsat height is the height at which the plume front apparent temperature drops below the camera’s saturation temperature

Full size image

Plume front apparent temperature and cooling rate

During image acquisition the camera was set at low-gain setting giving a sensor saturation temperature of 161°C. The majority of the emissions began with plume front apparent temperatures that exceeded saturation temperature. This precludes measurements of initial plume apparent temperatures and estimates of total heat content. A proxy for the ability of each plume to retain a high heat content can, however, be obtained by the noting the height at which apparent temperatures decline to below saturation (Fig. 3). Of the 25 plumes imaged, 23 dropped to apparent temperatures below saturation by 100 m. The two remaining plumes, both of which were type A, maintained plume front apparent temperatures above saturation to heights of 150 and 260 m, respectively (Fig. 3). All plumes experienced temperature decay as the plume ascended, with the rate of decay being greatest over the first 100 m, i.e. in the gas thrust region (Fig. 2c). Above 100 m, the cooling rate oscillates and shows positive cooling rates—implying an increase in plume front apparent temperature with ascent. These temperature increases correspond to multiple pulses during emission so that bursts of hot material are supplied into a generally cooling plume front, via the steady feeder plume, to temporarily reverse the cooling trend. Average cooling rates for all the plumes ranged from −0.7 to −4.0°C s−1 (Fig. 3), with type A plumes generally having the highest apparent temperatures in the buoyant region (Fig. 2c).

Comparison of two (pyroclastic-flow-generating) plumes

Two of the plumes generated short pyroclastic flows that extended a few hundred meters down a well-defined chute to the right of the image (Fig. 1). The plume parameters for these two plumes are highlighted in Figs. 2 and 3. The two plumes were quite different in character. The first was a higher (type A) plume and was easily the largest of all the plumes imaged. This plume had an at-vent velocity of 22 m s−1, and was able to maintain highest velocities in its buoyant region (10–12.5 m s−1). It also displayed the highest apparent temperatures of all plumes, maintaining saturated temperatures up to a height of 265 m (Fig. 3). The second plume was a lower (type B) plume but possessed an at-vent velocity of 38.5 m s−1. However, as with most type B plumes, it experienced significant deceleration during its buoyant phase, declining to 4 m s−1 by 800 m. The cooling rate for this plume were also higher, so that it maintained saturation temperatures only to 75 m (Fig. 3).

Discussion and conclusion

Plume dynamics over the initial 1 km of ascent were recorded for 25 vertical ash plumes at Santiaguito. For lower (<1 km high) plumes, their entire evolution could be tracked from emission to dispersal. These plumes underwent complete transitions from jets to starting plumes (Fig. 1a). As the emission ceased, the feeder plume died out and the plume head detached to form a thermal (Fig. 1b). All (but one) of the plumes had a minor (<50 m high) gas thrust region and a dominant convective region. Whether a plume ascended above 1 km did not appear to be controlled by initial exit velocity. Instead, higher (>1 km) plumes were characterized by higher average buoyant ascent velocities, lateral spreading rates and feeder plume radii (Fig. 3). Higher buoyant ascent velocities can be maintained by a steady influx of hot material from the vent, through the center of the feeder plume, and into the plume front. Larger feeder plumes may facilitate transport and insulation of hot material feeding the plume front. This, in turn, reduces cooling and maintains, or enhances, the buoyancy of the thermal by maintaining a high heat content. Our preliminary results from Santiaguito show that high-temporal resolution thermal imaging of plumes is an effective means of extracting plume ascent dynamic and apparent temperature parameters. Further imaging of a larger number plumes (at higher gain settings) will allow rigorous parameterization and understanding of plume dynamics, thermal properties, and their relationships. These can then be used to support laboratory and numerical modeling.