JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, F02029, doi:10.1029/2006JF000598, 2007 Analysis of the 2001 lava flow eruption of Mt. Etna from three-dimensional mapping M. Coltelli,1 C. Proietti,1,2 S. Branca,1 M. Marsella,2 D. Andronico,1 and L. Lodato1 Received 14 June 2006; revised 17 November 2006; accepted 21 December 2006; published 1 June 2007. [1] The 2001 Etna eruption was characterized by a complex temporal evolution with the opening of seven eruptive fissures, each feeding different lava flows. This work describes a method adopted to obtain the three-dimensional geometry of the whole lava flow field and for the reconstruction, based on topographic data, of the temporal evolution of the largest lava flow emitted from a vent located at 2100 m a.s.l. Preeruption and posteruption Digital Elevation Models (DEM) were extracted from vector contour maps. Comparison of the two DEMs and analysis of posteruption orthophotos allowed us to estimate flow area, thickness, and bulk volume. Additionally, the two-dimensional temporal evolution of the 2100 flow was precisely reconstructed by means of maps compiled during the eruption. These data, together with estimates of flow thickness, allowed us to evaluate emitted lava volumes and in turn the average volumetric flow rates The analysis performed in this paper provided, a total lava bulk volume of 40.1
106 m3 for the whole lava flow field, most of which emitted from the 2100 vent (21.4
106 m3). The derived effusion rate trend shows an initial period of waxing flow followed by a longer period of waning flow. This is in agreement not only with the few available effusion rate measurements performed during the eruption, but also with the theoretical model of Wadge (1981) for the temporal variation in discharge during the tapping of a pressurized source. Citation: Coltelli, M., C. Proietti, S. Branca, M. Marsella, D. Andronico, and L. Lodato (2007), Analysis of the 2001 lava flow eruption of Mt. Etna from three-dimensional mapping, J. Geophys. Res., 112, F02029, doi:10.1029/2006JF000598. 1. Introduction [2] On active volcanoes each new effusive eruption alters the local topography and builds new morphological features. This requires to modify existing maps for keeping topographic maps up-to-date and collecting accurate quantitative data such as flow area, volume and effusion rate. The availability of detailed maps contributes to monitoring efforts and it is valuable to understand and model the eruptive processes. Currently, several numerical lava flow models are available [Costa and Macedonio, 2005; Crisci et al., 2003; Hidaka et al., 2005; Vicari et al., 2007] with the aim of providing real-time lava flow simulation tools for predicting areas under threat. These models need to be verified and validated by simulating and checking the lava flow emplacement of test-case eruptions. In addition, to run properly the models and to give realistic output values, accurate preeruption topography and effusion rate estimates have to be used together with appropriate physical-chemical parameters of the lava [Vicari et al., 2007]. [3] Because Mount Etna is one of the most active volcanoes on the Earth, well known for its frequent lava flow 1 Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Catania, Catania, Italy. 2 Dipartimento Idraulica Trasporti e Strade, Università di Roma ‘‘La Sapienza,’’ Rome, Italy. Copyright 2007 by the American Geophysical Union. 0148-0227/07/2006JF000598 eruptions, the problem of keeping maps up-to-date is an old and critical issue. The first geological map of the volcano was published by Waltershausen [1880], and documented the presence of several lava flow fields produced by recent activity. Since then, geologists have mapped every new lava flow and have generated, in some cases, very detailed maps, such as that showing the flows of the 1910 eruption [Vinassa de Regny, 1911]. These early maps represent the evolution from a pictorial approach (the use of drawing and painting the shape of lava flows dates back to 15th century) toward topographic maps, though their geometrical content rarely included detailed altimetric information [Riccò, 1902]. [4] During the 20th century, ground surveying and aerial photogrammetry methods were used to update topographic maps during and after the emplacement of new lava flows. However, initially such techniques were not systematically adopted because they were considered too expensive for mapping the final lava morphology and, especially, for frequently updating (ideally every day) the area covered by new lava. In the last few decades, the possibility to estimate lava flow volume from topographic data, and to document the evolution of the flow field, has increasingly supported scientific and civil protection applications, such as numerical simulations of lava flow emplacement [e.g., Crisci et al., 1986; Ishihara et al., 1989; Wadge et al., 1994]. [ 5 ] The well-documented 1983 eruption of Etna [Frazzetta and Romano, 1984; Guest et al., 1987; Kilburn and Guest, 1993] initially raised the problem of uncertainty F02029 1 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION associated with lava-volume estimation from different sources, because different authors reported the same planimetric but different volumes. Frazzetta and Romano [1984], for example, calculated a volume of 100.2
106 m3 from daily measurements of lava flux near the vent; conversely Murray [1990] compared data collected by topographic field surveying to the 1969 IGM 1:25,000 contour map and estimated a volume of 78.5
106 m3. [6] Since the early 1990s, the availability of handheld GPS receivers has opened up the possibility to easily collect three-dimensional (3D) point measurements to contribute to the geological mapping of an ongoing eruption. Different 3D ground surveying techniques, for example, GPS and EDM [Calvari et al., 1994; Stevens et al., 1997] gave comparable results for evaluating the volume of the 1991 – 1993 Etna’s flow field. [7] Lava flow volume can be derived indirectly from observed or extrapolated geometrical parameters, or directly by subtracting preeruption and posteruption surfaces. The former method, also known as the planimetric approach [Stevens et al., 1999] requires measurements of the flow area, using remote sensing techniques or field data, and mean lava thickness from field surveys. The accuracy of this method strongly depends on the quality of the mapping and on the uncertainties of the thickness measurements. The availability of field-measured 3D points across the entire flow is necessary to obtain reliable results [Calvari et al., 1994; Stevens et al., 1997]. The latter method, known as the topographic approach [Stevens et al., 1999], requires a 3D reconstruction of the topographic surfaces (i.e., DEM generation) before and after the eruption. DEMs can be directly extracted from remote sensing data [Rowland et al., 1999, 2003] or derived from preexisting vector maps [Stevens et al., 1999]. The topographic approach can provide more detailed data whose accuracy can be easily assessed. [8] In this work a combination of the two approaches was utilized to evaluate the final area and volume of the 2001 Etna flow field (Figure 1). An error analysis permitted us to evaluate the accuracy associated with volume estimates derived from the different methods. The planimetric approach was also adopted to perform a multitemporal analysis of the flow emitted from the 2100 m vent (Figure 1) because syneruptive DEMs were unavailable. This temporal analysis was based on quasidaily mapping supported by photographs taken during helicopter overflights and allowed us to reconstruct the flow rate trend. [9] Several attempts to evaluate the flow rate trend of an Etna eruption have been carried out in the past [Frazzetta and Romano, 1984; Calvari et al., 1994; Behncke and Neri, 2003] without providing the necessary information to establish the methodology and data quality. In the paper we detail the methodology applied to the 2001 eruption, which required analysis of preeruption and posteruption DEMs, as well as of daily maps produced during the eruption. 2. Lava Flow Mapping and DEM Generation: A Review of Remote Sensing–Based Methods [10] A number remote of remote sensing techniques are now available for observing the evolution of an eruption in safety [Baldi et al., 2002] and collecting quantitative data useful for monitoring the morphological changes on the F02029 volcano surface. In order to monitor a specific eruption parameter (volume and area of the lava field, effusion rates, lava temperature, etc.) the observing system should be selected by considering its temporal and spatial resolution and the achievable accuracy of the measured parameters. [11] Regarding the geometrical parameters, most of the available satellite-borne systems provide data which, after being georeferenced and rectified, can be used for mapping 2D (planimetric) features, such as lava flow field limits. In this case, estimates of erupted lava volume may be obtained if thickness measurements are available from external data sources. If 3D coordinates of ground points can be measured, it is possible to directly estimate the lava volumes emplaced over a known time period. This can be achieved by reconstructing the surface topography through a Digital Elevation Model (DEM), for example by using SAR interferometry (InSAR) [Franceschetti and Lanari, 1999] techniques or processing stereopairs acquired by medium- to high-resolution sensors [Curlander and McDonough, 1991; Ridley et al., 1997]. [12] Limiting factors for the use of satellite data are not only the generally low spatial resolution (with the exception of the recently available high-resolution commercial satellites) but also the constraints stemming from a predefined acquisition schedules. Satellite overpasses may, for example, coincide with cloudy periods or may miss short-lived activity altogether. [13] A more flexible and accurate alternative to the satellite-based methods for DEM generation over large areas is aerial data collection using Light Detection And Ranging (LIDAR) systems [Fouler, 2001] and photogrammetric cameras [e.g., Baldi et al., 2002]. Both of these methods permit the acquisition of a large number of 3D points and generation of high-resolution DEMs (space grid density down to few point per square meters). The aerial surveys can be repeated, if logistical and weather conditions are favorable, many times during an eruption and thus appropriate for monitoring the spatial evolution of lava flows [e.g., Baldi et al., 2005; Honda and Nagai, 2002]. Recently, the application of LIDAR systems is increasing [e.g., Mouginis-Mark and Garbeil, 2005] owing to their capability of acquiring dense and accurate 3D point networks which permit accurate representation of terrain features, with less processing time than the photogrammetric technique. [14] A summary of the main characteristics of the available techniques to directly extract medium to high spatial resolution DEMs is reported in Table 1. If a DEM cannot be directly extracted from data collected during dedicated surveys, preexisting contour maps can be digitized and analyzed as described in this work and, for example, by Stevens et al. [1999]. In this case, the planimetric and vertical accuracies should be derived from the map scale. 3. The 2001 Eruption 3.1. Eruption Narrative [15] The narrative of the 2001 eruption is hereby reported with the support of six maps (Figure 2) which track the complex temporal evolution of the flow field. The six maps are drawn from digital aerial images that were rectified to posteruption 1:10,000 aerial orthophotos (Figure 1) of the Provincia Regionale di Catania (PRC). 2 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION Figure 1 3 of 18 F02029 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Table 1. Main Characteristics of the Techniques Used for Direct Extraction of Three-Dimensional Points Useful for DEM Generation Ground Resolution Elevation Accuracy Limitations Advantages Satellite InSAR 10 – 20 m 5 – 10 m Satellite stereopairs 1 – 10 m 1 – 10 m Aerial Photogrammetry 1 m to a few meters 1 m to a few meters all-day/weather acquisition DEM and orthophoto over large areas flexible schedule Airborne LIDAR 1 m to a few meters 1 m to 1 – 2 m geometrical and radiometric constraints fixed time schedule daylight acquisition daylight/weatherdependent acquisition weather-dependent acquisition 1 m to few meters 1 m to few meters Helicopter Photogrammetry/LIDAR These aerial images were then georeferenced for outlining the lava flow margins. [16] During the first three days of the eruption, activity evolved rapidly. The eruption began on 17 July with the almost simultaneous opening of several eruptive fissures on the south and northeast flanks of the volcano, extending between the summit and 2100 m a.s.l. (Figure 1). On the basis of structural data and the geochemical composition of the lavas [Calvari, 2001; Corsaro et al., 2007], the fissures were subdivided into two main groups each belonging to different eruptive systems (Figure 1). The Upper Fissure System (UFS) included those fissures that opened (1) at the foot of the South-East Crater (UFS1) and on the flanks of the South-East cone (UFS4 and UFS5); (2) on the southern flank between 2780 and 2640 m a.s.l. (UFS2); and (3) in Valle del Leone, on the northern flank (UFS3). The Lower Fissure System (LFS), the most hazardous owing to its location in relation to human settlements, opened on the southern flank at 2550 m a.s.l. (LFS2) and 2100 m a.s.l. (LFS1). 3.1.1. Activity During 17– 18 July (Figure 2a) [17] The eruption began around 7:00 on 17 July, when the eruptive fissure UFS1 opened at the base of the South-East Crater, feeding a lava flow that moved SE. Late in the evening two new fissures (together called UFS2) opened at about 2700 m a.s.l., producing lava flows that spread across the Piano del Lago. A few hours later, at 02:20 on 18 July, the LFS1 fissure opened between 2100 and 2150 m a.s.l., close to Mt. Calcarazzi. The most active effusive vent of the 2001 eruption became established at the lower end of this fissure, emitting a lava flow that extended around Mt. Silvestri and rapidly reached the SP92 road. At the same time, intense phreatomagmatic activity began at the higher portion of the LFS1 fissure and lasted three days. 3.1.2. Activity on 19 July (Figure 2b) [18] While the UFS1 lava flow continued to propagate toward the Belvedere area, the UFS2 flows moved south- limited area coverage weather-dependent acquisition partial penetration under vegetation flexible schedule ward and reached 2400 m a.s.l., close to La Montagnola. Late in the afternoon, two pit craters (LFS2) opened in the Piano del Lago area at 2550 m a.s.l., where an increasing phreatomagmatic ash emission began [Taddeucci et al., 2002; Scollo et al., 2007]. At this time, the main flow from the LFS1 vent had extended below 1450 m a.s.l. 3.1.3. Activity During 20– 23 July (Figure 2c) [19] Early on 20 July, the eruptive fissure UFS3 opened at 2600 m a.s.l. in Valle del Leone, feeding a new lava flow. On 23 July one of the flows fed by the UFS2 fissures continued to move approaching the Rifugio Sapienza. A lava flow extended from the UFS1 fissure toward the Valle del Bove. In addition two new short fissures opened on the southern (UFS4) and northern flanks (UFS5) of the SouthEast Crater. Both fed flows at modest effusion rates. The lava flow fed by LFS1 had extended to 1048 m a.s.l. by the early afternoon of 23 July. 3.1.4. Activity During 24– 25 July (Figure 2d) [20] Between 24 and 25 July, most of the flows emitted from the UFS had reached their maximum lengths. At UFS2, continuous overflows covered the upper portions of the previously emplaced lava flow field above 2400 m a.s.l. On 24 July, powerful Strombolian activity gradually built a scoria cone at 2550 m a.s.l. (UFS2). Weak lava effusion characterized activity at UFS5 on 25 July. The UFS3 lava flow in Valle del Leone reached 2100 m a.s.l., partially covering Mt. Rittmann, and the effusive activity at UFS4 ceased. The lava flow originating at the LFS1 vent attained its lowest elevation of 1040 m a.s.l., while a few overflows piled up on the proximal area of the flow field. 3.1.5. Activity During 26– 27 July (Figure 2e) [21] At the LFS2 fissure, on 26 July, new lava flows began to issue from the SW base of the cone that had developed on this fissure segment, reaching the SP92 road during the evening and connecting with the LFS1 lava field at 1840 m a.s.l. On 27 July a lava flow emerged from a new vent located at the southern tip of the LFS2 fissure and Figure 1. Posteruption orthophotos obtained from the 2001 photogrammetric surveys showing relevant topographic features, lava flows limits (red), eruptive fissures (yellow), scoria cones (light blue), and GPS cross sections (green). The Upper Fissure System (UFS) consists of five fissures: UFS1: 2950 m a.s.l.; UFS2: 2780– 2640 m a.s.l.; UFS3: 2600 m a.s.l.; UFS4: 3050 m a.s.l.; UFS5: 3050 m a.s.l. The Lower Fissure System (LFS) consists of two fissures: LFS1: 2100 m a.s.l.; LFS2: 2550 m a.s.l. The summit craters are: Voragine (VOR), Bocca Nuova (BN), Southeast Crater (SEC) and Northeast Crater (NEC). VdB, Valle del Bove; VDL, Valle del Leone. Contour lines are drawn every 200 m between 1000 and 3200 m a.s.l. Insets on the left locate Mt. Etna in the eastern part of Sicily and the study area on the volcano edifice. 4 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Figure 2. Flow field temporal evolution on 18, 19, 23, 25, 27, and 31 July traced on a shaded relief representation of the 1999 DEM. Eruptive fissures (1) are shown in yellow, and active (2) and inactive (3) flows are shown in red and orange, respectively. Contour lines are drawn every 500 m between 1500 and 3000 m a.s.l. extended eastward into the Valle del Bove. Explosive activity also built a scoria cone around the higher portion of the LFS1 fissure. At the same time, a marked decrease in the effusion rate caused the most advanced lava front to stop and the emplacement of new flows that overlapped the older flows down to 1400 m a.s.l. On 27 July, a new lava flow extended SSE from UFS1 toward the old Cisternazza pit. At the same time the effusive activity at UFS2 shifted down to 2640 m a.s.l. and ceased at the UFS5 fissure. 3.1.6. Activity During 28– 31 July (Figure 2f) [22] By 28 July, the UFS2 lava flow had extended 2 km SW to reach Mt. Nero. Two new lava flows originated from LFS2. The first extended from the NE base of the scoria cone for a short distance eastward. The second extended southwestward from the NW base of the scoria cone. The flow from the southern tip of the LFS2 fissure was still being fed, but its front appeared immobile. The most advanced front of the LFS2 western lava flow reached 5 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION 1700 m a.s.l. on 29 July and stopped. Likewise, the effusive activity at UFS3 completely ceased on 30 July, at which point the front of the UFS1 flow directed toward the Belvedere appeared motionless. Lava flows fed by LFS1 had active fronts that extended from 1180 m a.s.l. on 28 July to 1060 m a.s.l. on 30 July. At that point a lateral eastern branch developed at 1490 m a.s.l. (close to Mt. Gemellaro) which moved toward Mt. Grosso. A new overflow began from the SW base of the 2550 m scoria cone (LFS2), again threatening Rifugio Sapienza on 31 July. 3.1.7. Activity During 1 – 9 August [23] The eruption began to wane following 1 August. The LFS1 lava flow became tube-contained downslope of 1700 – 1600 m a.s.l. for a distance of about 500 m, before emerging and flowing to 1080 m a.s.l. Several ephemeral vents formed and emitted lava flows that were less than one hundred meters long. On 2 August two ephemeral vents, located west of Mt. Gemellaro at 1470 and 1460 m a.s.l., produced lava flows that moved toward Mt. Grosso and southwestward, respectively. While the main flow reached 1200 m a.s.l. on 7 August, the lava flow moving eastward toward Mt. Grosso stagnated at about 1240 m a.s.l. The lava flow extending from the LFS2 scoria cone suffered a marked decrease in activity on 1 August and stopped on 2 August. Also the explosive activity at this cone became drastically reduced, being replaced by minor ash emission that entirely ended on 6 August. The front of the UFS1 flow directed toward the Belvedere was moving slowly on 1 August and stopped on 2 August. The flows from UFS2 continued to propagate very slowly toward Mt. Nero, but their fronts stopped on 7 August. However, several overflows remained active until 9 August. Finally, on 8 August the Strombolian activity at the top of the LFS1 fissure ceased and active overflows remained confined above 1900 m a.s.l. The eruption ended during the late evening of 9 August and its final flow field is delimited in Figure 1. 3.2. Morphological Features of the Lava Flow Field [24] The lava flow field of the 2001 eruption was mapped using a series of color 1:10,000 scale PRC (Provincia Regionale di Catania) orthophotos acquired on 3 December 2001 (Figure 1). The color orthophotos allowed detailed mapping of the lava flow field and description of its main morphological features. In this way it was possible to distinguish the 2001 lava flows from the adjacent fresh lavas (i.e., those of the 2000, 1999, 1989, 1985 and 1983 eruptions) and to reconstruct the 2001 flow field boundaries with a high accuracy. The only limitation of the orthophotos was snow cover toward the volcano summit. In particular, the reconstruction of the lava flows emplaced in the summit area, i.e., above 2600– 2800 m a.s.l., were generally characterized by a planimetric error of up to 10 m as a consequence of the poor orthophoto quality and snow cover. Below 2600 – 2800 m a.s.l., however, the 2001 lava flows were clearly distinguishable and could be mapped with a planimetric error of less than 5 m. [25] The final lava flow field produced during the 2001 eruption was the result of lava flow emplacement related to seven fissure systems. The LFS1, LFS2-west, UFS1 and UFS2 vents fed compound lava flow fields. Conversely the LFS2-east, UFS3, UFS4 and UFS5 vents generated simple lava flows (terminology of Walker [1971]). F02029 [26] The long-lasting (23 days) effusive activity at the LFS1 vent produced a narrow lava flow 6.4 km long, with a maximum width of 545 m, that reached 1040 m a.s.l. (Figure 1). This lava flow field was characterized by aa morphology and a large axial lava channel, up to 90 m wide, developed between the vent and about 1700 m a.s.l. The lava flow field down to 1700 m a.s.l. was characterized by the superimposition of several lava flow units that were mainly related to overflows from the main lava channel. Between 1460 and 1080 m a.s.l. the main lava channel was divided into several secondary channels, with maximum widths of about 100 m. These formed during the uphill regression of the lava flows that began on 26 July. The frontal portion of the lava flow field was produced by the superposition of the lava flow units that drained from the 1460– 1080 m lava channels. A secondary branch that developed between 1400 m and 1250 m a.s.l., surrounding Mt. Grosso, was formed by the juxtaposition of several lava flow units between 30 July and 6 August. [27] The LFS2 fissure comprised four effusive vents located at the base of the large scoria cone built by the strombolian activity, as well as at the southern tip of the fissure. The western vents in this system fed prolonged effusive activity that lasted 15 days and generated a lava flow field that was 3.3 km long, had a maximum width of 265 m and reached 1720 m a.s.l., where it partially overlapped the LFS1 lava flow. This narrow lava field was built by the superposition of several lava flow units that filled a gully on the west slope of La Montagnola down to Rifugio Sapienza. The flow units were supplied from lava channels extending from 2500 m a.s.l., near the vent, to 1900 m a.s.l. In general they showed aa morphology and well developed flow fronts down to a break in slope at 1900 m a.s.l. [28] The UFS1 vent, located at the south base of SouthEast Crater, formed a fan-shaped lava flow field that was 2.7 km long and 430 m wide (Figure 1). It was active for 14 days and had aa morphology. The lava flow field was the result of the juxtaposition of several single flow units that extended eastward down the western wall of the Valle del Bove and southward (UFS1-LB) toward the LFS2 scoria cone, partially overlapping the UFS2 lava flow. Other flow units piled up in the central portion of the lava field close to the vent. [29] The 750-m-long UFS2 (2780 – 2640 m a.s.l.) fissure comprised a small spatter cone at its upper tip and was characterized by lava emission from different points. In 23 days of activity, it formed a lava flow field that was 4.1 km long, 360 m wide, and which reached 1890 m a.s.l., with the flow front reaching a point close to Mt. Nero degli Zappini (Figure 1). The lava flows had aa morphology with lava channels related to the emplacement of single flow units that piled up in the central portion of the lava flow field. The lateral and frontal portions were characterized by single lava flow units that partially overlapped the lava flow field from the LFS2 west-vents. [30] Simple lava flows were emitted from vents that were active for less than 10 days. In particular, the LFS2 eastvents generated two distinct lava flows, 0.78 and 2 km long, respectively, and both less than 150 m wide, that developed along the western wall of the Valle del Bove. The second one, emitted on the southern tip of the fissure, reached the Valle del Bove floor at about 1785 m a.s.l. (Figure 1). The 6 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION UFS3 fissure formed an arc that followed the morphology of the Valle del Leone wall (Figure 1). Two emission points established along this fissure generated a narrow lava flow that was 2 km long, had a maximum width of 170 m and which reached 2070 m a.s.l. in Valle del Bove. This flow had aa morphology with a small axial lava channel. The UFS4 vent, located on the SE flank of South-East Crater, formed a lava flow with aa morphology that was 1.1 km long, less than 150 m wide and which extended to 2670 m a.s.l. The UFS5 vent, on the NE flank of South-East Crater, generated a 0.54-km-long lava flow with aa morphology (Figure 1). 3.3. Syneruption and Posteruption Field Data Collection [31] During the 2001 eruption, daily surveys were carried out to map the propagation of ground surface fractures and eruptive fissures as well as the evolution of the lava flows. The collected data were plotted onto the 1:10,000 vector map issued in 1999 by the PRC. This allowed us to produce preliminary estimates of the daily areas covered by lava flows and to evaluate the rate of advance of the flow fronts. [32] The lava flows were mapped using digital photos acquired during helicopter overflights and hand-held GPS measurements collected during ground surveys along the active flow margins and fronts. At each GPS checkpoint we measured the lava flow thickness using a laser rangefinder with one meter accuracy. [33] In addition, we carried out seven estimates (on 18, 20, 22, 24 July, August 1, 4, 8) of the effusion rate close to the LFS1 vent by measuring the main channel width (w) along with the depth of the molten lava inside the channel (d) and a flow surface velocity (v). Effusion rate Er was estimated from the relation Er = wdv as described by Frazzetta and Romano [1984], Guest et al. [1987] and Calvari et al. [1994]. The measurements were all performed in an area of flat morphology, immediately below the vent, that did not show significant changes during the eruption. We evaluated the depth of the lava channel from the preeruption and posteruption topography. The lava channel depth was considered constant, except for possible thermomechanical erosion effects on the lava channel floor, that we assumed limited because of the flat substrate. The maximum flow surface velocity was determined by measuring the time taken by a marker at the center of the flow to travel between two selected natural targets. The distance between the two targets, the channel width and the depth of the molten lava in the main channel were measured with the laser rangefinder. Several marker speeds were taken during these experiments to obtain a stable average value. Effusion rates between 30 m3/s toward the beginning of the eruption and 1 m3/s at the end were obtained. [34] During September 2001, one month after the eruption ended, a survey was completed around the margins of the entire lava flow field to measure the final thickness of the flows. As part of this survey, the average thicknesses were obtained for every flow. These were used to calculate the volume of those lava flow field portions, located above 2700 m a.s.l., where the DEM data are not reliable and/or updated. [35] Finally, for the lava flow field generated by the LFS1 vent, two flow-transverse sections were carried out using a F02029 kinematic GPS receiver (Trimble 4700 Geodetic Surveyor). This allowed us to measure the local thickness of lava flows with decimeter precision and to obtain the shape and size of lava channels. The first section (S-S’ in Figure 1) was located in the upper portion of the lava flow field at 1890 m a.s.l. close to Rifugio Sapienza, about 800 m below the vent. This section also crossed the main lava channel. The second section (R-R’ in Figure 1) was located at 1065 m a.s.l., close to Mt. Rinazzi, about 50 m upslope from the lava front in an area with a nearly flat morphology, characterized by the accumulation of lava flows extending from the main lava channel. 4. Topographic Analysis [36] The volume and morphology of the flows forming the 2001 lava field were extracted using both topographic and planimetric approaches as described below. 4.1. Data Collection [37] The 1999 vector map of the PRC was chosen to characterize the preemplacement topography because it was based on an aerial survey performed on 9 November 1998. The posteruption map of PRC was obtained from an aerial survey performed on 3 December 2001. The two vector contour maps were derived from photogrammetric surveys, whose original data were not available to us. The map scales are 1:10,000 for the 1999 and 1:2000 for the 2001. The maps contain spot height data and contour lines with intervals of 10 and 2 m for 1999 and 2001, respectively. Contour lines every 10 m were extracted from both contour maps in order to provide consistent data for DEM extraction. Both maps, originally referred to the national projection system (GAUSS-BOAGA-Datum Roma40), were converted into the UTM-WGS84 coordinate system by applying the necessary transformation. [38] The 1999 map covers the whole province of Catania, subdivided in 7.5
5.5 km tiles, whereas the 2001 is limited to the eruption area and has an irregular shape (Figure 3). Their comparison revealed a geometric inconsistency both in the planimetric and vertical components, probably due to an inaccurate photogrammetric processing of the 2001 map. A procedure for improving the matching with the 1999 map was thus applied to the 2001 map before DEM extraction, as described in section 4.3. 4.2. DEM Extraction [39] The TIN (Triangular Irregular Network) method, based on Delaunay triangulation [Lee and Schachter, 1980], was utilized to interpolate elevation data. The TIN method partitions a surface into a set of contiguous, nonoverlapping triangles. A height value is recorded for each triangle node. A mask delimiting the 2001 map area was drawn and the TIN method was applied across this mask to avoid triangulation in areas with no height data. [40] The DEMs interpolated from the TIN were used to reconstruct the 1999 and 2001 topography in a grid format, which is more appropriate for conducting spatial analysis, such as volume estimation. The DEM grid size was set to 10 m which was considered appropriate given the data 7 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Figure 3. Shaded relief images of the (left) 1999 and (right) 2001 DEMs and tiles that comprise the corresponding cartographies. spacing and the minimum flow widths, i.e., about 50 m for the main channels and 20 m for the front fingers. 4.3. Improvement in DEMs Coregistration by Residual Analysis [41] A detailed comparison of the two DEMs was carried out on the whole map area in order to evaluate the method accuracy and check their coregistration. Elevation residuals outside the lava flow margin (terrain residuals) were used to assess horizontal and vertical misalignments between the two DEMs. Figure 4 shows the elevation differences on the whole map area and the histograms represent the distribution of the terrain residuals. Terrain residuals in the upper portion of the map are not useful for assessing the method accuracy owing to the presence of lava flows emplaced between 1999 and 2001; thus the following analysis is limited only to the rectangular area delimited on Figure 4a. [42] The terrain residuals in Figure 4a show the presence of horizontal misalignments and vertical shifts that were particularly severe at the southern edge of the map. The residuals are bimodally distributed, showing a first peak around zero and another between 15 and 10 m (Figure 4d). In order to improve the matching in the southern portion (rectangular box in Figure 4b), the 1999 and 2001 maps were superimposed, tile by tile and corresponding points were used to estimate the rotation and translation parameters and the vertical shift (about 10 m) to be applied to the 2001 map. The extracted 2001 DEM was compared to the 1999 DEM, resulting in the residual map of Figure 4b which shows a symmetric distribution of the terrain residuals (Figure 4e) having a mean value of 1.01 m and a standard deviation of 4.15 m. Unfortunately, such uncertainties have the same magnitude as the expected lava thickness, thus additional improvements were applied before computing lava volumes. [43] The procedure consisted of (1) definition of masks with homogeneous terrain residuals, (2) evaluation of average terrain residuals inside every mask and (3) subtraction of the residual mean values from the 2001 DEM. The 2001 DEM was again compared with the 1999 DEM resulting in the residual map of Figure 4c which shows a symmetric distribution of the terrain residual (Figure 4f) with a mean value of 0.25 m and a standard deviation of 2.69 m. This result is in accordance with the expected elevation accuracy (about 2 m) of a 1:10,000 scale map. [44] To estimate DEM quality using external data, two GPS cross sections (SS’ and RR’ in Figure 1) were compared with the corresponding sections extracted from the 2001 DEM. The GPS- and DEM-derived cross sections show good agreement, although some small-scale height 8 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Figure 4. Elevation residual analyses. (a) First evaluation between the 2001 and 1999 DEMs. The black box limits the selected area for residual distribution study. (b) Second evaluation after rotation, translation and height correction of the 2001 DEM. The black box limits the area where the 2001 cartography was modified. (c) Final evaluation, between the 2001 DEM, after masks application, and the 1999 DEM. Terrain residual and flow height color scales are the same for the three evaluation steps. White flows are those for which the planimetric volume evaluation was carried out. Histograms on right show terrain residual distributions (class interval 1 m): (d) first evaluation, (e) second evaluation, and (f) final evaluation. variations did not appear on the DEM-derived profiles (Figure 5). The estimated difference between the two data sets had a mean of 0.99 m with a standard deviation of 1.86 m for the SS’ profile, and a mean of 0.64 m and a standard deviation of 2.09 m for the RR’ profile. [45] Unfortunately, the estimated accuracies, although in good agreement with that obtained from terrain residual distribution, could not be applied to the whole data set because they were only valid for the LFS1 flow. Thus the terrain residual standard deviation of 2.69 m was adopted to represent the vertical accuracy of our lava thickness calculations. 4.4. Lava Flow Volume Evaluation [46] Lava flow volumes were calculated by subtracting the 1999 and 2001 DEMs in regions where the two data sets were considered reliable and updated. DEMs are not updated in the area covered by the lava emitted after the survey date and before the 2001 eruption. DEMs are reliable where the terrain residuals, evaluated after the improvement in DEMs coregistration, are characterized by a sufficient accuracy (i.e., comparable with the standard deviation of 2.69 m). On the contrary, the volumes of flows covering regions where the quality of the DEMs provided to be not acceptable or where the DEMs are not updated were obtained by multiplying flow area by the corresponding average lava thickness. The equations used to estimate the volumes and the relative standard deviations are given in Appendix A. [47] The DEM subtraction technique could only be applied to flows located below 2700 m a.s.l., i.e., those emitted from the LFS1 and LFS2 vents, the lower branch of that emitted from the UFS1 vent and the upper branch of Figure 5. Comparison between the GPS and DEM derived cross sections located near Rifugio Sapienza (SS’) and Mt. Rinazzi (RR’). See Figure 1 for locations. 9 of 18 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 F02029 Table 2. Lava Flow Volumes and Areas Evaluated by Comparing the 1999 and 2001 DEMsa Lava Flow Cut Area,
106 m2 Unchanged Area,
106 m2 LFS1 LFS2* UFS1-LB UFS2-UB 0.02 0.92% 0.01 1.43% <0.01 0.68% <0.01 0.78% <0.01 0.10% <0.01 0.02% <0.01 0.07% 0.01 0.39% Fill Area,
106 m2 1.93 0.90 0.14 0.76 98.98% 98.55% 99.25% 98.83% Total Area,
106 m2 Total Volume,
106 m3 Standard Deviation of Total Volume,
106 m3 Relative Error, % 1.95 0.91 0.15 0.77 21.40 6.39 0.82 4.71 0.37 0.16 0.05 0.12 1.7 2.5 6.1 2.6 a Both the Effective and percent values are reported for the areas. LFS2* includes the volume of the UFS2 lower branch (UFS2-LB in Table 3). that emitted from the UFS2 vent (respectively UFS1-LB and UFS2-UB in Table 2). Analysis of the UFS1-LB flow was complicated by the fact that it partially overlapped the UFS2-UB, similarly the front of the LFS2 flow partially overlapped the LFS1 flow. Because the overlapped areas were very small, the errors due to not considering the UFS2UB volume lying under the UFS1-LB and the LFS1 volume lying under the LFS2 flow were considered negligible. [48] Volume estimation by DEM subtraction (Table 2) was carried out using a mass balance analysis, included in a GIS toolset, and permitted evaluation of surface loss (cut) and gain (fill). In order to restrict the volume computation to the flow field area, the lava flow field limits, mapped on the 2001 orthophotos, were used to mask the 2001 DEM and dagala (Sicilian name for areas not covered by lava within the flow borders) limits were taken into account. The areas with zero or negative elevation change within the lava flow field limits were not considered in the volume estimation. The elevation changes in these areas most probably were within the DEM uncertainty of 2.69 m, or the flow limits were not correctly defined. However, their contributions represent less than 2% of the total flow areas (Table 2), so that their exclusion has a negligible impact on the volume estimation. [49] The volumes of the other lava flows cannot be estimated from the DEM subtraction. Lava flows erupted from the fissures that opened on the South-East Crater cone (UFS5 and UFS4) and from the UFS1 vent (excluding its lower branch UFS1-LB), overlap lava emitted between late 1998 and 2001. The Valle del Leone lava flow (UFS3) lay within an erroneously georeferenced part of the 2001 map. Volumes for these flows (Table 3) were thus estimated by means of the planimetric approach, that is by multiplying the corresponding areas by their average thicknesses, obtained from levée heights measured after the end of the eruption. Moreover the lower branch of the UFS2 flow (UFS2-LB in Table 3) is partially covered by lava emitted from the LFS2 vent. Its volume was evaluated by multiplying its area, reconstructed from the aerial photos, by its thickness, estimated from cross sections extracted outside and inside the overlapping area. [50] Table 4 reports the total volumes obtained for the seven flows composing the 2001 lava field by combining the results of Tables 2 and 3. The volume of the whole lava field was estimated to be 40.1
106 m3; about 53% of this volume (21.4
106 m3) was emitted from the LFS1 vent. 5. Reconstruction of the Temporal Evolution of the LFS1 Lava Flow [51] Helicopter surveys were carried out almost every day during the 2001 eruption to collect digital photos. These allowed us to reconstruct the lava flow evolution in plan view and integrating field data, to estimate partial and cumulative volumes. Our attempt to perform a daily reconstruction of the lava flow evolution is unfortunately limited to the lava flow emitted from the LFS1 vent because the field mapping performed during the eruption was not detailed enough (in space and in time) to extend it to the whole lava flow field. 5.1. Daily Map Preparation [52] Daily maps were drawn on the basis of the photo availability, quality and usefulness (in Table 5 ‘‘not useful’’ means that no significant modifications of the lava flow had occurred since the previous mapping). To check the mapping accuracy, a retroactive procedure was carried out whereby every map was cross-checked. This involved Table 4. Average Thicknesses, Areas, and Volumes of the Seven Composite Flows Forming the 2001 Lava Fielda Table 3. Average Thicknesses, Areas, and Volumes (Evaluated by Means of the Planimetric Approach) of the UFS1 Upper Branch, UFS2 Lower Branch, and the UFS3, UFS4, and UFS5 Flowsa Lava Flow UFS1-UB UFS2-LB UFS3 UFS4 UFS5 a Area, Average Volume,
106 m2 Thickness, m
106 m3 0.70 0.19 0.22 0.09 0.03 7.0 6.0 7.0 2.9 3.3 4.87 1.14 1.53 0.26 0.10 Standard Deviation of Relative Volume,
106 m3 Error, % 1.18 0.35 0.51 0.13 0.07 24.2 30.7 33.3 50.0a 70.0a These large relative errors are due to areas and thicknesses values being of the same magnitude of their associated uncertainties. Lava Flow Area,
106 m2 Average Thickness, m Volume,
106 m3 LFS1 LFS2 UFS1 UFS2 UFS3 UFS4 UFS5 Total 1.95 0.91 0.85 0.96 0.22 0.09 0.03 5.01 11.0 5.8 6.7 6.1 7.0 2.9 3.3 8.0 21.40 5.25 5.69 5.85 1.53 0.26 0.10 40.08 Standard Deviation of Volume,
106 m3 0.37 0.15 0.97 0.12 0.51 0.13 0.07 Relative Error, % 1.7 2.8 17.1 2.0 33.2 a LFS2 volume is the difference between LFS2* and UFS2-LB; UFS1 volume is the sum of UFS1-LB and UFS1-UB; and UFS2 volume is the sum of UFS2-UB and UFS2-LB. 10 of 18 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Table 5. List of the Vector Maps and Helicopter Photo Sets Available for the LFS1 Lava Flow Datea Vector Map Helicopter Photo 17/07/01 18/07/01 19/07/01 20/07/01 21/07/01 22/07/01 23/07/01 24/07/01 25/07/01 26/07/01 27/07/01 28/07/01 29/07/01 30/07/01 31/07/01 01/08/01 02/08/01 03/08/01 04/08/01 05/08/01 06/08/01 07/08/01 08/08/01 09/08/01 no lava flow yes yes yes no yes not useful no not useful yes not useful yes not useful yes not useful not useful yes not useful yes not useful yes yes no yes yes yes yes yes no flight yes photos shot too far photos not utilizable yes yes yes yes flow front photos not utilizable yes yes yes yes yes yes yes yes yes flow front photos not available yes a Date is given as dd/mm/yy. F02029 starting with the final map (9 August) obtained from the orthophotos, checking the previous map against it and moving backward in time through the sequence to the first map, corresponding to the flow of 18 July. [53] The daily map reconstruction permitted us to mark off active areas inside the lava flow after 26 July, when the regression of the active flow front began. Information on flow front position was also obtained from the daily INGV Sezione di Catania reports and utilized as an additional check. Figure 6 shows the temporal evolution of the LFS1 lava flow between 18 July and 09 August. On the basis of the daily maps the flow emplacement can be divided in three phases. The 1st phase (18 –26 July) involved lava flow lengthening. The second phase (28 July to 2 August) and third phase (4 – 9 August) respectively revealed slow and then fast regression of the active flow fronts, accompanied by development of minor branches. 5.2. Daily Volume Evaluation [54] The planimetric approach was utilized to evaluate the daily volumes of the LFS1 flow. Active flow areas were measured on the daily maps (Figure 6) while the daily average thicknesses were derived from a combined analysis of syneruption and posteruption data. Syneruption thicknesses were mainly measured on the flow front whereas final flow thicknesses, extracted from the 2001 DEM, were available for the whole lava flow. Figure 6. LFS1 lava flow temporal evolution. Colored areas are active while, after 26 July, white areas inside flow limits are not active. 11 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Figure 7. LFS1 lava flow final thickness. (a) Black boxes limit the seven zones (defined on the basis of both the lava flow emplacement history and the analysis of the flow final morphology) in which the flow was divided. (b) Dotted lines are the 20 cross sections tracked along the flow. Cross sections on the eastern branch are shorter than those on the main flow in order to limit the thickness evaluation to the secondary flow. [55] On the basis of both the lava flow emplacement history (see section 3.1) and the analysis of the flow final morphology, the LFS1 flow was divided into seven homogeneous zones (Figure 7a). The quantitative analysis of the flow evolution was then performed separately for each zone, providing the results (daily active areas and average thicknesses) listed in Tables 6 and 7, respectively. [56] Daily thicknesses were evaluated by examining twenty cross sections distributed along the flow (Figure 7b). Some of these sections (Figure 8) allowed us to extract the thickness of the first emplacement unit, corresponding to the flow maximum planimetric expansion reached on 26 July. On the cross sections in Figure 8 the widest zones, having a thickness of 10– 15 m, correspond to the first emplacement phase (18 – 26 July). Overlapping layers and localized accumulation peaks are related to the piling up of flow units during the active front regression (28 July to 9 August). During this period the active areas were restricted to the 12 of 18 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Table 6. Daily Evaluation of Active Areas on the Seven Zones Composing the LFS1 Flowa Active Area,
106 m2 Date 18/07/01 19/07/01 20/07/01 22/07/01 26/07/01 28/07/01 30/07/01 02/08/01 04/08/01 06/08/01 07/08/01 09/08/01 Zone Zone Zone Zone Zone Zone Zone 1 2 3 4 5 6 7 0.10 0.10 0.10 0.10 0.11 0.07 0.07 0.06 0.10 0.08 0.06 0.08 0.02 0.24 0.25 0.25 0.27 0.18 0.18 0.15 0.18 0.15 0.10 0.00 0.00 0.07 0.29 0.36 0.43 0.41 0.47 0.36 0.42 0.42 0.22 0.00 0.00 0.00 0.00 0.13 0.14 0.15 0.14 0.15 0.08 0.08 0.00 0.00 0.00 0.00 0.00 0.16 0.26 0.08 0.06 0.18 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.28 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.03 0.17 0.10 0.00 0.00 Total Active Area,
106 m2 0.12 0.40 0.54 1.02 1.54 0.88 0.94 0.93 0.95 0.84 0.38 0.08 a Date is given as dd/mm/yy. Last column shows the daily total active area, i.e., the sum of the seven active areas. central portion of the channel zone (Figure 6), thus the flow thickening is limited to this part as evidenced by the reduced width of the overlapping layers (Figure 8). [57] Considering that the lava flow undergone a continuous expansion between 18 and 26 July, the daily thicknesses, in every zone, were supposed to be equal to those observed at 26 July (Table 7). [58] After 26 July the lava began to pile up on the older flow, thus the additional thickness values were simply added to those of 26 July. The seven zones experienced very different emplacement histories (Table 7). Zone 6 was not active after 26 July and so underwent no additional thickness change. Zones 3 to 5 show a progressive decrease in additional deposition as a consequence of the active front regression. Zone 7 corresponds to the eastern branch of the flow that formed between 30 July and 6 August, and records only a single flow event. Finally, zone 1 did not experience additional emplacement until 9 August, when two lateral branches overflowed near the vent. [59] Daily volumes, evaluated in the seven zones, are shown in Table 8; the first five rows (cumulative volumes) correspond to the lava accumulated from the beginning of the eruption, while the last seven rows (additional volumes) quantify the lava added between two consecutive periods on the top of the older flow. [60] In order to verify the correctness of the performed reconstruction we compared, in Table 9, reconstructed versus DEM evaluated average thicknesses. Reconstructed thicknesses are those derived from the analysis of daily thicknesses and active areas. For each zone we compute an average thickness by dividing the final volume, derived from Table 8, by the corresponding area. These reconstructed thicknesses, as well as the final volumes, are in a good agreement with those obtained from DEM comparison (observed thicknesses and volumes). 5.3. Effusion Rate Estimation [61] Finally the lava flow emplacement reconstruction was used to estimate the temporal evolution of the lava discharge from the LFS1 vent. Effusion rate is commonly used in volcanology to indicate the instantaneous volumetric flux at which lava is erupted from a source vent or F02029 fissure. This is usually referred to an entire eruption which can have multiple vents. In this work we used ‘‘effusion rate’’ to indicate lava discharge rate from a single vent (in this case LFS1) instead of other more specific but not frequently used terms such as ‘‘volumetric flow rate’’ [Rowland and Walker, 1990]. We also defined daily effusion rate the average effusion rate during an observation period, obtained by dividing the emitted volume by the corresponding time interval. [62] Table 10 summarizes the main results of the analysis described above which allowed us to compute daily effusion rate for the whole flow. Figure 9 shows the temporal evolution of cumulative volumes and ‘‘daily effusion rate,’’ as well as some instantaneous ‘‘effusion rate’’ measurements collected during the eruption by INGV staff which are in good agreement with our estimates. The daily effusion rate trend shows a rapid increase from an initial rate of 10 m3/s to a peak value of about 30 m3/s between 20 and 22 July 2001 (three days after the beginning). This was followed by slow decline over the next 16 days, leading to an effusion rate value lower than 1 m3/s on 7 August. 6. Discussion [63] In this work on the 2001 Etna eruption, special attention was devoted to the use of rigorous methods for extracting quantitative data, to improve the congruency of preeruption (1999) and posteruption (2001) data sets, and to the evaluation of the associated uncertainties. In particular, we used two different analysis devoted to the quantitative evaluation of the final flow field area, volume and thickness, and we reconstructed the temporal evolution of the main lava flow from the LFS1 vent by using a semiquantitative approach. Final lava flow volumes were obtained by building and subtracting preeruption (1999) and posteruption (2001) DEMs or, where the two DEMs are not reliable and/or updated, by multiplying the area by the average measured thickness. The first and second approaches provided a relative error between 2 and 6% and 25 and 30%, respectively. The bulk volume (not corrected for vesicles or other voids) of the 2001 lava flow field was 40.1
106 m3 and the total covered area was 5.01
106 m2. We compared our results with those, based Table 7. Daily Values of Thickness Evaluated on the Seven Zones Composing the LFS1 Flow Datea Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Zone 7 Thickness, m 0.0 0.0 8.0 0.0 8.0 0.0 10.0 10.0 10.0 10.0 18/07/01 19/07/01 20/07/01 22/07/01 26/07/01 3.0 3.0 3.0 3.0 3.0 3.5 3.5 3.5 3.5 3.5 0.0 0.0 0.0 13.0 13.0 0.0 0.0 0.0 14.0 16.5 0.0 0.0 0.0 0.0 0.0 28/07/01 30/07/01 02/08/01 04/08/01 06/08/01 07/08/01 09/08/01 0.0 0.0 0.0 0.0 0.0 0.0 1.0 Additional Thickness, m 0.5 3.0 3.0 3.0 1.0 1.5 1.5 4.0 1.0 1.0 1.0 4.0 0.0 0.5 1.0 0.0 0.0 0.5 1.0 0.0 0.0 0.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 3.0 3.0 3.0 3.0 0.0 0.0 a 13 of 18 Date is given as dd/mm/yy. F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION Figure 8. Selected cross sections across the LFS1 lava flow. See Figure 7b for locations. The left axes shows the elevation of the 1999 and 2001 DEMs, and the right axes shows the elevation differences between them, representing the lava flow thicknesses. Black arrows mark the thickness of the first emplacement phase; accumulation peaks above the black arrows belong to the second and third phases. 14 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 Table 8. Daily Volumes of the Seven Zones Composing the LFS1 Flowa Date Total Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Zone 7 Volume 18/07/01 19/07/01 20/07/01 22/07/01 26/07/01 0.30 0. 30 0. 30 0. 30 0.33 Cumulative Volumes,
106 m3 0.07 0.00 0.00 0.00 0.00 0.84 0.56 0.00 0.00 0.00 0.88 2.32 0.00 0.00 0.00 0.88 3.60 1.30 2.08 0.42 0.95 4.30 1.40 3.38 4.62 0.00 0.00 0.00 0.00 0.00 0.37 1.70 3.50 8.58 14.98 28/07/01 30/07/01 02/08/01 04/08/01 06/08/01 07/08/01 09/08/01 0.00 0.00 0.00 0.00 0.00 0.00 0.08 Additional Volumes,
106 m3 0.09 1.23 0.45 0.24 0.00 0.18 0.71 0.21 0.24 0.00 0.15 0.36 0.15 0.72 0.00 0.00 0.21 0.08 0.00 0.00 0.00 0.21 0.08 0.00 0.00 0.00 0.11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.09 0.51 0.30 0.00 0.00 2.01 1.37 1.47 0.80 0.59 0.11 0.08 a Date is given as dd/mm/yy. Daily volumes are active areas multiplied per evaluated thicknesses. Volumes until 26 July 2001 are cumulative while subsequent volumes are partial. Last column shows the daily volumes of the whole flow. only on field measurements, reported in previous works on the 2001 Etna eruption [Calvari, 2001; Behncke and Neri, 2003]. This comparison illustrated the problem of the unreliability of preliminary and unverified data, in particular with regard to thickness measurements. Calvari [2001] used a preliminary lava flow mapping to estimate a bulk volume of 48.0
106 m3 and an area of 4.7
106 m2 for the 2001 flow field. The estimated area agrees fairly well with that obtained in this work on the basis of the posteruption orthophotos. However, the volume overestimation of 7.9
106 m3 may be due to the use of thickness values measured on flow levees and fronts and then extrapolated over the whole flow. Behncke and Neri [2003] provided, for the 2001 lava flow field, a dense rock equivalent (DRE) volume of 25.191
106 m3 and an area of 5.567
106 m2. The conversion from DRE to bulk volume (assuming the vesicularity of 20% utilized by the authors; B. Behncke and M. Neri personal communication, 2006) gives a bulk volume of 31.489
106 m3. The discrepancies observed with respect to our estimates (underestimation by 21% on the volume, and overestimation of the area by 15%) may be addressed to the use of inaccurate field data. [64] About 53% of the total 2001 volume was emitted from the LFS1 vent, producing a lava volume of 21.4
106 m3 and covering an area of 1.95
106 m2. Its effusion rate trend (Figure 9) is in accordance with the discharge model discussed by Wadge [1981] for eruptions from pressurized sources. Following the terminology of Wadge [1981], the effusion rate curve shows a brief initial period of ‘waxing flow’ followed, after the peak value, by a longer period of ‘waning flow.’ Consequently, the effusion rate curve fits well with the model of a magma source depressurization mainly owing to the reservoir relaxation (elastic contraction of the magma body). Moreover, the quite high effusion rate (up to 30 m3/s) and the relatively short duration (only 23 days) suggests a reservoir drainage more efficient than that expected solely from the reservoir relaxation. This behavior can be due to the gas expansion of the volatile-rich F02029 2001 magma, testified by the strong explosive activity observed during the eruption [Taddeucci et al., 2002; Scollo et al., 2007]. [65] Figure 9 also compares the LFS1 flow effusion rates presented in this work with those obtained from Figure 4 of Behncke and Neri [2003] for the same flow (their F4) and with the instantaneous effusion rates measured in the field. The three data sets are in fairly good agreement only in the final period of the eruption (i.e., after 1 August). In the first period, the effusion rates obtained in this work are in accordance with the fieldderived instantaneous effusion rates, but both are systematically higher than the values of Behncke and Neri [2003], derived from daily thickness and area measurements. The observed discrepancies highlight the importance of verifying the geometrical data used for volumetric effusion rate computation. This can be done by means of comparative and cumulative analysis using constraints: for example the final volume (derived from preeruption and posteruption 3D maps) have to correspond to the sum of the partial volumes. Finally, effusion rate trends not showing the lava emission peak between the waning and waxing flow periods may conduct to completely different simulated lava flow paths when adopted as input data in real-time forecasting applications aimed at hazard mitigation during the eruption. 7. Conclusive Remarks [66] In this work, we calculated a total bulk volume of 40.1
106 m3 for the complex lava field emitted by seven effusive vents during the 2001 Etna eruption. An error analysis was conducted to estimate the relative errors of all the estimated lava flow volumes and our lowest expected error was about 2% for the DEM-derived volume. Then, we focused on the LFS1 flow which emplaced a bulk volume of 21.40
106 m3 (53% of the total) forming a fairly simple lava flow. Table 9. Total Area, Measured From the 2001 DEM; Observed Volume Measured From the Comparison of the 1999 and the 2001 DEMs; Reconstructed Volume, Evaluated From Table 8 by Adding Partial Volumes After 26 July 2001 to the Volumes at 26 July 2001; Observed Average Thicknesses Measured From the Comparison of the 1999 and the 2001 DEMs; and Reconstructed Average Thickness Evaluated by Dividing the Reconstructed Volume by the Total Areaa Values at 09 August 2001 Total area,
106 m2 Observed volume
106 m3 Reconstructed volume,
106 m3 Observed average thickness, m Reconstructed average thickness, m Zone Zone Zone Zone Zone Zone Zone 1 2 3 4 5 6 7 0.13 0.28 0.63 0.20 0.29 0.28 0.17 0.38 1.49 7.25 2.26 4.62 4.46 0.95 0.40 1.36 7.11 2.37 4.63 4.54 0.97 3.0 5.5 11.7 11.4 15.8 16.2 5.5 3.2 4.9 11.4 11.8 16.0 16.5 5.6 a Every measure characterizes the LFS1 flow at the end of the eruption (9 August 2001), and it is evaluated inside the seven zones composing the LSF1 flow. 15 of 18 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 F02029 Table 10. Partial and Cumulative Volumes of the Whole LFS1 Lava Flow for Each Acquisition Datea Acquisition Date and Local Time 18/07/2001, 18/07/2001, 19/07/2001, 20/07/2001, 22/07/2001, 26/07/2001, 28/07/2001, 30/07/2001, 02/08/2001, 04/08/2001, 06/08/2001, 07/08/2001, 09/08/2001, 03:00 13:00 16:00 13:00 11:00 12:00 16:00 11:00 10:00 07:00 11:00 07:00 10:00 Eruption Day Acquisition Time, s Cumulative Volume,
106 m3 0 1 2 3 5 9 11 13 16 18 20 21 23 0 36,000 133,200 208,800 374,400 723,600 910,800 1,065,600 1,321,200 1,483,200 1,670,400 1,742,400 1,926,000 0.00 0.37 1.70 3.50 8.58 14.98 16.99 18.35 19.82 20.62 21.21 21.32 21.40 Time Span, s Partial Volume,
106 m3 Daily Effusion Rate, m3/s 0 36,000 97,200 75,600 165,600 349,200 187,200 154,800 255,600 162,000 187,200 72,000 183,600 0.00 0.37 1.33 1.80 5.08 6.40 2.01 1.37 1.47 0.80 0.59 0.11 0.08 0.00 10.28 13.68 23.81 30.68 18.33 10.74 8.85 5.75 4.94 3.15 1.53 0.44 a Date is given as dd/mm/yyyy. Partial volume is the volume emitted between two subsequent acquisition times (time span). Daily effusion rates were evaluated by dividing partial volumes by time spans. [67] The temporal evolution of the LFS1 lava flow, as well as its effusion rate trend, was reconstructed by means of a semiquantitative method using daily maps. These depicted the LFS1 flow expansion and were used, together with lava thicknesses evaluated in the field and from the analysis of preeruption (1999) and posteruption (2001) DEMs, to obtain volumes emplaced over known time periods. Volumes were then converted to time-averaged effusion rates by dividing their values by the corresponding emplacement times. The derived effusion rates were in good agreement with field measurements acquired during the eruption by INGV staff. Moreover, the observed trend is in agreement with the theoretical effusion rate curve of Wadge [1981] for a pressurized eruption, showing an initial period of waxing flow followed by a longer period of waning flow. [68] This work shows that flow volume and area can be evaluated with higher accuracy if sufficiently detailed topographic data are available before, during, and after an eruption. It points out that special attention must be devoted to the assessment of the accuracy of the data mapping, if they have to be used for quantitative processes and analysis such as modeling of the lava flow emplacement. Moreover volumes based solely on field data, such as those evaluated from the planimetric approach, can be affected by large errors and are not adequate to reconstruct a lava flow evolution. In order to apply a completely quantitative approach for the reconstruction of a lava flow evolution, data for generating DEMs should be daily collected (ideally) for example by means of photogrammetric or LIDAR surveys. and the 2001 DEM. The sum is limited to cells inside lava flow limits. [70] The standard deviation associated with this volume is calculated from the variance propagation law, vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !ffi u  2 uX  @V 2 2 @V s2Dx sDz þ sV ¼ t @zij @x ij sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi X ¼ Dx4  s2Dz þ 4  Dz2ij  Dx2  s2Dx ðA2Þ ij where sDx = 10 m is the planimetric accuracy and sDz = 2.69 m is the lava residual vertical accuracy. Appendix A [69] The volumes (V) computed from the DEM subtraction (Table 2) were calculated from V¼ X 2 Dx  Dzij ðA1Þ ij where Dx = 10 m is the linear dimension of the square DEM cells and Dzij is the height variation between the 1999 Figure 9. Left axis shows the temporal evolution of daily effusion rates evaluated in this work (gray bars) and in work by Behncke and Neri [2003] (squares), as well as field measurement of the instantaneous effusion rate (stars) made during the eruption by INGV staff. The right axis shows the cumulative volumes of the LFS1 lava flow evaluated in this work. 16 of 18 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION F02029 [71] The volumes obtained from the planimetric approach (Table 3) were calculated from V ¼AH ðA3Þ where A is the area covered by lava, evaluated by means of a mass balance analysis, and H is the average flow thickness. The related standard deviation is sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2  2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi @V @V 2 þ H 2  s2 ðA4Þ s2A ¼ A2  sH sV ¼ s2H þ A @A @H where sH = 1 m is the accuracy of the thickness measurements, and sA = nDx2 is the area accuracy, with n being the number of cells forming the perimeter of the lava flow and of its dagala. [72] Total volumes for the LFS2, UFS1 and UFS2 flows (Table 4) were obtained by summing or subtracting two independently estimated volumes: from DEM comparison (V1) and by means of a planimetric approach (V2). The accuracy of the total volume is then obtained adopting weights to take into account the different contribution to the total volume, sV ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P1  s2V 1 þ P2  s2V 2 ðA5Þ where P1 and P2 are the percent areas corresponding to the DEM and planimetric volumes respectively. [73] Acknowledgments. The authors especially thank G. Garfi and S. Di Mauro for their help in the preparation of the daily maps during the eruption, and M. Pompilio for his precious support during the fieldwork. We are indebted to the colleagues of INGV from Catania, Napoli, Pisa, and Roma involved in the lava-flow field monitoring. We are grateful to the pilots and technicians of the Italian Civil Protection helicopters for their professional work that permitted the daily overflight of the eruption. We wish to thank G. Calı̀ head of the VII Dipartimento, 1° Servizio Area Pianificazione Territoriale, of the Provincia Regionale di Catania that furnished us the posteruption vector maps and the orthophotos. The manuscript benefited from the suggestions of A. J. L. Harris, S. Calvari, and P. Baldi. References Baldi, P., S. Bonvalot, P. Briole, M. Coltelli, K. Gwinner, M. Marsella, G. Puglisi, and D. Remy (2002), Validation and comparison of different techniques for the derivation of digital elevation models and volcanic monitoring (Vulcano Island, Italy), Int. J. Remote Sens., 23, 4783 – 4800. Baldi, P., M. Fabris, M. Marsella, and R. Monticelli (2005), Monitoring the morphological evolution of the Sciara del Fuoco during the 2002 – 2003 Stromboli eruption using multi-temporal photogrammetry, ISPRS J. Photogramm. Remote Sens., 59, 199 – 211. Behncke, B., and M. Neri (2003), The July – August 2001 eruption of Mt. Etna (Sicily), Bull. Volcanol., 65, 461 – 476, doi:10.1007/s00445003-0274-1. Calvari, S. (2001), Multidisciplinary approach yields insight into Mt. Etna 2001 eruption, Eos Trans. AGU, 82(52), 653. Calvari, S., M. Coltelli, M. Neri, M. Pompilio, and V. Scrivano (1994), The 1991 – 1993 Etna eruption: Chronology and lava flow-field evolution, Acta Volcanol., 4, 1 – 14. Corsaro, R. A., L. Miraglia, and M. Pompilio (2007), Petrologic evidence of a complex plumbing system feeding the July – August 2001 eruption of Mt. Etna, Sicily, Italy, Bull. Volcanol., 69, 401 – 421. Costa, A., and G. Macedonio (2005), Numerical simulation of lava flows based on depth-averaged equations, Geophys. Res. Lett., 32, L05304, doi:10.1029/2004GL021817. F02029 Crisci, G. M., S. Di Gregorio, O. Pindaro, and G. Ranieri (1986), Lava flow simulation by a discrete cellular model: First implementation, Int. J. Modell. Simul., 6, 137 – 140. Crisci, G. M., S. Di Gregorio, R. Rongo, M. Scarpelli, W. Spataro, and S. Calvari (2003), Revisiting the 1669 Etnean eruptive crisis using a cellular automata model and implications for volcanic hazard in the Catania area, J. Volcanol. Geotherm. Res., 123, 211 – 230. Curlander, J. C., and R. N. McDonough (1991), Synthetic Aperture Radar: Systems and Signal Processing, 647 pp., John Wiley, Hoboken, N. J. Fouler, R. (2001), Topographic lidar, in Digital Elevation Model Technologies and Applications: The DEM Users Manual, edited by D. F. Maune, pp. 207 – 236, Am. Soc. for Photogramm. and Remote Sens., Bethesda, Md. Franceschetti, G., and R. Lanari (1999), Synthetic Aperture Radar Processing, 307 pp., CRC Press, Boca Raton, Fla. Frazzetta, G., and R. Romano (1984), The 1983 Etna eruption: Event chronology and morphological evolution of the lava flow, Bull. Volcanol., 47, 1079 – 1096. Guest, J. E., C. R. J. Kilburn, H. Pinkerton, and A. M. Duncan (1987), The evolution of lava flow field: Observation of 1981 and 1983 eruptions of Mount Etna, Sicily, Bull. Volcanol., 49, 527 – 540. Hidaka, M., A. Goto, S. Umino, and E. Fujita (2005), VTFS project: Development of the lava flow simulation code LavaSIM with a model for three-dimensional convection, spreading, and solidification, Geochem. Geophys. Geosyst., 6, Q07008, doi:10.1029/2004GC000869. Honda, K., and M. Nagai (2002), Real-time volcano activity mapping using ground-based digital imagery, ISPRS J. Photogramm. Remote Sens., 57, 159 – 168. Ishihara, K., M. Iguchi, and K. Kamo (1989), Numerical simulation of lava flows on some volcanoes in Japan, in Lava Flows and Domes, IAVCEI Proc. Volcanol., vol. 2, edited by J. Fink, pp. 174 – 207, Springer, New York. Kilburn, C. R. J., and G. E. Guest (1993), Aa lavas of Mount Etna, Sicily, in Active Lavas: Monitoring and Modeling, edited by C. R. J. Kilburn and G. Luongo, pp. 73 – 101, UCL Press, London. Lee, D. T., and B. J. Schachter (1980), Two algorithms for constructing a Delaunay triangulation, Int. J. Comput. Inf. Sci., 9, 219 – 242. Mouginis-Mark, P. J., and H. Garbeil (2005), Quality of TOPSAR topographic data for volcanology studies at Kilauea Volcano, Hawaii: An assessment using airborne lidar data, Remote Sens. Environ., 96, 149 – 164. Murray, J. B. (1990), High-level magma transport at Mount Etna volcano, as deduced from ground deformation measurements, in Magma Transport and Storage, edited by M. P. Ryan, pp. 357 – 383, John Wiley, Hoboken, N. J. Riccò, A. (1902), Rilevamento topografico della lava dell’eruzione Etnea del 1892, Boll. Accad. Gioenia Sci. Nat. Catania, 75, 5 – 8. Ridley, H. M., P. M. Atkinson, P. Aplin, J.-P. Muller, and I. Dowman (1997), Evaluating the potential of the forthcoming commercial U.S. high-resolution satellite sensor imagery at the ordnance survey, Photogramm. Eng. Remote Sens., 63, 997 – 1005. Rowland, S. K., and G. P. L. Walker (1990), Pahoehoe and aa in Hawaii: Volumetric flow rate controls the lava structure, Bull. Volcanol., 52, 615 – 628. Rowland, S. K., M. E. MacKay, H. Garbeil, and P. J. Mouginis-Mark (1999), Topographic analyses of Kilauea Volcano, Hawai’i, from interferometric airborne radar, Bull. Volcanol., 61, 1 – 14. Rowland, S. K., A. J. L. Harris, M. J. Wooster, F. Amelung, H. Garbeil, L. Wilson, and P. J. Mouginis-Mark (2003), Volumetric characteristics of lava flows from interferometric radar and multispectral satellite data: The 1995 Fernandina and 1998 Cerro Azul eruptions in the western Galapagos, Bull. Volcanol., 65, 311 – 330. Scollo, S., P. Del Carlo, and M. Coltelli (2007), Tephra fallout of the July – August 2001 Etna eruption: Deposit features and analysis of plume dispersion, J. Volcanol. Geotherm. Res., 160, 147 – 164. Stevens, N. F., J. B. Murray, and G. Wadge (1997), The volume and shape of the 1991 – 1993 lava flow field at Mount Etna, Sicily, Bull. Volcanol., 58, 449 – 454. Stevens, N. F., G. Wadge, and J. B. Murray (1999), Lava flow volume and morphology from digitised contour maps: A case study at Mount Etna, Sicily, Geomorphology, 28, 251 – 261. Taddeucci, J., M. Pompilio, and P. Scarlato (2002), Monitoring the explosive activity of the July – August 2001 eruption of Mt. Etna (Italy) by ash characterization, Geophys. Res. Lett., 29(8), 1230, doi:10.1029/ 2001GL014372. Vicari, A., A. Herault, C. Del Negro, M. Coltelli, M. Marsella, and C. Proietti (2007), Modeling of the 2001 lava flow at Etna Volcano by a cellular automata approach, Environ. Modell. Software, doi:10.1016/j. envsoft.2006.10.005, in press. 17 of 18 F02029 COLTELLI ET AL.: THE 2001 ETNA LAVA FLOW ERUPTION Vinassa de Regny, P. (1911), L’eruzione Etnea del 1910, Parte quinta: Osservazioni geologiche e morfologiche, Atti Acc., 5, 4 – 11. Wadge, G. (1981), The variation of magma discharge during basaltic eruptions, J. Volcanol. Geotherm. Res., 11, 139 – 168. Wadge, G., P. A. V. Young, and I. J. McKendrick (1994), Mapping lava flow hazards using computer simulation, J. Geophys. Res., 99, 489 – 504. Walker, G. P. L. (1971), Compound and simple lava flows and flood basalts, Bull. Volcanol., 35, 579 – 590. F02029 Waltershausen, W. S. (1880), Der Etna, 2 vols., 317 pp., Engelman, Leipzig, Germany.  D. Andronico, S. Branca, M. Coltelli, L. Lodato, and C. Proietti, Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Catania, Piazza Roma, 2, I-95125 Catania, Italy. (andronico@ct.ingv.it; branca@ct.ingv.it; coltelli@ ct.ingv.it; lodato@ct.ingv.it; proietti@ct.ingv.it) M. Marsella, Dipartimento Idraulica Trasporti e Strade, Università di Roma ‘‘La Sapienza,’’ Via Eudossiana, 18, I-00184 Rome, Italy. (maria.marsella@uniroma1.it) 18 of 18