The mud volcanoes at Santa Barbara and Aragona ( Sicily , 1 Italy ) : Their potential hazards for a correct risk assessment 2 3

3 Alessandro Gattuso , Francesco Italiano, Giorgio Capasso, Antonino D’Alessandro, Fausto 4 Grassa, Antonino Fabio Pisciotta, Davide Romano. 5 6 1 Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Palermo, via Ugo La Malfa 153, 90146, 7 Palermo Italy 8 2 Istituto Nazionale di Geofisica e Vulcanologia, Osservatorio Nazionale Terremoti, Via Vigna Murata 605, 9 00143 Roma Italy 10 3 Dipartimento di Scienze Matematiche e Informatiche, Scienze Fisiche e Scienze della Terra (MIFT), 11 Università di Messina, Viale Stagno D’Alcontres, 98166 Messina, Italy 12 13 *Correspondence to: Alessandro Gattuso (alessandro.gattuso@ingv.it) 14 15 Abstract 16 The Santa Barbara and Aragona areas are affected by mud volcanism (MV) phenomena, consisting of continuous or 17 intermittent emission of mud, water and gases. This activity could be interrupted by paroxysmal events, with an eruptive 18 column composed mainly by clay material, water and gases. They are the most hazardous phenomena and, nowadays, 19 it is impossible to define the potential parameters for modelling the phenomenon. In 2017, two DSM were performed 20 by drone in both areas, thus allowing the mapping of the emission zones and the covered areas by the previous events. 21 Detailed information about past paroxysms was obtained from historical sources and, with the analysis of the 2017 22 DSMs, a preliminary hazard assessment were carried out, for the first time at two sites. Two potentially hazardous 23 paroxysm surfaces of 0.12 km and 0.20 km for Santa Barbara and Aragona respectively, were defined. On May 2020, 24 at Aragona a new paroxysm covered a surface of 8,721 m. After this, a new detailed DSM was collected with the aim 25 to make a comparison with the 2017 one. Since 2017, a seismic station was installed at Santa Barbara. From preliminary 26 results, both seismic events and ambient noise showed a frequency of 5-10 Hz. 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

DSMs, a preliminary hazard assessment were carried out, for the first time at two sites. Two potentially hazardous 23 paroxysm surfaces of 0.12 km 2 and 0.20 km 2 for Santa Barbara and Aragona respectively, were defined. On May 2020, 24 at Aragona a new paroxysm covered a surface of 8,721 m 2 . After this, a new detailed DSM was collected with the aim 25 to make a comparison with the 2017 one. Since 2017, a seismic station was installed at Santa Barbara. From preliminary 26 results, both seismic events and ambient noise showed a frequency of 5-10 Hz. https://doi.org/10.5194/nhess-2020-369 Preprint. Discussion started: 13 November 2020 c Author(s) 2020. CC BY 4.0 License.
volcanism. Low permeability of clays in mud-volcano areas (Kopf, 2002) suggests that, in the lack of large mud outflow 66 (typical of quiescent phases), gas propagation from the reservoir mainly occur by the uprising of gas bubbles (Etiope and  and shapes of the Italian mud volcanoes vary considerably. According to (Martinelli et al., 2004), only a small proportion 76 (20%) can be described as 'large' with a surface area >500 m 2 , while only 5% exceed 2 m in height.

77
In Sicily, mud volcanoes are mostly located within Caltanissetta and Agrigento Provinces (S. Barbara and Aragona 78 locations respectively). The name of these phenomena is known as "maccalube" (or macalube), that derives from Arabic 79 and it means, "overturning". In some cases, a violent and instantaneous explosion called "paroxysm" could occurr and,

86
The majority of the mud eruptions occurred in the absence of any earthquake, suggesting that mud volcanoes may erupt 87 in response to a seismic input only if the internal fluid pressure approaches the lithostatic one. A repose time is needed 88 for triggering an eruption, related to the production rate of the driving gas to overcome the permeability of the system at 89 depth (Bonini et al., 2009).

90
In this paper, we have gathered some historical information about the pre and post paroxysmal events occurred in the past 91 at both study areas as starting point for a correct hazard assessment.

92
In October 2017, a seismic monitoring station was installed at Santa Barbara, in order to collect some seismic information 93 of the site. Moreover, a number ofdrone surveys were performed both at Santa Barbara and Aragona. Finally, at Aragona 94 a drone survey has been carried out a few days after the last paroxysm event occurred on 19 th may 2020, with the aim of 95 mapping the surface of the erupted material and estimating volume and thickness.

96
Moreover, a Digital Surface Model (DSM) has been elaborated and the emission points at the Earth's surface were 97 mapped. Based on the DSM analysis and our historical information, two main hazardous paroxysm areas at Santa Barbara 98 and Aragona have been elaborated, in this paper, for the first time.

111
At Santa Barbara, the mud volcanism is located eastward of the Caltanissetta town, near the "Santa Barbara village". The 112 composition of its deposits, consists essentially in clay, clayey-marly and sandy composed. Around the main mud 113 emission, in the northern sector, different residential buildings are present which were built mainly in the 60's while, in 114 the southern sector, twenty mono-familiar houses (Fig.2). Several public facilities are present at the western side of the 115 mud volcano and, electric pipelines, roads and services for about 4,000 resident people should be considered for a correct 116 risk assessment of the entire area.

144
The old naturalists and geologists, have described the activity of the mud volcano at Santa Barbara, since 1800, reporting 145 some of their major paroxysmal events (Carnemolla, 2017). The first scientific document was produced in 1823 with a  166 recorded up to 3-5 cm of progressive movements accumulating just before the event in the direction towards the satellite.

167
As a consequence of these phenomena heavy damages to factories, roads, residential buildings and public facilities (water,  where an imposing fountain of mud raised, which in its ascent dragged blocks of marl mixed with sandstones and gypsum.

211
This fountain reached ten to fifteen meters in height.

212
Only  (Fig.3B) covering an area of about 16,350 m 2 (Fig.3A). It is interesting to note that a strong correlation exists between 225 the erupted material and the covered surface areas for the paroxysms occurred from 1998 to 2012 (no volume data are 226 available for the 2014 paroxysm) as is demonstrated by the high correlation coefficient (R 2 =1) and showed in figure 3C.

227
From the same plot, the 2020 paroxysm event falls far from the general trend previously highlighted covering a smaller

238
They are quite common, especially at Aragona, and therefore, it is likely to hypothesize that others hazardous events, with 239 the same magnitude or higher, could repeat in the future.

240
In all of the paroxysmal events occurred in the past, both at Santa Barbara and Aragona (Tables 1-2), diffuse soil fractures 241 and deformations, even at a considerable distances from the mud volcanism area, occurred during a pre-paroxysm period.

242
In particular, at Santa Barbara the population has felt several seismic events before the 2008 paroxysm.

243
Another important element that emerges from historical descriptions is that, following the paroxysms, people approaching 244 the mud volcano areas, usually detected a strong acrid smell of gas, reasonably being H2S.

253
Before conducting drone mapping, we planned the flight paths and areas for each flight mission. The drone was set to 254 take aerial photographs using "autopilot mode" with a camera facing directly downwards for a hilly terrain. The surveys 255 were conducted with the camera mounted 90° sideways. We selected 75% forward and sideways overlap of images.

256
The acquisition of field data requires the determination of several control points on the ground, known as GCPs (Ground

257
Control Points). Therefore, 11 points distributed within the defined area, were recorded using a GPS NAVCOM SF-3040 258 with angular accuracy of 1 cm.

259
The images were processed with a Structure-from-Motion (SfM) and multi-view stereo approach, in order to produce a

268
In order to define the potential paroxysm hazardous scenarios for both areas, in this paper, we consider the maximum real

284
In this preliminary phase, in order to model the potential hazard scenarios, we assumed that both areas, in a next future,

287
For these reasons, at first, from our 2017 DSM, we identified the mud volcanoes and bubbling pools in both areas (Fig.5) 288 as the potential emission points for generating a future paroxysmal event. With the aim of the kernel density tool in 289 ArcGIS 10.5, we defined different clusters maps (Fig.4), with two main directions, appeared mostly highlighting a NW-

290
SE and NE-SW directions at Aragona while, at Santa Barbara, the distribution at the surface seems to be inhomogeneous. Secondly, we calculated on each emission points checked, the greatest distance reached by the erupted material at Santa

297
Barbara and Aragona respectively, through the elaboration in ArcGis 10.5 Software, of an omni-directional potential 298 hazardous area, considering a + 30% surplus rounded up, due to the creation of a safety limits for both areas.

299
The final potential paroxysmal hazardous areas, in both areas, are considered as the envelope among the entire buffer 300 circumferences elaborated. For the hazard assessment, we elaborated 117 and 165 buffer circumferences with a radius of 301 180 m and 195 m at Santa Barbara and at Aragona respectively (Fig.6).

307
The application of the methodology for the hazard assessement in both study areas, inevitably, is based on assumptions 308 which could give us some uncertainties. At the same time, the absence of a modelling approach for the paroxysm events

316
Since October 2017, a seismic INGV station was installed at Santa Barbara (see Fig.2  The hazardous paroxysm areas for both areas were created through the envelope of all buffer circumferences of Fig.6. An 327 area of 0.12 km 2 and 0.20 km 2 , potentially exposed to possible paroxysmal events was calculated for the Santa Barbara

328
and Aragona site respectively (Fig.7). In these two hazardous paroxysm areas, different geophysical phenomena as well 329 as deformation, fracturing and seismic events together with geochemical ones could occur. For that reason, these two 330 exposed areas should be interdict to visitors, residential or public activities, due to their correlated hazardous phenomena 331 that could occur before, during and after a paroxysm event. In both areas, a dedicated safe path, outside the hazardous paroxysm areas of Fig.7 should be created in order to permit the safety observations of these geological phenomena to 333 visitors.

334
The The maximum distance reached by the erupted materials, according to our analysis is around 130 meters. The 2020 356 paroxysm occurred in a medium -high density area of emission points detected from our 2017 survey, where a NE-SW 357 structural lineament has been highlighted ( Fig. 5and Fig. 9). In particular, the eruptive centre for the 2020 event is located, In this paper, for the first time a preliminary hazard assessment of two main mud volcanoes area of Sicily was evaluated.

389
We calculated the hazard scenarios based on the most recent paroxysm events at Santa Barbara and Aragona, in order to 390 define a realistic dimension for a correct risk assessment. It is evident that the hazardous paroxysm areas that we have 391 computed, should be implemented with a probabilistic modelling approach, deriving from the real measured parameters 392 on both areas. For these reasons, it should be important to implement in terms of acquistion frequency as well as number

393
of parameters, the actual discrete multidisciplinary surveys, with a new technological geochemical and geophysical 394 observatory, in order to minimize the knowledge gaps in these two areas. In light of this, therefore, it is appropriate to 395 realize and maintain an high frequency multidisciplinary data acquisition system to allow the construction of a forecast 396 model able to best represent the real conditions and, on the basis of which, a monitoring system should be implemented.

397
Nowadays, it is impossible to define "when" the next paroxysm will occur and how much will be its intensity. This is 398 because currently there are not enough information to recognize the parameters that could potentially change before a 399 paroxysm as well as a modelling approach of the phenomenon does not exist.

400
In this work, our hazard assessment for the Santa Barbara and Aragona areas, represent a picture of the 2017 survey. The

404
It is important to underline that we cannot exclude that these paroxysmal events, could occur out of the restricted area in 405 which most of the emission points are located at the surface. At the same time, an update of the actual hazard maps for 406 the two areas must be implemented.

407
From hystorical informations, we know that different phenomena could occur before a paroxysm in the mud volcanoes 408 areas, in particular deformations, soil fractures, increasing of seismicity and, from geochemical point of view only at