Liquefaction is one of the most complex earthquake-induced phenomena in geotechnical earthquake engineering. Commonly the term liquefaction, though used to indicate a variety of different but related phenomena (e.g., flow slide failures, lateral spreading, sand boils), is associated with soil permanent deformations produced by a rapid excess pore pressure increase (and a concurrent reduction in effective stresses) in saturated cohesionless soils subjected to earthquake loading (e.g., Kramer, 1996). Under these conditions, the pore water pressure in the soil mass increases in response to the seismic shaking, and the soil loses its shear strength and behaves temporarily as a viscous liquid.

The assessment of liquefaction hazards at a site always involves two questions: (1) are the geological conditions prone to liquefaction and (2) will liquefaction be triggered by future earthquakes? The latter is usually asked in case of a positive answer to the first question, namely the following: if the site is susceptible to soil liquefaction, will liquefaction be triggered by future seismicity? While answering the first question may involve costly in situ investigations mainly aimed at defining the depth of the water table and soil granulometry, the answer to the second question can be simply obtained by querying national hazard maps via online web services, at least at the screening level. The scope is to get the value of the two ground-motion parameters commonly used in liquefaction hazard assessments: peak horizontal acceleration (in the present work, the terms “peak horizontal acceleration” and “peak ground acceleration”, as well as the respective acronyms PHA and PGA, are used interchangeably) at the ground surface for the return period of interest and earthquake magnitude (e.g., maximum magnitude, mean, or modal magnitude from hazard disaggregation). These parameters are related to the duration and intensity of ground shaking and are key factors of the well-known simplified approach of Seed and Idriss (1971) for evaluating the liquefaction resistance of soils (see also Youd et al., 2001). Thus, it may be convenient to reverse the question above: if liquefaction can be triggered by seismic activity at the site of interest, is the site susceptible to soil liquefaction? Indeed, if the seismic hazard is deemed too low to trigger liquefaction (i.e., if the values of the surface PGA and earthquake magnitude for that site are below some threshold values), then one can avoid recommending specific investigations aimed at the geological and geotechnical characterization of the subsoil. This reversed decision scheme can be useful to guide land-use planners in deciding whether liquefaction is a hazard that needs to be considered within the planning processes or not.

With the increased availability of data in recent years, substantial research has been carried out to establish thresholds of magnitude and PGA at the ground surface below which the possibility of triggering liquefaction can be discounted. Although the use of a minimum threshold magnitude is controversial (Musson, 2020; Green and Bommer, 2019), available studies converge to a value of 5.0 (e.g., Atkinson et al., 1984; Green and Bommer, 2019), while a threshold of about 0.1 g has been suggested for the PGA at the ground surface (e.g., Santucci de Magistris et al., 2013; Santucci de Magistris, 2015). However, exceptional liquefaction phenomena induced by minor events (i.e., with magnitude less than 5.0) have been observed in extremely susceptible soils (e.g., Martino et al., 2014; Santucci de Magistris, 2015, and references therein; Zimmaro et al., 2019; Brandenberg et al., 2020; Bozzoni et al., 2021). Such values are adopted by the Italian building code (Ministero delle Infrastrutture e dei Trasporti, 2018

It is worth noting that, compared to the original version released in 2008 (Ministero delle Infrastrutture e dei Trasporti, 2008), the 2018 update of the Italian building code does not specify a lower-bound magnitude for liquefaction triggering.

The scope of the present work is to define a liquefaction triggering map of Italy that can be used for seismic microzonation purposes and, therefore, can guide land-use planning within the framework of risk mitigation programs. According to the Italian Catalogue of Earthquake-Induced Ground Failures – CEDIT (Martino et al., 2014; Caprari et al., 2018), more than 300 liquefaction events have occurred in Italy since 1000 CE, the last of which are associated with the 2012 Emilia seismic sequence (e.g., Scognamiglio et al., 2012; Minarelli et al., 2022). Therefore, it is of paramount importance to define a map that classifies sites as a function of their liquefaction triggering potential. To this end, we combine data from the reference Italian seismic hazard maps (MPS Working Group, 2004; Stucchi et al., 2011) and hazard disaggregation (Barani et al., 2009), which is here expressed in terms of magnitude contributions to the hazard. Note that the results presented in this work do not address the issue of liquefaction assessments for structural design, for which the hazard corresponding to multiple return periods should be taken into account. In this work, we focus on the 475-year return period according to the ICMS. The reliability of the results is checked by comparing them with observations of past liquefaction events in Italy (Martino et al., 2014; Caprari et al., 2018), as well as in light of recent case studies associated with the 2012 Emilia seismic sequence.

Although we are aware that liquefaction hazard requires site-specific geological and geotechnical investigations (liquefaction is indeed a highly localized phenomenon), our study may serve as a reference guide for identifying sites where the possibility of triggering liquefaction can be discounted. Moreover, it provides basic ground-motion data for evaluating liquefaction resistance of soils at specific sites and for defining liquefaction susceptibility maps (e.g., Zumpano et al., 2022). Such data are freely available through the web service at https://distav.unige.it/rsni/milq.php (last access: 28 April 2023).

According to the ICMS, liquefaction is expected to occur if the following conditions are met (SM Working Group, 2008, 2015).

The lithological sequence presents layers of non-cohesive, saturated soil (sandy silts, sands, silty sands, gravely sands, clayey sands, and sandy gravels) down to 20 m depth.

In this study, our focus is on the third condition. Contrary to the first two ones, it does not require local data, at least at a screening level (corresponding to the so-called level 1 microzonation in the ICMS). We use national hazard maps along with the corresponding hazard disaggregation results to define those sites in Italy where this condition is not met – that is, where the triggering of liquefaction phenomena is unlikely and can be excluded. If the condition is not met, then we can avoid recommending specific geotechnical investigations aimed at verifying the first two criteria, with consequent savings in efforts and money.

In order to define those areas in Italy where seismic activity could trigger liquefaction (according to the third condition listed in the previous section), we analyze the reference seismic hazard data for the national territory. First, the PGA hazard for rock conditions and flat topography associated with a return period of 475 years (MPS Working Group, 2004; Stucchi et al., 2011) is modified to incorporate site amplification due to local geology and irregular topography (recall that the 475-year return period is assumed as the single, reference return period by the ICMS). Essentially, the resulting amax map allows us to define areas with high (i.e., amax≥0.1 g) and low (i.e., amax<0.1 g) local seismic hazard. Then, 1D disaggregation results (i.e., magnitude disaggregation) are produced based on the reference work of Barani et al. (2009) and analyzed to define those sites where the hazard is controlled by scenarios with magnitude Mw≥5. As sites respond at specific characteristic frequencies (depending on local geological characteristics) and disaggregation results may vary significantly with response period (T), disaggregation analysis is carried out for different values of T. Particular attention is paid to the choice of mean or modal magnitude values as reference scenarios in order to avoid both non-conservative and over-conservative scenarios. A cross analysis of hazard and disaggregation data is finally performed via the QGIS software (https://www.qgis.org/en/site/, last access: 28 April 2023; QGIS.org, 2021) to define the sites where liquefaction triggering is expected to occur.

The three maps shown in Fig. 5 are used in conjunction with the map of site classification (Fig. 9b in Forte et al., 2019) to define the final map of M^ (from here on, we drop the dependence of M^ on T to indicate the preferred magnitude) according to the following criteria:

M^=M^T=0.01s for ground type A sites (i.e., rock sites or stiff soils with VS≥800 m s-1);

Combining the maps in Figs. 6 and 2b leads to the liquefaction triggering map shown in Fig. 7, which is referred to with the Italian acronym MILQ – Mappa del potenziale d'Innesco della LiQuefazione (map of liquefaction triggering potential). The color scale has been chosen to divide sites into classes of increasing liquefaction triggering potential (LTP), from no potential (i.e., amax<0.1 g and M^<5.0) to very high potential (i.e., amax≥0.2 g and M^≥6.0). Analyzing the map, we observe the following:

9.8 % of nodes fall in areas with no liquefaction triggering potential (class LTP-0: amax<0.1 g and M^<5.0); they are mainly concentrated in very few areas in the northwest and in almost all of the island of Sardinia.

To make our results available to land-use planners and practitioners, we have developed a web service, freely accessible at https://distav.unige.it/rsni/milq.php (last access: 28 April 2023). The web service data are stored in a “PostgreSQL” database. All the online components were developed in “PHP” and “HTML5” languages to ensure adherence to current web standards. Online maps are based on “Leaflet”, an open-source JavaScript library for mobile-friendly interactive maps.

For a specified location (defined by a pair of geographic coordinates), the web service provides the values of amax and M^ computed according to the site classifications adopted in the present study. Specifically, the values associated with the nearest node are returned (no interpolation is performed). In addition, as the actual ground type and topographic class at a site (e.g., resulting from site-specific data) can differ from those considered here, the service allows the user to change them through a user-friendly interface and returns the updated values of amax and M^ as output. Alternatively, in line with the ICMS document (level 2 and 3 seismic microzonation) and the Italian building code (for sites that can not be classified in one of the subsoil classes mentioned in the previous section, and/or in the case of complex topography), the web service allows the amplification factor value (S) obtained from specific studies (e.g., ground response analysis) to be entered. In this case, the user is also requested to select the value of the spectral period T of interest among those considered in the disaggregation analysis (e.g., as a function of site response) to obtain the value of M^.

Given the practical importance of liquefaction triggering potential maps, they should be subjected to testing before application. In the present study, we examine the reliability of the map in Fig. 7 by analyzing the geographic distribution of past liquefaction phenomena reported in the Italian Catalogue of Earthquake-Induced Ground Failures – CEDIT (Martino et al., 2014; Caprari et al., 2018). These events have been superimposed on that map (blue and light blue dots), which is now displayed in Fig. 8 using a more intuitive red and white scale, with the red color indicating the sites where liquefaction triggering is expected to occur (see the caption for further details). As is evident, most observations fall in red areas, thus indicating the consistency of our results. The matched observations are 314 out of 328. A more quantitative analysis of the results is provided by the histogram in Fig. 9, which shows the percentage of CEDIT observations for each class of liquefaction triggering potential defined above (from LTP-0 to LTP-5). Among the observations that fall in the red areas of the map in Fig. 8, 92 % of them correspond to nodes with high or very high liquefaction triggering potential (classes LTP-4 and LTP-5), while 4 % have a moderate potential (class LTP-3). Only 4 % of observations correspond to nodes with very low or low triggering potential (classes LTP-1 and LTP-2). No observations fall in regions with no triggering potential (class LTP-0).

Despite the good agreement between our map and past observations, a few questions need to be answered. To what extent does this matching derive from the criterion adopted to define M^ in relation to the soil resonance? In other words, what is the sensitivity of our results to the choice of M^T for different ground types? Can one obtain similar conservative results just assuming the disaggregation results for a single spectral period (e.g., T=0.01 s in line with the definition of amax) and increasing the return time (e.g., to 2475 years)? To answer these questions, we focus on the Po Plain area, which experienced liquefaction in several locations during the 2012 Emilia seismic sequence (Scognamiglio et al., 2012; Minarelli et al., 2022). The Quaternary deposits, consisting mainly of an alternation of sands, silts, and clays, are rather homogeneous throughout this area. Indeed, according to the soil classification of Forte et al. (2019) (see Fig. 9b therein), most sites in the Po Plain basin can be classified as ground type C or D (i.e., sites characterized by deposits of loose-to-medium cohesionless soil with thickness greater than 30 m and 100 ≤VS,eq<360 m s-1). However, deep stratigraphic discontinuities can be identified (e.g., Mascandola et al., 2019). Such discontinuities are responsible for significant ground-motion amplification at long periods (>1 s) (e.g., Luzi et al., 2013; Abraham et al., 2015; Mascandola et al., 2021b).

For the Po Plain area, Fig. 10 compares alternative liquefaction triggering scenarios by adopting different assumptions: precisely, Fig. 10a is just a zoom of Fig. 8; Fig. 10b is the same as Fig. 10a but is obtained by assuming M^=M^T=0.01s regardless of the ground type; finally, Fig. 10c is the same as Fig. 10b but for a return period of 2475 years. As expected, comparing the map in Fig. 10b with the reference one in Fig. 10a reveals that disaggregating only the PGA hazard can lead to an underestimation of the liquefaction triggering potential at soil sites. Such an underestimation can be significant at sites characterized by thick and soft soil deposits, as they generally have resonance periods much larger than 0.1 s. This justifies the use of the disaggregation of the hazard associated with either short or long spectral periods depending on site response. Especially for soft soil sites or sites characterized by thick soil deposits, the use of hazard disaggregation corresponding to longer periods prevents the adoption of non-conservative provisions by land-use planners or practitioners. As for the map in Fig. 10c, which refers to a return period of 2475 years, it still provides a conservative scenario like the reference map in Fig. 10a, but that scenario is justified neither by the local geology (see above) nor by the return period of a Mw=5.9 earthquake, such as the main shock of the 2012 Emilia seismic sequence with its epicenter near the town of Ferrara. According to the MPS Working Group (2004), indeed, the return period of a Mw=5.9 earthquake in that area is about 360 years.

Besides ground-motion amplification effects, which are undoubtedly the most important of all seismic (site) effects because of their impact on the environment and society, instability phenomena induced by earthquake ground shaking (i.e., surface faulting, ground failure, and soil liquefaction) also play a key role in defining risk mitigation strategies and land-use planning. In recent years, the identification of unstable areas and the subsequent quantification of the instabilities have become primary activities of seismic microzonation studies. In this context, our study has focused on the identification of areas susceptible to soil liquefaction in Italy. We have analyzed ground-motion hazard maps and the associated hazard disaggregation to define areas where liquefaction triggering due to seismic activity can not be excluded and where, therefore, further efforts are required to evaluate liquefaction susceptibility. The final result is a liquefaction triggering map showing areas with different triggering potential (Fig. 7). The information contained therein (particularly the value of M^) can be considered as an alternative to the results one can obtain by applying other approaches, particularly those proposed by the SM Working Group (2008, 2015; interested readers may also refer to the note of Technical Commission on Seismic Microzonation, 2018) for the determination of the reference magnitude for liquefaction susceptibility assessment. It is worth noting that, regardless of the approach used, soil liquefaction is a highly localized phenomenon whose occurrence is intimately related to site-specific geological and geotechnical conditions, which deserve focused investigations. Therefore, studies like the one presented here primarily serve as a basic guide to identifying sites where the possibility of triggering liquefaction can be discounted within the land-use planning process.

Despite the reliability of our results, which was examined by analyzing the distribution of past liquefaction events over the Italian territory, they can certainly be refined as soon as site-specific studies become available. On the one hand, site-specific hazard analyses (e.g., Cramer et al., 2014; Barani et al., 2020; Mascandola et al., 2023) allow the refinement of the values of amax. On the other hand, in situ measurements provide a more accurate definition of the ground type for the site of interest. At least as far as this issue is concerned, practitioners and land-use planners can interrogate our results through the web service at https://distav.unige.it/rsni/milq.php (last access: 28 April 2023) and refine the level of liquefaction triggering potential of the site of interest by changing the ground type and the topographic class through a user-friendly interface or by entering the site-specific amplification factor value at hand (e.g., determined from ground response analysis) along with the value of the spectral period of interest. The service returns as output the updated values of amax and M^, thus allowing the refinement of the triggering potential level.