Assessing subsidence at fuel-storage tanks in Moa Port, Cuba, using interferometric synthetic aperture radar and engineering-geological interpretation

c_u

undrained shear strength

m_a

mean amplitude

m_v

coefficient of volume compressibility

w_L

liquid limit

γ

unit weight of soil

σ_a

amplitude standard deviation

φ′

effective angle of shearing resistance

Land subsidence is a major geotechnical hazard in coastal and industrial environments because progressive downward ground movement can impair serviceability, damage foundations and compromise the safe operation of critical infrastructure (Sosa Rojas, 2020). Assessing its effects, therefore, requires not only reliable deformation measurements but also a sound understanding of the ground conditions that control settlement behaviour. Among the monitoring techniques currently available, interferometric synthetic aperture radar (InSAR) has become especially valuable because repeated satellite acquisitions can detect ground motion over broad areas and long observation periods, complementing conventional topographic, geodetic, photogrammetric and field-based surveys (Ali et al., 2020; Bai et al., 2022; Cárdenas and Hernández, 2017; Castellazzi et al., 2016; Del Conte and Falorni, 2019; Dong et al., 2023; Hu et al., 2019; Pepe and Calò, 2017; Zafar et al., 2024).

In Cuba, land subsidence has been associated with several geological and environmental factors, including tectonic activity, faulting, landslides, high ground moisture and clay-rich soils (Galeana Pérez et al., 2022; Glazunov et al., 2022; Mazurov et al., 2019; Pospehov et al., 2024; Rosabal Domínguez et al., 2021; Ustyugov et al., 2024). These factors are especially relevant at Moa Port, in Moa, Holguín province, north-eastern Cuba, a strategic coastal-industrial facility that supports nickel production and provides fuel-storage services for the Cuban Petroleum Union (Volohov et al., 2023). Within the port, the structural condition of fuel-storage tanks TK-5 and TK-6 has become a matter of concern, particularly because two comparable tanks were decommissioned in 2015 after exhibiting similar subsidence-related distress (Afanasev et al., 2024).

The storage tank facility is located on reclaimed coastal ground with unfavourable foundation conditions. According to the 2019 site investigation prepared by Cuba’s Ministry of Construction (MICONS, 2019), the subsoil beneath TK-5 and TK-6 consists of marine sedimentary materials of silty-sandy composition with scarce gravels, overlain by lateritic and serpentinic reclamation materials, including redeposited lateritic soils with a predominance of clays, gravelly sand and silts. A field measurement identified the groundwater table at approximately 0.8 m below ground surface, hydraulically linked to sea level and affected by tidal fluctuations. These conditions promote persistent saturation, low bearing capacity and high compressibility, all of which are unfavourable for shallow foundation performance. The site is also located within a structurally complex regional setting influenced by the Mao, Moa and Cabañas faults and by recurrent low-magnitude seismicity associated with Caribbean–North American plate interaction. In this study, however, regional seismicity is treated as a geological context rather than as a demonstrated direct trigger of the observed deformation.

The decommissioning of adjacent tanks in 2015 underscored the need for continued assessment of the remaining operational facilities. For that reason, this study evaluates deformation affecting TK-5 and TK-6 by integrating multi-temporal InSAR, existing geotechnical information and field observations. Radar measurements were initially processed in the satellite line-of-sight (LOS) and then combined through ascending/descending decomposition to derive east–west horizontal and vertical deformation components.

The objectives of this study are to quantify the rate and spatial pattern of deformation affecting tanks TK-5 and TK-6, to relate the observed behaviour to the engineering-geological conditions beneath the foundations, to compare the remote-sensing results with field evidence of structural distress and to demonstrate the practical value of an InSAR-based monitoring framework for assessing subsidence risk in critical coastal infrastructure at Moa Port.

The port of Moa is located in Moa, Holguín province, north-eastern Cuba (Figure 1), at geographic coordinates 20°39′06.4″ N and 74°54′28.9″ W, referenced to the World Geodetic System 1984 (WGS 84) datum (Lafita Cobas, 2021). The port is a strategic coastal-industrial facility associated with nickel production and fuel storage. The fuel base examined in this study, which includes tanks TK-5 and TK-6, was constructed in 1956 on an artificial dock developed in connection with local nickel exploitation.

From a regional geological standpoint, the port is situated in a structurally complex setting influenced by the first-order Mao fault and by the Moa and Cabañas faults. The broader area is also affected by recurrent low-magnitude seismicity associated with the interaction between the Caribbean and North American plates (Cotilla Rodríguez, 2019; Galbán Rodríguez et al., 2021). In the context of this study, seismicity is considered part of the regional geological setting and site vulnerability, rather than a demonstrated direct trigger of the deformation documented at the tanks (Afanasev et al., 2024).

At the site scale, TK-5 and TK-6 are founded on reclaimed coastal ground placed over natural marine deposits. According to the 2019 site investigation cited in this study, the subsoil beneath the tanks is composed of marine sedimentary materials of silty-sandy composition with scarce gravels, overlain by reclamation materials derived from lateritic and serpentinic sources. More specifically, the reclaimed layer includes redeposited lateritic soils with a predominance of clays, gravelly sand and silts (MICONS, 2019). Borehole information used elsewhere in this study indicates a generalised vertical succession consisting of heterogeneous fill in the upper profile, a very soft cohesive layer at intermediate depth, underlying alluvial soils and marine sediments below the investigated interval. The detailed lithological breakdown and the engineering-geological interpretation are presented later in Section 3.1 and are not repeated here.

Groundwater conditions further increase the geotechnical vulnerability of the site. A field measurement identified the groundwater table at approximately 0.8 m below ground surface. This shallow groundwater level relates to marine influence and tidal fluctuations, which implies persistently saturated foundation soils. In combination with the described soil composition, these conditions are unfavourable for shallow foundation performance because they are associated with low bearing capacity and high compressibility. For clarity, the present section is limited to the geographic and physical setting of the site.

This study combined engineering-geological characterisation, satellite-based deformation monitoring and field validation to assess subsidence affecting fuel-storage tanks TK-5 and TK-6 at the port of Moa. The methodological framework included three complementary components: characterisation of the foundation soils from available borehole and site-investigation data, multi-temporal InSAR processing of Sentinel-1 imagery and field verification of the structural deformation observed at the tanks.

The foundation system of TK-5 and TK-6 is located on an artificial platform created during land reclamation associated with the development of the port of Moa. According to the available site documentation, the upper 2.75 m of the profile is composed of heterogeneous engineered fill. This upper sequence includes a compacted lateritic surface layer from 0 to 0.50 m and an underlying reclamation layer from 0.50 to 2.75 m composed of serpentinised rock fragments within a gravelly sandy matrix. Although these materials form the construction platform, the available documentation does not indicate modern compaction-control procedures; accordingly, they are interpreted as heterogeneous reclamation fill with variable stiffness and strength characteristics (MICONS, 2019).

Below the fill, a very soft cohesive layer was identified between 2.75 m and 6.20 m depth. This unit is described as a dark, bluish, clayey silty sand with plastic behaviour, very low consistency and high moisture content. In engineering terms, it represents a highly compressible stratum with limited shear strength and is therefore regarded as a critical control on short-term settlement and differential deformation beneath the tank foundations.

At depths between 6.20 m and 7.80 m, the profile includes alluvial deposits consisting of sandy silt with gravel inclusions and low compaction. Below that interval, marine sedimentary soil was identified from 7.80 m to 9.00 m depth; within the logged interval considered in this study, the corrected thickness of this unit is 1.2 m. These deeper materials are described as sandy silt with scarce gravels, low plasticity and medium to dense compactness. A field measurement identified the groundwater table at approximately 0.8 m below ground surface, relating this shallow level to marine influence and tidal fluctuations. In combination, these conditions indicate a persistently saturated and mechanically unfavourable foundation environment.

Based on the available borehole information, an engineering-geological model was developed to summarise the stratigraphic arrangement and its relevance to foundation performance. The resulting interpretation consists of a relatively stiff but heterogeneous reclamation fill overlying a compressible sequence formed by a very soft cohesive layer, transitional alluvial deposits and marine sediments. Lateral variability in the thickness of the soft and marine units was considered a plausible explanation for the different settlement responses observed at TK-5 and TK-6. Table 1 summarises the lithological units and their thicknesses, and Table 2 presents the engineering-geological interpretation of the foundation subsoil.

Because the availability of site-specific laboratory and in situ testing was limited, the geotechnical parameters reported in Table 3 are presented as indicative ranges only. These values were compiled from available investigation reports, borehole descriptions and literature ranges for comparable soils in coastal and reclaimed settings. The table includes representative values of unit weight, liquid limit, plasticity index, undrained shear strength, effective friction angle and coefficient of volume compressibility for the main soil units. They are intended to support engineering interpretation of foundation behaviour and should not be treated as a substitute for detailed geotechnical design data.

Ground deformation was quantified through InSAR, a remote-sensing method that estimates surface displacement from phase differences between repeated radar acquisitions over the same area (Beladam et al., 2019; Jiang et al., 2021; Mohamadi et al., 2020; Vystrchil et al., 2025; Zhang and Lu, 2022; Zwieback et al., 2024). Processing was carried out in the Sarproz software program, using the version accessed on 1 October, 2023. Sentinel-1 images were obtained from the Copernicus platform of the European Space Agency and from the Alaska Satellite Facility. The data set consisted of single look complex (SLC) products acquired in interferometric wide swath mode and vertical transmit and receive polarisation. A total of 294 scenes were processed, including 170 ascending and 124 descending images, covering the period from 1 November, 2014 to 27 July, 2021.

The processing sequence, summarised in Figure 2, followed a persistent scatterer (PS)-InSAR workflow (Roccheggiani et al., 2019; Tamayo Duque et al., 2023). The first stages included import of the ascending and descending image stacks, selection of master scenes and application of precise orbits and co-registration. Acquisitions dated on 22 December, 2018 and 15 May, 2019 were selected as the master images. Reflectivity maps and the amplitude stability index (ASI) were then computed using the shuttle radar topography mission digital elevation model (Bryn et al., 2024). The ASI was calculated as

where $σa$ is the amplitude standard deviation and $ma$ is the mean amplitude.

Subsequent processing steps included the generation of interferograms for multi-temporal InSAR analysis and estimation and removal of the atmospheric phase screen (APS) using filters. PSs were selected using ASI thresholds greater than 0.75 for APS estimation and greater than 0.6 for denser spatial coverage. A reference network was then constructed by Delaunay triangulation, followed by estimation of deformation velocity and residual topographic error. Phase residuals were inverted with one PS fixed as the reference point.

The initial displacement estimates were handled in the satellite LOS. After LOS-based processing of the ascending and descending stacks, ascending/descending decomposition was applied to derive east–west horizontal and vertical deformation components (Dong et al., 2021). The final stage of the processing included geocoding and time-series generation for PSs with temporal coherence greater than 0.6–0.7 (Fárová et al., 2019).

To verify the deformation patterns identified by the InSAR analysis, field validation campaigns were conducted in the study area between October and November 2022. These inspections included direct ground-truth measurements using tape measurements and visual documentation of structural distress at the tanks. The field survey focused on displacement at the metal-to-concrete joints at the tank bases, where persistent cracking was observed, and on measurement of vertical differential displacements in the concrete foundations.

The inspections were performed according to standard geodetic monitoring practice for vertical fuel-storage tanks in order to maintain methodological consistency during the field campaign (Kuzin and Filippov, 2024). The integration of borehole-based site characterisation, long-term satellite monitoring and on-site structural inspection provided the basis for the subsequent interpretation of settlement behaviour at TK-5 and TK-6.

This section presents the deformation results and their engineering interpretation by integrating satellite monitoring, geotechnical characterisation and field observations. As described in Section 3.2, the radar observations were initially processed in the satellite LOS and were subsequently combined through ascending/descending decomposition to derive east–west horizontal and vertical components. In the present section, the deformation map is discussed as a vertical-component result, whereas the two representative displacement time-series are discussed in descending LOS.

The PS-InSAR analysis identified persistent and spatially non-uniform deformation across the fuel-storage area during the 2014–2021 monitoring period. The vertical cumulative deformation map shows that most pixels are close to 0 mm, but localised subsidence reaches approximately −10 mm south of TK-5 and approximately −75 mm at the site of a previously decommissioned eastern tank (Figure 3). This historical case, in which pronounced subsidence was associated with structural failure and demolition, supports the interpretation that the phenomenon observed at Moa Port is progressive rather than isolated. The area around TK-6 is also identified as presenting average deformation rates of around −2.5 mm/year, indicating that it is currently the most affected of the remaining operating tanks.

Representative persistent scatterer points extracted from the two tanks confirm that TK-5 and TK-6 behaved differently over the monitoring period. For PS ID 1526, located on TK-5, the analysis reports a velocity of −0.9 mm/year and a cumulative displacement of −6.4 mm (Figure 4). For PS ID 1395, located on TK-6, the reported values are −2.2 mm/year and −15.1 mm (Figure 5). Across the reported results, TK-5 is characterised by a lower level of deformation, on the order of approximately −1.0 to −1.4 mm/year, whereas TK-6 reaches values of up to approximately −2.5 mm/year. These values are moderate in absolute terms, but they are significant in light of the site’s history of tank decommissioning related to subsidence.

The detailed time-series provides further insight into temporal behaviour. For PS ID 1526 on TK-5, the descending LOS series is dominated by values between approximately +5 mm and −10 mm, with an average displacement velocity of −1.4 mm/year between 2014 and 2021 (Figure 4). Peak displacements of approximately −12 mm in 2016 and −17.2 mm in 2020 also note a temporal association with the wettest months of the year in Cuba. For PS ID 1395 on TK-6, the descending LOS series shows larger cumulative motion, with the highest displacements between approximately −20 mm and −25 mm. During the 2016–2021 interval, the TK-6 series maintains a persistent sinking trend, with values mostly between −10 mm and −20 mm (Figure 5).

Overall, the InSAR results show that deformation at the site is neither spatially uniform nor strictly linear through time. The cyclical character of the displacement relates to seasonal rainy periods and to the influence of average sea level on the shallow groundwater table. In practical terms, the remote-sensing results identify TK-6 as the more severely affected operating structure and place both tanks within a sector where subsidence has already contributed to the loss of adjacent infrastructure.

The deformation pattern identified by InSAR is consistent with the engineering-geological setting established for the site. The initial 2019 investigations indicated that the foundation soils beneath the tanks are primarily of marine sedimentary origin, with a silty-sandy composition and scarce gravels, overlain by reclamation materials derived from serpentine and laterite (Figure 6). The ground is also described as consisting of redeposited lateritic soils with a predominance of clays, gravelly sand and silts. A field measurement identified the groundwater table at approximately 0.8 m below ground level. This shallow groundwater level is hydraulically connected to sea level and affected by tidal oscillations, which implies persistent saturation conditions unfavourable for foundation performance.

Field inspections conducted between October and November 2022 confirmed structural damage consistent with the subsidence trends detected by InSAR. Persistent cracking was measured at the metal-to-concrete joints at the tank bases, with openings of approximately 5.0 cm, and vertical differential settlements of up to approximately 6.0 cm were also observed in the concrete foundations (Figure 7). These observations demonstrate that the measured deformations are not only detectable in the radar record but also expressed as clear structural distress at the site.

The observed deformation of fuel-storage tanks TK-5 and TK-6, therefore, cannot be explained solely by uniform vertical settlement, because the magnitude of damage documented during field inspections significantly exceeds the average cumulative displacements derived from InSAR analysis. This apparent discrepancy highlights the importance of considering differential settlement mechanisms and foundation–soil interaction effects. It is also important to note that InSAR-derived displacements represent spatially averaged vertical movements of coherent scatterers and therefore may not fully capture localised differential settlement occurring over short distances within the foundation area. Structural distress in rigid systems such as fuel-storage tanks is primarily governed by differential settlement and rotation rather than absolute vertical displacement. Consequently, relatively small average settlements detected by InSAR can still produce cracking, joint opening and distortion of the tank base when settlement gradients are sufficiently high.

The engineering-geological model indicates that the foundation system consists of a heterogeneous reclamation fill resting on a sequence of very soft cohesive soils, loose alluvial deposits and marine sediments. Such stratigraphy is highly susceptible to excessive settlement under sustained static loading. The presence of a very soft, highly compressible cohesive layer immediately beneath the engineered fill suggests that localised strain concentrations may develop under the applied loads from the fuel-storage tanks. Verified project information further indicates that the tanks are supported by reinforced concrete foundation slabs resting directly on the reclamation fill, with slab thicknesses on the order of 0.8–1.2 m. The tanks have large diameters and heights of several tens of metres and are designed to store petroleum products with densities close to 9–10 kN/m³. Under full operational conditions, the resulting average contact pressures transmitted to the foundation soils are of the order of several tens of kilopascals. These sustained static loads, when applied to heterogeneous fill materials overlying soft saturated soils, are sufficient to induce significant consolidation and differential settlement, particularly where lateral variability in soil stiffness exists.

From a geotechnical perspective, the observed deformation pattern is consistent with settlement mechanisms approaching bearing-capacity-controlled behaviour in the underlying marine and reworked lateritic sediments. Although no sudden collapse has occurred, progressive shear deformation and punching-type behaviour at the base of the foundation slab may develop in soft saturated soils when contact stresses approach the available bearing resistance. This mechanism is particularly relevant for large-diameter storage tanks, where relatively small average settlements can induce significant differential movements across the foundation footprint. In the present study, however, this mechanism is treated as a prudent interpretive hypothesis rather than as a demonstrated failure mode.

The larger cumulative settlement observed at TK-6 compared to TK-5 can be plausibly attributed to lateral variability in the thickness and mechanical properties of the soft cohesive and marine sedimentary layers. Even modest differences in sediment thickness or stiffness may result in markedly different settlement responses under similar loading conditions. This interpretation is supported by the greater amplitude of InSAR-derived displacements and the more pronounced structural damage observed at TK-6.

The shallow groundwater table plays a critical role in controlling soil behaviour. Persistent saturation reduces effective stress within the soft cohesive layers and marine sediments, leading to reduced shear strength and increased compressibility. Seasonal fluctuations in groundwater level associated with tidal oscillations and rainfall events promote cyclic changes in pore water pressure, which may accelerate consolidation and progressive settlement over time (Gezgin, 2022). Within that framework, the cyclical displacement behaviour observed in the time-series is consistent with hydrologically modulated settlement superimposed on a longer-term subsidence trend.

The structural loading imposed by the fuel-storage tanks represents an additional contributing factor. Large-diameter tanks apply sustained vertical stresses over extensive foundation areas. When such loads are transferred to heterogeneous reclamation fills overlying soft natural deposits, stress concentrations and non-uniform deformation may develop. Regional tectonic structures, including the Mao, Moa and Cabañas faults, provide the broader geological context of the site, together with recurrent low-magnitude seismicity in eastern Cuba; however, in accordance with the scope of this study, that seismicity is treated only as a regional context rather than as a demonstrated direct cause of the measured settlement. Overall, subsidence at the site is governed by the interaction between soft saturated soils, shallow groundwater conditions, heterogeneous foundation materials and sustained structural loading. None of these factors alone is sufficient to explain the observed deformation; rather, it is their combined effect that controls the magnitude, spatial variability and progressive nature of settlement at the Moa Port fuel-storage facility.

This study presents an integrated assessment of land subsidence affecting fuel-storage tanks at the port of Moa in Holguín province, north-eastern Cuba. The results provide new insight into the deformation behaviour of critical coastal infrastructure founded on soft reclaimed materials and marine soils. An engineering-geological model was developed using the available investigation data and borehole records. It confirms that the tanks are underlain by a heterogeneous reclamation fill over very soft cohesive soils and marine sediments. These ground conditions are unfavourable for the applied static loads and high groundwater levels: the combination of a thin structural slab (0.8–1.2 m thick) with sustained vertical loading and a shallow water table (∼0.8 m) promotes excessive settlement. In other words, the foundation configuration is not well suited to limit consolidation under the tanks’ operational loads and saturated conditions.

Multi-temporal InSAR analysis quantified average vertical deformation rates on the order of −0.9 to −1.4 mm/year for Tank TK-5, and up to −2.2 to −2.5 mm/year for Tank TK-6, over the 2014–2021 period. The cumulative displacements reach approximately −6.4 mm at the reference point on TK-5 and −15.1 mm on TK-6. These results reflect the influence of seasonal hydrological conditions on settlement: higher rates coincided with the wettest months of each year, consistent with a partially water-controlled consolidation process.

Field inspections conducted in late 2022 documented significant structural distress at the tanks. Persistent cracking of about 5.0 cm was measured at the metal-to-concrete base joints, and vertical differential settlements of up to 6.0 cm were observed in the foundations. The observed deformation is clearly attributable to differential settlement and rotation of the foundations rather than uniform vertical sinking. This explains why relatively small average displacements from InSAR are associated with large damage: the radar measurements capture overall vertical movements of scatterers, but structural damage responds to localised gradients and rotations.

Interpretation of the combined data set indicates that the dominant cause of the subsidence is the presence of very soft, highly compressible soils of both reworked lateritic and marine origin beneath the foundations. Lateral variations in the thickness or stiffness of these layers provide a plausible explanation for the more severe deformation observed at TK-6 compared to TK-5. Specifically, even modest differences in subsoil conditions can lead to significantly different settlement responses under the same loading. Subsidence at the site is governed by the interaction of multiple factors: the soft saturated soil stratigraphy, the shallow groundwater table and the sustained tank loads. Alone, none of these factors fully explains the measured settlement; together they control the magnitude and spatial pattern of deformation. The tectonic framework of the region (including nearby faults and recurrent low-magnitude seismicity) is noted for completeness, but no direct co-seismic effects were observed. Thus, seismicity is considered a regional context rather than a demonstrated cause of the subsidence in this case.

In summary, the results demonstrate the value of integrating PS-InSAR data with geotechnical site characterisation and field surveys for evaluating subsidence risk. The deformation behaviour at Moa Port reflects the fundamental geotechnical setting of large, fully loaded tanks on soft reclaimed ground. This combined approach yields actionable information for monitoring and managing similar coastal infrastructure built on weak, water-saturated soils.

Vystrchil M. G. prepared the original manuscript draft. Ramirez-Melendez L. A. performed the satellite image processing using Sarproz and contributed to writing the manuscript. Herrera-Blanco W. contributed to drafting the manuscript and to the interpretation of the results. Acosta-González L. E. conceived the research idea and contributed to data interpretation. Ruiz-Armenteros A. M. processed the synthetic aperture radar data using Sarproz, contributed to data interpretation, and revised the manuscript. All authors reviewed and approved the final manuscript.