Satellite Radar Reveals Rainfall Drives the Speed of Cyprus Landslides

Slow-moving landslides rarely make headlines the way earthquakes or volcanic eruptions do, yet they quietly destroy roads, fracture buildings and swallow farmland across the Mediterranean every year. Now a team of researchers from the University of Cyprus and the Istituto Nazionale di Geofisica e Vulcanologia in Rome has combined satellite radar measurements with geological and geotechnical data to answer a deceptively simple question: what actually controls how fast a landslide moves? Their study, published in Scientific Reports, focuses on southwestern Cyprus, an island whose soft, clay-rich rocks have long made it a natural laboratory for slope failure. By statistically linking radar-derived ground velocities to the activity class of mapped landslides and to a suite of causal factors, the team has produced one of the clearest pictures yet of the interplay between geology and climate in shaping ground instability.

The technological heart of the study is Multi-Temporal Interferometric Synthetic Aperture Radar, or MT-InSAR, a technique that has transformed how scientists monitor ground deformation. Synthetic aperture radar satellites bounce microwave signals off the Earth’s surface and record the phase of the returning waves. When the same point on the ground is imaged repeatedly over months or years, tiny changes in the distance between the satellite and the surface shift the phase of the reflected signal. By comparing these phase differences across many acquisitions, researchers can measure ground movement with millimetre-level precision, without ever setting foot on the unstable slope. The Cyprus team processed both ascending and descending orbit datasets, meaning the satellite flew over the study area from different directions, which allowed them to sample the movement of the ground along two distinct lines of sight and to build a more complete picture of the three-dimensional deformation field.

The researchers applied this technique to landslides already catalogued in an existing inventory of southwestern Cyprus, covering slopes classified as active, dormant and relict. This classification matters because it reflects the state of a landslide at the time of mapping: active landslides are currently moving, dormant ones have not moved recently but retain the potential to reactivate, and relict ones are ancient features formed under different climatic conditions that are generally considered stable. The team extracted the Line-of-Sight, or LOS, velocities for each landslide from the radar data and then asked whether the measured rates differed systematically between these activity classes. The answer was emphatic. The largest LOS velocities were observed in landslides characterised as active, followed by dormant and then relict landslides, a hierarchy that confirms the radar measurements are capturing genuine differences in slope behaviour rather than noise.

That validation is more important than it might sound. One of the persistent challenges in landslide science is that inventories are often compiled from aerial photographs, historical records and field visits, all of which involve a degree of subjective judgement. If satellite-derived velocities align cleanly with independently assigned activity classes, it suggests both datasets are reliable and, more intriguingly, that radar could in future be used to update and refine inventories automatically. For a country like Cyprus, where development continues to push onto marginal slopes, an objective, repeatable measure of which landslides are truly moving could directly inform planning decisions and early-warning strategies. The study effectively demonstrates that the velocity signal from space carries real diagnostic information about the state of a landslide.

The second major strand of the analysis asked which materials move fastest, and here the geology of Cyprus provides a striking answer. The largest displacement rates corresponded mostly to outcrops of fine-grained clastic rocks, including marls, radiolarian mudstones and mélange. These rock types share a critical property: weak shear strength. Marls, which are mixtures of clay and calcium carbonate, are notorious in engineering geology because they swell when wetted, slake when dried and lose strength dramatically as pore water pressure rises. Radiolarian mudstones and the chaotic, mixed rock assemblages known as mélange behave similarly, deforming readily under stress. When such weak materials sit on a slope, the resistance to sliding is low, and once movement begins, the ground can creep steadily for years. The radar data confirmed that these lithologies host the fastest-moving terrain in the study area, tying the observed deformation directly to the mechanical properties of the bedrock.

To move beyond simple correlations, the team turned to a machine-learning interpretability tool called Shapley Additive Explanations, or SHAP. Borrowed from cooperative game theory, SHAP analysis quantifies how much each input variable contributes to a model’s prediction for any individual case, distributing credit among factors in a way that is mathematically consistent. In the context of this study, the method allowed the researchers to rank the influence of geological, geomorphological, environmental and geotechnical factors on the measured LOS velocities. The headline finding was unambiguous: rainfall emerged as the primary factor influencing the increase in LOS velocity. In other words, while many ingredients conspire to create a landslide, precipitation is the dominant throttle controlling how quickly the ground moves once instability is established.

This result carries a profound implication for how landslide hazard should be understood and managed. The geological and geotechnical characteristics of a site determine whether a landslide can occur at all; they set the stage by providing weak materials, favourable orientations and susceptible slope geometries. But once a landslide has been initiated, the tempo of its movement is governed largely by the water delivered from above. Rainfall infiltrates the slope, raises pore water pressures within the sliding mass, reduces the effective stress that holds the material together and lubricates the failure surface. The Cyprus findings suggest that monitoring rainfall and incorporating precipitation records into deformation forecasts could substantially improve predictions of when slow-moving slides will accelerate, which is precisely the information that engineers and emergency managers need most.

Among the geotechnical indices examined, two stood out. The Plasticity Index, or PI, a standard measure of how much water a soil can absorb before it transitions between consistency states, generally exhibited a positive correlation with ground velocity. High-PI clays are more compressible, more prone to swelling and shrinkage, and more sensitive to moisture changes, so it makes intuitive sense that slopes founded on such materials move faster. Conversely, the Geological Strength Index, or GSI, a qualitative scale that characterises the overall strength of a rock mass based on its structure and the condition of its discontinuities, demonstrated a weak negative correlation with velocity. Stronger, more intact rock masses move more slowly, as expected, though the weakness of the correlation suggests that intact rock strength plays a secondary role compared with moisture conditions and the intrinsic weakness of fine-grained units.

The choice of Cyprus as a case study gives the work practical weight well beyond the island. Southwestern Cyprus is underlain by the Mamonia Complex and related units, a tectonically jumbled assemblage of weak sedimentary rocks that has produced some of the island’s most problematic ground. Landslides in this region have damaged roads and infrastructure for decades, and the Geological Survey Department of Cyprus, whose scientists provided the geological and geotechnical data underpinning the study, maintains detailed records of the affected areas. The research was funded through the European Union’s Horizon 2020 programme under a Marie Skłodowska-Curie grant, reflecting the European dimension of the problem: weak, clay-bearing formations and intensifying rainfall extremes threaten slopes across the entire Mediterranean basin, from Italy and Greece to Turkey and North Africa.

What makes the study resonate beyond its regional focus is the methodological template it offers. By fusing freely available satellite radar data with conventional geological mapping and laboratory-derived geotechnical indices, and by interrogating the relationships with a modern interpretability framework, the researchers have shown how the pieces of the landslide puzzle can be assembled quantitatively rather than anecdotally. The approach requires no expensive ground instrumentation, can be repeated as new satellite acquisitions arrive, and scales naturally from a single slope to an entire country. As climate change intensifies rainfall variability across the Mediterranean and elsewhere, the ability to identify which slopes are moving, how fast, and what drives their acceleration will become an essential tool for protecting communities. The Cyprus study suggests that the answer, at least for the fastest-moving hazards, may literally be falling from the sky.

Subject of Research: Statistical correlation of InSAR-derived landslide ground movement rates with activity class and geological, geotechnical and climatic causal factors in southwestern Cyprus

Article Title: Correlating InSAR-derived ground movement rates with landslide activity class and causal factors: a case study from Cyprus

Article References: Tzampoglou, P., Loukidis, D., Tolomei, C., & Svigkas, N. (2026). Correlating InSAR-derived ground movement rates with landslide activity class and causal factors: a case study from Cyprus. Scientific Reports. https://doi.org/10.1038/s41598-026-74387-6

Image Credits: AI Generated

DOI: 10.1038/s41598-026-74387-6

Keywords: InSAR, landslides, Cyprus, MT-InSAR, rainfall, SHAP analysis, geotechnical indices, Plasticity Index, Geological Strength Index, marls, remote sensing, natural hazards