Sub-pixel Image Correlation Research Articles

When image correlation is needed: Unravelling the complex dynamics of a slow-moving landslide in the tropics with dense radar and optical time series

Slow-moving landslides exhibit persistent but non-uniform motion at low rates which makes them exceptional natural laboratories to study the mechanisms that control the dynamics of unstable hillslopes. Here we leverage 4.5+ years of satellite-based radar and optical remote sensing data to quantify the kinematics of a slow-moving landslide in the tropical rural environment of the Kivu Rift, with unprecedented high spatial and temporal resolution. We measure landslide motion using sub-pixel image correlation methods and invert these data into dense time series that capture weekly to multi-year changes in landslide kinematics. We cross-validate and compare our satellite-based results with very-high-resolution Unoccupied Aircraft System topographic datasets, and explore how rainfall, simulated pore-water pressure, and nearby earthquakes control the overall landslide behaviour. The landslide exhibited seasonal and multi-year velocity variations that varied across the landslide kinematic units. While rainfall-induced changes in pore-water pressure exerts a primary control on the landslide motion, these alone cannot explain the observed variability in landslide behaviour. We suggest instead that the observed landslide kinematics result from internal landslide dynamics, such as extension, compression, material redistribution, and interactions within and between kinematic units. Our study provides, a rare, detailed overview of the deformation pattern of a landslide located in a tropical environment. In addition, our work highlights the viability of sub-pixel image correlation with long time series of radar-amplitude data to quantify surface deformation in tropical environments where optical data is limited by persistent cloud cover and emphasize the importance of exploiting synergies between multiple types of data to capture the complex kinematic pattern of landslides.

Geometric Variation in the Surface Rupture of the 2018 Mw7.5 Palu Earthquake from Subpixel Optical Image Correlation

We obtained high-resolution (10 m) horizontal displacement fields from pre- and post-seismic Sentinel-2 optical images of the 2018 Mw7.5 Palu earthquake using subpixel image correlation. From these, we calculated the curl, divergence, and shear strain fields from the north-south (NS) and east-west (EW) displacement fields. Our results show that the surface rupture produced by the event was distributed within the Sulawesi neck (0.0974–0.6632°S) and Palu basin (0.8835–1.4206°S), and had a variable strike of 313.0–355.2° and strike slip of 2.00–6.62 m. The NS and EW displacement fields within the Palu basin included fine-scale displacements in both the near- and far-fault, the deformation patterns included a small restraining bend (localized shortening), a distributed rupture zone, and a major releasing bend (net extension) from the curl, divergence, and shear strain. Surface rupture was dominated by left-lateral strike-slip from initiation to termination, with a localized normal slip component peaking at ~3.75 m. The characteristics and geometric variation of the ruptured fault controlled both the formation of these surface deformation patterns and sustained supershear rupture.

Using Daily Observations from Planet Labs Satellite Imagery to Separate the Surface Deformation between the 4 July Mw 6.4 Foreshock and 5 July Mw 7.1 Mainshock during the 2019 Ridgecrest Earthquake Sequence

Abstract On 4 July 2019, the Ridgecrest earthquake sequence began with a series of foreshocks including an Mw 6.4 event near Searles Valley, California. This was then followed 34 hr later by an Mw 7.1 mainshock located just 15 km to the north, with the earthquake sequence resulting in a complex array of intersecting faults. This earthquake sequence poses several interesting questions including, did the stress changes induced by the Mw 6.4 foreshock trigger the Mw 7.1 mainshock and what possible mechanism(s) could explain the occurrence of widespread secondary faulting surrounding both surface ruptures? However, most of the geodetic data (such as Interferometric Synthetic Aperture Radar, light detection and ranging, and optical satellite imagery) were acquired after both events had occurred making it difficult to discern which surface fractures happened when and their possible triggering mechanism. Here, we provide a dataset composed of high-resolution optical imagery, pixel-value difference maps, .kmz fracturing mapping, and horizontal deformation maps derived from subpixel image correlation, which can uniquely separate the surface fracturing and deformation between the foreshock and mainshock events that can help answer these questions. Separate imaging of the events is made possible by the daily acquisition of optical imagery by the Planet Labs cubesat constellation, which acquired data between the two earthquakes, in the morning of 4 and 5 July, at 11.13 a.m. and 05.12 p.m. PST, respectively, with the images acquired just 40 min after the foreshock and 56 min before the mainshock, respectively. Analysis from this optical imagery reveals the location of surface faulting that allows us to map their spatial extent and determine their timing. These data which we provide here can help guide and validate field survey observations to help understand which faults ruptured when, and constrain slip inversion models for more accurate estimates of stress changes induced by the foreshock imposed on the surrounding faults.

Three‐Dimensional Surface Deformation in the 2016 MW7.8 Kaikōura, New Zealand, Earthquake From Optical Image Correlation: Implications for Strain Localization and Long‐Term Evolution of the Pacific‐Australian Plate Boundary

AbstractWe generated dense, high‐resolution 3‐D ground displacement maps for the 2016 MW7.8 Kaikōura, New Zealand earthquake—the most geometrically and kinematically complex rupture yet recorded—from stereo WorldView optical satellite imagery using a new methodology that combines subpixel image correlation with a ray‐tracing approach. Our analysis reveals fundamental new details of near‐field displacement patterns, which cannot easily be obtained through other methods. From our detailed correlation maps, we measured fault slip in 3‐D along 19 faults at 500‐m spacing. Minimum resolvable horizontal slip is ~0.1 m, and vertical is ~0.5 m. Net slip measurements range from <1 to ~12 m. System‐level kinematic analysis shows that slip on faults north of the Hope fault was oriented primarily subparallel to the Pacific‐Australian plate motion direction. In contrast, slip on faults to the south was primarily at high angle to the plate motion and secondarily parallel to plate motion. Fault kinematics are in some locations consistent with long‐term uplift patterns, but inconsistent in others. Deformation within the Seaward Kaikōura Range may indicate an attempt by the plate boundary fault system to geometrically simplify. Comparison of published field measurements along the Kekerengu fault with our correlation‐derived measurements reveals that ~36% of surface displacement was accommodated as distributed off‐fault deformation when considering only field measurements of discrete slip. Comparatively, field measurements that project previously linear features (e.g., fence lines) into the fault over apertures >5–100 m capture nearly all (~90%) of the surface deformation.

Displacement measurement with nanoscale resolution using a coded micro-mark and digital image correlation

A method for simple and reliable displacement measurement with nanoscale resolution is proposed. The measurement is realized by combining a common optical microscopy imaging of a specially coded nonperiodic microstructure, namely two-dimensional zero-reference mark (2-D ZRM), and subsequent correlation analysis of the obtained image sequence. The autocorrelation peak contrast of the ZRM code is maximized with well-developed artificial intelligence algorithms, which enables robust and accurate displacement determination. To improve the resolution, subpixel image correlation analysis is employed. Finally, we experimentally demonstrate the quasi-static and dynamic displacement characterization ability of a micro 2-D ZRM.

Surface slip and off‐fault deformation patterns in the 2013 MW 7.7 Balochistan, Pakistan earthquake: Implications for controls on the distribution of near‐surface coseismic slip

AbstractComparison of 398 fault offsets measured by visual analysis of WorldView high‐resolution satellite imagery with deformation maps produced by COSI‐Corr subpixel image correlation of Landsat‐8 and SPOT5 imagery reveals significant complexity and distributed deformation along the 2013 Mw 7.7 Balochistan, Pakistan earthquake. Average slip along the main trace of the fault was 4.2 m, with local maximum offsets up to 11.4 m. Comparison of slip measured from offset geomorphic features, which record localized slip along the main strand of the fault, to the total displacement across the entire width of the surface deformation zone from COSI‐Corr reveals ∼45% off‐fault deformation. While previous studies have shown that the structural maturity of the fault exerts a primary control on the total percentage of off‐fault surface deformation, large along‐strike variations in the percentage of strain localization observed in the 2013 rupture imply the influence of important secondary controls. One such possible secondary control is the type of near‐surface material through which the rupture propagated. We therefore compared the percentage off‐fault deformation to the type of material (bedrock, old alluvium, and young alluvium) at the surface and the distance of the fault to the nearest bedrock outcrop (a proxy for sediment thickness along this hybrid strike slip/reverse slip fault). We find significantly more off‐fault deformation in younger and/or thicker sediments. Accounting for and predicting such off‐fault deformation patterns has important implications for the interpretation of geologic slip rates, especially for their use in probabilistic seismic hazard assessments, the behavior of near‐surface materials during coseismic deformation, and the future development of microzonation protocols for the built environment.

Surface reconstruction and landslide displacement measurements with Pléiades satellite images

Recent advances in image-matching techniques and VHR satellite imaging at submeter resolution theoretically offer the possibility to measure Earth surface displacements with decimetric precision. However, this possibility has yet not been explored and requirements of ground control and external topographic datasets are considered as important bottlenecks that hinder a more common application of optical image correlation for displacement measurements. This article describes an approach combining spaceborne stereo-photogrammetry, orthorectification and sub-pixel image correlation to measure the horizontal surface displacement of landslides from Pléiades satellite images. The influence of the number of ground-control points on the accuracy of the image orientation, the extracted surface models and the estimated displacement rates is quantified through comparisons with airborne laser scan and in situ global navigation satellite measurements at permanent stations. The comparison shows a maximum error of 0.13m which is one order of magnitude more accurate than what has been previously reported with spaceborne optical images from other sensors. The obtained results indicate that the approach can be applied without significant loss in accuracy when no ground control points are available. It could, therefore, greatly facilitate displacement measurements for a broad range of applications.

Fusion of D-InSAR and sub-pixel image correlation measurements for coseismic displacement field estimation: Application to the Kashmir earthquake (2005)

In geophysics, the uncertainty associated with model parameters or displacement measurements plays a crucial role in the understanding of geophysical phenomenon. An emerging way to reduce the geodetic parameter uncertainty is to combine a large number of data provided by SAR images. However, the measurements by radar imagery are subject to both random and epistemic uncertainties. Probability theory is known as the appropriate theory for random uncertainty, but questionable for epistemic uncertainty. Fuzzy theory is more adapted to epistemic uncertainty. Moreover, in a context of random and epistemic uncertainties, the conventional joint inversion in the least squares sense cannot be considered any more as the best scheme to reduce uncertainty. Therefore, in this article, in addition to joint inversion, two other fusion schemes, pre-fusion and post-fusion, are proposed. We consider here the conventional approach and an original fuzzy approach for handling random and epistemic uncertainties of D-InSAR and sub-pixel image correlation measurements. Joint inversion and pre-fusion are then applied to the measurement of displacement field due to the 2005 Kashmir earthquake by fusion of these data. The behaviours of these two fusion schemes versus uncertainty reduction are highlighted through comparisons of results.

Study of the performance of different subpixel image correlation methods in 3D digital image correlation

The three-dimensional digital image correlation (3D-DIC) method is rapidly developing and is being widely applied to engineering and manufacturing. Despite its extensive use, the error caused by different image matching algorithms is seldom discussed. An algorithm for 3D speckle image generation is proposed, and the performances of different subpixel correlation algorithms are studied. The advantage is that there is no interpolation bias of texture in the simulation before and after deformation, and the error from the interpolation of speckle can be omitted in this algorithm. An error criterion for 3D reconstruction is proposed. 3D speckle images were simulated, and the performance of four subpixel algorithms is addressed. Based on the research results of different subpixel algorithms, a first-order Newton-Raphson iteration method and gradient-based method are recommended for 3D-DIC measurement.

Deformation between 1989 and 1997 at Piton de la Fournaise volcano retrieved from correlation of panchromatic airborne images

We processed two panchromatic airborne images in the attempt to measure deformation on Piton de la Fournaise volcano between 1989 and 1997. Piton de la Fournaise erupted three times during 1989-1997 with fissures opening near Dolomieu summit crater. During this period no Synthetic Aperture Radar Interferometry data were available and only few geodetic points surveyed the summit crater. We therefore, apply for the first time the subpixel image-correlation technique on airborne images for monitoring ground deformation on Piton de la Fournaise volcano. Correlogram analysis reveals cumulative deformation in the summit area. We observe ESE spreading of the summit caused by the inflation-deflation cycle during 1989-1997. We calculated relative ground displacement along both sides of the main eruptive fissures, finding a maximum of 1.8 (±0.42) m opening. Direct comparison between deformation retrieved by correlation and available geodetic data at four locations shows qualitative agreement. We demonstrate the potential of subpixel image correlation technique, using panchromatic airborne images, for deformation monitoring on volcanoes. We also illustrate the capability of this technique for remote lava flow mapping.

Sub-pixel image correlation: an alternative to SAM and dye penetrant for crack detection and mechanical stress localisation in semiconductor packages

Sub-pixel image correlation has been applied to the measurement of surface deformation, in particular to the measurement of thermally induced strain in microchips. A Fourier shift technique coupled with successive approximation is used to determine the sub-pixel component of the deformation. This reduces the effects of systematic errors on data associated with interpolation or curve-fitting present in other sub-pixel techniques. For all image correlation techniques, defocus errors can have serious effects on the accuracy of the technique. These errors are eliminated in this case through use of a set of images at different axial positions, where only the in focus information is used. This allows determination of the out-of-plane component of the displacement to within roughly 1/40 of the depth of focus (250 nm at ×10 magnification). An in-plane deformation/displacement uncertainty of roughly 2% of a pixel (<15 nm at ×10 magnification) is achievable on real-world samples.