PSI - Issue 44

Andrea Nettis et al. / Procedia Structural Integrity 44 (2023) 1996–2003 Andrea Nettis et al. / Structural Integrity Procedia 00 (2022) 000 – 000

1999

4

March 2001. Additionally, Macchiarulo et al., (2022) and Farneti et al., (2022) analyse the collapse of the Himera viaduct in Sicily and the Albiano Magra bridge in Tuscany, respectively, whose collapses were (likely) caused by ongoing landslides. The use of MTInSAR as an early warning for the Polcevera bridge in Genova which collapsed in 2019 is discussed by (Milillo et al., 2019). 3. Application of MTInSAR for the bridges of the primary road network in Rome This study is focused on exploring the potentials of MTInSAR for monitoring bridge portfolios. To this aim, the highway networks of Roma in Central Italy is selected as a case study. First, a description of the case-study database and the methodology adopted for interferometry analysis is provided. Subsequently, the results are discussed with reference to the entire bridge portfolio and to the PS time series on several selected test bridges. 3.1. Description of the case study and input SAR dataset Within this study, the highway network in the metropolitan area of Roma (Lazio, Italy) is selected as a case study. Geospatial data of the case-study network is derived in the format of shapefiles by using the OpenStreetMap database (OpenStreetMap contributors, 2017). It includes several attributes such as geographical position and type of the roads in Rome (Fig. 1). The QGIS software package is used as a geographical information system platform for the post processing elaborations described as follows. Line-type features representing bridges are extracted from this database by using the appropriate query. The case-study network includes 244 motorway bridges and 268 highway bridges (labelled as “primary” and “trunk” in OpenStreetMap database). Both Sentinel-1 (C-band) and COSMO-SkyMed (X band) SAR datasets collected in the five-year period between January 2015 and September 2020 are used for the interferometry analysis. The X-band COSMO-SkyMed dataset is characterized by a high spatial resolution (3 m) and allows for a precise (< 1 m) estimation of the PS height. C-band Sentinel-1 dataset offers a poor spatial resolution (20 m) but is characterised by a high temporal resolution (6 days) which is more suitable for monitoring non-linear displacements with respect to the former dataset. 3.2. Methodology for interferometry analysis The interferometry analysis is carried out through the Stable Point INterferometry even over Un-urbanised Areas technique (synthetically defined as SPINUA) by Bovenga et al., (2005). A thorough analysis of the SPINUA processing chain is considered out-of-topic in this study and interested readers are referred to the appropriate abovementioned references. Generally, with an appropriate definition of the processing parameters, the algorithm reaches an accuracy higher than 1÷2 mm/year for measurement points with a distance under five kilometres from the selected reference point. Finally, SPINUA produces a table containing for each PS: (1) geographic position (latitude, longitude, altitude), (2) average velocities (for each observation year and for the whole observation period), (3) several parameters describing non-linear trends (e.g. acceleration, seasonality), (4) temporal displacements trend, (5) accuracy of the displacements (long term coherence and standard deviation of the measurements). In this study, once the SPINUA algorithm is applied, a post-processing phase for associating PS displacement time-series to specific bridge footprints and identifying a deformation scenario is performed. This phase consists of a spatial resampling phase, performed by using a clustering approach oriented to associate PSs with several bridge longitudinal sections. This can be performed in a GIS environment for all the investigated bridges considering the following steps. 1) Each bridge line-type feature is discretised in several segments by using point features. 2) A buffer polygon is created around each of the generated points representing a bridge region. The total area of the buffer polygons should cover the superstructure width plus a value of tolerance related to the accuracy in PS positioning. 3) PSs and their displacement time series are associated with each bridge region based on their geospatial position and their height information (i.e. clustering). An input Digital Surface Model may be suitable to preliminary assign altitude information to regions of the bridge footprints. 4) The average annual velocity of PSs clustered in a given bridge region is calculated. It is indicated as ƒ† is assigned as an attribute to the centroid of the region.