Surface Motion Prediction and Mapping for Road Infrastructures Management by PS-InSAR Measurements and Machine Learning Algorithms

Abstract

This paper introduces a methodology for predicting and mapping surface motion beneath road pavement structures caused by environmental factors. Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR) measurements, geospatial analyses, and Machine Learning Algorithms (MLAs) are employed for achieving the purpose. Two single learners, i.e., Regression Tree (RT) and Support Vector Machine (SVM), and two ensemble learners, i.e., Boosted Regression Trees (BRT) and Random Forest (RF) are utilized for estimating the surface motion ratio in terms of mm/year over the Province of Pistoia (Tuscany Region, central Italy, 964 km2), in which strong subsidence phenomena have occurred. The interferometric process of 210 Sentinel-1 images from 2014 to 2019 allows exploiting the average displacements of 52,257 Persistent Scatterers as output targets to predict. A set of 29 environmental-related factors are preprocessed by SAGA-GIS, version 2.3.2, and ESRI ArcGIS, version 10.5, and employed as input features. Once the dataset has been prepared, three wrapper feature selection approaches (backward, forward, and bi-directional) are used for recognizing the set of most relevant features to be used in the modeling. A random splitting of the dataset in 70% and 30% is implemented to identify the training and test set. Through a Bayesian Optimization Algorithm (BOA) and a 10-Fold Cross-Validation (CV), the algorithms are trained and validated. Therefore, the Predictive Performance of MLAs is evaluated and compared by plotting the Taylor Diagram. Outcomes show that SVM and BRT are the most suitable algorithms; in the test phase, BRT has the highest Correlation Coefficient (0.96) and the lowest Root Mean Square Error (0.44 mm/year), while the SVM has the lowest difference between the standard deviation of its predictions (2.05 mm/year) and that of the reference samples (2.09 mm/year). Finally, algorithms are used for mapping surface motion over the study area. We propose three case studies on critical stretches of two-lane rural roads for evaluating the reliability of the procedure. Road authorities could consider the proposed methodology for their monitoring, management, and planning activities.

Can Machine Learning and PS-InSAR Reliably Stand in for Road Profilometric Surveys?

“…The technique can be efficiently employed for detecting surface motion patterns in the context of slow or relatively slow movements due to human-related [ 18 , 19 ] or environmental-related activities, e.g., subsidence [ 20 , 21 , 22 ], sinkholes [ 23 , 24 ], and landslides [ 25 , 26 , 27 , 28 , 29 , 30 ]. Readers may find additional valuable sources in Reference [ 8 ].…”

“…The high capabilities of SAR sensors in terms of measurement reliability, accuracy, speed of execution, high coverage, and the possibility of data processing in near real-time and back in time have meant a bursting development of SAR-based products as high-performance NDT in monitoring and inspection activities of linear infrastructures. In the literature, there are several SAR-based applications, mainly PS-InSAR surveys, of infrastructure monitoring activities, e.g., road infrastructures [ 6 , 7 , 8 , 31 , 32 , 33 , 34 , 35 ], rail infrastructures [ 36 , 37 , 38 , 39 , 40 ], airport runways [ 41 ], and bridges [ 42 , 43 , 44 , 45 ]. In these research studies, the common objective is to identify critical infrastructural sections by processing SAR images and comparing these surveys with other techniques (e.g., leveling or GPS surveys).…”

“…Such roads are monitored and managed by the TRRA, which provided all the required environmental data for developing the previously mentioned SAR-based MLAs [ 7 , 8 ]. Two-lane roads are defined in the Highway Safety Manual [ 80 ] as single carriageway roads with one lane for each travel direction.…”

“…In our previous works [ 7 , 8 ], we calibrated some MLAs for predicting and mapping the average vertical displacement of the road pavement surface as a result of exogenous phenomena occurrence. Exogenous events are defined as external factors to the infrastructure.…”

“…We aim to check if PS-InSAR is sufficiently able to detect surface irregularities highlighted by a profilometric survey. The proposed methodology stems from two previous studies [ 7 , 8 ], in which we calibrated several Machine Learning Algorithms (MLAs) for predicting vertical deformation of road pavement surfaces, in terms of [mm/year], by linking PS-InSAR-based outcomes and several environmental parameters. This modeling phase was due to insufficient PS coverage on the road surface of two-lane rural roads.…”