Crevasse Detection in Ice Sheets Using Ground Penetrating Radar and Machine Learning

Abstract

This paper presents methods to automatically classify ground penetrating radar (GPR) images of crevasses on ice sheets. We use a combination of support vector machines (SVMs) and hidden Markov models (HMMs) with down sampling, a preprocessing step that is unbiased and suitable for real-time analysis and detection. We perform modified crossvalidation experiments with 129 examples of Greenland GPR imagery from 2012, collected by a lightweight robot towing a GPR. In order to minimize false positives, an HMM classifier is trained to prescreen the data and mark locations in the GPR files to evaluate with an SVM, and we evaluate the classification results with a similar modified cross-validation technique. The combined HMM-SVM method retains all of the correct classifications by the SVM, and reduces the false positive rate to 0.0007. This method also reduces the computational burden in classifying GPR traces because the SVM is evaluated only on select prescreened traces. Our experiments demonstrate the promise, robustness, and reliability of real-time crevasse detection and classification with robotic GPR surveys.Index Terms-Geophysical signal processing, ground penetrating radar (GPR), robotic sensing systems.

Estimation of Snow Accumulation on Samudra Tapu Glacier, Western Himalaya Using Airborne Ground Penetrating Radar

“…However, ground penetrating radar (GPR), a non-destructive technique, can be used to collect both point as well as spatial distribution of snow thickness. It has been widely used in ground mode, for snow and ice thickness measurements [8][9][10][11][12][13][14][15] , snowpack stratigraphic delineation 16 and to study the subsurface properties of other strata 17 . Forte et al 18,19 reported the applications of GPR data in determining the density and electromagnetic (EM) wave velocity for snow, firn and ice.…”

Automated Target Detection for Geophysical Applications

“…Many studies were performed on detecting different type of anomalies: mine detection ( [5][6][7][8]) sinkholes [9], subterranean voids and tunnel detection [10,11], buried utilities [12] cracks in ice sheets [13]) as well as for bridges and roads assessment [14]. The vast majority of detection methods involve some pre-processing stage (for example: Qin et Al.…”

Foreword to the Special Issue on Pattern Recognition in Remote Sensing

“…Such study is important to assess the factors controlling the development of debris flows and to identify the areas susceptible to their occurrences. In [24], methods are presented to automatically classify GPR images of crevasses on ice sheets using a combination of SVMs and hidden Markov models (HMMs). The combined HMM-SVM method retains all of the correct classifications by the SVM, reduces the false positive rate, and also reduces the computational burden in classifying GPR traces.…”