Detecting spatio-temporal outliers in crowdsourced bathymetry data

Abstract

The widespread availability of Internet access and locationacquisition technologies, such as the global positioning system (GPS), has given rise to the growing phenomenon of Volunteered Geographic Information (VGI). Our work presents the use of VGI in bathymetry and hydrographic surveying and demonstrates that crowdsourced bathymetry data (CSB) can yield valuable knowledge for the maritime community. In this study, CSB data collected from 2012 to 2013 within the Baltimore Inner Harbor was used to locate anomalous depth measurements that could indicate the presence of submerged debris. To this end, we explored two approaches for detecting spatio-temporal outliers in the CSB data. In the first approach, we combined Local Outlier Factor and DBSCAN in an ensemble method to find spatio-temporal clusters of anomalous measurements that could indicate the presence of submerged debris. In the second approach, we calculated a measure of local spatial autocorrelation over time to identify "hotspots" or specific areas that consistently have low depth measurements compared to their immediate neighbors (i.e. "low-high" outliers). Results from both approaches revealed locations within the Fort McHenry Channel whose depth measurements may be indicative of the presence of submerged marine debris and, as such, may pose a threat to the safety of mariners operating in that region. Our results indicate that CSB data can not only help to improve the safety of mariners, but also serve to alert authorities in a timely manner that channel maintenance, a re-survey, and/or changes to the nautical chart may be needed.

A Review of Data Cleaning Approaches in a Hydrographic Framework with a Focus on Bathymetric Multibeam Echosounder Datasets

“…Four approaches are based on simple cascading filter [46][47][48][49]. The number of filters varies from 3 to 5 depending on the approach.…”

“…Most of these filters operate in swath mode with no need to define the size of the neighborhoods, which are directly fixed by the approach. The only parameter to be set by the operator is, if necessary, the size of the group of pings to be processed [46][47][48]. The common basic filters make use of the depth or cross track distance gates [46,47,49] or the number of non-zero neighboring soundings [46] to detect erroneous soundings.…”

“…The only parameter to be set by the operator is, if necessary, the size of the group of pings to be processed [46][47][48]. The common basic filters make use of the depth or cross track distance gates [46,47,49] or the number of non-zero neighboring soundings [46] to detect erroneous soundings. The statistical filters implemented in these approaches are primarily based on the calculation of mean or standard deviation values.…”

“…Two approaches make use of a density-based scheme [32,47]. The first method consists of projecting the data along the orthogonal direction and side direction respectively in the MBES frame.…”

Vessel to shore data movement through the Internet of Floating Things: A microservice platform at the edge

“…Scientific and business projects involve, for example, the measurement of seafloor bottom depth (bathymetry) via data crowdsourcing through data collected by ferries, sport vessels, and cruise ships 11 . Unfortunately, data crowdsourcing in marine applications is limited by the scarce availability of stable, reliable, and cheap data connectivity.…”