New study uses mathematical analysis of walking patterns for early detection of Parkinson’s disorder.

Combining satellite imagery and sonar data to measure the depth of shallow water bodies

Read time: 2 mins
Mumbai
14 Jun 2018

A team of researchers from Indian Institute of Technology Bombay, Mumbai, UNESCO-IHE Institute for Water Education, Netherlands, and the Department of Harbour, Coastal and Offshore engineering, Netherlands have come up with a cost effective and efficient method to measure the depth of shallow water. Combining satellite imagery and echo-sounding data, along with a machine learning technique, the team has produced an economical solution to the depth measurement challenge.

Oceans cover around 70 percent of the surface of our planet and hold around 90 percent of its water. They also play a major role in human survival, not only providing a constant source of food but also means for cheaper transportation. Ocean floors, however are yet to be completely mapped, making charting trade routes or exploring oceans a challenge. Bathymetry—the study of underwater depths of water bodies like lakes and oceans, conducted using sonar techniques are the preferred methods for mapping lake beds and ocean floors. Mapping the entire bed of rock under the oceans using methods like echolocation and sonar, would require an enormous investment that no single country could afford by itself. An alternative would be to use images from satellites orbiting over the oceans. This has been a preferred method to map the ocean floors and to measure the depth of water, but lacks the precision of measurements done using sonar methods.

To overcome the shortcomings of both the methods, the new study proposes a combination of the two methods. The study uses a machine learning technique called Support Vector Machine (SVM), and data from echo-sounding (a type of sonar) measurements and satellites, to successfully measure the depths of Sint Maarten Island and Ameland Inlet in Netherlands.

Researchers used Landsat Enhanced Thematic Mapper Plus (Landsat 7 ETM+) and Landsat 8 Operational Land Imager (OLI) imagery with 30 m spatial resolution for the study. Of the echo-sounding measurements obtained, 80% of the data was used to train the machine learning algorithm, SVM, while 20% was used to test.

The results are encouraging, with a very low error rate of around 8.26% and 14.43% for Sint Maarten Island and Ameland Inlet respectively. The results were also comparable, and in some cases better, than conventionally used methods linear transform model and ratio transform model.

“Based on the results, it is evident that SVM provides a comparable or better performance for shallow depth ranges and can be used effectively for deriving accurate and updated medium resolution bathymetric maps” remark the authors about the results of the study.