NASA’s earthquake prediction coding ‘match’ almost complete

10 Aug 2015

A NASA crowdsourcing exercise to seek a way to predict earthquakes using software algorithms is set to conclude, with US$25,000 up for grabs for the best performers in the contest.

The project is all based on a theory that electromagnetic pulses (EMPs) are precursors to quakes.

As a starting point, NASA released swathes of EMP data collected over a three-month period, some from areas that have suffered earthquakes and some from areas that have not, giving contestants two weeks to establish an applicable code.

It is hosted on Appirio, which NASA partnered with, and the top performer will bag US$10,000 by the conclusion of the project today (10 August).

QuakeFinder, a humanitarian group that’s been trying to predict earthquakes for over a decade, is providing the data for NASA, with fault lines in the US, Greece, Indonesia, Chile, Peru and Taiwan revealing 65 terabytes of data.

With more than 800,000 people in the TopCoder community – of which Appirio operates and is utilising for this contest – NASA’s thinking is sound. Software development is all about crowd-thinking, so why not incentivise those who can help?

A long relationship with the community

NASA has been looking at these types of platforms for a number of years now, with Mike Morris, TopCoder CEO, claiming it’s an easy way to find out the feasibility of problems.

By getting hundreds of people at any one time to put their minds to a project, he argues, dozens of approaches are suggested for investigation. If they don’t work, hosts can then “be confident that it was an unsolvable problem”, he said.

For this project, for example, nobody really knows the full relationship between EMPs and earthquakes.

One theory suggests that fracturing rock in the earth’s crust creates electrical pulses, these rise to the surface offering clues that an earthquake is on the way.

However, there are a number of natural and man-made EMP ‘noise’ sources. These can be natural (lightning, solar storms) or man-made (traffic), meaning it’s difficult to discern what’s relevant and what’s not.

The graphic below, provided by NASA, gives a basic explanation of what it is that the contestants must investigate – click to view it in a larger format.

Earthquake prediction - NASA

“Developing a reliable approach that can separate potential earthquake-induced electromagnetic pulses from the myriad of natural and anthropogenic sources has been a significant challenge,” said Craig Dobson, programme scientist at NASA headquarters in Washington.

“We look forward to seeing the innovative ideas from this competition and learning more about this controversial phenomenon.”

Main image via Shutterstock

Gordon Hunt was a journalist with Silicon Republic