Data Science Insight #2- GIS and Machine Learning with Earthquakes and Disaster Managment

Weak building infrastructure is certainly an issue in developing countries of the world. When people in these countries don’t have an abundance of money, corners tend to be cut, and safety becomes a problem. Not only is the physical safety of the buildings an issue, but disaster management is also an obstacle for these countries. However, these ailments can be improved, at the very least to an extent, using geographic information systems (GIS) and machine learning.

GIS compiles data previously collected from land areas of a country, like soil conditions, information about bodies of water, and previous performance of buildings against earthquakes to make various models that officials can study and direct funding through. Machine learning for earthquakes in developing countries, and even developed ones, is about utilizing time and location sensitive information in real time to help first responders get to the highest risk and highest damage areas. Saving time during an emergency can save people’s lives. That is the most important benefit of using data science in this area of disaster management.

One of the deadliest earthquakes in history was the 2005 Kashmir earthquake. It killed upwards of 80,000 people, in the disputed Kashmir region. Over 4 million people were estimated to have lost their homes. One of the people with a first hand account of the earthquake is Ahmad Wani. He was not harmed in the earthquake, but witnessed the destruction around him. Later in his life, he was able to succeed academically and reached Stanford’s graduate business school in 2013. Using GIS and machine learning, he created a preliminary project to see if using these data science methods could accurately predict locations and amount of damages to areas or buildings. He was successful, attracted investors, and eventually founded the company named One Concern.

Seismic Concern, the company’s web platform in 2016, uses GIS for their predictive maps. In real time, they can give rapid hotspot reports. These reports, according to Wani, are about 80 percent accurate. However, after using social media like twitter, they can increase the accuracy rate using the information embedded in tweets to gauge the damages. Wani says, “if there is even a moderate 6.5 magnitude earthquake, you can expect to see around 500,000 tweets… We basically squeeze out all of the relevant possible information we can from the tweet and enrich our map using it.”

Now, One Concern is expanding into other types of natural disasters, like hurricanes, with the aim of helping people and decreasing risks and amounts of damage from these natural disasters. In June, Wani was interviewed by Medium.com and said this, “In a few months, in between Summer and Autumn, they (Japan) will be at the peak of their cyclone season. They have a significant senior citizen population rate, and if they have to evacuate during a cyclone, we might be potentially looking at a spike in COVID-19 cases — potentially by several orders of magnitude. This is a similar situation to several cities in the US. So we have been interested in calculating the risk of COVID-19 spread during an evacuation, first by estimating how many people would potentially be at risk of evacuation based on data from previous US hurricanes. Then we started to layer in our pandemic models, to understand what kind of shelters we should actually prioritize, what sort of mitigative actions we should take in those shelters to curb an outbreak, and what sort of people should be taken to which shelters. We were really running this as a data-driven, machine learning exercise, to see how cities can avoid a spike — that they evacuate appropriately.” These types of data science methods are being used to potentially save lives everywhere, and are important to keep pursuing and improving upon.

In addition to GIS and machine learning being used to help with disaster management and prevention, data science tools are being used to identify precursors to certain types of earthquakes. A new study in Science Advances, Columbia researchers say they are using these tools to analyze 46,000 earthquakes in California. The analysis of these earthquakes is revealing that (at least in the Geysers geothermal field) in California, they are finding patterns matching the “seasonal rise and fall of water-injection flows into the hot rocks below, suggesting a link to… triggering an earthquake,” according to the Earth Institute at Columbia University.

Here are some sources for full information about the topics above and for further research:

https://gcn.com/blogs/emerging-tech/2016/05/earthquake-damage-predict.aspx

https://medium.com/@oneconcerninc/a-q-a-with-ahmad-wani-ceo-one-concern-eb89cf2c0a8b

https://towardsdatascience.com/earthquake-prediction-faffd7160f98

https://blogs.ei.columbia.edu/2018/05/23/69994/