Natural disasters tend to be more severe nowadays because of the worsening climate change. AI can learn from disaster data gathered by related organizations. It was trained to analyse and predict the possibility of the next disaster which is quite accurate and reliable. The more in advance we know the prediction, the more advanced we can prepare how to deal with it. This goes especially for the related staff and rescue personnel, who can set a plan on how to help people effectively through the prediction of AI.
Geospatial Data together with weather data, aerial photograph, drone photography, IoT sensor, and disaster statistics were fed to train the AI model. Some models are trustable and accurate which can predict the next disaster 2-3 years in advance such as earthquake, flood, storm, avalanche, etc. One example of one of the most damaging disasters in the World is a Hurricane. If we already able to predict of when it will happen next, we can announce the notice in the danger zone to prepare to evacuate on given date and time and inform where is a safe place to quickly move to, level of damage, as well as inform people in closer proximity to monitor and be aware of the situation.
Under McKinsey’s Nobel Intelligence, one interesting project developed was based on the potential of AI to evaluate the damage of buildings after a disaster by using geo data, weather data and aerial photography. Normally, with manpower, it takes time around weeks but this model can show the result in just a few seconds.
WSL, the Institute for Snow and Avalanche Research in Switzerland, has used seismic sensors combined with a supervised machine learning algorithm to detect the tremors that precede avalanches. Avalanche signals are different from other earthquake signals so the algorithm can detect them automatically. Of course, if we keep gathering avalanche data, the model will be more accurate for prediction next time. This model is very important in terms of saving human lives and property. The weather stations also use the real-time data around the Swiss Alps to develop the snowpack stratigraphy simulation model to monitor and predict avalanches.
Another essential AI role that we can use for disaster management is “Social Listening Tools”. Normally we use this advanced technology for monitoring customer’s opinion through social media including Facebook, YouTube, Twitter, and blogs in order to create a marketing strategy plan. With this idea, we can use the same model to listen & monitor disaster information by setting key words, monitoring hashtag, gathering real time posts and comments from people in the disaster area. The rescue personnel will use the result from social listening tools to help people as fast and efficiently as they can.
Our society is driven by data and AI. Disaster is not as far as we think. The more advanced we are in predicting a natural disaster, the more time we have to prepare.. To train AI models, we need cooperation from related departments to gather & share data which can create accurate AI models.