St Edward's Academic Review 2025

ST EDWARD’S, OXFORD

Conclusion

In particular, the prediction of extreme weather events has proved difficult over the years. When asked about this, Mathilde Bøttger Sørensen (Professor of Geophysics at the University of Bergen in Western Norway) claimed that an earthquake can only be measured after it has happened. She went on to say that many monitoring stations can be built but results can only be retrieved when the earthquake has occurred. As a result, it means that it is almost impossible to warn of earthquakes and other extreme weather events in advance. Fortunately, the monitoring systems mentioned by Bøttger Sørensen can help to understand earthquakes and to create statistics which could be used to better prepare for the next earthquake and to assess where it is likely to occur (Kaste, 2023). communicate the news of the detection in real time, it can be said that the use of data science in predicting natural disasters is still ineffective (Goswami, Chakraborty, Ghosh, Chakrabarti, & Chakraborty, 2015). A national seismological network was set up in the Haiti following the earthquake in 2010 which was maintained by the Bureau of Mines and Energy (BME), a governmental institution (Calais, et al., 2020). The BME operates five broadband seismic stations. During an earthquake in Haiti in October 2018, none of the seismic stations were functional and this meant that the national civil protection agency and the population had to rely on information from the United States Geological Survey (USGS). This shows that there is still a long way to go in maintaining the operability of such a system and providing quick and independent information to the public. The report by Calais, et al. asserted that it is important that seismologists monitor earthquakes with high-quality, expensive sensors located at carefully chosen sites where environmental noise is minimal and try to ensure constant real-time data communication, for instance via satellite links. In order to improve making predictions, it is important for stakeholders in natural disaster research to collaborate on their research to find the most efficient way to predict the events. It is evident that it would be important for countries with high risks of extreme weather events to take heed of warnings in the future. Finally, it is important for these countries to put measures in place to reduce the damage caused by these events. Due to the fact that meteorological observatories detect natural disasters but are unable to

Computational modelling faces an almost insurmountable challenge when predicting

disasters: the events are so rare that there is not enough data to use in predictive models and to accurately forecast when they will happen next. However, researchers have suggested that by using knowledge of how the disasters occur (for example, knowledge about the movement of tectonic plates and the collapse of rock formations on Earth’s surface) coupled with data from previous disasters, predictions can be made. Research by Siliezar (2023) has revealed that the researchers of a study in Nature Computational Science combined statistical algorithms (which need less data to make accurate, efficient predictions) with a powerful machine learning technique and trained it to predict scenarios, probabilities and sometimes the timeline of rare events despite the lack of historical record on them. It is also important to note that all natural disasters are different which makes them difficult to predict. For example, no two earthquakes have the same magnitude on the Richter scale and other extreme weather events occur on different scales. In conclusion, it is clear that data science can aid the prediction of natural disasters by using historical data as input to predict new output values without being programmed to perform these tasks. From this, patterns from data could be extracted which can help with predicting natural disasters. This could be achieved theoretically. However, there are limitations to this due to the weakness of computers in processing massive amounts of data. Computers are still very slow in doing this and, although Artificial Intelligence assists with finding hidden dependencies in data, the models produced are still far from perfect. Data mining is still ineffective in predicting natural disasters, however significant progress has been made aided by data science and this is expected to improve with time.

56