St Edward's Academic Review 2025

ACADEMIC REVIEW 2025

difference in magnitude of earthquakes and the different categories of hurricanes. Warburton’s book Atmospheric Processes and Human Influence published in 2001 outlines how several natural disasters such as hurricanes and earthquakes are formed. For example, the book says that in a year, four or five in about forty seeds (hurricanes in the early stage of development) in the Atlantic develop into hurricanes. Warburton goes on to explain that specific conditions are necessary for the formation of hurricanes. For example, a large area of more or less uniform temperature, humidity and pressure is needed to turn a tropical storm into a hurricane. This, as well as the need for warm seas with a surface temperature of about 27 Celsius and a water layer 60m deep among other factors, explains why only four or five of about forty seeds in the Atlantic may develop into full hurricanes. As a result, it can be deduced that if researchers can monitor the activity in regions that are prone to these natural disasters and observe patterns by comparing with historical data, predictions can be made. One of the most devastating natural disasters to have occurred in recent history is the earthquake which occurred in Haiti on 12th January 2010, at 4.35pm local time. Although it lasted only 35 seconds, it had a magnitude of 7 on the Richter scale and led to 300,000 casualties and left a further 300,000 injured. The earthquake displaced 1.5 million people, who were forced to live in internally displaced person camps and buried about 206 years of government records and archival documents. This earthquake was soon followed by two aftershocks of magnitude 5.9 and 5.5 (Pallardy, 2023). According to Reid (2022), the 2010 earthquake was the most devastating natural disaster in Haiti’s history and, as a result, Haiti faced the greatest humanitarian need in its history. There was tremendous financial damage caused by the earthquake, coming to a figure between $7.8 billion and $8.5 billion, which is about 120% of the country’s Gross Domestic Product (GDP). The earthquake in 2010 was initially attributed to the eastward movement of the Caribbean tectonic plate along the Enriquillo-Plantain Garden (EPG) strike-slip fault system. However, when no surface deformation was observed, the rupture of the main strand of the fault system was ruled out as a possible cause. The EPG fault system forms a transform Case studies Haiti earthquake: 12th January 2010

more effective in predicting natural disasters but not as effective as it should be. According to Lavrskyi, abnormally large amounts of data are collected worldwide by researchers and the data has a lot of parameters. Lavrskyi argues that models created to predict disasters are still far from perfect. Data needs computational resources and having this large amount of data means that computational models are slower and more complex. Artificial Intelligence has been of huge assistance to data science more recently because it has been able to find hidden dependencies in data, particularly machine learning, which is one of the many Artificial Intelligence techniques, which teach computers to learn from experience. Rather than employing a pre-existing equation as a model, machine learning algorithms use computational methods to “learn” information directly from data. Machine learning teaches computers to think in a similar way to how humans do, by learning and improving upon past experiences. It works by exploring data and identifying patterns and involves minimal human intervention (Data Robot, 2020). As the number of samples available for learning increases, the algorithms’ performance improves adaptively (The MathWorks, 2023). Fortunately, research has proved that the use of Artificial Intelligence techniques, including data mining and machine learning, can assist in understanding how disasters form and the idiosyncrasies of each of them. As a result of this, it is possible to make predictions about the occurrence of natural disasters. Another study by Burns (2021) has shown that machine learning, one of the various Artificial Intelligence techniques, allows software applications to become more accurate at predicting outcomes without being explicitly programmed to do so. Additionally, researchers in Nature Computational Science have confirmed in a recent study that combining statistical algorithms with powerful machine learning techniques has helped to train systems to predict scenarios, probabilities and sometimes even the timeline of rare events despite the lack of historical records (Brown University, 2023). This is a significantly more efficient method of predicting disasters than the method proposed by Lavrskyi of using historical data with machine learning algorithms as input to predict new output values, because even though each natural disaster has its own features and unique ways of how it is formed, there may be some anomalies. In my opinion, using historical data may not be useful due to the differences in the nature of certain extreme weather events, such as the

53