Is the number of extreme meteorological events increasing due to global climate change? (On the example of hurricanes in North America)

Inna Rybalka,Halyna Tereshchenko,Oksana Lenevych,Yurii Vergeles

DOI: https://doi.org/10.31861/geo.2024.847.5-22

2024-06-09

Abstract:Tropical cyclones are the deadliest natural disaster in the United States, with enormous financial losses. Scientists estimate that between 1980 and 2023 alone, they caused nearly 6,700 deaths and $2.6 trillion in damage. Traditionally, tropical cyclones in the Atlantic Ocean are called hurricanes, and those in the Pacific Ocean are called typhoons. We focus on hurricanes that hit the US mainland. It is believed that the intensity of these extreme meteorological events will increase by the end of the 21st century, and the purpose of our work was to test this hypothesis. We obtained data on the number of hurricanes on the 5-point Saffir-Sampson scale that hit the US mainland from the website of the US National Hurricane Center. The general scheme for analyzing the original data used in the study included three main stages: (1) calculation of basic statistics and assessment of the reliability of the means; (2) analysis of time series (autocorrelation analysis, assessment of the stationarity of the series, assessment of the normality of the distribution of observations in time series, calculation of the regression line); (3) comparison of distributions using the nonparametric Kolmogorov criterion. The initial data cover the period from 1851 to 2022, i. e. 171 years in total, and were grouped by decade. In total, 304 hurricanes of varying destructive power were recorded during the period under consideration including: category 1 hurricanes – 127 (42 % of the total number of hurricanes), category 2 – 81 (27 %, respectively), category 3 – 64 (21 %), category 4 – 28 (9 %), category 5 – 4 (1 %). At the first stage of working with the data for all samples, it was statistically substantiated that the arithmetic means for them were reliable. The first and most critical stage of time series analysis is the search for autocorrelation. To verify that the series is not random, we calculated the parametric autocorrelation coefficient. Taking into account the sample size (17 decades), the maximum lag at which we can try to establish autocorrelation is the fourth: 17/4 = 4.3. The total number of time series that we obtained after analyzing for autocorrelation is four. The longest time series were obtained for the number of hurricanes belonging to category «3» according to the Saffir-Simpson scale and categories «1», «2», «3», «4», «5» (16 decades, i. e. 160 years), and the shortest - for the number of hurricanes belonging to category «2» (13 decades, 130 years). The analysis was completed for three samples at this stage of the work, all of which represented random stochastic fluctuations. The next step was to test the four time series for stationarity with respect to their central tendency. To do this, we used the method of sequential grouping, i. e., each time series was divided into three consecutive groups and compared with each other using the Student's t-test. The analysis showed that when it comes to hurricanes in general that hit the US mainland, we are right about the non-stationary time series. In terms of individual categories, the picture is different: the series characterizing the number of low strength hurricanes (category 2) is non-stationary, while both series characterizing the number of more powerful hurricanes (categories 3 and 5) are stationary in time. The normality assessment revealed that, out of the four we studied, only for one of them, the number of category 5 hurricanes, was the actual value of C less than the lower limit of the critical table values. For the remaining three time series, it can be stated with high reliability that the observations are normally distributed. To identify the trend shape for the four time series, we used MS Excel® tools. To do this, each time series was plotted as a line graph in a rectangular coordinate system. For the most accurate alignment of the empirical series, we used the least squares method. To verify the reliability of the relationship between the studied parameter (Y-axis) and time (X-axis), the Pearson correlation coefficient with correction was determined. At the significance level of P < 0.10, the hypotheses of a linear trend for the time series characterizing the number of weak hurricanes (category «2») and hurricanes in general (categories «1», «2», «3», «4», «5») were accepted as valid. In addition, our study also showed that against the background of a decrease in the number of weak hurricanes and their total number, the number of rather destructive hurricanes (category 5) is increasing. This is generally in line with the conclusions of the International Panel on Climate Change (IPCC AR6). Based on the calculation of the Kolmogorov criterion, it was found that there are no differences in the distributions of the average number of hurricanes for the periods in question (1851–1910, 1911–1960, 1961–2020, and for 1851–1900, 1901–1940, 1941–1990, 1991–2020), which indicates a certain stability of these extreme meteorological events. Keywords: climate change, hurricanes, North America, Saffir-Simpson scale, tim series analysis.