Statistical Analysis of Long-term Hydrological Data of Shovi for Determining the Expected Amount of Precipitation

Author: Tamar Tsamalashvili
Co-authors: Tamar Tsamalashvili
Keywords: Precipitation, statistical analysis, recurrence interval, Shovi
Annotation:

The work presents a thorough analysis of the extreme indicators of the meteorological station in the Shovi resort. Particular attention was paid to the calculation of the maximum massive debris flow in the narrow valley of the Bubishkhevi River on the southern slope of the Caucasus, an event that had no analog in the hundred-year history of the existence of the Shovi resort. This event highlights the significance of understanding the interrelation of geological and meteorological processes and underscores the need to study the causes of such processes to determine the likelihood of similar catastrophic events in the future and to adopt appropriate measures. Extensive multi-year observations of daily extreme indicators were conducted with the aim of determining the expected maximum precipitation amounts for various recurrence periods. The initial phase involved the discovery and testing of anomalies in the data base obtained as a result of many years of daily observations of precipitation. The initial analysis included visualization (histograms and light diagrams) to consider the trends of annual maximums over the observation period. For the initial modeling, the Gumbel distribution was used, but the Kolmogorov-Smirnov (KS) test showed a poor fit. Subsequently, alternative distributions were used; Lognormal, Log-Pearson Type III, Weibull, and Exponential, to find a suitable model. According to the Kolmogorov-Smirnov (KS) test, the best fit was shown in the case of the Lognormal and Type III Log-Pearson (Gamma) distributions. These distributions were used to calculate recurrence intervals of 25, 50, 100, 150, and 200 years. Despite the widespread use of the Gumbel distribution in hydrological analyses, alternative distributions such as Lognormal and Log-Pearson Type III may also be used and provide a more accurate representation of precipitation data in certain contexts. The result is significant for calculating precise models of catastrophic events, which forms the basis for the development of effective strategies and adaptation plans in the context of climate change. It is important to test various statistical models and improve methods for calculating natural risks.