The ordinary least squares regression is a parametric technique that is commonly used for analysis of time series. This method is very popular when the long-term trend of the change in a sea-level has to be estimated on basis of mareograph data. The main advantage of ordinary least squares regression is that this approach allow us to estimate the slope and the intercept of a regression line by parametric methods that are relatively simple. However, the ordinary least squares regression relies on the assumption about the normality of data. If this assumption is not met, the regression results may not be correct. Another disadvantage of the method is that it is very sensitive to impact of outliers in data. Analyzing the records of the Black Sea level, gathered by tide gauges in Varna in the period from 1929 to 2011, we can see that in 2005 and 2006 years we registered significantly low sea-levels, in comparison with those in the preview and in the next few years. Similarly, the Black Sea level near the tide gauge in Burgas, we also recorded significantly low sea-level in 2005 and 2006. The purpose of the current research is to recalculate these trends by the use of non-parametric regression, i.e., by the Theil-Sen robust estimator of simple linear regression. Using non-parametric regression, we calculated the slopes of the linear upward trends of the Black Sea in Varna and Burgas equal to 2.35 mm/year and 2.03 mm/year, respectively. The slopes of the linear trends of the Black Sea in Varna and Burgas, estimated by other Bulgarian researches, are about 1.2-1.4 mm/year and 1.5-1.6 mm/year, respectively. Contrary to the results, published by other researchers, we found that sea-uplift in Varna is greater than that in Burgas. Therefore, when we analyze oceanic data, which is likely to contain extreme data points, it is preferable to give more trust to non-parametric technics in order to obtain results that are more relevant than results obtained by ordinary regression under violation of its assumptions.

Robust regression, ordinary least squares regression, sea level, black sea, time series