Accurate estimation of reference evapotranspiration (ET₀) is essential for effective water resources management, particularly under changing climatic conditions. This study aimed to evaluate the performance of 14 empirical methods for estimating ET₀ against the standard FAO Penman–Monteith (PM-FAO56) method and to investigate their temporal variations at the Hashemabad synoptic station in Golestan Province, Iran. For this purpose, descriptive statistics, distribution analysis, Pearson’s correlation coefficient, and time-series analyses based on the Mann–Kendall, Sen’s slope, and Pettitt tests were employed. The results showed that the ABT method performed well in terms of the similarity of its estimated mean and distribution characteristics to those of PM-FAO56. The ROM method exhibited the highest temporal correlation with PM-FAO56 (r = 0.848), whereas PAP showed relatively good agreement and error characteristics, with a higher Lin’s concordance correlation coefficient (LinCCC = 0.702), low bias (−14.64 mm), and the lowest RMSE (105.05 mm). In contrast, BC showed substantial overestimation, while MAK exhibited considerable underestimation. Trend analysis indicated that all ET₀ series had significant increasing trends at the 5% significance level, with the highest Sen’s slope observed for PM-FAO56 (11.7 mm year⁻¹). The Pettitt test also identified significant change points for most methods in 1997, 1999, and 2011. The concurrence of these changes with increases in temperature, solar radiation, and wind speed, along with a decrease in mean relative humidity, indicates substantial changes in the climatic conditions associated with evaporative demand at the study station. These findings highlight the importance of a multi-criteria evaluation of ET₀ estimation methods; however, the generalization of the results to other regions requires independent temporal and spatial validation