Document Type : Research Paper
Authors
1 PhD Student, Hydrology of land, water resources, hydrochemistry, Russian State Hydrometeorological University, Saint Petersburg, Russia
2 Candidate of Technical Sciences, Associate Professor at the Department of Engineering Hydrology of the RSHU, Saint Petersburg, Russia.
3 Kurdistan Agricultural and Natural Resources Research and Education Center, AREEO, Sanandaj, Iran
Abstract
This study explores agricultural drought variability by applying both classical and rank-based statistical methods within two distinct probabilistic frameworks, aiming to assess how distribution choice affects the interpretation of drought patterns. Monthly meteorological data from 1998 to 2022 were processed into four drought indices: SPI, aSPI, RDI, and eRDI. Effective precipitation was calculated using the United States Department of Agriculture (USDA) method and CROPWAT, and potential evapotranspiration (ET0) was determined with the FAO Penman-Monteith approach. Each index was standardized with both gamma and log-normal distributions in DrinC to evaluate the influence of distribution choice on trend detection. Trend analysis was conducted using linear regression with Pearson correlation for parametric tests and the Mann–Kendall test with Sen’s slope for non-parametric tests. Results from both methods and distributions were consistent: median Sen slopes were within ±0.03 index units per year, and Mann–Kendall Z scores ranged from −0.82 to 0.63, indicating no significant monotonic change. Regression slopes supported this, remaining below 0.03 with p-values above 0.25. The close agreement between parametric and non-parametric results, and between gamma and log-normal distributions, shows that model selection does not bias drought trend analysis. This multi-index, dual-distribution framework provides a robust and transferable methodology for drought monitoring, particularly in data-scarce and semi-arid regions worldwide.
Keywords
)