Re-analysis of Drought Trend Statistics: Figure 5 in Dukat et al. (2022)

Overview This dataset contains a re-analysis of data extracted from Figure 5 of Dukat et al. [1]. The original figure plots the number of months with drought conditions (organized by pentads), including the linear trend lines and regression parameters for the SPEI-1, SPI-1, SPEI-3, SPI-3, SPEI-6, and SPI-6 indices.Purpose The purpose of this re-analysis is to verify the linear trend lines and regression parameters reported in the original publication. This dataset serves as accompanying material for a post-publication review on PubPeer. Recalculations based on the plotted bar values indicate significant discrepancies; specifically, the underlying data diverges from both the visual trend lines and the published regression metrics (slope, intercept, and R²) for most indices.MethodsData ExtractionThe data points (bar heights representing the number of months) and the start/end points of the trend lines were digitally extracted from the high-resolution bitmap of Figure 5 in the original publication using the plot digitization software WebPlotDigitizer [2].Re-calculationa) Linear regression analysis was performed on the extracted bar data to calculate the actual slope, intercept, and coefficient of determination (R2). These parameters serve as reference. b) Slope and intercept of the visual trend lines were calculated with their start/end points. ComparisonThe regression statistics calculated from the extracted bar data were established as the reference. Both the parameters published in the figure text and the digitized parameters of the visual trend lines were benchmarked against this reference to identify discrepancies.Statistical Analysis To verify the trends, two statistical methods were applied to the extracted data:Mann-Kendall Test: A non-parametric test (mkt user defined function) used to detect monotonic trends. It assesses whether a consistent upward or downward trend exists without assuming a linear relationship or normal distribution.Linear Regression Significance: A parametric test (pvaluetrend user defined function) that calculates the two-tailed p-value for the linear regression slope. It utilizes a t-statistic derived from the slope and its standard error to determine if the observed linear trend is statistically significantly different from zero.File DescriptionsDukat_2022_Figure_05_v03.xlsx: The master summary table. It lists the original parameters reported by Dukat et al. ("Dukat regression"), the parameters derived from this re-analysis ("re-analysis regression"), and the calculated difference between the two. It covers indices: SPEI-1, SPI-1, SPEI-3, SPI-3, SPEI-6, and SPI-6.SPEI-1_SPI-1_pentad.csv (and similar): Raw bar heights representing the number of months extracted from the figure.*_trend.csv: Coordinates representing the start and end points of the trend lines as originally plotted.References [1] Dukat, P., Bednorz, E., Ziemblińska, K. et al. Trends in drought occurrence and severity at mid-latitude European stations (1951–2015) estimated using standardized precipitation (SPI) and precipitation and evapotranspiration (SPEI) indices. Meteorol Atmos Phys 134, 20 (2022).https://doi.org/10.1007/s00703-022-00858-w[2] Rohatgi, Ankit. WebPlotDigitizer. Version 5.2. https://automeris.io/ (accessed 2025-11-05).