Changes in Thermal Growing‐Season Characteristics in Estonia During 1965–2025: Regional Variability and Methodological Sensitivity

ABSTRACT Thermal growing‐season characteristics in Estonia were analysed for 1965–2025 using daily mean air temperature observations from 15 meteorological stations. Start of season (SOS), end of season (EOS), and growing‐season length (GSL) were determined using both the traditional Raik–Tarand method and the internationally standardised ETCCDI definition to evaluate their comparability in spatial patterns, annual variability, and long‐term change. Three thermal growing‐season regimes were identified across Estonia: maritime, transitional, and southern inland. Maritime stations showed the latest SOS (mean 25 April) and EOS (9 November) and the longest growing seasons, whereas southern inland stations had the earliest spring onset (mean SOS 16 April; EOS 24 October). Across Estonia, GSL increased by 4.3 days decade −1 on average, driven mainly by earlier SOS, while EOS changes were more spatially variable. In southern inland Estonia, growing‐season extension resulted primarily from earlier spring onset, whereas in maritime Estonia both earlier SOS and later EOS contributed substantially. GSL increased significantly in all three regimes, with the strongest increase in the maritime regime. Although the Raik–Tarand and ETCCDI definitions produced similar regional regimes, spatial patterns, and overall directions of long‐term change (national mean SOS/EOS/GSL: 21 April/28 October/189.0 days for Raik–Tarand vs. 23 April/29 October/188.8 days for ETCCDI), they differed substantially in annual SOS, EOS, and GSL estimates (mean bias +1.3 days for SOS, +1.2 days for EOS, and near‐zero for GSL, reflecting partial cancellation of the SOS and EOS biases; RMSE 8.5–13.2 days), particularly during years with unstable temperatures around the 5°C threshold. This reflects a structural difference between the definitions: ETCCDI requires six consecutive days above or below the threshold, whereas the Raik–Tarand method identifies a sustained seasonal transition using a cumulative thermal‐balance criterion that is less sensitive to brief interruptions. ETCCDI is therefore suitable for regional climate‐change analyses and international comparisons, whereas the Raik–Tarand method may better represent sustained thermal transitions under Estonian conditions.