Innovative trend analysis methods using multi-duration and variable length sub–series

Climate change is among the most critical challenges of our time and is clearly reflected in the rise of global mean temperatures. This study investigates global mean temperature trends for 1854–2023 using innovative trend analysis (ITA), frequency-based ITA, multi-duration ITA (MD-ITA), and the multi-duration different-length ITA (MDL-ITA) across 10-, 20-, 30-, 40-, and 50 year timescales. The ITA and F-ITA results reveal a pronounced upward shift in the global temperature distribution, with very low temperatures (13.0 °C–13.5 °C) disappearing in the later period high (14.5 °C–15.0 °C) and very high (15.0 °C–15.5 °C) temperatures appearing more frequently. The MD-ITA shows that warming estimates are strongly dependent on the selected timescale. The MD-ITA analysis short-duration periods are more sensitive to internal variability and may produce weak or even negative trends, whereas longer periods yield more stable and consistently significant warming signals. The MDL-ITA analysis complements this perspective by showing how recent short-term temperature conditions depart from the broader historical background, while the proposed trend-length metric quantifies the extent of this distributional separation. Bootstrap confidence intervals indicate that the strongest recent warming trends are statistically robust. These findings demonstrate that multi-timescale ITA-based analysis improves the interpretation of global warming by distinguishing persistent long-term change from shorter-term fluctuations and by showing that recent warming is not only stronger in magnitude but also more widespread across the temperature distribution.