Climate Change and Indigenous Knowledge: Traditional Weather Forecasting Practices in a Changing Environment

Background: Rain-fed farming communities in semi-arid India have long anticipated monsoon behaviour through locally accumulated observation of the sky, wind, plants, animals and seasonal lore. Climate change is altering the regularities on which these practices depend, yet the status of such knowledge is rarely examined alongside instrumental records. Objective: This study documents the repertoire of traditional weather forecasting practices in eight villages, examines who holds this knowledge and how its perceived reliability has changed, and tests indicator classes against observed rainfall. Methods: A convergent mixed-methods design combined a structured survey of 240 farmers, 32 interviews and focus group discussions analysed thematically, and a 35-year (1990–2024) monsoon-season rainfall and onset series analysed with the Mann–Kendall test and Sen’s slope. Indicator skill was assessed as the proportion of 25 seasons (2000–2024) in which each indicator class matched the observed rainfall tercile. Results: Respondents knew on average 7.6 of 20 catalogued indicators; knowledge rose strongly with age (Spearman ρ = 0.71, p < 0.001) and fell slightly with schooling. Mean perceived reliability of traditional indicators fell from 4.04 to 3.00 on a five-point scale (Wilcoxon p < 0.001). Monsoon onset shifted later by 0.35 days per year (Mann–Kendall p = 0.003), whereas seasonal totals showed no significant monotonic trend (p = 0.35). Atmospheric (hit rate 60.0%) and plant-based (56.0%) indicators exceeded chance, whereas celestial and almanac-based indicators did not. Conclusion: Traditional forecasting is neither uniformly reliable nor uniformly obsolete: its more physically grounded components retain skill while its culturally transmitted components are eroding with the climate and across generations. Participatory, indicator-level co-validation with meteorological services is recommended.