Abstract:

Accurate prediction of the initial flowering period (IFP) of lavender is essential for optimizing field management and enhancing tourism services in Yining of Xinjiang. Based on the observed data of IFP from 2001 to 2024 at the Yining horticultural station and the meteorological data from February to May of the same period, the correlations between the IFP and meteorological factors were analyzed. Building on the biological characteristics of lavender, pre−flowering climatological year types were classified via fuzzy clustering, and meteorological pathways were analyzed using path analysis to identify key predictors. Subsequently, multiple linear regression models were developed and validated. The results showed that: (1) significant correlations were observed between IFP and meteorological factors such as the average temperatures in late February, early March, and mid−March, precipitation in early February, and accumulated growing degree−days (GDD0 ) from the start of spring to May 31. GDD0 had the greatest influence, surpassing both precipitation and decadal scale variables. This dominance of GDD0 as a predictor was consistent and statistically significant (P<0.01) across all year−type classifications. (2) The year-type-classified model demonstrated superior accuracy (ME=0.5d, RMSE=0.8d, RE=0.6%) compared to the non-classified model (ME=0.9d, RMSE=1.2d, RE=0.9%). The implementation of the year-type classification model is recommended for operational phenological forecasting. Therefore, it is recommended to adopt the year-type classification–based model for operational forecasting of the IFP of lavender onset in Yining.

Key words: Lavender, Meteorological condition, Year-type classification, Initial flowering period prediction