Abstract: To clarify the impact mechanism of meteorological factors on the yield of the first harvest of geo-authentic Qiai (Artemisia argyi), daily meteorological observation data from the Qichun National Basic Weather Station from 1995 to 2025 were analyzed. After examining the interannual variation trends of various meteorological indicators, the main meteorological factors were identified through correlation analysis between the county-wide first-harvest yield from 2018 to 2025 and multiple meteorological elements during the key growth period (January to May). Seven machine learning or regression algorithms, including LightGBM, Multiple Linear Regression, Ridge Regression, Lasso Regression, Random Forest Regression, SVR, and XGBoost, were selected to construct yield prediction models. The results showed that from 1995 to 2025, the average temperature, active accumulated temperature ≥0 ℃, and active accumulated temperature ≥10 ℃ during the first-harvest growing season of Qiai all increased significantly, with climatic tendency rates of 0.40 ℃/10 years, 46.73 ℃·d/10 years, and 33.40 ℃·d/10 years, respectively. The diurnal temperature range showed a weak increasing trend (climatic tendency rate of 0.11 ℃/10 years). The number of precipitation days showed no significant trend, mainly characterized by interannual fluctuations. Sunshine hours showed a significant decreasing trend, with a climatic tendency rate of -43.66 h/10 years. Average relative humidity showed a slight decreasing trend, with a climatic tendency rate of -0.09%/10 years. The average number of frost-free days was 137.4 days, with a climatic tendency rate of 0.767 5 d/10 years. Through a comprehensive evaluation of the fitting of the seven models, the LightGBM model was identified as the best, with a validation set R² of 0.999 9 and a relative error of only 0.48%, significantly outperforming all other models. Based on the correlation analysis heatmap and combined with the feature importance analysis from LightGBM, it was determined that thermal conditions (active accumulated temperature ≥10 ℃) were the most critical factor affecting the yield of the first harvest of Qiai, with a contribution of 22.5%.

Key words: climatic background, geo-authentic Qiai (Artemisia argyi), first-harvest Qiai, yield per unit area, meteorological factors, response, prediction