Abstract:

This study used the different cotton growth date and meteorological data from 22 agricultural meteorological stations in different cotton regions of Xinjiang from 2000 to 2023. The daily variation of temperature and the warming effect of plastic mulch had been introduced to improve the conventional climatic suitability method by incorporating indicators of temperature development at different growth stages of cotton. A prediction model of cotton growth stages applicable to different cotton regions of Xinjiang has been constructed. The mean absolute error (MAE), root mean square error (RMSE) and normalized root mean square error (NRMSE) were used to evaluate the improved model. The aim was to address the large simulation bias in conventional climatic suitability method, which resulted from neglecting the daily variation of temperature and the warming effect of plastic mulching, and thereby to improve the accuracy of cotton growth stage prediction in Xinjiang. The results showed that: (1) from 2000 to 2023, the sowing date and emergence date of cotton in Xinjiang cotton region showed advanced trends, which tendency rates were −2.16d·10y⁻¹ and −1.44d·10y⁻¹ in northern Xinjiang, and −0.47d·10y⁻¹ and −0.96d·10y⁻¹ in southern Xinjiang, respectively. However, the boll opening date showed a delayed trend with tendency rates of 1.76d·10y⁻¹ and 0.36d·10y⁻¹ in northern Xinjiang and southern Xinjiang, respectively. Budding and flowering date were advanced in northern Xinjiang, but delayed in southern Xinjiang. (2) From 2000 to 2023, in northern Xinjiang the main growth stages (sowing to emergence, emergence to squaring, squaring to anthesis, anthesis to boll opening) as well as the overall sowing to boll opening were prolonged (P<0.05). In southern Xinjiang, only emergence to squaring (P<0.01) and sowing to boll opening were extended, while other growth stages showed insignificant shortening trends (P>0.05). (3) From 2000 to 2020, based on the unimproved climatic suitability model, the MAE, RMSE and NRMSE of cotton at different growth stages from 2000 to 2020 were 2.0−24.6d, 2.7−30.1d and 15.7%−32.3%,respectively. After the introduction of the daily variation of temperature and the warming effect of plastic film mulching, the improved model significantly reduced errors to 1.9−9.6d (MAE), 2.6−11.8d (RMSE) and 6.4%−23.0% (NRMSE). (4) From 2021 to 2023, compared with the unimproved climatic suitability model, the improved MAE and RMSE of cotton growth stage were reduced by 4.2%−79.3% and 3.2%−82.2%, respectively. The accuracy of anthesis to boll opening and sowing to boll opening reached an excellent level (NRMSE≤10.0%), while other growth stages were at a good level (10.0 %<NRMSE≤20.0%). These findings demonstrate that the improved climatic suitability model can improve the simulation accuracy and stability of cotton growth stages in Xinjiang, with good regional adaptability and practical application prospects. 

Key words: Cotton, Growth stage, Climatic suitability, Plastic mulching, Daily variation of temperature