| Flue-cured tobacco is an important economic crop in China,its yield and quality characteristics are mainly influenced by climate resources and ecological environment.Under the background of global climate change,studying the quantitative relationship between flue-cured tobacco quality and climate resources and ecological factors can make rational use of the natural resources of Yuxi City,improve the fineness of small-scale flue-cured tobacco zoning,meet the demand for precise procurement of flue-cured tobacco raw materials of Chinese cigarette enterprises,and have important significance for further improving the intelligent evaluation of flue-cured tobacco quality.Based on the measured data of Yuxi City’s meteorology and flue-cured tobacco quality,the trend analysis and mutation test of the meteorological factors and extreme climate events in the field period of flue-cured tobacco in the past 70 years were carried out.Based on the data of automatic weather stations in various towns and townships,the suitable climate zoning of flue-cured tobacco in Yuxi county was described by using anuplin interpolation;Based on the quality characteristics of 15 kinds of flue-cured tobacco,the quality of flue-cured tobacco in Yuxi City was divided into four types by principal component analysis and K-means clustering;Based on the correlation between ecological factors and flue-cured tobacco quality indexes screened by Lasso,a multivariate statistical and neural network prediction model for flue-cured tobacco quality characteristics was established..The main results are as follows:(1)The average temperature of flue-cured tobacco in Yuxi showed an upward trend,while the temperature difference showed a downward trend,and the abrupt change mostly occurred near 1982 and 1992.Precipitation showed a decreasing trend,and only the precipitation in field period changed in 2004(p<0.05).The sunshine duration showed a decreasing trend,and there were significant mutations in both field and mature periods in1984(p<0.01).The number of extreme high temperature,extreme precipitation and continuous rainy days showed an increasing trend,while the number of continuous drought days showed a decreasing trend.The frequency and interannual fluctuation of each event were small from the 1870s to the early 20th century.Most of Yuxi is in the most suitable area for the growth of flue-cured tobacco,and more than half of Yuxi is in the most suitable area for the climate of flue-cured tobacco.The most suitable areas account for about one-third of the counties,mostly in Chengjiang,Eshan,Hongta and Jiangchuan.(2)The flue-cured tobacco in Yuxi city was divided into four types,with good chemical scores and poor physical scores.The physical and chemical scores of typeⅠwere the highest,with an average of 0.48 and 1.01,respectively,which were mostly distributed in Tonghai,Jiangchuan and Chengjiang.The physical and chemical scores of flue-cured tobacco in typeⅡarea were poor,the ratio of chlorine to potassium chloride,folial density,stem content and tensile strength were all in the unsuitable range,which were distributed in Yuanjiang and Xinping.TypeⅢhad the best chemical quality and the worst physical quality.The physical quality of typeⅣwas the worst,the chemical quality of flue-cured tobacco was the best,the leaf density,stem rate,tension were all in the inappropriate range,the chemical composition were all in the appropriate range,more distributed in Yimen.(3)The RMSE of multivariate statistical prediction models for quality characteristics of flue-cured tobacco was higher than that of neural network prediction models.The average n RMSE of multivariate statistical prediction models for physical characteristics and chemical composition were 14%and 19%,respectively,which were at a high level,and R~2was about0.3.The average n RMSE of the neural network prediction model for physical properties and chemical composition was 9%,and R~2was higher than 0.7,indicating better simulation effect.From the point of view of simulation effect,the neural network prediction model has the better simulation effect and can provide accurate results in practical application. |
