Abstract:

 In recent years, climate change has had a significant impact on agricultural ecosystems, particularly on crop diseases. To further understand the effects of early meteorological factors on tobacco target spot disease, this study collected data of tobacco target spot disease index and diseased plant rate in the Tianzhu county Pingpu city tobacco region from 2022 to 2023, collected early meteorological data to analyze correlations between the tobacco target spot disease index, diseased plant rate and the meteorological factors affecting them, the key factors were screened. The support vector machine (SVM) model and multiple regression models was established to simulation model of the tobacco target spot disease and validate, respectively. The results showed that: (1) the initial outbreak of tobacco target spot disease in the tobacco-growing area of Pingfu village, Tianzhu county, Guizhou province was from the end of May to the first ten days of June. This was followed by a fluctuating increase in both disease index and disease incidence, culminating in a peak period of incidence in mid−July. (2)The key meteorological factors influencing tobacco target spot disease include the average ground temperature 15 days prior to the disease survey date, the cumulative precipitation 30 days prior, and the average relative humidity 15 days prior. These factors showed a significant positive correlation with both the disease index and disease incidence rate of tobacco−targeted endemic diseases.. Specifically, higher soil temperatures, greater precipitation, and increased relative humidity 15−30 days prior to the date of disease investigation were associated with more severe outbreaks of tobacco target spot disease and a faster field transmission rate. (3) Based on the aforementioned key meteorological factors, a multiple linear regression model and an SVM model for tobacco target spot disease were established. The average fitting degrees (R²) of the two models were 0.95 and 0.93, respectively, indicating good simulation results. Upon testing, it was found that in the simulation of disease index, the average accuracy of the multiple linear regression model was 87%, higher than that of the SVM model, which was 75%. In the simulation of disease plant rate, the average accuracy of the multiple linear regression model was 80%, higher than that of the SVM model, which was 73%. The simulation results of the multi linear regression model outperform those of the nonlinear SVM model, indicating that the multilinear regression model is better suited to model the occurrence and development of tobacco−targeted scrofula. 

Key words: Tobacco, Target spot disease, Meteorological factor, Support vector machine, Multiple linear regression