Introduction: Cigar stacking fermentation is a key step in tobacco aroma enhancement and miscellaneous gas reduction, which both have a great influence on increasing cigar flavor and improving industrial availability.Methods: To analyze the effect of cigar tobacco leaf (CTLs) microbial community on volatiles flavor compounds (VFCs), this study used multi-omics technology to reveal the changes in microbial community structure and VFCs of different cigar varieties during stacking fermentation, in addition to exploring the interaction mechanism of microbiome and VFCs.Results: The results showed that the dominant microbial compositions of different CTL varieties during stacking fermentation were similar, which included Staphylococcus, Corynebacterium 1, Aerococcus, and Aspergillus. These dominant microbes mainly affected the microbial community structure and characteristic microorganisms of CTLs through microbial interactions, thereby influencing the transformation of VFCs. Characteristic microorganisms of different CTLs varieties such as Trichothecium, Trichosporon, Thioalkalicoccus and Jeotgalicoccus, were found to positively correlate with characteristic VFCs like megastigmatrienone 4, pyrazine, tetramethyl-, geranyl acetone, and 2-undecanone, 6,10-dimethyl-, respectively. This in turn affected the aroma and sensory quality of the CTLs.Discussion: This study provides theoretical support for the analysis of the mechanism of microorganisms on VFCs and aroma, and development of microbial agents during cigar stacking fermentation.

Accurate forecasts of tropical cyclone (TC) track and intensity with a sufficient lead time are critical for disaster preparedness and risk mitigation. Traditional numerical weather prediction models, while fundamental to operational forecasting, often exhibit systematic errors due to limitations in observations, physical parameterizations, and model resolution. In recent years, machine learning (ML) and deep learning (DL) approaches have emerged as promising data-driven alternatives for improving TC forecasts. This study presents a comparative evaluation of six ML and DL models—Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Categorical Boosting (CatBoost), Artificial Neural Network (ANN), and Convolutional Neural Network (CNN)—for forecasting TC track and intensity in the North Atlantic basin. The models are trained using the National Hurricane Center’s (NHC) HURDAT2 best-track dataset for storms from 1990 to 2019 and evaluated on an independent test set from the 2020 season. Model performance is compared across all models and benchmarked against the 2020 mean Decay-SHIFOR5 intensity error, CLIPER5 track errors, and the NHC official forecast (OFCL) errors. Forecast skill is assessed using mean absolute error (MAE) with 95% bootstrap confidence intervals and the coefficient of determination ($R^2$) across lead times of 6, 12, 18, 24, 48, and 72 h. The results show that: (1) several ML and DL models achieve intensity forecast performance that is broadly comparable in magnitude to the 2020 mean OFCL benchmarks, with an average error reduction of 5–11% at the 24 h lead time; (2) among the ML models, XGBoost and CatBoost slightly outperform LightGBM and RF in accuracy, while LightGBM demonstrates the highest computational efficiency; and (3) among the DL models, CNNs outperform ANNs in predictive accuracy and intensity forecasting efficiency, while ANNs exhibit lower computational cost for track forecast. Bootstrap confidence intervals indicate relatively low variability in model errors, supporting the statistical stability of the results within the 2020 season. However, these results reflect within-season variability and do not necessarily generalize across different years or climatological conditions. Overall, the findings demonstrate the potential of ML/DL-based approaches to complement existing operational forecast systems and enhance TC track and intensity forecasting in the North Atlantic basin.

Hurricane Florence (2018) was one of the most destructive storms of the 2018 hurricane season. This storm produced a substantial amount of precipitation, which caused immense flooding along the coast. As a result, billions of dollars of damage were done to the coast. This study explored approaches to improve the prediction of the track and intensity of Hurricane Florence (2018) by utilizing the deterministic Numerical Weather Prediction (NWP) models and a statistical modeling-based ensemble technique. The Global Forecast System (GFS) data is employed to initialize the Weather Research and Forecasting (WRF) and Hurricane WRF (HWRF) models to produce numerous simulations with various scheme options and starting times. The simulation data from five different NWP (Numerical Weather Prediction) models including the HWRF, WRF, ECMWF (European Centre for Medium-Range Weather Forecasts), and GFS models, were then interpolated to prepare for the statistical models. With the interpolated data, a hybrid method with multiple linear regression (MLR), random forest, and simple ensemble (SE) was developed. This hybrid method used multiple linear regression and random forest to identify the significant factors for hurricane prediction in the training set, and an averaging ensemble was then applied to the significant factors' data. As verified in the testing data sets, the errors from the hybrid method were reduced, indicating the improvement of the predictability. It is found that our numerical simulations using the HWRF model with a statistical modeling-based ensemble technique improved the accuracy of the track and intensity prediction of Hurricane Florence (2018). Overall, these tools and methods can greatly improve the accuracy of the track and intensity prediction of future hurricanes like Florence and can help ensure better civilian preparedness for a hazardous storm.