Using tropical cyclone data along with sea surface temperature data (SST) and atmospheric circulation reanalysis data during the period of 1980–2019, the seasonal variation of tropical cyclone genesis (TCG), and the related oceanic and atmospheric environments over the Arabian Sea (AS) and Bay of Bengal (BOB) are compared and analyzed in detail. The results show that TCG in both the BOB and AS present bimodal seasonal variations, with two peak periods in the pre-monsoon and post-monsoon season, respectively. The frequencies of TCG in the BOB and AS are comparatively similar in the pre-monsoon season but significantly different in the post-monsoon season. During the post-monsoon season of October–November, the TCG frequency in the BOB is approximately 2.3 times higher than that of the AS. The vertical wind shear and relative humidity in the low- and middle-level troposphere are the two major contributing factors for TCG, and the combination of these two factors determines the bimodal seasonal cycle of TCG in both the AS and BOB. In the pre-monsoon season, an increase in the positive contribution of vertical wind shear and a decrease in the negative contribution of relative humidity are collaboratively favorable for TCG in the AS and BOB. During the monsoon season, the relative humidity factor shows a significant and positive contribution to TCG, but its positive effect is offset by the strong negative effect of vertical wind shear and potential intensity, thus resulting in very low TCG in the AS and BOB. However, the specific relative contributions of each environmental factor to the TCG variations in the AS and BOB basins are quite different, especially in the post-monsoon season. In the post-monsoon season, the primary positive contributor to TCG in the AS basin is vertical wind shear, while the combined effect of vertical wind shear and relative humidity dominates in the BOB TCG. From the analysis of environmental factors, atmospheric circulations, and genesis potential index (GPI), the BOB is found to have more favorable TCG conditions than the AS, especially in the post-monsoon season.

Postprocess correction is essential to improving the model forecasting result, in which machine learning methods play more and more important roles. In this study, three machine learning (ML) methods of Linear Regression, LSTM-FCN and LightGBM were used to carry out the correction of temperature forecasting of an operational high-resolution model GRAPES-3km. The input parameters include 2 m temperature, relative humidity, local pressure and wind speed forecasting and observation data in Shaanxi province of China from 1 January 2019 to 31 December 2020. The dataset from September 2018 was used for model evaluation using the metrics of root mean square error (RMSE), average absolute error (MAE) and coefficient of determination (R²). All three machine learning methods perform very well in correcting the temperature forecast of GRAPES-3km model. The RMSE decreased by 33%, 32% and 40%, respectively, the MAE decreased by 33%, 34% and 41%, respectively, the R² increased by 21.4%, 21.5% and 25.2%, respectively. Among the three methods, LightGBM performed the best with the forecast accuracy rate reaching above 84%.