Tropical overshooting convection has a strong impact on both heat budget and moisture distribution in the upper troposphere and lower stratosphere, and it can pose a great risk to aviation safety. Cloud-top height is one of the essential concerns of overshooting convection for both the climate system and the aviation weather forecast. The main purpose of our work is to verify the application of the machine learning method, taking the random forest (RF) model as an instance, in overshooting cloud-top height retrieval from Himawari-8 data. By using collocated CloudSat observations as a reference, we utilize several infrared indicators of Himawari-8 that are commonly recognized to relate to cloud-top height, along with some temporal and geographical parameters (latitude, month, satellite zenith angle, etc.), as predictors to construct and validate the model. Analysis of variable importance shows that the brightness temperature of 6.2 um acts as the dominant predictor, followed by satellite zenith angle, brightness temperature of 13.3 um, latitude, and month. In the comparison between the RF model and the traditional single-channel interpolation method, retrievals from the RF model agree well with observation with a high correlation coefficient (0.92), small RMSE (222 m), and small MAE (164 m), while these metrics from traditional single-channel interpolation method shows lower skills (0.70, 1305 m, and 1179 m). This work presents a new sight of overshooting cloud-top height retrieval based on the machine learning method.

We have developed a canonical correlation analysis (CCA) model for improving seasonal winter rainfall prediction. It uses the anomalies of sea surface temperature (SST), vertically integrated vapor transport (IVT), and geopotential height at 250 hPa (Z250) in October and November, respectively, as the predictors for winter rainfall prediction. These predictors represent the processes that influence winter rainfall over California as documented in the literature, but their potential for improving predictability was previously unclear. This statistical model shows prediction skills higher than those of the baseline autoregressive model, the CCA‐based prediction model using only the SST anomalies, and the dynamic predictions by the North American Multi‐Model Ensemble (NMME). Averaged over California, the Pearson correlation (R) is 0.64, root mean squared error (RMSE) is 0.65, and Heidke skill score (HSS) is 0.42 when the CCA‐based model is initialized by the three predictor fields (SST, IVT, and Z250) in November. These skills are higher than those of the NMME predictions initialized in November (R, RMSE, and HSS are 0.30, 0.83, and 0.15, respectively) and those of the autoregressive baseline (R, RMSE, and HSS are 0.10, 0.79, and 0.08, respectively). Hindcasts of winter rainfall initialized by October observations show R, RMSE, and HSS of 0.53, 0.81, and 0.39, respectively, also higher than those of the NMME seasonal prediction initialized in October (0.32, 0.79, and 0.22 for R, RMSE, and HSS, respectively) and the autoregressive model (0.30, 0.75, and 0.16 for R, RMSE, and HSS, respectively).