Search

Producing an accurate and calibrated probabilistic forecast has high social and economic value. Systematic errors or biases in the ensemble weather forecast can be corrected by postprocessing models whose development is an urgent challenge. Traditionally, the bias correction is done by employing linear regression models that estimate the conditional probability distribution of the forecast. Although this model framework works well, it is restricted to a prespecified model form that often relies on a limited set of predictors only. Most machine learning (ML) methods can tackle these problems with a point prediction, but only a few of them can be applied effectively in a probabilistic manner. The tree-based ML techniques, namely, natural gradient boosting (NGB), quantile random forests (QRF), and distributional regression forests (DRF), are used to adjust hourly 2-m temperature ensemble prediction at lead times of 1–10 days. The ensemble model output statistics (EMOS) and its boosting version are used as benchmark models. The model forecast is based on the European Centre for Medium-Range Weather Forecasts (ECMWF) for the Czech Republic domain. Two training periods 2015–18 and 2018 only were used to learn the models, and their prediction skill was evaluated in 2019. The results show that the QRF and NGB methods provide the best performance for 1–2-day forecasts, while the EMOS method outperforms other methods for 8–10-day forecasts. Key components to improving short-term forecasting are additional atmospheric/surface state predictors and the 4-yr training sample size. Significance Statement: Machine learning methods have great potential and are beginning to be widely applied in meteorology in recent years. A new technique called natural gradient boosting (NGB) has been released and used in this paper to refine the probabilistic forecast of surface temperature. It was found that the NGB has better prediction skills than the traditional ensemble model output statistics in forecasting 1 and 2 days in advance. The NGB has similar prediction skills with lower computational demands compared to other advanced machine learning methods such as the quantile random forests. We showed a path to employ the NGB method in this task, which can be followed for refining other and more challenging meteorological variables such as wind speed or precipitation. [ABSTRACT FROM AUTHOR]

Globally, extreme rainfall has intense impacts on ecosystems and human livelihoods. However, no effort has yet been made to forecast the extreme rainfall indices through machine learning techniques. In this paper, a new extreme rainfall indices forecasting model is proposed using a random forest (RF) model to provide effective forecasts of monthly extreme rainfall indices. In addition, RF feature importance is proposed in this study to identify the most and least important features for the proposed model. This study forecasts only statistically significant extreme rainfall indices over Bangladesh including consecutive dry days (CDD), the number of heavy rain days (R10mm; rainfall ≥ 10 mm), and the number of heavy rain days (R20mm; rainfall ≥ 20 mm) within 1–3 months of lead time. The proposed model uses monthly antecedent CDD, R10mm, and R20mm including atmospheric parameters and ocean–atmospheric teleconnections, namely, convective available potential energy (CAPE), relative humidity (RH), air temperature (TEM), El Niño–Southern Oscillation (ENSO), Indian Ocean dipole (IOD), and North Atlantic Oscillation (NAO), as the inputs to the model. Results show that the proposed model yields the best performance to forecast CDD, R10mm, and R20mm with only the antecedent of these indices as input. Ocean–atmospheric teleconnections (IOD, ENSO, and NAO) are useful for CDD forecasting, and local atmospheric parameters (CAPE, RH, and TEM) are useful for R10mm and R20mm forecasting. The results suggest that adding atmospheric parameters and ocean–atmospheric teleconnections is useful to forecast extreme rainfall indices. [ABSTRACT FROM AUTHOR]

Recent research has shown that random forests (RFs) can create skillful probabilistic severe weather hazard forecasts from numerical weather prediction (NWP) ensemble data. However, it remains unclear how RFs use NWP data and how predictors should be generated from NWP ensembles. This paper compares two methods for creating RFs for next-day severe weather prediction using simulated forecast data from the convection-allowing High-Resolution Ensemble Forecast System, version 2.1 (HREFv2.1). The first method uses predictors from individual ensemble members (IM) at the point of prediction, while the second uses ensemble mean (EM) predictors at multiple spatial points. IM and EM RFs are trained with all predictors as well as predictor subsets, and the Python module tree interpreter (TI) is used to assess RF variable importance and the relationships learned by the RFs. Results show that EM RFs have better objective skill compared to similarly configured IM RFs for all hazards, presumably because EM predictors contain less noise. In both IM and EM RFs, storm variables are found to be most important, followed by index and environment variables. Interestingly, RFs created from storm and index variables tend to produce forecasts with greater or equal skill than those from the all-predictor RFs. TI analysis shows that the RFs emphasize different predictors for different hazards in a way that makes physical sense. Further, TI shows that RFs create calibrated hazard probabilities based on complex, multivariate relationships that go well beyond thresholding 2–5-km updraft helicity. [ABSTRACT FROM AUTHOR]

Approximately seven years of daily initializations from the convection-allowing National Severe Storms Laboratory Weather Research and Forecasting Model are used as inputs to train random forest (RF) machine learning models to probabilistically predict instances of excessive rainfall. Unlike other hazards, excessive rainfall does not have an accepted definition, so multiple definitions of excessive rainfall and flash flooding—including flash flood reports and 24-h average recurrence intervals (ARIs)—are used to explore RF configuration forecast sensitivities. RF forecasts are analogous to operational Weather Prediction Center (WPC) day-1 Excessive Rainfall Outlooks (EROs) and their resolution, reliability, and skill are strongly influenced by rainfall definitions and how inputs are assembled for training. Models trained with 1-yr ARI exceedances defined by the Stage-IV (ST4) precipitation analysis perform poorly in the northern Great Plains and Southwest United States, in part due to a high bias in the number of training events in these regions. Increasing the ARI threshold to 2 years or removing ST4 data from training, optimizing forecast skill geographically, and spatially averaging meteorological inputs for training generally results in improved CONUS-wide RF forecast skill. Both EROs and RF forecasts have seasonal skill—–poor forecasts in the late fall and winter and skillful forecasts in the summer and early fall. However, the EROs are consistently and significantly better than their RF counterparts, regardless of RF configuration, particularly in the summer months. The results suggest careful consideration should be made when developing ML-based probabilistic precipitation forecasts with convection-allowing model inputs, and further development is necessary to consider these forecast products for operational implementation. Significance Statement: Machine learning (ML) models can deduce statistical relationships between a set of predictors and meteorological events. In this work, ML models are developed to predict excessive rainfall events. Since excessive rainfall is difficult to uniformly define across the United States, multiple ML models are built from a variety of rainfall datasets with predictors gathered from output of a high-resolution numerical weather prediction model and forecasts are made from each model. Forecasts made from these models are highly sensitive to both the definitions of excessive rainfall (e.g., 100 mm of rain in a day may cause flooding in a usually dry area, but not in a wet area) and the predictors used. Forecast skill can increase when excessive rainfall events are rarer and when predictors synthesize the surrounding environment rather than characterize specific geographical points. ML-based models have great potential for excessive rainfall prediction, but careful attention to the configuration of these models is required. [ABSTRACT FROM AUTHOR]