Large-ensemble experiments with global and regional climate models enable us to assess changes in the risks of local-scale heavy snowfall due to anthropogenic global warming. We conduct 100-ensemble historical and non-warming global climate experiments forced by oceanic conditions in 2021/22 when La Niña phenomena occurred, and conduct dynamical downscaling using regional climate models with 20 km and 5 km grid intervals. The 10-year return values of total winter snowfall decrease in most of Japan due to anthropogenic global warming, while they increase at high elevations and the northern parts of Japan. The winter-maximum daily snowfall is enhanced not only over high elevations but also over low elevations in Japan. Tsunan Town is located in an inland area of central Japan where the winter-maximum daily snowfall is enhanced by anthropogenic global warming. Composite analyses of winter-maximum daily snowfall events at the Tsunan weather station indicate that the enhancement of daily snowfall due to anthropogenic global warming is related to deeper troughs at 500 hPa and warmer and more humid air in the lower atmosphere in the historical 2021/22 winter than those in the non-global-warming 2021/22 winter.

On 15 August 2023, Typhoon Lan (2023) struck the Kinki region in western Japan, bringing record precipitation to the Kinki and Chugoku regions. This study investigates a turbulent transport scheme that can predict precipitation more accurately using the Weather Research and Forecasting model. The turbulent transport schemes compared are the Yonsei University scheme, the Mellor–Yamada–Nakanishi–Niino (MYNN) scheme, and the eddy-diffusivity mass-flux (EDMF) scheme, which is a blend of the MYNN scheme and a mass-flux scheme. Simulations are performed for a domain with a horizontal resolution of 5 km. The results show that the simulated track and central pressure of the typhoon over the Sea of Japan vary depending on the turbulent transport schemes, the MYNN scheme reasonably reproduces the distribution of heavy precipitation areas, the EDMF scheme even improves the quantitative prediction of precipitation, and the formulation of the turbulent length scale is also a key factor for the better prediction using the EDMF scheme. The EDMF scheme is expected to become a leading turbulent transport scheme in operational forecast models.

 Since the end of continuous rocket-sonde observations, which had been conducted until the 1990s, direct observations at altitudes higher than 30 km have been conducted only intermittently, so there are fewer observation data than in lower altitude regions.

 In the present study, we conducted radiosonde observations with large rubber balloons to obtain vertical structures of wind velocity and temperature at altitudes higher than 30 km from 27^th September to 3^rd October 2022 at the University of the Ryukyus, Okinawa Island, Japan.

 During the observation period, temperatures from 20 to 40 km altitude basically increased monotonically, including small perturbations. However, an observation at 1730 JST on 28^th September showed a remarkable continuous decrease with altitude in temperature at 30-36 km altitude. This was also confirmed by Constellation Observing System for Meteorology, Ionosphere and Climate (COSMIC)-2 GNSS-RO temperature data observed near Okinawa Island and ERA5 reanalysis data. Using the ERA5 reanalysis and radiosonde observation data, we found that the temperature-depleted layer is caused by a planetary-scale wave and upward energy propagating inertia gravity wave.

Climate experiments using global cloud-system resolving models (GCRMs) are expected to realistically simulate precipitation diurnal cycle (PDC) in the tropics, which is important for better representation of influences of cumulus convection on the climate system. This study examines how three series of decade-long climate experiments with Nonhydrostatic ICosahedral Atmospheric Model (NICAM), one of the GCRMs, realistically simulate the PDC over tropical coastal regions. Analyses reveal that it is more difficult to reproduce the PDC over coastal waters than that over coastal land, the former of which is characterized by nighttime offshore migration of precipitation areas. A comparison with in situ shipborne observations further reveals that biases in the offshore migration feature are associated with poor representation of convective cold pools; experiments with poor reproducibility of the offshore migration underestimate overall intensity of cold pools. The underestimation of the intensity may be associated with overestimation of environmental moisture in the lower free troposphere. As reproducing the environmental field is a difficult task particularly for climate experiments with global models, it seems more challenging for the climate experiments to realistically simulate the PDC over the coastal waters than for short-term experiments and regional climate experiments.

The observation of wind over oceans remains challenging. This results in difficulty in predicting the wind speed and direction. In this study, we examined the accuracy of upper-tropospheric wind speed forecasts along the flights between Tokyo International Airport and Los Angeles International Airport. We compared the Global Spectral Model data from the Japan Meteorological Agency (as forecast data) with the observation data from the aircraft's Quick Access Recorder (as true values). The forecast errors are highest over the North Pacific Ocean, not at the end of flight when the elapsed hours are longest, with meridional winds having a larger forecast error than zonal winds. Analysis of the meteorological field where a large meridional wind forecast error occurred using ERA5 indicates that the convergence-divergence of the jet stream or the blocking of westerly winds by the upper trough may have affected the large forecast error.

Afternoon thunderstorms, mesoscale convective systems, and other short-duration rainfall events threaten property and transportation. Recent deep learning techniques have been proven effective in nowcasting for rainfall accumulation (rain maps), but predicting occurrences of intense convective cells can add additional value to decision-making procedures. This study develops a deep-learning model that predicts the locations of cell occurrences in the next 60 minutes. The training data include reflectivities from the Taiwanese radar network and convective cell trajectories from the System for Convection Analysis and Nowcasting (SCAN). The label is the SCAN cell occurrence (1 or 0) within a 7.5 × 7.5 km² area in the next hour. In addition to providing occurrence probabilities, the post-analysis procedure deploys a threshold mask to convert the probabilistic forecast into deterministic forecasts; it achieves a ∼40% improvement in the critical success index compared with the baseline method. Furthermore, the new model informs users about the risks under the chosen threshold selected based on their risk tolerance. This study provides proof of concept that replacing the predicting objectives (“cell occurrence” instead of “rainfall”) of the model may help forecasters' decisions and the integration of deep learning into operational forecasting.

Climate change impact modelling studies often require not only mean temperature and precipitation but also other climate variables (e.g., solar radiation and wind speed) and extreme indices as input data. However, studies on observational constraints (emergent constraints) about these variables and indices are limited. Based on linearities of future climate change as functions of global warming levels and biases in recent global mean temperature trends in the simulations of 40 Earth system models (ESMs), the upper bounds of uncertainties in future changes of various variables (annual mean temperature, annual maximum daily maximum temperature, mean specific humidity, mean downward longwave radiation and specific humidity on days when annual maximum daily precipitation (Rx1day) events occur) are successfully lowered in most regions of the world. We can also reduce inter-model variances of regional changes in mean precipitation, Rx1day, mean downward shortwave radiation, mean sea level pressure and mean surface wind speed in some areas. These results would be useful for climate change impact studies to consider whether they should weight ESMs or exclude some ESMs to prevent possible biases in impact assessments.