The objective of this study is to improve forecast accuracy by using low-precision floating-point arithmetic when performing ensemble weather forecasting. Low-precision floating-point arithmetic is reproduced using a software emulator developed to allow the mantissa bit length of floating-point numbers to be adjusted in one-bit increments. First, two different methods of generating an ensemble forecast using low-precision techniques were compared with a conventional ensemble-generation approach. For one, the precision of the initial conditions is reduced (called initial value ensemble), and for the other, the precision of the model calculations is reduced (called model ensemble). Then, it is found that the former technique is inadequate for generating sufficient ensemble spread, but the latter gives an ensemble spread comparable to the reference. In order to further evaluate the ensemble method using low-precision floating-point arithmetic in accordance with the model ensemble method, ensemble forecasting experiments were conducted in combination with the conventional ensemble method. As a result, the combined ensemble forecast had a higher spread evaluation index than the ensemble forecast using only the low-precision floating-point arithmetic and the conventional ensemble method. The reasons why the ensemble forecasts have higher index when incorporating low-precision floating-point ensemble methods are considered as follows: weather forecast models do not reproduce weather phenomena below the grid scale due to their low spatio-temporal resolution, and some models incorporate statistical assumptions to reduce computational load, which suppress the random nature of weather phenomena rather than actual weather events. On the other hand, ensemble methods using low-precision floating-point arithmetic can compensate for this randomness, and thus are expected to have higher evaluation index. This suggests that low-precision floating-point arithmetic, implemented in hardware by using Field Programmable Gate-Arrays (FPGAs) for example, may allow for faster operations without compromising forecast accuracy in ensemble forecasting.

 In this study, our objective is to identify the appropriate cold pool scales over Taiwan's complex topography during predominant afternoon thunderstorms under a local-circulation dominated weather regime in summer. We utilize semi-realistic TaiwanVVM simulations, which cover the entire area of Taiwan, to investigate this phenomenon. Our findings reveal that when buoyancy is defined using a conventional environmental scale (109 km), the cold pool locations do not align with the precipitation areas, instead being concentrated mainly along the mountain ridges. We hypothesize that this discrepancy arises from the environmental scale at which cold pool buoyancy operates. To assess this, we conducted systematic analyses and the results show that an optimal environmental scale of approximately 7 to 11 km (about 3 times of the 75^th and 90^th percentile of the precipitation object length) can be identified. The statistics of cold pool frequency better align with precipitation hotspots, characterized by evaporative cooling over the plains and increased water loading within the core of precipitation objects over the mountains. We demonstrate that this method effectively captures the shift in cold pools associated with precipitation responses in a warming climate in Taiwan. This work highlights the importance of using an appropriate environmental scale when estimating buoyancy over complex topography.

 Linear eigenvalue analysis of the primitive equations is performed to study atmospheric free oscillations under the influence of a zonal mean field. The model for the primitive equations is based on a three-dimensional spectral formulation, and the zonal mean field is produced by averaging reanalysis data over 10 years. The frequencies and latitudinal/vertical structures of the eigenmodes obtained by the analysis are compared with the results of the classical tidal theory and with those of the free oscillation modes detected from reanalysis data by a recent study. The frequencies and vertical structures of the eigenmodes obtained in the present study are consistent with those of the eigenmodes detected in the recent study, while the obtained latitudinal structures do not differ significantly from those of the classical tidal theory. It is shown that the deviation from the frequency obtained from the classical tidal theory is mainly due to the effect of the zonal mean flow, but partly also to the latitudinal variation of the temperature field. The present study also shows that the vertical phase structure of the obtained eigenmodes, which is inconsistent with the classical tidal theory, can be understood qualitatively by using the wave dispersion relation.

 In recent years, Atmospheric Rivers (ARs) have been recognized to influence the Antarctic ice sheet via extreme snowfall, latent and sensible heat transports, and anomalous changes in radiation balance. ARs are defined as extreme moisture transport events and are thought to account for a significant fraction of total moisture transport from mid to high-latitude regions, such as Antarctica. While previous studies have investigated ARs associated with extreme events over Antarctica and the Southern Ocean, their climatological features remain poorly understood. We investigate the climatology of ARs in the south polar region such as their geographical distribution and their role in moisture transport, by using an AR detection method that extracts the area with a localized moisture transport at 6-hourly intervals for JRA55. Notably, our method effectively describes the geographical distribution of ARs, contrasting with conventional methods that use temporal fixed criteria. We find that the contours of climatological AR frequency display a zonally asymmetric, spiral-like structure extending from mid-latitudes in the Atlantic to high-latitudes in the Pacific Ocean. This distribution produces a zonal asymmetry in meridional moisture transport, which may contribute to the observed zonally asymmetric distribution of Antarctic precipitation. We also suggest that the dominant meteorological systems associated with the ARs differ geographically: extra-tropical cyclones in the Atlantic and blocking events in the Pacific Oceans. At 60°S, we find that the AR detection number has not had a significant trend over recent decades, but the typical intensity of individual ARs in austral summer has increased over the last 41 years.

 Tropical cyclones (TCs) are a threat to coastal regions in countries and areas situated in the tropics to, at times, mid-latitudes, and their threat is expected to escalate due to factors like global warming and urbanization. This emphasizes imperative need that warnings based on accurate and reliable forecasts be delivered to those who need them in order to prevent or mitigate TC impacts effectively. While conventional Numerical Weather Prediction (NWP) models have traditionally dominated TC forecasting at short to medium range lead times (i.e., up to two weeks), the emergence of Artificial Intelligence (AI) models, i.e., Machine Learning (ML) models trained on global reanalysis, has raised the possibility of such models competing and thus supplementing NWP models. Here, we examine the potential of ML models in operational TC forecasting, comparing them with conventional NWP models. The ML model used in this study is Pangu-Weather and TC forecasts by this ML model are compared with those from the operational global NWP model at the Japan Meteorological Agency, especially focusing on the track. All 64 named TCs for a period of 2021 to 2023 in the western North Pacific basin are verified. Results indicate that the ML forecasts exhibit smaller position errors compared to the NWP model, alleviate the westward bias around Japan, and retain its forecast accuracy for TCs with unusual paths, offering potential operational utility. Another benefit would be the ability to deliver forecast results to forecasters quicker than before, since the ML model's forecast takes less than a minute. Meanwhile, challenges such as forecast bust cases and TC intensity, which are also present in NWP models, persist. A proposed way to utilize ML models at current operational systems would be to add ML-based track forecasts as one independent member of consensus forecasts.

 The earliest attempts to study the global normal mode oscillations of the atmosphere used time series of barometric in situ observations, but such approach is limited by the spatial and temporal inhomogeneity of meteorological station data. A major advance on the subject was recently made by applying a zonal-time spectral analysis to the surface pressure field in hourly gridded ERA5 reanalysis data, which disclosed an array of spectral peaks at theoretically predicted zonal wavenumber-frequency pairs, including many peaks with periods between 2 and 12 hours. However, this result relies on adequate representation of the modes in ERA5, which (i) ingests data sources that cannot explicitly resolve high frequency modes (e.g., radiosondes and polar satellite observations), and (ii) employs a numerical forward model that potentially introduces spurious effects. The present study provides “ground truth” for the reanalysis by a simple analysis of hourly barometric observations taken at ∼ 3800 stations over the globe. For each putative global mode, a time series of its index is computed by filtering the hourly ERA5 pressure fields. This index is then regressed onto the station data, revealing, for each mode, a characteristic, globally coherent spatial pattern of regression coefficients. The meridional structures of the regression patterns agree fairly well with the corresponding Hough functions, not only for low-frequency Rossby and Rossby-gravity modes, but also for high-frequency modes such as Kelvin and inertia-gravity modes. Even the Pekeris resonance is identified for a couple of Kelvin modes. These findings both solidify the evidence for a rich spectrum of global normal modes in the real atmosphere and also lend credence to their representation in ERA5. It is impressive that ERA5, by combining a numerical model with scattered meteorological observations, even reproduces the tiny (∼ 0.1–1 Pa amplitude) pressure signals of the high-frequency global normal modes.

 The Japanese Reanalysis for Three Quarters of a Century (JRA-3Q) with top at 0.01 hPa (high-top) is investigated focusing on the semiannual oscillation (SAO) in the tropical middle atmosphere, together with the other high-top reanalyses, ERA5 and MERRA-2, and the MLS and SABER satellite data. By removing the annual component and using the SAO component alone in the SABER data spanning the recent two decades, the seasonal cycle of the mesospheric SAO (MSAO) at 0.01 hPa is found to have significantly larger first cycle than the second cycle in a year with the largest easterly wind in boreal spring. The seasonal cycle of the stratospheric SAO (SSAO) at 1 hPa shows commonly in both satellite data that the easterly wind amplitude in boreal winter is double as large as that in boreal summer, while the westerly wind amplitudes in boreal spring and autumn are nearly the same. The two satellite data exhibit that the MSAO amplitude has significant and negative trend, about −5 and −7 m s⁻¹ decade⁻¹ at 0.01 hPa in MLS and SABER, respectively. JRA-3Q reproduces well the seasonal cycle of the SAO, i.e., the calendar-locked downward propagation of the SAO from 0.01 hPa to 10 hPa with clear separation between the MSAO and SSAO, despite the MSAO being substantially underestimated compared to the satellite observations. The SSAO amplitude at 1 hPa is significantly increasing in JRA-3Q over about three decades from 1970s to 2000s, and it exhibits slight decreasing trend over the recent two decades from 2000s. Before 1970s the SSAO wavelet spectra are less concentrated around 6 months and the wavelet spectra around the annual component are significantly larger than those after 1970s in JRA-3Q and ERA5. None of the reanalyses show any hint of the MSAO significant and negative trend at 0.01 hPa.