Search

Two global atmospheric circulation datasets (ERA5 and NCEP FNL) with horizontal resolutions of 0.25° × 0.25° are investigated in terms of kinetic energy (KE) spectra at 200 hPa (roughly between 11 and 12 km). The horizontal KE (HKE) in NCEP FNL is larger and flatter than that in ERA5 at subsynoptic scales and mesoscales. Restoring the energy of this wavenumber range to the physical space shows that the HKE in NCEP FNL is larger than that in ERA5 over most areas but smaller mainly in the Indo-Pacific warm pool. The spectral budgets show that at these scales, the positive contribution from net vertical flux in ERA5 is stronger than that in NCEP FNL, while the negative contribution from available potential energy (APE) conversion is smaller; assuming that the atmosphere is in a quasi-stationary state, more dissipation is found in ERA5 than in NCEP FNL, which should be responsible for the HKE spectrum in ERA5 to be steeper and weaker than that in NCEP FNL. Our formulation shows that the APE conversion and net vertical flux are related to the pressure vertical velocity (PVV). The APE conversion and net vertical flux differences between the two datasets, like the PVV difference, are mainly from the tropical region. At large scales, the vertical motion in ERA5 is larger than that in NCEP FNL. The amplitude differences of the PVV spectra between two datasets are consistent with those of the large-scale precipitation spectra associated with microphysics parameterizations. These results support that vertical motion is a key dynamical factor explaining energy discrepancies at mesoscales. Significance Statement: The atmospheric kinetic energy spectrum reflects energy distribution characteristics in atmospheric motion at different scales. According to the Helmholtz decomposition, the horizontal wind field can be decomposed into two parts: rotational wind field and divergent wind field. We explore the dynamical causes of the atmospheric energy spectra differences among different datasets from the perspective of rotational and divergent components of motion. Our results reveal that the differences in energy spectra and their energy budget at scales below a few hundred kilometers are relatively significant in tropical regions and are closely related to vertical motion. A better understanding of the differences in kinetic energy spectra will contribute to the improvement of atmospheric models. [ABSTRACT FROM AUTHOR]

To study the multiscale interactions between rotational and divergent components of atmospheric motion, a new formulation of spectral budget of rotational kinetic energy (RKE) and divergent kinetic energy (DKE) based on the primitive equations in the pressure coordinate is derived, with four main characteristics: 1) horizontal kinetic energy (HKE) spectral transfer is exactly divided into spectral transfer of RKE and DKE, 2) the exact spectral conversion term between DKE and RKE is constructed, 3) the Coriolis term is considered, and 4) both the baroclinic conversion from available potential energy (APE) and the vertical flux of HKE act only on DKE. With this new formulation, outputs from ERA5 global reanalysis are investigated. At planetary scales, HKE spectral transfer, mainly attributed to β effect, is dominated by downscale DKE transfer. At synoptic scales, it is dominated by an upscale transfer of RKE energized by conversion of DKE mainly due to the Coriolis effect. The ultimate source of DKE in the upper troposphere is conversion of APE, while in the stratosphere it is the vertical flux. At mesoscales, the spectral transfers of RKE and DKE are both downscale, and conversion from RKE to DKE exists at sub-800-km scales in the upper troposphere, which is mainly attributed to the contribution from relative vorticity. At different heights, the intersection scales of RKE and DKE spectra are affected by the scales of positive peaks of the local spectral conversion from DKE to RKE around total wavenumber 10. Significance Statement: The purpose of this study is to explore more physical insights on the dynamics underlying the atmospheric energy spectra from the perspective of rotational and divergent components of motion. We derive a new formulation of the spectral rotational and divergent kinetic energy budget in the pressure coordinate for the global atmosphere, with application to ERA5 global reanalysis. Our results reveal the differences of spectral energy budget between rotational and divergent motions at different heights and scales. This new formulation provides a good tool for revealing the multiscale cascade and interaction between atmospheric rotational and divergent motions. Future work should investigate these dynamical processes with higher-resolution simulations and datasets. [ABSTRACT FROM AUTHOR]

The forecast uncertainty, particularly for precipitation, serves as a crucial indicator of the reliability of deterministic forecasts. Traditionally, forecast uncertainty is estimated by ensemble forecasting, which is computationally expensive since the forecast model is run multiple times with perturbations. Recently, deep learning methods have been explored to learn the statistical properties of ensemble prediction systems due to their low computational costs. However, accurately and effectively capturing the uncertainty information in precipitation forecasts remains challenging. In this study, we present a novel spatiotemporal transformer network (ST-TransNet) as an alternative approach to estimate uncertainty with ensemble spread and probabilistic forecasts, by learning from historical ensemble forecasts. ST-TransNet features a hierarchical structure for extracting multiscale features and incorporates a spatiotemporal transformer module with window-based attention to capture correlations in both spatial and temporal dimensions. Additionally, window-based attention can not only extract local precipitation patterns but also reduce computational costs. The proposed ST-TransNet is evaluated on the TIGGE ensemble forecast dataset and Global Precipitation Measurement (GPM) precipitation products. Results show that ST-TransNet outperforms both traditional and deep learning methods across various metrics. Case studies further demonstrate its ability to generate reasonable and accurate spread and probability forecasts from a single deterministic precipitation forecast. It demonstrates the capacity and efficiency of neural networks in estimating precipitation forecast uncertainty. [ABSTRACT FROM AUTHOR]