Asia-Pacific Journal of Atmospheric Sciences最新文献

Beta gyre and Rossby wave train induced by tropical cyclones were identified from the ERA5 global-reanalysis data of the recent 30 years using a composite method. To pick up the disturbances relevant to the beta gyre and Rossby wave train surrounding tropical cyclones, the disturbances were decomposed into three distinct horizontal scales including small, intermediate, and planetary-scale. Composite map of the disturbances containing small- and intermediate-scale showed a well-organized Rossby wave train. The orientation of wave train was found to depend on the translation direction of tropical cyclones, and also appeared to split into two orientations except for those translating in the west-northwestward direction. The wave energy of the wave train was shown to propagate along the wave train axis, which was inferred from the amplitude change with time within the wave train. The wave train shows a weak upward-westward tilt and increasing amplitude with height, implying the wave energy propagating upward. A dipole circulation cell, bearing a close resemblance to the beta gyre depicted in the theories and numerical models, was found from the Rossby wave train. The strength and orientation of the beta gyres were revealed to vary with the translation direction of the tropical cyclones, with the weakest and strongest amplitudes being found for the westward- and northward-translation cases, respectively. It was shown that the orientation of the beta gyre obtained by a lag-composite method rotates clockwise with time regardless of the translation direction of tropical cyclones.

The development of large-amplitude planetary waves (PWs) in March in the upper stratosphere during the easterly phase of the equatorial quasi-biennial oscillation (QBO) was investigated using ERA-interim reanalysis data for 1979–2019. During the 10-hPa easterly QBO, the amplitude at 3 hPa was significantly larger than that during the westerly QBO for cases of large-amplitude PWs. Case studies were conducted for individual events of the wave number 1 (wave-1) PW growth: an easterly case in 1994 and a westerly case in 1995. During the easterly QBO in March 1994, a developing perturbation at middle latitudes moved rapidly northeastward to replace the decaying high-latitude wave. In the early stage, conversion from the zonal mean to eddy kinetic energy in the subtropical region was crucial for wave development. This energy conversion was dependent on the sign of the meridional shear of the zonal wind in the middle latitudes. Negative shear was produced by the secondary meridional circulation associated with the equatorial QBO. After the perturbation started to develop in the middle latitudes, it moved northeastward over a few days due to potential vorticity flux, and the growth of the high-latitude waves was enhanced. A composite analysis also showed that the meridional shear of the zonal wind in the middle latitudes was negative during the easterly QBO in March. This study improves our understanding of the dynamic mechanism underlying the equatorial-polar relationship in the stratosphere in March.

The Global/Regional Integrated Model system (GRIMs) is upgraded to version 4.0, with the advancement of the moisture advection scheme and physics package, focusing on the global model program (GMP) for seasonal simulation and climate studies. Compared to the original version 3.1, which was frozen in 2013, the new version shows no Gibbs phenomenon in the moisture and tracer fields by implementing the semi-Lagrangian advection scheme with a better computational efficiency at higher resolution. The performance of the seasonal ensemble simulation (June–August 2017 and December 2016–February 2017) is significantly improved by new physics and ancillary data. The advancement is largest in the stratosphere, where the cold bias is dramatically reduced and the wind bias of the polar jets is alleviated, especially for the winter hemisphere. Noticeable improvements are also found in tropospheric zonal mean circulation, eddy transport, precipitation, and surface air temperature. This allows GRIMs version 4.0 to be used not only for long-term climate simulations, but also for subseasonal-to-seasonal climate prediction.

The National Institute of Environmental Research, under the Ministry of Environment of Korea, provides two-day forecasts, through AirKorea, of the concentration of particulate matter with diameters of ≤ 2.5 μm (PM_2.5) in terms of four grades (low, moderate, high, and very high) over 19 districts nationwide. Particulate grades are subjectively designated by human forecasters based on forecast results from the Community Multiscale Air Quality (CMAQ) and artificial intelligence (AI) models in conjunction with weather patterns. This study evaluates forecasts from the long short-term memory (LSTM) algorithm relative to those from CMAQ-solely and AirKorea using observations from 2019. The skills of the one-day PM_2.5 forecasts over the 19 districts were 39–70% for CMAQ, 72–79% for LSTM, and 73–80% for AirKorea; the AI forecasts showed comparable skills to the human forecasters at AirKorea. The one-day forecast skill levels of high and very high PM_2.5 pollution grades are 31–98%, 31–74%, and 39–81% for the CMAQ-solely, the LSTM, and the AirKorea forecasts, respectively. Despite good skills for forecasting the high and very high events, CMAQ-solely forecasts also generate substantially higher false alarm rates (up to 86%) than the LSTM and AirKorea forecasts (up to 58%). Hence, applying only the LSTM model to the CMAQ forecasts can yield reasonable forecast skill levels comparable to the operational AirKorea forecasts that elaborately combine the CMAQ model, AI models, and human forecasters. The present results suggest that applications of appropriate AI models can greatly enhance PM_2.5 forecast skills for Korea in a more objective way.

Air pollution modeling and forecasting over a national level scale for a country as large as India is a very challenging task due to the large amount of data involved in a limited spatial frequency. Often the air pollution and pollutant dispersion process depend on underlying meteorological conditions. Recently, Graph Neural Networks emerged as an effective deep learning model for discovering spatial patterns for various classification and regression tasks. This study proposes to employ a cluster-based Local Hybrid-Graph Neural Network (HGNN) methodology instead of using a single global Graph Neural Network for monitoring station-wise multi-step PM_2.5 concentration forecasting across India’s states. This methodology respects sudden changes in PM(_{2.5}) concentration due to the local meteorological variations. However, the local Hybrid GNN models consist of two parts: a spatio-temporal unit containing a Graph Neural Network layer along with a Gated Recurrent Unit layer to model the influence of wind speed and other meteorological variables on PM_2.5 concentration. The other part is a station wise feature extraction unit to extract station-wise meteorological feature impact on PM_2.5 concentration, along with the temporal dependency between historical records. The results from the two units are fused in step-wise manner for multi-step PM_2.5 forecasting. The proposed methodology was used to develop separate PM_2.5 concentration forecasting models, +24, +48 and +72 hours ahead. Subsequently, a detailed analysis is carried out to unfold the advantages of the proposed methodology. Results demonstrate the proposed models perform better than the state-of-the-art with significantly lesser computation time.

The x and y components of wind (U and V, respectively) are widely used as control variables in radar assimilation; therefore, it is common to choose (U, V) as the control variables for multi-scale data assimilation (DA) in convective-scale. When the model resolution reaches the convective scale, whether (U, V), as the momentum control variables, are still more suitable than the stream function (ψ) and unbalanced velocity potential (χ_u), it needs to be studied further examination. This study uses 3-km resolution forecast samples to calculate the background error covariance (B) with two different pairs of momentum control variables ((ψ, χ_u) and (U, V)) by the National Meteorology Center (NMC) method. In single-observation experiments, the analysis wind field is most sensitive to the two pairs of B, and the temperature is insensitive. When using (U, V) as the control variables, the local characteristic is more evident according to vertical and horizontal wind increments. The study assimilates low- resolution conventional observations to compare different momentum control variables, (ψ, χ_u) and (U, V), in numerical simulation experiments of the torrential rainfall in North China. In addition, the impacts of the two control variables options are also compared in terms of the 15 continuous days of cases in flood season. The main results are as follows: (1) the wind field is the critical difference between the two assimilation experiments at the analysis time. Using (U, V) as the control variables, the analysis field of wind from both the surface and different vertical levels is superior. The analysis field closer fits the wind observation; (2) the use of (U, V) control variables improves the short term (0 ~ 3-h) in surface wind prediction; and (3) the use of (U, V) control variables enhances the 24-h TS (threat score) in moderate rain and heavy rain.

Abstract

Current climate models still have considerable biases in the simulation of the East Asian summer monsoon (EASM), which in turn reduces their reliability of monsoon projections under global warming. We hypothesize that a higher-resolution coupled climate model with atmospheric and oceanic components at horizontal resolutions of 0.25° and 0.1°, respectively, will better capture regional details and extremes of the EASM. Present-day (PD), 2 × CO₂ and 4 × CO₂ simulations are conducted with the Community Earth System Model (CESM1.2.2) to evaluate PD simulation performance and quantify future changes. Indeed, our PD simulation well reproduces the climatological seasonal mean and intra-seasonal northward advancement of the monsoon rainband, as well as climate extremes. Compared with the PD simulation, the perturbed CO₂ experiments show an intensified EASM response to CO₂-induced warming. We find that the precipitation increases of the Meiyu-Baiu-Changma band are caused by comparable contributions from the dynamical and thermodynamical components in 2 × CO₂, while they are more driven by the thermodynamical component in 4 × CO₂ due to stronger upper atmospheric stability. The regional changes in the probability distribution of the temperature show that extreme temperatures warm faster than the most often temperatures, increasing the skewness. Fitting extreme precipitation values with a generalized Pareto distribution model reveals that they increase significantly in 4 × CO₂. Changes of temperature extremes scale with the CO₂ concentrations over the monsoon domain but not for precipitation extreme changes. The 99^th percentile of precipitation over the monsoon region increases at a super Clausius-Clapeyron rate, ~ 8% K^–1, which is mainly caused by increased moisture transport through anomalous southerly winds.

As a dominant pattern of the North Pacific sea surface temperature decadal variability, the Pacific Decadal Oscillation (PDO) has remarkable influences on the marine and terrestrial ecosystems. However, the PDO is highly unpredictable. Here, we assess the performance of the Coupled Model Intercomparison Project Phase 6 (CMIP6) models in simulating the PDO, with an emphasis on the evaluation of CMIP6 models in reproducing a recently detected early warning signal based on climate network analysis for the PDO regime shift. Results show that the skill of CMIP6 historical simulations remains at a low level, with a skill limited in reproducing PDO’s spatial pattern and nearly no skill in reproducing the PDO index. However, if the warning signal for the PDO regime shift by climate network analysis is considered as a test-bed, we find that even in historical simulations, a few models can represent the corresponding relationship between the warning signal and the PDO regime shift, regardless of the chronological accuracy. By further conducting initialization, the performance of the model simulations is improved according to the evaluation of the hindcasts from two ensemble members of the Decadal Climate Prediction Project (NorCPM1 and BCC-CSM2-MR). Particularly, we find that the NorCPM1 model can capture the early warning signals for the late-1970s and late-1990s regime shifts 5–7 years in advance, indicating that the early warning signal somewhat can be captured by some CMIP6 models. A further investigation on the underlying mechanisms of the early warning signal would be crucial for the improvement of model simulations in the North Pacific.

This study proposes a new method of estimating snow depth by using a moment (({M}_{n})) of snow particle size distribution ((SPSD)). We assumed that estimated snow depth ((ESD)) is given by a simple relationship: (ESD) (cm) = (A)×({M}_{n}), where the parameters, (A) and (n) are a proportional coefficient and an exponent in the moment formula, respectively. They were determined by a regression analysis between the observed snow depths (OSD) by laser snow depth meter, and the values of ({M}_{n}) from (SPSD) observed by Parsivel, installed at three observation sites: Cloud and Physics Observation Site (CPOS), Yongpyeong (YP) and Mokpo (MP) in South Korea. Snow observations were made from November to April: CPOS (2012 to 2015), YP (2015 to 2017) and MP (2005 to 2015). The analysis results indicate that the optimized value of A ranges from 2.16 × 10^–5 to 2.28 × 10^–5, and the optimized range of n is 2.21 to 2.68. The average values of A and n are 2.47 × 10^–5 and 2.21, respectively. The coefficient of determination (R²) between (OSD) and (overline{ESD})(obtained by using average values of (A) and (n)) was 0.81, indicating a fairly good correlation between them. This indicates that (overline{ESD}) does appear to have potential for estimating operationally, timely information on snow depth. This study suggests that (SPSD) observed by disdrometer (Parsivel or 2DVD) can be also used as an alternative of the typical snow measuring instruments such as snow stake and ultra-sonic snow depth meter.