Asia-Pacific Journal of Atmospheric Sciences最新文献

This study examined the observed PM_2.5 concentration across 247 underground stations  consisting of Line-1 to Line-8 of the Seoul Metro from April 2021 to March 2023 in order to understand general characteristics of underground PM_2.5 air quality. Approximately, in one-thirds of underground stations (85 stations), annual averaged PM_2.5 concentration are over 35 µg m⁻³. Moreover, in 30 underground stations (approximately 12%), it exceeds 50 µg m⁻³, the recommended 24-hour maintenance standard for PM_2.5 concentration in underground stations. We found that PM_2.5 concentration is considerably influenced by both internal and external factors. Among the internal factors (i.e., depth, the number of passengers and operation frequency), the frequency of subway operation significantly affects changes in PM_2.5 concentration however, various internal factors may act in combination. In terms of external factor, there are positive correlation coefficients (r = 0.15–0.95) between daily averaged PM_2.5 concentration in underground station and that of the outdoor observatory closest to each underground station. In particular, in underground stations with high PM_2.5 concentration, the correlation with outdoor PM_2.5 air quality was low, suggesting that for better air quality in underground stations, we need to focus more on reducing the inherent emission from underground stations in highly polluted stations, but for less polluted stations, we need to improve outdoor air quality as well. We believe that this study may provide insights for effective future PM_2.5 air quality management in underground stations.

Variations in the wind speed intensity significantly impact evapotranspiration, water cycle processes, air quality and wind utilization. Previous studies have focused primarily on changes in mean wind speed, with little research on variations in different wind speed intensities. In this paper, we defined five wind speed indices to quantify the changes in different wind speed intensities and analyzed their associations with atmospheric circulation based on daily wind speed data collected from 601 meteorological stations across China from 1960 to 2018. The wind speed indices we defined include the annual mean wind speed, the annual maximum daily mean wind speed, the number of heavy wind days, the number of gentle breeze days and the number of light breeze days. The results showed that from 1960 to 2018, the annual mean wind speed, the annual maximum daily mean wind speed, the number of heavy wind days and the number of gentle breeze days exhibited significant decreasing trends (P < 0.05). The number of light breeze days exhibited a significant increasing trend (P < 0.001) in China during the same period. Large-scale atmospheric circulation patterns were one of the main factors affecting the changes in wind speed intensity. The Arctic Oscillation (AO) and the West Pacific Subtropical High Intensity Index (WPSHI) were significantly negatively correlated with the annual mean wind speed, the annual maximum daily mean wind speed, the number of heavy wind days and the number of gentle breeze days (P < 0.01), and the Asian Polar Vortex Intensity Index (APVI) was extremely significantly positively correlated with these four wind speed indices (P < 0.001). This suggests that monitoring and analyzing these atmospheric circulation indices can enable more accurate predictions of wind speed. These findings will provide information for climate change forecast, air pollution risk assessments and wind energy utilization.

Exponential-random vertical overlap of clouds is applied for radiative processes in a research version of the Korean Integrated Model (KIM) to replace the maximum-random vertical overlap of clouds. The cloud radiative effect (CRE) increases overall when the exponential-random overlap is used. This is because vertically continuous clouds, which are assumed to overlap maximally under the maximum-random overlap assumption, can be relaxed to random overlap depending on the vertical distance between cloud layers and the specified decorrelation length of clouds. CRE is more enhanced by considering the latitudinal dependency of cloud decorrelation length based on previous observational studies. This alleviates biases in CRE, which is underestimated overall, except in the low latitudes where the CRE is overestimated in the present simulations. The interaction between radiative and convective processes plays a role in decreasing CRE over the tropical western Pacific region, where strong convections develop, although the direct impact of applying the exponential-random overlap is to decrease the vertical overlap between ice clouds. The simulation of temperature in the lower troposphere is improved owing to the changes in cloud overlap. The warm bias over the Eurasian continent, in particular, is alleviated as more shortwave fluxes are reflected due to increased CRE.

The Weather Research and Forecasting (WRF) model has numerous model parameters that significantly affect rainfall forecasts. However, the multitude of parameters makes it challenging to identify which of these are critical for rainfall forecasting and optimization. This study utilizes the Morris One-At-a-Time (MOAT) Global Sensitivity Analysis (GSA) to ascertain the sensitivity of the simulated rainfall and other key atmospheric variables to 23 tunable model parameters across seven physics schemes in the WRF model. The MOAT mean and standard deviation were used as sensitivity measures and calculated for two Tropical Cyclone (TC)-enhanced southwest monsoon events in August 2012 and 2013 that resulted in catastrophic flooding over Metro Manila, Philippines. Results show that of the 23 model parameters, the ones more critically important to simulating rainfall are parameters that are related to cumulus schemes such as the multiplier for downdraft mass flux rate (P3), multiplier for entrainment mass flux rate (P4), starting height of downdraft over updraft source layer (P4), and mean consumption time of convective available potential energy (P6). To investigate the optimum parameter for the simulation of rainfall for each of the two events, the root mean square error (RMSE) is computed between the simulated rainfall over Metro Manila and observed data from the Global Satellite Mapping of Precipitation (GSMaP). The best performing set of parameters was able to reduce the RMSE of rainfall over Metro Manila by about 42% and 27% for the 2012 and 2013 enhanced monsoon events, respectively, relative to the default runs. For the first time, this study provides insight into which model parameters in the WRF model are critically important to the simulation of enhanced monsoon events. The results of this study may serve as a basis for future optimization studies of extreme weather events over the Philippines.

We investigated the microphysical characteristics of low-level Arctic clouds using four cloud microphysics parameterization schemes (Morrison, WDM6, NSSL, and P3) implemented in the Polar-optimized Weather Research and Forecasting (PWRF) model. Our assessment was based on a comparison with data collected during the Arctic Cloud Observations Using Airborne Measurements during the Polar Day (ACLOUD) experiment, which occurred near Svalbard between May and June 2017. During the ACLOUD campaign, a substantial number of clouds were observed, primarily influenced by adiabatic motions and sensible/latent heat fluxes that led to air masses warming up by 4 °C as they traversed over the sea ice and ocean transition zone. Among the parameterization schemes tested, the Morrison and WDM6 schemes demonstrated superior performance overall, showing frequency bias (FB) values closer to 1 (1.07 and 1.13) and high log-odds ratios (0.50 and 0.48) in cloud occurrence predictions, indicating good agreement with observed data. In contrast, the NSSL and P3 schemes exhibited higher FB values (1.30 and 1.56) with lower log-odds ratios (0.17 and 0.16), indicating an overestimation of cloud occurrence. The WDM6 scheme produced higher ice-mixing ratios compared to Morrison and NSSL schemes, while the latter two tended to generate more snow and graupel. The NSSL scheme showed the least bias in simulating ice water content (IWC) in mixed-phase clouds; however, all schemes generally underestimated both liquid water content (LWC) and IWC. Notably, significant deviations in IWC were observed at an altitude of 1.2 km compared to observations, attributed to differences in temperature thresholds for ice formation. This study emphasizes the importance of developing cloud parameterization in the Arctic based on observations to improve the accuracy of estimating cloud impacts on Arctic climate under rapid Arctic warming trends.

For climate risk assessments accurate gridded data sets are needed. An important aspect of such data sets is that they reliably represent the spatial and temporal characteristics of extreme events. This is particularly important for precipitation extreme events which are still not well represented in climate models. Here, we compare South Korean station data with two observation-based gridded data sets (APHRODITE and ERA5-Land) and data from global high-resolution Community Earth System Model (CESM) simulations with an atmospheric resolution of about 25km. We find that the two observation-based data sets have a lower level of the 99th percentile than the station data, but that CESM reproduces extreme events better. Our study provides evidence for an overall historical decrease in very large extreme events in the station data, which is not the case in the two gridded data sets. However, changes in extremes are locally dependent as shown by local quantile regression analysis; where local historical increases in precipitation extremes are statistically significant. The spatial dependence of extreme precipitation events is not well reproduced by the two gridded data sets but well by CESM. The temporal clustering of precipitation extremes is well reproduced by all data sets. Compared to the present day simulation, the CESM simulation of a warmer climate state shows an overall increase in mean precipitation and precipitation extremes and regionally dependent changes in temporal clustering. The model results also provide evidence for a change in spatial dependence in a warmer climate with spatially larger extreme precipitation systems possible. Our results highlight the need to produce better observation-based gridded data sets and also the need to adapt to more intense and frequent extreme precipitation events in the future in South Korea.

We developed a rice paddy model based on Noah land surface model (LSM) considering the standing water layer during the irrigation periods. In the model, we adopted a consistent subcanopy process from thin to thick canopy conditions and considered a small scalar roughness length of the water surface in the rice paddy fields. We evaluated the performance of the model using observations from three rice paddy sites with different leaf area index and water depth in Japan during the growing season. Two simulations were performed in an offline mode: a Noah LSM simulation with saturated soil moisture in the top two soil layers (IRRI) and a rice paddy model simulation (RICE). The average root mean squared errors of ground, sensible, and latent heat fluxes, and first soil layer temperature decreased by 20%, 16%, 17%, and 31%, respectively in the RICE simulation, compared to the IRRI simulation. The better performance of the RICE simulation was attributed to the consideration of the heat storage of the standing water layer during the irrigation periods and the realistic energy partitioning by the single-canopy model during the non-irrigation periods. Two sensitivity tests were performed related to the roughness length of the water and the constant mean water depth. When the small roughness length of the water surface during the irrigation periods was not considered, the subcanopy resistance decreased, which resulted in a cold bias in the daily mean ground and soil temperature and an overestimation of the daily mean latent heat flux under low leaf area index conditions. The use of constant mean water depth in the model did not significantly change simulated surface fluxes and ground and first soil layer temperature, implying that detailed information on temporally changing water depth is less important in the simulation.