Search

Yunlin County is located in the central part of western Taiwan with major emissions from the Mailiao industrial park, the Taichung Power Plants and heavy traffic. In order to understand the influence of meteorological conditions on PM 2.5 concentrations in Yunlin County, we applied a two-stage cluster analysis method using the daily averaged surface winds from four air quality monitoring stations in Yunlin County to classify the weather pattern. The study period includes 1095 days from Jan 2013 to December 2015. The classification results show that the low PM 2.5 concentration occurs when the synoptic weather in Taiwan is affected by the strong southwesterly monsoonal flow. The high PM 2.5 concentration occurs when Taiwan is under the influence of weak synoptic weather conditions and continental high-pressure peripheral circulation. A high PM 2.5 event was studied and the Weather Research and Forecasting (WRF) meteorological model was performed. The result indicated that due to being blocked by the Central Mountain Range, Yunlin County, which is situated on the leeside of the mountains, exhibits low wind speed and strong subsidence behavior that favors PM 2.5 accumulation.

A real-time air quality forecasting (AQF) system was developed in Taiwan using the Weather Research and Forecasting meteorological model and Community Multiscale Air Quality model framework. This study evaluated the performance of the one-year archived AQF PM2.5 forecasts (October 2018 to September 2019) and developed a bias-correction method to improve the accuracy of the PM2.5 forecasts. The bias-correction method incorporates a cluster-analysis-based synoptic weather pattern (WP) classification (one type of analog method). In principle, the historical model errors are categorized according to the synoptic WP and used to adjust the PM2.5 forecast. First, the synoptic WPs are determined using K-means cluster analysis (six WPs were identified). Second, the historical AQF PM2.5 bias at each surface station is estimated for each classified WP. Third, a linear-regression relationship between the AQF PM2.5 bias and PM2.5 forecasts for the six WPs is developed to postprocess the PM2.5 forecasts. The AQF PM2.5 bias is found to have a strong dependency on the synoptic WP. A performance assessment of the AQF PM2.5 forecasts reveals systematic PM2.5 underprediction, with the most pronounced underprediction occurring on days associated with weak synoptic weather conditions. Under these conditions, a severe PM2.5 event is likely to occur in Taiwan. The bias-correction method is able to reduce the PM2.5 forecast error and improve the root mean square error (RMSE) and mean bias (MB) calculations. The improvement is most significant on days associated with a weak synoptic WP and in regions where high PM2.5 concentrations are likely to occur. The method is shown to be effective at reducing the AQF PM2.5 bias.

The objective of this study is to understand the influence of regional climate change on local meteorological conditions and their subsequent effect on local ozone (O3) dispersion in Taiwan. The 33-year NCEP-DOE Reanalysis 2 (NNR2) data set (1979–2011) was analyzed to understand the variations in regional-scale atmospheric conditions in East Asia and the western North Pacific. To save computational processing time, two scenarios representative of past (1979–86) and current (2004–11) atmospheric conditions were selected but only targeting the autumn season (September, October and November) when the O3 concentrations were at high levels. Numerical simulations were performed using weather research and forecasting (WRF) model and Community Multiscale Air Quality (CMAQ) model for the past and current scenarios individually but only for the month of October because of limited computational resources. Analysis of NNR2 data exhibited increased air temperature, weakened Asian continental anticyclone, enhanced northeasterly monsoonal flow, and a deepened low-pressure system forming near Taiwan. With enhanced evaporation from oceans along with a deepened low-pressure system, precipitation amounts increased in Taiwan in the current scenario. As demonstrated in the WRF simulation, the land surface physical process responded to the enhanced precipitation resulting in damper soil conditions, and reduced ground temperatures that in turn restricted the development of boundary layer height. The weakened land–sea breeze flow was simulated in the current scenario. With reduced dispersion capability, air pollutants would tend to accumulate near the emission source leading to a degradation of air quality in this region. The conditions would be even worse in southwestern Taiwan due to the fact that stagnant wind fields would occur more frequently in the current scenario. On the other hand, in northern Taiwan, the simulated O3 concentrations are lower during the day in the current scenario due to the enhanced cloud conditions and reduced solar radiation.

Ammonia (NH3) is an important precursor of inorganic fine particulate matter (PM2.5). Without specific information on the temporal variation in NH3 emissions, a simplified temporal variation in NH3 emissions is often used in chemical transport models. To better characterize NH3 emissions in an air quality model simulation for Taiwan, a dynamical NH3 emissions parameterization was applied to improve the temporal profile of NH3 emissions from livestock operations, synthetic nitrogen fertilizers, and standing crops. The Community Multiscale Air Quality (CMAQ) modeling simulation with a fixed NH3 emissions rate (the CONST experiment) presents a large positive bias in simulated nitrate and NH3, particularly during the nighttime and winter months. On the other hand, the CMAQ simulation with a dynamical NH3 emissions approach (the DYN experiment) improves the diurnal and seasonal variations and reduces the simulated bias. Moreover, according to the Taiwan emissions inventory, NH3 emissions from sewage accounts for a large portion (37%) of total NH3 emissions in Taiwan. The CMAQ simulation with the dynamical NH3 emissions approach and with a reduced level of NH3 emissions from sewage (the DYN1 experiment) was conducted to assess the possibility that the existing Taiwan emissions inventory may overestimate sewage NH3 emissions. The evaluations with observed NH3, nitrate, and ammonium wet deposition concentrations indicate that the DYN1 experiment performs better than the CONST and DYN experiments.

This study is conducted to identify the synoptic weather patterns that are prone to cause high carbon monoxide (CO) concentrations observed at a mountain site, Lulin atmospheric background station (LABS), in Taiwan due to the biomass-burning activity in Southeast (SE) Asia. LABS is recognized as a clean background station. The study period targets the biomass-burning season (February to May) from 2007 to 2010. The synoptic weather patterns were classified using a two-stage clustering method with inputs from the Weather Research and Forecasting (WRF) meteorological model simulation result in a 27-km spatial grid. A 9-km resolution WRF modeling was performed additionally for 13 to 26 March 2007, when a high CO concentration reaching 500 ppb was observed at LABS. The simulation result indicates that not only the existence of the thermal forcing induced low pressure system formed in Indochina, but also the presence of the high terrain located in the northern part of SE Asia that further forced the uplift of the biomass-burning emissions. On the other hand, when the northeasterly monsoonal flow is strong enough and intruding into Indochina, this would hinder the development of the thermal low and weaken the upward movements, in turn preventing the transport of biomass-burning emissions from Indochina to the area of Taiwan. The simulation results also demonstrate that the location of the SE Asia high pressure system has a moderate effect on the particle dispersion path in the upper level.

Adequate air quality modeling is reliant on accurate meteorological simulations especially in the planetary boundary layer (PBL). To understand how the boundary layer processes affect the mixing and transport of air pollutants, the sensitivity of Weather Research Forecasting (WRF) model with different PBL schemes (YSU and MYJ) is utilized. Community Multiscale Air Quality (CMAQ) modeling system is performed subsequently to study the effects of the PBL physical processes on the meteorological and air quality simulations. A comparison is made of two distinct atmospheric conditions. Case 1 considers the influence of the Asian continental outflow where air pollutants carried by long-range transport (LRT) to Taiwan. The variation in ozone (O3) concentration between the two sensitivity runs is mainly caused by the PBL height difference with WRF–MYJ predicts much deeper PBL height near the frontal low-pressure region than does the WRF–YSU. Case 2 is associated with the land-sea breeze flow. In this situation O3 is locally produced from the western side of the country where major metropolitan cities and highways are located. Distinctions in O3 are caused by difference in the strength of the land-sea breeze flow between the two runs. At night the WRF–YSU predicts a weaker offshore land breeze than does the WRF–MYJ near the western coastline. During the day, the WRF–YSU predicts a stronger sea breeze near the offshore area than does the WRF–MYJ, while over the landside, the WRF–YSU predicts a lower wind speed than does the WRF–MYJ.

In the companion paper, we showed that MM5 simulation using a satellite-derived high resolution Texas Forest Service (TFS) land use and land cover (LULC) data set (M2), compared to the MM5 results with the default USGS-LULC (M1), improved representation of the complicated features of the atmospheric planetary boundary layer (PBL) in the Houston ship channel (HSC) area, where large industrial emission sources are concentrated. In the present paper, the study is extended to investigate these effects on air quality simulations. Two emission inputs, namely E1 and E2, are prepared with the M1 and M2 meteorology data, respectively, to reflect the differences in the point source plume rise estimates while keeping the biogenic and mobile emissions the same. Air quality simulations were performed with CMAQ using the M1E1 and M2E2 inputs. The simulation results demonstrate the importance of utilizing high resolution LULC data. In the default LULC data, the HSC area was classified as grass land cover, and MM5 predicted confined mixing, resulting in over-prediction of ozone (O3) precursors, such as NOx (NO plus NO2), and highly reactive volatile organic compounds (HRVOC) species, including ethylene and propylene, over the HSC area. In the TFS data, the area was classified as the impervious “urban” land use and MM5 predicted enhanced mixing of the precursor species, leading to better agreements with measurements. The high resolution LULC also resolves the location of water body near the HSC more accurately, predicting shallower PBL heights than the default LULC during daytime. With favorable wind conditions, the O3 precursors were transported from the HSC emission source towards the area, trapping the pollutants in a confined shallow mixing layer that occasionally led to a rapid photochemical production of O3. The above comparison includes the changes in both meteorological and plume-rise emissions inputs. We performed two additional CMAQ simulations using the same meteorological result (M2) but with different emission point sources E1 and E2 to determine the importance of emission changes on the air quality simulations. The sensitivity tests with the different plume-rise emission inputs due to two different meteorological inputs show little impact on the air quality simulations in this case.