Climate change, air pollution, and public health: Past, present, and future
Keynote Speaker: Susan Anenberg, George Washington University
Description: Climate change and air pollution are interrelated in several important ways. This talk will address how the composition of the Earth's air has changed since preindustrial times, impacts of climate change on air pollution, and approaches for improving public health through greenhouse gas and air pollution mitigation.
Turbulence Resolved and Fine Scale Processes
Towards Modelling Turbulence Intermittency in Atmospheric Flow
Presenter: Cedrick Ansorge, Universität zu Köln
Presentation Description: The challenge of turbulence intermittency in strong stratification is the spatio-temporal confinement of turbulent mixing to a subset—be it sparse or abundant—of the flow domain. Recent field observations and simulation of intermittency in atmospheric flow demonstrate that existing models for turbulent flow are well-suited to locally analyze and model fluxes in the turbulent partition, at least in simple configurations. However, because the turbulent partition is confined to a subset of the entire flow, a unifying approach to model fluxes of active and passive tracers in the average-flow remains elusive for the intermittent regime. A complete quantitative model requires to understand and quantify (1) the relative size of the turbulent partition, (2) the role of interaction in between turbulent and non-turbulent partitions. This interplay is accessible by fully turbulence-resolving simulation (DNS), but a quantitative transfer of results from DNS to atmospheric scale is currently hampered by a lack of observations and Large-Eddy simulations in the simplistic configurations that are accessible to DNS. We present here first results from a joint DNS-LES approach to this intermittently turbulent regime that is designed to eventually overcome this scale gap.
An introspection in the universe of omitted large secondary circulations
Presenter: Marc Calaf, Department of Mechanical Engineering, University of Utah
Presentation Description: Transport of momentum, mass, and energy has traditionally been interpreted through the lens of a mean contribution and turbulent fluctuations. The mean part represents the predominant, stable behavior of a given variable, and the turbulent fluctuation represents the sporadic departure from that mean. This a priori simplified interpretation of atmospheric flow processes has been broadly applied in both experimental analyses and numerical simulations. This view, promoted in part by the Reynolds decomposition, facilitated the study of the turbulent contributions, enabling a rather deep understanding of turbulent processes, their evolution, dependence on external forcing, and corresponding scaling or lack thereof. For example, development of Monin-Obukov’s similarity theory and successive development of scaling relations was one of the main successes in atmospheric boundary layer (ABL) theory of the last century. Also, this flow decomposition enabled the development of Large-Eddy simulations and the corresponding subgrid models used therein. Meanwhile, study of the mean contributions remains limited, perhaps in part because of the apparent lack of complexity, or because it was considered already well understood. It could also be because traditionally, momentum, mass, and energy exchanges in the ABL have been mostly considered in the vertical direction, under the overarching assumption of horizontal homogeneity. Yet, this dominant interpretation has two important weaknesses: what is the correct definition of the mean; and why are we content with overlooking the beautiful heterogeneity that characterizes this world? At present, it is not disputed that we remain unable to close the surface energy balance despite all our best efforts in measuring, accounting for data corrections, and using the most modern and advanced technology. In fact, most recent hypotheses point towards the influence of initially unaccounted large secondary circulations for the closure of the SEB. Interestingly, these secondary circulations happen to be not part of the mean flow behavior, nor of the turbulence fluctuations as traditionally described. Similarly, several recent works have also pointed to the problem that mesoscale models omit the representation of secondary circulations through their subgrid models because these do not necessarily behave as traditional inertial turbulence, and at the same time remain unresolved in the current numerical grids. As a result, mass flux corrections need to be added, or near surface processes remain unaccounted. In this work we will plunge into the universe of secondary circulations, both from an experimental and computational perspective, exploring those cases in which their contributions should be accounted for, and those where neglecting them is inconsequential.
Quantifying the Impact of Flow Unsteadiness on Momentum and Scalar Transfer in Urban Environments
Presenter: Marco Giometto, Columbia University, Department of Civil Engineering and Engineering Mechanics
Presentation Description: Advancing the current understanding and capability to predict atmospheric flow and related transport in urban areas is critical for many applications, including air quality assessment and modeling, urban climate, pedestrian comfort and structural resilience. Turbulence in these environments is rarely in equilibrium with the underlying surface and is typically characterized by strong departures from statistical stationarity. For example, the atmospheric boundary is often driven by a range of (sub)meso forcings that can evolve over sufficiently short time scales and result in unsteady flow conditions in cities. Yet, current theories describing mass, energy, and momentum transport in urban areas are largely established for equilibrium and statistically stationary flow. This presentation will provide an overview on the effects of a particular class of flow unsteadiness (flow pulsation) on the structure of mean flow and turbulence in urban areas. The study is based on results from large-eddy simulation of pulsatile open-channel flow over and within an array of cuboids, where the frequency and amplitude of the pressure gradient driving the flow have been programmatically varied to span a realistic range of conditions. The discussion will focus on the impact of flow unsteadiness on local and horizontally-averaged vertical transfer rates of momentum and kinetic energy, and on implications related to urban ventilation.
Near-Field Modeling of Transport and Dispersion of COVID-19 Virus Indoors and Outdoors
Presenter: Steven Hanna, Harvard TH Chan School of Public Health
Presenter Information: COVID-19 has been shown to be spread by transport and dispersion of aerosols emitted from infected person’s mouths. Infection may occur when more than a few hundred virons have been breathed in by a person. This inhalation dose could occur in a few seconds in the near-field as a cloud of aerosols from a sneeze passes by. The transport and dispersion process follows standard near-field formulations such as a Gaussian plume or puff model or a diffusivity (K) model, which require knowledge of the mean flow speeds, the turbulence intensity, and the distance scale of the turbulence. We show how a Gaussian model is more appropriate for cloud sizes less than the distance scale of the turbulence, while a K model is more appropriate for cloud sizes larger than the distance scale. Here we focus on simulation of observations from the TRANSCOM tracer experiment in Boeing 767 and 777 airplanes, where the ventilation system has about 30 air changes per hour. We assume a constant mean flow (10 cm/s) and a constant turbulence intensity (0.5), based on previous studies similar airplanes. The TRANSCOM tracer study involved a constant release rate of tracer over 60 s from a mannequin’s mouth and 40 samplers within about 5 m. The observed cloud spread was relatively large (over a 180° horizontal arc) due to large turbulence intensities. A simple near-field model is shown to agree with observations of maximum concentration and dose within a factor of two or three.
Benefits of a Three-Dimensional Planetary Boundary Layer Parameterization for Horizontally Heterogeneous Flows
Presenter: Timothy Juliano, National Center for Atmospheric Research (NCAR)
Presentation Information: Producing accurate forecasts of planetary boundary layer (PBL) properties is challenging in geographical areas that contain complex terrain or horizontal inhomogeneities in, for example, land characteristics. Although recent computational advances have led to a substantial improvement in the spatial resolution of numerical weather prediction (NWP) models, their horizontal grid cell spacing (dx) is currently within or approaching the “terra incognita” (i.e., dx ≈ 100 m – 1000 m). At this grid spacing, three-dimensional (3D) effects are non-negligible, since the most energetic turbulent eddies are neither fully parameterized (as in mesoscale simulations) nor fully resolved (as in large-eddy simulations). To address this modeling challenge, we have implemented a 3D PBL parameterization into the Weather Research and Forecasting model. The 3D PBL scheme is based on the level 2.5 algebraic model developed by Mellor and Yamada (MY) and explicitly calculates the momentum, heat, and moisture flux divergences. To illustrate the benefit of using a fully consistent turbulence closure framework during horizontally heterogeneous conditions, we present results from idealized cases focusing on a seabreeze front initiation and a mountain-valley thermal circulation, as well as a real case of a well-observed cold pool event. While the 3D PBL scheme reproduces the evolution of salient features well, we highlight the need to improve the original MY turbulent length scale formulation.
Application of urban climate model PALM-4U to investigate pollutant distribution in Stuttgart
Presenter: Abdul Samad, University of Stuttgart
Presentation Description: The air pollution situation in the German city of Stuttgart is of great importance as high pollutant concentrations are measured here as compared to other German cities. This is mainly due to its geographical location as it is situated in a basin covered by hills on the three sides. This leads to reduced wind speeds that inhibit pollutant dispersion. One of the main contributors to the pollutant concentrations in the Stuttgart is local traffic. In this study, the urban climate model PALM-4U was applied to obtain the pollutant distribution along the federal highway B14 of Stuttgart in order to evaluate the impact of traffic on air quality. The simulations were carried out in two areas of the city, namely Kaltental valley and Am Neckartor with the domain size of 1.6 × 2 km and 3.2 × 2 km respectively having grid size of 10 m for each domain. The federal highway B14 affects the air quality of both investigated areas. The influence of traffic emissions on the air quality of Stuttgart was studied for summer and winter days. The results show that air pollutants from traffic emissions on federal highway B14 are distributed in the investigated areas proving the significance of traffic on the air quality of the city.
Unique/Extreme Events and their Impacts on Meteorology and Air Quality
PM2.5 Impacts of the Record- Setting 2020 Wildfire Season in Southern California: The Bobcat and El Dorado Fires
Presenter: Melissa Maestas, South Coast Air Quality Management District
Presentation Description: The South Coast Air Basin (SoCAB) encompasses the greater Los Angeles Metropolitan area in Southern California, including 6,745 square miles with 17 million people and significant air quality challenges. In particular, the SoCAB does not attain the federal PM2.5 standards. While anthropogenic PM2.5 concentrations have declined substantially, in certain years such as 2020, wildfire smoke is a dominant source of PM2.5 in the region. The Clean Air Act allows the exclusion of PM2.5 measurements during exceptional events from attainment calculations. Exceptional events are natural or human-caused events unlikely to recur when high PM2.5 concentrations are not controllable. Without the influence of two exceptional events in 2020—the El Dorado and Bobcat fires—the SoCAB would attain the 24-hr federal standard. The El Dorado fire started on Sept. 5, 2020 in the San Bernardino National Forest and the Bobcat Fire started the next day in the Angeles National Forest. These fires burned a combined total of 138,540 acres. Influences of daily fire activity and meteorology on the spatial and temporal variations of PM2.5 concentrations within the SoCAB are analyzed. Smoke from these fires intermittently caused increased PM2.5 concentrations across the entire basin. During September 11-16, 2020, the SoCAB experienced 60 exceedances of the 24-hour PM2.5 standard (35 µg/m3) among 16 stations, with 24-hour average PM2.5 concentrations up to 103 µg/m3.
Australia’s Black Summer Pyrocumulonimbus Super Outbreak Reveals Potential for Increasingly Extreme Stratospheric Smoke Events
Presenter: David Peterson, U.S. Naval Research Laboratory
Presentation Description: The Black Summer fire season of 2019-2020 in southeastern Australia contributed to an intense ‘super outbreak’ of fire-induced and smoke-infused thunderstorms, known as pyrocumulonimbus (pyroCb). More than half of the 38 observed pyroCbs injected smoke particles directly into the stratosphere, producing two of the three largest smoke plumes observed at such altitudes to date. Over the course of three months, these plumes encircled a large swath of the Southern Hemisphere while continuing to rise, in a manner consistent with existing nuclear winter theory. We connect cause and effect of this event by quantifying the fire characteristics, fuel consumption, and meteorology contributing to the pyroCb spatiotemporal evolution. Emphasis is placed on the unusually long duration of sustained pyroCb activity and anomalous persistence during nighttime hours. The ensuing stratospheric smoke plumes are compared with plumes injected by significant volcanic eruptions over the last decade. As the second record-setting stratospheric pyroCb event in the last four years, the Australian super outbreak offers new clues on the potential scale and intensity of this increasingly extreme fire-weather phenomenon in a warming climate.
Using observations of Western U.S. wildfire smoke to improve fire emissions in air quality forecasting models
Presenter: Megan Bela, Cooperative Institute for Research in Environmental Sciences (CIRES) University of Colorado / NOAA Chemical Sciences Laboratory (CSL)
Presentation Description: Air quality forecasts using regional chemical models provide key information for affected communities and smoke management efforts, yet many models fail to accurately predict ozone (O3) and fine particulate matter (PM2.5) levels during fire events. Our research utilizes the dataset from the 2019 NOAA/NASA Fire Influence on Regional to Global Environments and Air Quality (FIREX-AQ) field campaign to improve process-level understanding and model representations of fire emissions, plume rise, and chemistry, with the aim of developing a better capability to predict air quality and weather in fire-affected regions. Satellite-based fire emission estimates for the FIREX-AQ period are compared with emissions estimates based on field measurements and satellite data. We simulate the FIREX-AQ period with the Weather Research and Forecasting with Chemistry (WRF-Chem) model. Model fire emissions amounts are constrained by observation-based estimates, and emission factors are updated based on laboratory and field observations. Simulated trace gas and aerosol fields and plume injection heights are compared with observations from FIREX-AQ and surface network observations. The simulations are used to explore chemical and physical processes affecting O3 and PM2.5 amounts.
Use of linear regression techniques for demonstrating the exceptional impact of the Covid-19 health emergency on ozone air quality in the District of Columbia
Presenter: Joseph Jakuta, District of Columbia Department of Energy an Environment
Presentation Description: Tropospheric ozone (O3) is an air pollutant harmful to human health and is formed in the atmosphere as a secondary gas through the interaction of precursor pollutants- nitrogen oxides and hydrocarbons, in the presence of sunlight. Tropospheric O3 is known to exacerbate respiratory ailments such as asthma.
During the 2020 Coivid-19 health emergency, the traffic volumes were greatly impacted leading to unusually lower ambient O3 levels in several areas, including the District. The objective of this work is to build a regression analysis forecast tool using 2013-2017 meteorological data that can reliably predict District O3 levels in 2018 and 2019. This tool is expected to be unreliable for the 2020 Covid-19 health emergency period.
To forecast O3 in the District, ordinary least squares regressions (OLS) and quantile regressions (QR) were used, where 1-hr O3 was the dependent variable, and the meteorological conditions and temporal traits as independent variables. QR is a regression tool that more fully captures the distribution of nonlinear variables such as O3. The QR model was observed to be more efficient for forecasting 2018 and 2019 hourly O3, while OLS tended to under-predict O3.
These models may prove useful for protecting District residents’ health and will aid in requesting that the 2020 health emergency be deemed an “exceptional event” by the U.S. Environmental Protection Agency so that the 2020 O3 data is not used for making regulatory and policy decisions.
COVID-19 lockdown and chemical characteristics of ozone formation in India
Poster Presenter: Behrooz Roozitalab, University of Iowa
Lightning Talk Description: Ozone pollution is an emerging air quality issue in India. It became more prominent during the COVID-19 lockdown period in which no major changes in concentrations were measured. We studied the air quality during the lockdown period in India. Using the Integrated Reaction Rate (IRR) package in WRF-Chem model, we performed a process analysis and studied the chemistry of ozone formation over different regions in northern parts of India. Indeed, we used the ratio of chemical loss of radicals with radicals and NOx to find the ozone production regimes. Furthermore, we estimated the corresponding formaldehyde to NO2 ratio (FNR) in each region. We found that urban regions and regions with many power plants are primarily in VOC-limited regime, while rural regions are in NOx-limited regime. As a result, large NOx reductions during the lockdown period increased ozone concentrations in some NOx-limited regions in India. Our results also showed that formaldehyde, isoprene, acetaldehyde, and ethylene as VOC species with highest contribution to ozone formation in all the regions. These results provide information for policy makers to implement efficient emission control scenarios to address ozone pollution in India.
Connecting fire behavior to air quality: a case study of the 2020 Northern California wildfire season
Poster Presenter: William Lassman, Lawrence Livermore National Lab
Lightning Talk Description: Wildfires are a major source of atmospheric aerosol, which can degrade regional air quality and impact the earth’s radiative balance. Due to the growing importance of wildfire, different tools for simulating smoke emission and evolution from wildfires, each with different applications in mind, have been developed. One approach is to use remote sensing with parameterized fire behavior to derive prescribed emissions; while this approach is well-suited for simulating many fires at once, the parameterized fire behavior may not be an accurate recreation of any specific event. Another approach is to couple the fire emissions to a dynamic fire spread model, which allows for a more accurate representation of the spatial and temporal behavior of emissions, but requires high-quality fuel data as an input and is challenging to scale to complex fire systems. In this study, we compare these two approaches for the case of the 2020 Northern California wildfire season. We present simulations of this event using WRF-Chem with FINN biomass burning emissions, as well as WRF-SFIRE-Chem where the smoke emissions are coupled to the online fire spread model. We will evaluate the meteorological forecast, fire spread rate, and smoke concentrations for the dispersion models. We will then identify strengths and weaknesses of the different approaches, and identify missing or unconstrained processes that impact the models’ abilities to simulate the complex phenomena that occurred.
Ensemble PM2.5 Forecasting During the 2018 Camp Wildfire
Presenter: Daniel Tong, George Mason University
Presentation Description: As drought becomes the new normal in many western states, wildfires have exerted increasing effects on air quality and human health and welfare in the United States, examplified by the 2018 Camp Fires in California. This study uses the HYSPLIT and CMAQ models and a suite of in-situ and remotely sensed observations to assess the accuracy of five fire emission products derived from satellites. These products include four top-down emission datasets (GBBEPx from NOAA, FLABME from Naval Research Lab, FEER from NASA, and GFAS from ECWMF) and one bottom-up emission dataset (BlueSky) from US Forestry Service). Comparisons of fire emisisons show large variations of PM2.5 estimates by these datatsets, caused by both emission algorithms and fire detection capability of different satellite sensors. We conducted 112 model simulations to investigate the sensitivity of plume injections to different meteorology inputs (GDAS, NAM12, NARR, and WRF), planetary boundary layer (PBL) conditions, plume rise schemes (Briggs 1969, and Sofiev 2012). Results show that the simulated injection height is highly dependent on the PBL height and that the injection height calculated using the Sofiev 2012 scheme is higher than using the Briggs 1969 scheme. Finally, we showed the results of evaluations of the 84 simulations against the MISR and CALIPSO observed plume heights, as well as the ground PM2.5 concentration observation from the EPA Air Quality System network.
Lighting the dark: first retrieval of fire combustion efficiency from space for air quality applications
Presenter: Jun Wang, University of Iowa
Presentation Description: Depending on the region and time of interest, existing inventories of biomass burning emissions for carbonaceous (BC or OC) aerosols can have uncertainty up to a factor of 10. One key source of this uncertainty is fire emission factors that depend on the surface type and combustion efficiency. For the same surface type or the same fire, the combustion efficiency may also vary with time depending on the meteorological conditions and other environmental factors. However, even in the top-down estimates of biomass burning emissions in which the total amount of fuel burned is constrained by the fire radiative power (FRP) measured by the satellite, the temporal variation of emission factors as a function of combustion efficiency are largely ignored. In many methods for inventory estimates, the emission factors are prescribed for 6-10 surface types and are used as constants for each season if not year round.
Here, we show the development of a novel approach to improve the estimate of fire emission factors by using satellite-based fire visible energy fraction (VEF) data that we are able to retrieve from Visible Infrared Imaging Radiometer Suite (VIIRS) at night. VEF is defined as the ratio between the fire radiative power in the visible spectrum and the total FRP and, therefore, is an indicator of fire combustion efficiency: the larger the VEF, the more complete the combustion. We will present the semi-operational product of modified combustion efficiency (MCE) derived from VIIRS over the continental U.S. (http://esmc.uiowa.edu:3838/fires_detection/). We will also illustrate the advantages of using VIIRS-based MCE to describe the life cycle of Camp fire in 2018 – the most devasting fire in the history of California. Finally, we will also show the validation of satellite-based MCE based on the findings from FIREX-AQ campaign.
Quantifying the impact of wildfires on air quality in Western US urban centers
Poster Presenter: Kai Wilmot, University of Utah
Lighting Talk Description: Wildfires are degrading air quality in the Western US during the months of August and September, as revealed by decadal scale trend analyses (2000-2019) of atmospheric composition, wildfire emissions, and fire area burned datasets. Overlap of increasing and statistically significant trends in upper quantile fine particulate matter, organic carbon, and absorption aerosol optical depth highlight the impact of wildfires on air quality in the Pacific Northwest during the month of August. September is characterized by emerging trends across the Pacific Northwest, western Montana, and Wyoming. Trend analyses of wildfire derived PM2.5/burned area are used to identify potential wildfire emission “hotspots” with relevance to human exposure to degraded air quality. The proximity of potential wildfire emission hotspots and extreme air quality trends, as well as their similar spatial shifts from August to September, further supports the hypothesis that wildfires are driving extreme air quality trends in Western US urban centers. Preliminary results from atmospheric transport modelling that explicitly parameterizes turbulence support partial attribution of Western US air quality trends to wildfire sources. Potential emission hotspots identified via observational analyses are further supported by time-reversed simulations providing source-receptor linkages between upwind fires and Western US urban centers.
Assessment of WRF-Fire’s Forecasting Skill on Large Wildfires
Presenter: Francis Turney, University of California, Los Angeles
Presentation Description: Large wildfires have become notoriously common and impactful on the earth system in recent years. Despite currently limited forecasting skill, predicting fire spread and heat output are crucial to predicting smoke emissions and transport, and planning emergency response efforts. Here we examine the potential of the weather/fire-spread model WRF-Fire in predicting burned area and heat flux for the 2019 Williams Flats Fire, the biggest fire sampled during the 2019 NOAA-NASA FIREX-AQ campaign. Using a novel re-gridding scheme, we compare WRF-Fire heat output and fire count to the GOES-17 satellite and other datasets to get the highest temporal resolution comparison of WRF-Fire that currently exists. Preliminary comparisons show that modeling containment efforts and prescribing accurate fuel density provide the largest sensitivities, and several solutions are developed and described to aid future forecasters in predicting heat flux and area burned.
Modeling of Processes Across Global to Regional and Local Scales
Generating an open biomass burning emission inventory using fire radiative power from VIIRS
Presenter: Gonzalo Ferrada, Center for Global and Regional Environmental Research
Presentation Description: Biomass burning emissions are one of the major contributors to climate change. Fires emit aerosols and many trace gases, including many hydrocarbons and nitrogen compounds that could lead to the formation of ozone or secondary organic aerosols. Several biomass burning inventories have been created to quantify emissions. Some of them, use a bottom-up approach which relies on the burned area of the fire as input. Others, use the fire radiative power (FRP) to estimate emissions (top-down approach). Almost all the inventories use hotspot detections from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument onboard of the Terra and Aqua satellites and provide emissions for a limited number of species and at resolutions no higher than 10 km. This can be a problem when conducting model simulations at high resolutions because fires can be misplaced. Thus, smoke plumes can be subject to wrong circulation patterns. Here, we generated a new fire emission using a top-down approach using FRP data from the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard of the Suomi NPP satellite. The result is an inventory with daily emissions for dozens of species at resolutions of around 500 m. We named this inventory as VIIRS-based Fire Emission Inventory (VFEI). VFEI emissions are comparable to other 4 major inventories. Simulations conducted with WRF-Chem for North America and Southern Africa showed a good agreement with MODIS aerosol optical depth observations.
The CAMS global atmospheric composition forecast system: Recent upgrades and impact of prognostic aerosols and ozone on weather forecasts
Presenter: Johannes Flemming, ECMWF
Presentation Description: To address environmental concerns about atmospheric composition, the European Union funds the Copernicus Atmosphere Monitoring Service (CAMS) as part of its Copernicus programme. CAMS is implemented by ECMWF and delivers a wide range of regional and global products on air quality, stratospheric ozone, emissions, solar radiation and climate forcing. Using ECMWFs operational weather modelling and data assimilation framework, the global CAMS system delivers two times daily 5-day global forecasts of atmospheric composition.
We will give an overview of recent upgrades of the model and data assimilation components of the CAMS global system as well as of the impact of newly added satellite retrievals from the TROPOMI instrument aboard the Sentinel5 P satellite. The potential of prognostic aerosol to improve weather forecasting will be demonstrated for specific cases such as the intensive wildfires in the Western USA in 2020 and desert dust transport events over Europe in Spring 2021.
We will also present the impact of prognostic ozone on stratospheric temperatures during the anomalously long-lasting Antarctic ozone hole in 2020. Besides the case studies, we will discuss the impact of prognostic aerosol and ozone in a more general way based on time- and spatially averaged NWP scores and highlight the specific challenges of that approach.
Exploring the pseudo-global warming method to quantify potential changes in extreme meteorological case studies
Presenter: Geneva Gray, North Carolina State University
Presentation Description: Pseudo-global warming (PGW) is a form of climate model downscaling, where information from global climate models is used to influence the initial and boundary conditions of a regional-scale model. PGW can be used on an array of timescales from multi-decadal to event-based. The event-based implementation of PGW may be used to explore extreme meteorological conditions (such as from temperature or precipitation) that lead to adverse air quality, impacts on human health, or flood risk.
This study uses a PGW approach to simulate potential changes in precipitation events under a warmer climate scenario. Two extreme rainfall events that resulted in heavy flooding are investigated: 1) Ellicott City, Maryland event on July 30-31, 2016, and 2) Portland, Oregon on October 31, 2015. The Weather Research and Forecasting (WRF) model is used to simulate the storm events for both historical and projected scenarios. For future scenarios, a vertical warming profile derived from an ensemble of CMIP5 RCP8.5 scenarios is added to the background environment. Analysis between the historical and PGW simulations shows that the warmer environmental temperature alters duration, intensity, and the spatial distribution of precipitation.
These extreme event examples demonstrate that PGW can be implemented on the event-scale. Future work with this method can explore how warmer meteorological conditions can influence modeled case studies of air quality action days.
Evaluation and Intercomparison of Modeled Atmospheric Deposition over North America and Europe – An Overview of Phase 4 of the Air Quality Model Evaluation International Initiative (AQMEII4)
Presenter: Christian Hogrefe, U.S. Environmental Protection Agency
Presentation Description: This talk will provide an overview of research performed under Phase 4 of the Air Quality Model Evaluation International Initiative (AQMEII) by more than 10 research groups from North America and Europe. The focus of this initiative is on analyzing deposition of trace gases simulated by regional-scale air quality models. This research initiative is divided into two complementary activities. The first activity—an analysis of deposition processes in annual simulations performed by grid-based regional air quality models—is based on adding detailed land use/land cover (LULC)-specific diagnostics to the algorithms used in the models’ dry deposition modules as well as archiving variables for comparing the relative influence of different pathways towards the net or total dry deposition. The second activity—an evaluation of dry deposition point models against ozone flux measurements at multiple towers with multiyear observations at a diverse set of northern hemisphere locations—allows for quantifying differences among dry deposition schemes driven by identical conditions at a variety of sites, minimizing input uncertainty in model evaluation, and identifying responses to meteorology, biophysics, and ecosystem characteristics. We will discuss the details of the design and objectives of this two-pronged approach and present examples of its application to simulations contributed by participating groups.
Incorporating satellite soil moisture data into dry deposition modeling: sensitivity to dry deposition parameterizations
Presenter: Min Huang, George Mason University
Presentation Description: Satellite soil moisture data from NASA’s Soil Moisture Active Passive mission are assimilated into a coupled regional-scale modeling system covering the southeastern US based on two different dry deposition schemes (i.e., the Wesely and “dynamic” schemes, in the latter of which the dry deposition parameterization is coupled with photosynthesis and vegetation dynamics). It is demonstrated that, when the "dynamic" scheme is applied, the modeled dry deposition velocities and ozone fluxes are larger and more sensitive to soil moisture data assimilation. It is also found that the choice of soil moisture factor controlling stomatal resistance (i.e., the beta factor) scheme affects strongly the quantitative results. Satellite and satellite-derived vegetation, latent/sensible heat fluxes and gross primary productivity data are also used to help assess the land surface model performance and indirectly indicate the usefulness of soil moisture data assimilation for improving the modeled flux estimates. This study is an extended work of Huang et al. (2020, https://doi.org/10.5194/acp-2020-499), in which the widely-used Noah land surface model was applied.
Reduced order modeling and source attribution with CMAQ-DDM-3D in California
Presenter: Zhen Liu, California Air Resources Board
Presentation Description: Modern air quality policy making and planning increasingly calls for efficient estimate and comprehensive analysis of source-receptor relationships. We present our ongoing effort of testing and applying CMAQ-DDM-3D for reduced order modeling and source attribution for the whole state of California. In this work, we have implemented and tested CMAQ-DDM-3D (v5.2) with the SAPRC07tic-AERO6i chemical mechanism. The model was run at 12 km resolution and first-order sensitivities were evaluated for the emissions of NOx and VOCs from all 58 counties in California. Specifically, two annual sensitivity runs were conducted with the emission inventories for 2017 and 2032 baselines, respectively, generating two sets of reduced order modeling results at two different emission levels. These reduced order modeling results were then used to project changes of air quality due to large and small emission perturbations at the 58 counties, respectively. The results were evaluated against those from the full CMAQ modeling, focusing on PM and toxic VOCs species in light of their non-linear response to emissions. Examples of results from county-level source attribution with reduced order modeling will be presented. Our experience from this relatively large application of CMAQ-DDM-3D, including computational needs, pre- and post- processing procedures will also be presented.
The potential impact of vehicle-induced turbulence on regional air pollution
Presenter: Paul Makar, Environment and Climate Change Canada
Presentation Description: Air-quality models require parameterizations for processes occurring below the scale that of the models’ horizontal and vertical resolution; an example of such a sub-grid-scale parameterization is buoyant plume rise from large stacks (plume temperature, exit velocity, and the atmospheric temperature gradient and stability are used to estimate plume rise, with the emitted mass added into the model atmosphere up to the estimated plume height). Moving vehicles also add at-source (kinetic) energy to fresh emissions. We develop a sub-grid-scale parameterization linking vehicle motion to at-source enhancements in turbulent kinetic energy, using a combination of observations and past large eddy simulation model results. We link this parameterization to the number of vehicle kilometers travelled within a model grid cell - and show how it may be applied within a model’s vertical diffusion operator. Similar to buoyant plume rise, the new parameterization simulates the effects of sub-grid-scale vehicle-induced turbulence, redistributing freshly emitted mobile sector pollutants in the vertical as they are emitted. The new parameterization has been tested using North American (10km grid cell size) and PanAm Games (2.5km grid cell size) domains within the GEM-MACH regional air-quality model: results show model forecast improvements significant at the 90% confidence level for NO2, O3, and PM2.5, with larger increases in performance in urban, high-population areas.
MUlti-Scale Infrastructure for Chemistry and Aerosols (MUSICA)
Poster Presenter: Gabriele Pfister, National Center for Atmospheric Research
Lightning Talk Description: To explore the various couplings across space, time and between ecosystems in a consistent manner, atmospheric modeling is moving away from the fractured limited-scale modeling strategy of the past towards a unification of the range of scales inherent in the Earth System. The MUlti-Scale Infrastructure for Chemistry and Aerosols (MUSICA) is intended to become the next generation community infrastructure for research involving atmospheric chemistry and aerosols. MUSICA is being developed collaboratively by the National Center for Atmospheric Research (NCAR) and university and government researchers, with the goal of serving the international research and applications communities. The capability of unifying various spatio-temporal scales, coupling to other Earth System components and process-level modularization will allow advances on topics ranging from fundamental research to air quality to climate and is also envisioned to become a platform that addresses the needs of policy makers and stakeholders. The overall design of MUSICA and some first results from the publicly released Version 0 will be presented. The community is invited to participate in MUSICA development and the opportunities for collaboration will be described.
Development and evaluation of the coupled MPAS-CMAQ model system
Presenter: Jonathan Pleim, US EPA
Presentation Description: The USEPA has embarked on the Advanced Air Quality Modeling System (AAQMS) project to enable modeling of air quality from global to regional to local scales. The system will have three configurations: Global meteorology with seamless mesh refinement and online (coupled) atmospheric chemistry; Regional online meteorology and chemistry; and Offline regional meteorology and chemistry. The global configuration includes the Model for Prediction Across Scales – Atmosphere (MPAS-A) v7.0, developed at the National Center for Atmospheric Research (NCAR), coupled with the latest version of the Community Multiscale Air Quality (CMAQv5.3.2) model. We will show comprehensive evaluation of meteorology simulated by the EPA-enhanced version of MPAS-A that includes the addition of four-dimensional data assimilation, the ACM2 PBL model, PX land surface model, and enhanced Kain-Fritsch convection on two global meshes with refinement over North America. We will also present our most recent testing and evaluation of air quality simulations by the coupled MPAS-CMAQ system. Global emissions from EDGAR-HTAP are combined with the 2016 EPA NEI for the U.S., supplemented by biogenic emissions provided by the inline Model of Emissions of Gases and Aerosols from Nature (MEGANv3.1). Global ozone fields from the Copernicus Atmosphere Monitoring Service (CAMS) are used for a “hot start” and for stratospheric ozone data assimilation in the upper layers of the model for the entire simulation.
Examination and Processing of MODIS Leaf Area Index (LAI) Product for Air Quality Modelling
Presenter: Junhua Zhang, Environment and Climate Change Canada
Presentation Description: Leaf Area Index (LAI) is used in air quality models for land surface processes and for calculating biogenic emissions. MODIS LAI product provided by NASA has been widely used in the air quality modeling community for such purposes. However, limitations of MODIS LAI product have been seen for some geographic areas, particularly unreasonably low LAI over the evergreen needleleaf boreal forests in the northern hemisphere during wintertime due to snow cover and low sun angle. Missing retrievals over urban areas and areas with persistent cloud cover are also seen. Considerable efforts have been made to improve the MODIS LAI product. However, some issues are still persistent, such as the very low LAI over boreal forests during wintertime. In order to solve these issues for supporting regional air quality modelling, the 8-day MODIS Collection 6 (C6) LAI product at 500m resolution (MCD15A2H) was examined for North America. Statistics were calculated by month and by land cover type defined in the “Land Cover Type 1” science data set (SDS) of the Collection 6 MODIS Land Cover (MCD12Q1) product. Comparisons with LAI calculated from the Biogenic Emissions Landuse Database, version 4 (BELD4) were also done. This presentation will show the analysis results and discuss approaches to address existing issues. A final LAI dataset for North America based on 17 years (2003-2019) of MODIS LAI product will also been shown.
Assessment of future wintertime meteorology over California using dynamical downscaling method with a bias correction technique
Presenter: Zhan Zhao, California Air Resources Board
Presentation Description: This study utilizes the Weather Research and Forecasting model to dynamically downscale a bias-corrected coarse-resolution global climate model dataset from the Coupled Model Intercomparison Project Phase 5 (CMIP5) to a grid size of 4×4 km2 over California for a present (2003– 2012) and a future (2046–2055) decade. Compared to the present climate, an increase in 2-m temperature (up to 2 K) and water vapor mixing ratio (up to 1 g/kg) and a decrease in planetary boundary layer height (up to 80 m) are projected by the 2050s for the entire state of California. The number of stagnant days over the San Joaquin Valley is expected to increase by approximately 6% in the future decade, indicating potential exacerbation of the winter PM issue in this region. The wintertime precipitation is projected to increase by up to 50% in northern California and, conversely, to decrease by up to 40% in southern California during 2046–2055. The solid phase precipitation is projected to decrease over mountain ranges with lower elevations despite an overall increase in total precipitation, while it is projected to increase over the eastern side of the Sierra Nevada with elevation over 2 km.
Utilizing MOS-based Gas Sensors with Algorithmic Temperature Fluctuation Correction for Local Ambient Pollutant Monitoring
Presenter: Akarsh Aurora Student, Ashland High School, MA
Presentation Description: Metal oxide semiconductor (MOS) gas sensors are a mature innovation with high sensitivity, fast response and recovery, and low-cost mass production, making them prudent for small-sized mobile applications. The known sensor technology is suitable for measuring volatile organic compounds and gasses such as NO2, CO, CH4, and H2S. Widespread application of MOS sensors can provide consistent streams of high accuracy data locally to help produce more comprehensive model evaluations for air quality trends.
In this study, we propose a solution to coping with the ambient temperature fluctuations, which can cause drift and inaccuracy in MOS-based sensors. Combining a novel curvature correction method with Taylor polynomials, we pre-model the variance in sensor response under various temperatures and humidities for active, algorithmic correction. We will present the correction method and provide statistical analysis demonstrating its functionality alongside the potential for employing MOS sensors to expand monitoring operations in ambient environments cost-effectively.
Merging Models with Observations
Effect of hygroscopic growth on aerosol light scattering - observed climatology and model evaluation
Presenter: Elisabeth (Betsy) Andrews, University of Colorado, CIRES and NOAA
Presentation Description: Knowledge of the scattering enhancement factor, f(RH), is important for an accurate description of direct aerosol radiative forcing. This factor is defined as the ratio between the aerosol scattering coefficient at enhanced relative humidity, RH, to a reference (dry) scattering coefficient. Here we first present an overview of f(RH) at more than 20 sites based on tandem nephelometer datasets (Burgos et al., 2019). We investigate relationships between f(RH) and other aerosol optical parameters (single scattering albedo and scattering Angstrom exponent) in a less than successful attempt to identify simple potential proxies for aerosol water uptake. Next, we use the observations of f(RH) to evaluate aerosol water content in global model simulations. We show there is a strong indication that differences in the model parameterizations of hygroscopicity and model chemistry are driving some of the observed diversity in simulated f(RH). Finally, we explore how models represent the relationship between f(RH) and other simulated optical properties and evaluate whether our findings provide information that can help improve model representation of aerosol water interactions.
Seasonal Variation Analysis of Ground Level Ozone in the Mekong Delta by Using the Coupled WRF/CMAQ Model
Presenter: Long Bui, University of Technology, Viet Nam National University Ho Chi Minh City (VNU-HCM)
Presentation Description: The Mekong River Delta (Mekong Delta) is the most important socio-economic center of Vietnam, and is considered one of the largest rice granaries in Southeast Asia. Although accounting for only about 12% of the country's total area, the region provides more than 50% of Vietnam's total annual food production, 95% of rice exports, and 65% of aquaculture production. However, in recent years, ground level ozone pollution has shown a worrying trend, leading to a negative impact on agriculture and rice production. The coupled WRF/CMAQ model is applied to find out the laws of the distribution of ground ozone in the region, the relationship between ozone concentration and emissions, meteorology, and terrain. The results show that the transport of ground ozone in the area is greatly influenced by the factors that emit precursor species (nitrogen oxides, NOx and volatile organic compounds, VOCs) from anthropogenic as well as natural activities. In addition, the emission load contributed from volatile organic compounds (VOCs) has a significant impact, more sensitive than NOx during the distribution of ground ozone concentrations. High ground ozone concentrations usually occurred in the coastal provinces East Sea, and in the Ca Mau peninsula area, which are also provinces that have contributed significant emissions in the region during the simulation period.
Advancing Aerosol Modeling across Air Quality, Weather and Climate Applications
Presenter: Gregory Carmichael, University of Iowa
Presentation Description: Atmospheric aerosols impact air quality and human health and play key roles in the Earth’s weather and climate systems. Aerosol amounts and physical and chemical properties determine their toxicity, radiative and microphysical impacts. Recent advances in observations and models are significantly enhancing our ability to quantify the distribution and properties of aerosols, understand their impacts on atmospheric radiation and cloud distributions and properties, and their impacts on human health. Improving air quality (and weather) predictions requires closer integration of observations and models. In this talk we present illustrative results from on-going inter-comparison studies evaluating current capabilities to model important aerosol properties (including MICS-Asia, ORACLES, KORUS-AQ and FIRE-AQ). We draw on these results to highlight areas where further advances in models and observations are needed to enhance seamless prediction of environmental, weather and climate services across relevant spatial and temporal scales.
WRF-Chem modeling of PM2.5 and AOD of Summertime Air Quality around Lake Michigan
Presenter: Megan Christiansen, University of Iowa
Presentation Description: In coastal environments with substantial anthropogenic emissions, peak concentrations of ozone and fine particulate matter (PM2.5) often coincide due to common sources and conditions responsible for secondary aerosol and ozone formation. Regions around Lake Michigan persistently record high ozone concentrations exceeding the National Ambient Air Quality Standards in spring and summer accompanied by transport and formation of PM2.5 through oxidation mechanisms related to ozone formation. The Lake Michigan Ozone Study 2017 (LMOS), a field campaign, occurred during May and June 2017 and employed aircraft, ship, mobile lab, and two enhanced monitoring stations to build an extensive dataset to address these issues. In situ and remote sensed aerosol sampling tracked the particles that occurred during ozone episodes.
To date, post-analysis of LMOS chemical transport modeling has focused on ozone and other gas-phase species. Here, we look at the performance of high-resolution (4km x 4km) Weather Research and Forecasting with Chemistry (WRF-Chem), with 4-bin MOSAIC aerosol scheme, simulations in capturing surface PM and aerosol optical depth. Observations from the LMOS campaign, AQS stations, AERONET, and satellite products are used to evaluate the model. Preliminary results give R=0.52 and mean bias of 1.59 µg m-3 for daily-PM2.5 along the west coast of Lake Michigan. These observations will also be used to determine areas of improvement in model configuration and aerosol schemes.
Observationally constrained source attribution modeling of air pollution health impacts
Presenter: Daven Henze, University of Colorado Boulder
Presentation Description: Air quality models are often used to generated quantitative estimates of the contributions of different pollutant sources to the health impacts associated with exposure to O3, PM2.5 and NO2. However, models themselves often struggle to robustly capture key factors such as fine-scale spatial variability in air pollutant exposure and the magnitudes and distributions of the emissions. Here we present work that integrates ambient observations from in situ measurements and remote sensing with the air quality models used for source attribution, through processes such as inverse modeling, data fusion, downscaling, and observationally constrained exposure estimates. In particular we consider methods for quantifying sources that contribute to national-scale exposure to PM2.5 and O3 in G20 countries. We also present observationally constrained source attribution for O3, PM2.5 and NO2 health impacts in Washington DC. In each case we examine the roles of local vs long-range sources, contributions from individual sectors and species, and how changes to emissions over time contribute to each regions air pollution health burden.
Fire Plume Injection Heights Estimated from Doppler Weather Radar Observations
Poster Presenter: Mansa Krishna, UCLA
Lightning Talk Description: The vertical distribution of wildfire smoke aerosols is useful in determining the effects of these smoke aerosols on surface air quality. Existing data sources, such as case study experiments or lidar measurements, generally do not possess the spatial and temporal resolution required to resolve aerosol profiles on a regional scale. With recent improvements in radar technology allowing for higher resolution spatial and temporal data, work has focused on detecting non-weather artifacts using the Doppler radar. Accordingly, we explore the use of Doppler radar data in estimating the injection heights of biomass burning debris (BBD) generated by fires. We detect BBD as a possible surrogate for smoke aerosol particles, which are often collocated with BBD but are too minute to be detected by the radar. The injection heights of BBD are extracted using 192 hours (August 2 to 10, 2019) of Weather Surveillance Radar-1988 Doppler (WSR-88D) radar data from Spokane, WA to study the Williams Flats Fire in Northeast Washington. Preliminary analyses of the data and test runs of the developed algorithm indicate that the maximum injection height was approximately within 7-9 km on August 7, 2019, which is consistent with airborne lidar data for the same day. Injection heights extracted using this approach represent an initial effort at providing inputs for machine learning models to predict smoke injection or other air quality forecasting applications at a much larger, regional scale.
Development of a chemical data assimilation system for air quality reanalysis over the CONUS
Poster Presenter: Rajesh Kumar, National Center for Atmospheric Research
Lightning Talk Description: The errors in air quality simulations can be partially addressed via assimilation of satellite observations of atmospheric composition. To improve long-term air quality simulations, we have developed a chemical data assimilation system to simultaneously assimilate aerosol optical depth (AOD) retrievals from the Moderate Resolution Imaging Spectroradiometer (MODIS), and carbon monoxide (CO) retrievals from the Measurement of Pollution in the Troposphere (MOPITT) in the Community Multiscale Air Quality (CMAQ) model. The Weather Research and Forecasting (WRF) model provides meteorological input for CMAQ simulations over the CONUS at 12 x 12 km2. WRF simulations are performed for 2005-2018 and evaluated against the ground-based and satellite observations. The WRF model has been found to capture the seasonal, interannual, and regional variability of key meteorological parameters very well. CMAQ air quality simulations with assimilation of MODIS AOD and MOPITT CO have been completed for 2005-2007 and are being run for 2008-2018 currently. A preliminary evaluation of the first three years of CMAQ simulations shows good performance in capturing the seasonal cycle of ozone and fine particulate matter at state-level in the U.S. This presentation will discuss the design of the chemical data assimilation system, its impact on air quality simulations, and evaluation of the trends in surface ozone and fine particulate matter from the chemical reanalysis against the observations.
Sub-city-scale air quality forecasts combining models, satellites, and surface measures
Presenter: Carl Malings, NASA/USRA
Presentation Description: Although there exist many sources of information on air quality, no single source currently allows for high accuracy, high spatial resolution, city-scale coverage, high temporal frequency, and the capability of forecasting air quality in the near future. Existing data sources must instead be combined to meet these objectives. In this work, we start with the relatively coarse spatial resolution air quality predictions from NASA’s Goddard Earth Observing System - Composition Forecasting (GEOS-CF) model. These predictions are downscaled using information from the ESA TROPOspheric Monitoring Instrument (TROPOMI) for tropospheric column concentrations of NO2, to extract typical concentration patterns at finer spatial resolution. Finally, we make use of ground-based measures of air quality to (1) correct for biases in the model- and satellite-derived predictions of surface concentrations and (2) account for transient local pollution events. We validate the method by forecasting surface NO2 concentrations across several US cities. In particular, we evaluate the impact that varying densities of surface-level monitoring have on the method’s performance, with a view to its potential application in regions of the world which lack dense ground-based air quality monitoring. We also evaluate the capabilities of both linear and non-linear machine learning regression techniques in establishing the relationships between the various data sources needed in this forecasting method.
Assimilation of multiple satellite retrievals and emissions adjustment to improve high resolution air quality forecasting
Presenter: Arthur Mizzi, USRA/NASA
Presentation Description: We will present results from a high spatiotemporal resolution (3 km, 3 hr cycling) ensemble air quality (AQ) forecast/assimilation system based on WRF-Chem/DART.
WRF-Chem/DART integrates the Weather Research and Forecast (WRF) model with on-line chemistry (WRF-Chem) into the Data Assimilation Research Testbed (DART). It assimilates AirNow CO, O3, NO2, SO2, PM10, and PM2.5 measurements, MOPITT CO, IASI CO, O3, and MODIS AOD total/partial column and/or profile retrievals. We are extending WRF-Chem/DART to assimilate OMI O3, NO2, SO2, TROPOMI CO, O3, NO2, SO2, synthetic TEMPO O3, NO2, and MAIA AOD retrievals. WRF-Chem/DART uses: (i) the state augmentation method for adjusting emissions, and (ii) state-space localization.
We will discuss five experiments: (i) CNTL EX: assimilates only meteorology; (ii) CHEM EX: same as CNTL EX also assimilates MOPITT, IASI, MODIS, and AirNow; (iii) TEMPO/MAIA EX: same as CHEM EX also assimilates TEMPO and MAIA; (iv) CHEM-ADJ EX: same as CHEM EX with emissions adjustment; and (v) TEMPO/MAIA-ADJ EX: same as TEMPO/MAIA EX with emissions adjustment.
We expect that: (i) assimilating chemical observations will increase AQ forecast skill; (ii) including emissions adjustment will increase forecast skill/predictability time; and (iii) including assimilation of TEMPO and MAIA will increase forecast skill/predictability time.
Using In-situ Surface Measurements of Aerosol Optical Properties to Evaluate Model Simulations - the AeroCom INSITU Experiment
Presenter: Lauren Schmeisser, Cooperative Institute for Research in Environmental Sciences; CU Boulder/NOAA
Presentation Description: Aerosols absorb and scatter incoming sunlight and are thus an important part of the climate system. Despite their key role, aerosols are still the largest uncertainty in estimates of Earth’s energy budget. Global climate models provide a way to evaluate the effects of aerosols on climate; however, these model simulations make assumptions about aerosol characteristics and processes that need to be continuously evaluated against observational datasets in order to confidently interpret model output. Using long-term in-situ surface measurements provides unique advantages over comparisons using satellite and remote sensing products. In-situ measurements can be related to physical standards, can be made more reliably at low aerosol loadings, and can be made at night and during cloudy conditions. Here we use high-quality in-situ aerosol optical property measurements from over 350 surface monitoring stations worldwide to evaluate the suite of models from the AeroCom (Aerosol Comparisons between Observations and Models, http://aerocom.met.no/) INSITU experiment. We explore model/measurement comparisons of aerosol scattering coefficient, aerosol absorption coefficient, scattering and absorption Ȧngström coefficients, as well as single scattering albedo, and find large inter-model variability in representation of aerosol amount, seasonality, and characteristics. The biases outlined here provide a path forward to improve the predictive capability of global climate models.
Benchmark on methodologies to integrate low-cost sensor networks with official measurements and modelled data: first results
Presenter: Joost Wesseling, National Institute for Public Health and the Environment
Presentation Description: The European Forum for Air quality Modeling (FAIRMODE) aims to bring together air quality modelers and users in order to promote and support the harmonized use of models by EU Member States, with emphasis on model application under the European Air Quality Directives.
Low-cost air quality sensors are becoming very relevant for the air quality community, especially concerning methodologies to integrate results of sensor networks with modelled data and official measurements.
In FAIRMODE, a benchmark has been organized to discuss and understand the strengths and weaknesses of the different ways low-cost sensors can be used. Present experiences suggest important roles for data fusion and assimilation approaches, and possibly other techniques with similar scopes. The focus points for this benchmark are to 1) Exchange potential concepts and best practices about the integration of sensor network data in air quality mapping methods and 2) Explore how air quality modelling can contribute to the exploitation and validation of an air quality sensor network.
Since end of 2020, groups from some ten European countries are working on the benchmarking exercise using data from some 1500 stationary PM2.5 sensors. The first steps are data validity and calibration. After that, focus is on different approaches to combine the calibrated sensor data with official measurements and models. The first results of the benchmark will be presented and discussed.
Evaluating modeled smoke plume heights against airborne lidar observations for multiple fires during FIREX-AQ
Presenter: Laura Thapa, University of California, Los Angeles
Presentation Description: Wildfire smoke degrades air quality, health, visibility, and climate. Smoke plume injection height can determine the extent of these impacts, yet injection models remain ill-constrained. We evaluate the Freitas plume rise model in WRF-Chem and HRRR-Smoke for fires sampled with the DIAL-HSRL lidar and MASTER multispectral imager during the Boise portion of the Fire Influence on Regional to Global Environments and Air Quality (FIREX-AQ) field campaign. We sampled the models along the DC-8 flight route, compared modeled PM2.5 with observed backscatter, and classified smoke as non-injection (in the PBL) or injection (in the free troposphere). Our model sampling method handles fires being shifted in space and time relative to observations and doubles the WRF-Chem sample size. In both models, injection is modeled more often than observed, and this overprediction tends to occur during the daily peak of fire activity and for mid-range values of PBL height, fire radiative power (FRP), and smoldering area fraction, suggesting a problem with how energy released from the fires is modeled. Our key finding is that the heat fluxes (energy per unit area) used in the Freitas model and the FRP to fire size conversion factors used in HRRR-Smoke can be off by up to a factor of ten with respect to MASTER observations. We hypothesize that this leads to injection overprediction and suggest sensitivity studies be performed using heat flux and FRP to fire size conversion factors derived from MASTER.
Complex Terrain and Coastal Zone Meteorology
Improved Prediction of Cold-Air Pools in the Weather Research and Forecasting Model Using a Truly Horizontal Diffusion Scheme
Presenter: Robert Arthur, Lawrence Livermore National Laboratory
Presentation Description: Cold-air pools, caused by the buildup of cold air within topographic basins, are associated with urban pollution, low wind energy output, and risks to transportation and agricultural production. Furthermore, they are seen as a major forecasting challenge because their dynamics are generally not well represented in mesoscale atmospheric models. This is due, in part, to errors arising from the calculation of horizontal gradients over steep slopes, where skewed grid cells occur on commonly used terrain-following model grids. To limit these errors within the Weather Research and Forecasting (WRF) model, we have implemented a scheme that uses Taylor series approximations to vertically interpolate variables to the level necessary for a “truly horizontal” gradient calculation. Applied to the horizontal diffusion of potential temperature in WRF, the scheme is shown to improve cold-air pool prediction in both idealized and realistic tests cases. The idealized test case demonstrates a reduction in spurious numerical mixing within a quiescent cold pool. Then, a realistic case study in the Columbia River basin shows a reduction in positive wind speed bias by up to roughly 20% compared to observations from the Second Wind Forecast Improvement Project (WFIP2). The truly horizontal diffusion scheme presented here has the potential to improve complex terrain models over a range of scales, and to be extended to additional model variables such as velocity, moisture, and pressure.
Does a representative dynamics exist for the Grenoble valley during winter anticyclonic episodes?
Poster Presenter: Enzo Le Bouëdec, Université de Grenoble Alpes / LEGI
Lightning Talk Description: In the scope of a study on air quality within the Grenoble valley (French Alps), we investigate the winter anticyclonic episodes, for which the highest PM10 concentrations are observed. Numerous studies have shown how the formation of cold air pools in the valley generates a stably stratified atmosphere which allows for the accumulation of pollutants. In addition, the atmospheric dynamics in the valley is often assumed to be decoupled from the synoptic flow. This study has two objectives: (I) to address the validity of the latter assumption, (ii) to investigate how the PM10 concentration field close to the valley bottom depends upon the characteristics of the synoptic forcing. To this end, Large Eddy Simulations at high resolution (1/9 km horizontally) of four winter anticyclonic episodes are run using the WRF model. Implementation of the valley’s PM10 emission inventory enables for a realistic simulation of the PM10 concentration (assuming to behave as passive scalars). We find that the general dynamics observed in the valley are impacted by the synoptic winds but that the ciculation observed near the surface is dominated by thermally driven flows for all episodes. This leads to very similar PM10 distributions close to the valley bottom from one episode to another, implying that a part of the population is highly exposed to poor air quality during winter.
Impacts of Saharan mineral dust on air‐sea interaction and coastal cloud activities over the Eastern North Atlantic Ocean using a fully coupled regional model
Presenter: Shu-Hua Chen, University of California, Davis
Presentation Description: The modifications of air-sea coupling processes by dust-radiation-cloud interactions in summer 2015 over the Eastern North Atlantic Ocean are examined using a high-resolution coupled atmosphere-ocean-dust regional model. The dust-induced mechanisms that are responsible for changes of sea surface temperature (SST), latent and sensible heat fluxes (LHF/SHF), and cloud activities, in particular near the West African coastal region, are also examined. Two three-month numerical experiments are conducted, and they differ only in the activation and deactivation of dust-radiation-cloud interactions. Model results show that the dust reduces surface downward radiation fluxes over the ocean with the maximum change of 20-30 W m^-2. Over the dust plume region, the dust effect creates a low-pressure anomaly and a cyclonic circulation anomaly, which drives a positive wind stress curl anomaly, thereby reducing sea surface height and mixed layer depth. However, the SST change by dust, ranging from -0.5 to 0.5 K, has a great spatial variation different from the dust plume shape. Dust cools SST around the West African coast but unexpectedly warms SST over a large area of the western tropical North Atlantic. Unlike the SST change pattern, the LHF and SHF changes are mostly reduced underneath the dust plume region. These SST, LHF, and SHF changes are controlled by different mechanisms and will be discussed. In addition, how dust-induced mechanisms modify cloud activities and rainfall over the West African coastal region will be presented as well.
Measurement and modeling of pollutant dispersion in highly complex terrain: the Bolzano Tracer Experiment (BTEX)
Presenter: Lorenzo Giovannini, University of Trento, Department of Civil, Environmental and Mechanical Engineering
Presentation Description: The simulation of pollutant dispersion over complex terrain is still a challenging task, due to the inherent difficulties in accurately reproducing both atmospheric and dispersion processes. Moreover, only few suitable observational datasets are available to be used as benchmark for testing dispersion models over complex terrain.
This contribution presents results from the Bolzano Tracer Experiment (BTEX), which was performed to characterize the impact of pollutants emitted by a waste incinerator close to the city of Bolzano, located in a basin in the Italian Alps. The experiment included two controlled releases of a tracer from the stack of the incinerator, under different meteorological conditions. Meteorological variables were monitored by means of a high-resolution network of ground-based instruments, which allowed the evaluation of the atmospheric factors controlling dispersion processes in the whole basin, including a low-level valley-exit jet entering the basin from one of the tributary valleys. The dispersion of the tracer was simulated by means of a modeling chain composed of the WRF meteorological model, coupled with CALPUFF, a gaussian puff dispersion model, and SPRAY-WEB, a lagrangian particle dispersion model.
Data from BTEX represent one of the rare examples of datasets characterizing dispersion processes in a typical mountainous environment, and offers a remarkable benchmark for testing meteorological and pollutant dispersion models in complex terrain.
Drivers of severe air pollution events in complex terrain during wintertime stable atmospheric conditions.
Presenter: Julián Quimbayo-Duarte, Goethe University Frankfurt
Presentation Description: During the winter season, mountainous areas are commonly affected by episodes of severe air pollution. This occurs when atmospheric stability increases due to the formation of a temperature inversion that suppresses mixing in the lower atmosphere. The Arve river valley, an airshed located in the northern the French Alps, experiences particularly severe air pollution during wintertime stable atmospheric conditions associated with persistent cold-air pools observed in the area. The recordings of the air monitoring network indicate that the urbanized area of the central basin-shape section of the valley is generally the most polluted (PM10), with a harmful impact on the health of inhabitants. We examine the air pollution transport potential of the Arve river valley airshed using results from high-resolution numerical simulations of a cold-air pool observed during the Passy-2015 field campaign. Passive tracers were used to model PM10 with emissions provided by a detailed inventory developed by the local air-quality agency. The observed differential in PM10 levels between valley sections was well captured by the numerical model. The stagnation, recirculation and ventilation potential of the airshed was evaluated spatially and temporally using integral quantities. This study allows to identify the origin of the strong pollution episodes in the Arve river valley, through the link between the local topography, emission sources and pollutant transport.
Met-Chemistry Interactions: Aerosol Direct & Indirect Feedbacks and Aerosol-Cloud Interactions, Aerosol Chemistry, Radiative Impacts of Gases
Influence of Different Atmospheric Aerosol Compositions on the Life Cycle of Stratocumulus Clouds over Southern West Africa
Presenter: Lambert Delbeke, LAERO
Presentation Description: Low-Level Stratiform Clouds (LLSC) appear frequently over Southern West Africa (SWA). During the West African Monsoon (WAM) period, both local (air pollution) and remote (dust and biomass burning aerosols from North and Central Africa, respectively) aerosol sources can play a significant role in LLSC life cycle. The Dynamics-Aerosols-Chemistry-Cloud Interactions In West Africa (DACCIWA) campaign has produced a considerable number of clouds and aerosols measurements during the WAM in 2016. To better understand the influences of aerosols from different sources on the diurnal cycle of LLSC, we have conducted an in-depth analysis to correlate the evolution of LLSC with the appearance and abundance of particularly dust and biomass burning aerosols. Numerical simulations using a Large Eddy Simulations (LES) model with detailed aerosol and cloud microphysical processes and constrained by DACCIWA observations have also been conducted, driven by different atmospheric aerosol compositions in order to study the impacts of different aerosols on the key processes (formation, break-up, transition to cumulus) of LLSC. Detailed results alongside preliminary conclusion will be presented.
The Challenges of Modeling Wintertime Particulate Matter Within Basins
Presenter: Jerome Fast, Pacific Northwest National Laboratory
Presentation Description: High particulate matter concentrations are frequently observed within urban and agricultural basins throughout the world during the winter. These events arise due to emissions of aerosols and their precursors into stably-stratified air masses sheltered from lateral and vertical mixing by the surrounding terrain during persistent cold-air pools that can last from days to weeks. Air mass stagnation and chemical aging alters aerosol properties as fresh emissions accumulate, mix, and react with aged aerosols. Ammonium nitrate is often the largest fraction of inorganic material within basins of the western U.S., but predictions of ammonium nitrate are highly uncertain because of the range of chemical mechanisms responsible for the formation, growth, and fate of ammonium nitrate and the dependance of those mechanisms on meteorology that varies within basins during persistent cold-air pools. For example, complex dispersion patterns recirculate aerosols over regions with higher and lower emission rates. Models often do not adequately represent these circulations as well as the strength and duration of cold pool events, in part, because of the high resolution needed to represent local topography and the vertical gradients in meteorology and chemistry. This talk therefore describes the primary challenges associated with representing meteorological and chemical processes, including their interactions, when simulating wintertime particulate matter in basins.
Confronting the uncertainties in simulating elevated wintertime air pollution concentrations in mountainous regions
Presenter: Heather Holmes, University of Utah
Presentation Description: Many areas in the western U.S. are currently violating the federal air pollution standards. This is exacerbated by unique air pollution sources (e.g., windblown dust and wildfire smoke) and the local meteorological and orographical effects. The mountainous terrain and synoptic weather patterns lead to complex winds that impact air pollution transport, dispersion, and accumulation. This talk will focus on one distinct condition, during wintertime, with complex atmospheric physics that impacts air quality throughout the western U.S. It occurs when local stagnation events cause pollutants to become trapped in cold dense air on valley floors, often referred to as temperature inversions. Research advancements in quantifying the turbulent fluxes during stable atmospheric conditions and comparing with numerical models will be shown, using observations from the wintertime Persistent Cold-Air Pool Study (PCAPS) in the Salt Lake Valley, Utah. Uncertainties associated with the simulated meteorology under stable conditions over complex terrain hinder realistic air quality simulations. Chemical transport modeling results during PCAPS will be shown to illustrate the meteorology-chemistry interactions and their uncertainties. We use a thermodynamic approach to investigate limiting reagents for the wintertime ammonium nitrate formation and relate the findings to the underestimated particulate matter concentrations from the PCAPS air quality simulations.
Examining the Impacts of an Interactive Fire Plume-Rise Model in E3SM on Aerosol Indirect Effects
Presenter: Zheng Lu, Texas A&M University
Presentation Description: In this study, we replace the fixed vertical profiles of monthly biomass burning (BB) aerosol emissions in the DOE Energy Exascale Earth System Model (E3SM) with an interactively fire plume-rise model that was developed by Freitas et al. (2007). The vertical distribution of BB aerosol emissions for each grid is calculated as a function of ambient thermodynamic conditions obtained from the host E3SM, and distributions of fire sizes and heat fluxes from observations. In the model, we use the maximum-fire radiative power (Max-FRP) technique to determine the active fire size and scaled MODIS FRP observations for the heat release. The performance of the fire plume-rise model is evaluated by comparing against campaign observations and satellite retrievals.
To examine to what extent the incorporation of the plume-rise model affects aerosol indirect effects, we conduct two sets of experiments. In the first one we still use fixed vertical profiles of BB aerosol emissions, while using plume-rise model in the second one. Both experiments are conducted for 20 years. By contrasting the clean-sky cloud radiative forcing, we quantify the role of interactive plume-rise mode on simulating aerosol indirect effect and examine the regional signatures.
New particle formation over the oceans: Results from recent field campaigns
Presenter: Jian Wang, Washington University
Presentation Description: With their extensive spatiotemporal coverage, marine clouds exert substantial net cooling effect on global climates. Because of their relatively low optical thickness and background aerosol concentration, marine clouds are particularly susceptible to the perturbation of cloud condensation nuclei concentration. New particle formation, namely the formation of particles from gas-phase molecules, strongly influences the concentration of aerosol particles, and therefore their impact on clouds and climate. There have been numerous observations of new particle formation in the free troposphere over ocean, and models indicate that the subsidence of the particles formed in the free troposphere can contribute to a large fraction of the cloud condensation nuclei in the marine boundary layer. In contrast, it has long been thought that new particle formation rarely occurs in the marine boundary layer over the open oceans. In this presentation, we investigate the new particle formations over mid-latitude and tropical oceans using recent airborne measurements and long-term surface observations. The conditions leading to the new particle formation are quantified, and the vertical distribution, frequency, and seasonal variation of the new particle formation are examined. We will discuss the potential mechanisms of the new particle formation over the oceans and the influence of the new particle formation on the population of cloud condensation nuclei in the marine boundary layer.
Modeling reactive ammonia uptake by secondary organic aerosol in a changing climate: a WRF-CMAQ evaluation
Presenter: Shupeng Zhu, University of California, Irvine
Presentation Description: In addition to the well-constrained inorganic acid-base chemistry of ammonia in PM2.5 formation, ammonia can also react with certain organic compounds in secondary organic aerosol (SOA) to produce nitrogen-containing organic compounds. In this study, the potential meteorology and air quality impacts of the heterogeneous uptake of NH3 by SOA are investigated with the WRF-CMAQ two-way coupled model, which calculates the two-way radioactive forcing feedback caused by aerosol between meteorology and chemistry in a single simulation. Simulations over the continental US are performed for July 2014 and July 2050 under the RCP 8.5 IPCC scenario to study the potential combined impact due to climate change. In total 6 simulation scenarios are generated, with four 2014 scenarios and two 2050 scenarios. First, the results show that the two-way coupled process has much more impact on air quality than the SOA-NH3 uptake process. Without the two-way feedback, the SOA-NH3 uptake process along causes a 2.4% averaged increase on PM2.5 concentration, while the two-way feedback alone is causing a 12.3% averaged decrease, with a peak change almost 5 times higher. Secondly, the inclusion of the two-way feedback results in some significant impacts on meteorology conditions, which is much large compares to those caused by the SOA-NH3 uptake through the two-way feedback. Finally, the impact caused by adding the SOA-NH3 uptake is found to be smaller in 2050 when compared to the results of 2014.
Can we reduce the persistent uncertainty in aerosol effects on climate?
Presenter: Ken Carslaw, University of Leeds
Presentation Description: Aerosol radiative forcing has persisted through all IPCC assessments as one of the most uncertain factors in climate change. Why is this the case despite enormous investment in more-sophisticated models and observations as well as substantial increases in our knowledge of aerosol processes? In this presentation I will describe how perturbed parameter ensembles of an aerosol-climate model provide some insight into the causes of uncertainty in aerosol radiative forcing and the challenges involved in reducing the uncertainty. The uncertainty range of our single model is comparable to the multi-model spread. Observations help to constrain the forcing uncertainty, but by much less than we expected. The two main limitations are model 'equifinality' (compensating effects of uncertainties) and the representation error associated with the sparse in situ observations. Our results suggest there may be a natural limit to how much we can reduce forcing uncertainty, although it's not yet clear what that limit is.
Quantifying the role of aerosol to precipitation variability in East Asia through aerosols to cloud interactions
Poster Presenter: Paul Adigun, University of Tsukuba
Lightning Talk Description: Using a high resolution regional model weather research and forecasting (WRF-Chem), the impact of the East Asian aerosol load on precipitation variability was assessed through sensitivity experiment Based on a monthly climatology, model simulations compare satisfactory with wind field from reanalysis data, cloud observations, daily average particulate matter (PM), temperature, relative humidity and satellite retrieved CO mixing, Long term impact of aerosols on precipitation are identified over East Asia through the analysis of 10 years measurement of precipitation 2005-2015, We Investigate the comparism of regional climate model simulation over East Asia precipitation and perform dynamical and statistical downscale for the East Asia region, to access climate change impact using multi-model dataset from the ensembles prediction of climate change of east Asian monsoons using multi-method, WRF-Chem model with a moment bulk microphysical scheme is employed to simulate monsoon rainfall in these area and elucidate the effect of aerosols on cloud process through aerosols to cloud interactions, The study delineate the uncertainty and complexity of aerosols effect on east Asian monsoon systems different scales and furthermore stress the significance of aerosol forcing for future climate projection
Aerosol-Cloud-Longwave Radiation Interactions in Stratocumulus Clouds
Presenter: Adele Igel, University of California, Davis
Presentation Description: Aerosol particles are well known to lead to smaller, more reflective cloud droplets in stratocumulus clouds assuming there is no change to the liquid water path. However, their impacts on longwave radiation have been less studied, in part because many clouds are well-approximated as blackbodies. We use large eddy simulations and observations of stratocumulus clouds to better understand the response of longwave radiation to aerosol concentration. While indeed the longwave emission from clouds is independent of liquid water path (LWP) and aerosol concentration for LWP greater than about 100 g/m2, LWP in stratocumulus clouds is often less than this. As a result, aerosol concentration can have a large influence on the surface energy budget. Moreover, for LWP < 100 g/m2, the radiative cooling rate at cloud top is impacted by aerosol concentration and influences the dynamics of the cloud. Finally, contrary to many previous studies we do not find a robust aerosol dispersion effect in stratocumulus clouds.
Evaluation of Aerosol effects in GEOS5-S2S-V2 Seasonal-Subseasonal hindcast during the Dust season and Biomass burning seasons
Poster Presenter: Zhao Li, NASA-GSFC GMAO
Lightning Talk Description: GEOS-S2S-V2 seasonal and subseasonal hindcasts from 2003 to 2019 are analyzed for the month of May and September using three types forecast runs. The total effect run include both direct and indirect aerosol effects. The direct effect run eliminated the feedbacks from the aerosol cloud interactions. The observed aerosol run uses monthly varying aerosol concentration from observations. We found both direct and indirect aerosol effects vary significantly for different regions and seasons. The Total effects of aerosol on near surface temperature are enhanced over high latitudes in May due to aerosol-cloud interactions. Removing aerosol-cloud interaction will result in increased low level cloud a much colder Northern Hemisphere and the opposite in the Southern Hemisphere. The indirect effects also reduce precipitation in ITCZ and increased aerosol optical depth in the Northern Hemisphere.
Exploring the association of meteorology and atmospheric composition to snow cover changes: A case study over High-Mountain Asia and Central Andes
Poster Presenter: Chayan Roychoudhury, The University of Arizona
Lightning Talk Description: Light-absorbing particles (LAPs) have been known to be instrumental in accelerating snow-melting rate and reducing snow albedo (i.e., snow darkening). Changes in snow cover over High-mountain Asia (HMA) and central Andes (CA) are known to be associated with deposition of LAPs (both black carbon and dust) during spring in both hemispheres. A correlation study is performed between snow cover and albedo (from Moderate-Resolution Imaging Spectroradiometer (MODIS) aboard Terra and Aqua satellites) and surface parameters, both at a spatial resolution of 0.1o like air temperature, downward surface solar radiation, relative humidity, and mean sea level pressure (from ERA5 Reanalysis), precipitation (from Integrated Multi-satellite Retrievals or Global Precipitation Measurement (IMERG) dataset) along with black carbon and dust aerosol loading (from CAMS global reanalysis EAC4) during 2003 - 2018, to understand the meteorological and snow pollution relationships in the two distinct glacier regions. The aerosol dataset from CAMS Reanalysis is used to examine the possible impact of LAPs on snow-melting. Furthermore, multivariate regression is performed to elucidate the relative contribution of the meteorological and pollution parameters on snow cover and their variation across the summer and winter hemispheres. Principal Component Analysis (PCA) is utilized to explore the dominant patterns in the snow cover and the independent variables of interest.
Just Scratching the Surface : Improving the Representation of Aerosol Size Dependent Properties in a Thermodynamic Mixing Model
Presenter: Ryan Schmedding, McGill University
Presentation Description: Atmospheric aerosols are ubiquitous across the planet and are among the largest sources of uncertainty in climate change. Many aerosols in North America form from atmospheric chemistry of semi-volatile organic compounds, which interact with inorganic salts, acids, and water. Recently, there has been interest in the effects of surfactants cloud condensation nuclei (CCN) activation.
The Aerosol Inorganic–Organic Mixtures Functional groups Activity Coefficients (AIOMFAC) model is a thermodynamic mixing model for aqueous solutions of ions and organics. Our model predicts the gas-particle partitioning of atmospheric species. The standard AIOMFAC model also predicts liquid–liquid phase separation (LLPS) within an aerosol using a size independent approach. LLPS aerosols exhibit a radially symmetric morphology with a spherical, inorganic core surrounded by an organic shell or a partially engulfed structure, with an incomplete organic lens around the inorganic core.
We extended AIOMFAC to account for size-dependent interfaces and morphology effects on equilibrium partitioning and cloud droplet formation. I will discuss the treatment of curved interfaces between distinct phases and the partitioning of species to said interfaces. Aerosols modeled with the updated method exhibit size-dependent morphologies relevant in the sub-100nm diameter range. I will further discuss improved predictions of critical CCN properties.
Composition and Operational Forecasting from Daily to Seasonal Scales
Overview of operational air quality forecasting systems in France and Europe
Presenter: Joaquim Arteta, Météo-France / CNRM
Presentation Description: Over the last decades, various operational forecasting systems have been created at national and European levels.
In France, the Prév'Air forecasting system is based on the forecasts of 2 french transport-chemistry models, including the CTM MOCAGE, developed and operated at Météo-France. Forecasts of the Prév'air system are used to trigger actions in case of pollution peaks.
Since the early 2000s, various national initiatives including Prév’Air have converged to create a European forecasting system, funded by the European Union. This work led in 2014 to the operationalization of the regional CAMS system, initially gathering 7 models from 6 different European countries within a median ensemble. The system produces daily forecasts for the main regulated air pollutants as well as the analysis of the previous day and a reanalysis of the past year. This system is constantly evolving. Thus, 2 new models have been integrated in 2018, and new species are regularly added, such as new types of pollens. Finally, a tool (Evaltools software) was developed and operationalized to provide daily statistical evaluations of the quality of the forecasts and analyses with respect to NRT surface concentration measured by the European air quality monitoring stations and gathered by EEA (European Environment Agency).
Forecasts and analyses produced by the regional CAMS system are accessible via a web site and the data are freely available for research and commercial use.
An Improved National Air Quality Forecasting Capability Using the NOAA Global Forecast System. Part I: Model Development and Community Application
Presenter: Patrick Campbell, George Mason University/NOAA-Air Resources Affiliate
Presentation Description: Here we describe the forward-thinking vision and development of the NOAA Finite Volume Cubed Sphere (FV3)-based Global Forecast System Version 16 (GFSv16) coupling with the “state-of-the-science” Community Multiscale Air Quality CMAQ model version 5.3.1. The GFSv16-CMAQv5.3.1 coupling is based on the seminal version of the NOAA-EPA Atmosphere-Chemistry Coupler (NACC), which will ultimately form the next National Air Quality Forecasting Capability (NAQFC) system (dubbed “NACC-CMAQ”) and includes numerous scientific advancements that will be highlighted. Such advancements include an optimized model framework extending out to a 72-hr forecast, updated anthropogenic and biogenic emissions, bi-directional ammonia fluxes, dynamic aerosol boundary conditions, satellite-based data acquisition technology to improve land cover and soil characteristics, and inline wildfire smoke and windblown dust predictions that are vital to predictions of gas and aerosol concentrations during hazardous events. The advanced NAQFC is slated to become operational at NOAA/National Weather Service in July 2021, and will include improved web applications for forecasters. We also will demonstrate the capability of the NACC-CMAQ system as an additional community research tool for atmosphere-chemistry coupling, as well as other ways that NOAA-ARL can facilitate NACC-CMAQ community usage for air quality modeling applications.
Updating and Evaluating Emissions in NOAA’s Global Ensemble Forecast System (GEFS)-Aerosols
Presenter: Gill-Ran Jeong, George Mason University - NOAA Air Resources Laboratory
Presentation Description: Anthropogenic emissions of gases and particles have vital impacts on the predictions of atmospheric constituents in global atmospheric aerosol models, and consequently impact climate, air quality, and health. Recently, NOAA’s Global Ensemble Forecast System (GEFS) – Aerosols is undergoing significant upgrades to improve the predictions of global aerosols and further development under the Unified Forecast System (UFS) for expansion to seasonal-to-sub-seasonal (S2S) weather forecasts with aerosol feedbacks. GEFS-Aerosol is a global aerosol model currently based on NASA’s GOCART model and is driven by meteorological predictions from NOAA’s Global Forecast System (GFS) that uses the Finite Volume Cubed-Sphere (FV3) dynamical core. This study focuses on experimental GEFS-Aerosol simulations using updated global anthropogenic emission inventories that are produced by the NOAA Emission and eXchange Unified System (NEXUS). Here we show comparisons of updated emissions datasets and analysis of GEFS-Aerosol sensitivity simulations (SENS) compared to the BASE that uses an obsolete emissions pre-processor. The GEFS-Aerosol BASE and SENS simulations of aerosol optical depth and absorbing aerosol optical depth are compared with global observations such as AERONET, MISR, MODIS, and VIIRS for September-November (SON) 2019. The fundamental reasons for their consistencies and differences between model and observations will be discussed based on a sectoral analysis of the emissions data.
Near real-time air quality forecasts using the NASA GEOS model
Presenter: Emma Knowland, USRA/GESTAR, NASA GMAO
Presentation Description: NASA's Global Modeling and Assimilation Office (GMAO) produces high-resolution global forecasts for weather, aerosols, and air quality. The NASA Global Earth Observing System (GEOS) model has been expanded to provide global near-real-time 5-day forecasts of atmospheric composition at unprecedented horizontal resolution of 0.25 degrees (~25 km). This composition forecast system (GEOS-CF) combines the operational GEOS weather forecasting model with the state-of-the-science GEOS-Chem chemistry module (version 12) to provide detailed analysis of a wide range of air pollutants such as ozone, carbon monoxide, nitrogen oxides, and fine particulate matter (PM2.5). Satellite observations are assimilated into the system for improved representation of weather and smoke. The assimilation system is being expanded to include chemically reactive trace gases. We discuss current capabilities of the GEOS Constituent Data Assimilation System (CoDAS) to improve atmospheric composition modeling and possible future directions, notably incorporating new observations (TROPOMI, geostationary satellites) and machine learning techniques. We show how machine learning techniques can be used to correct for sub-grid-scale variability, which further improves model estimates at a given observation site.
Progress and Plans for Advances in Air Quality and Aerosol Modeling in UFS
Presenter: Jeff McQueen, NOAA/NWS/NCEP/EMC
Presentation Description: NOAA is developing the Unified Forecast System (UFS) (https://ufscommunity.org/) as a source system for operational numerical weather prediction applications. Integration of global aerosol predictions based on the NASA Goddard Chemistry Aerosol Radiation and Transport (GOCART) scheme into the UFS has been successful, within a single member of the Global Ensemble Forecast System V12 (GEFS-Aerosols), which was implementation during Fall 2020. The global atmospheric composition dust component in GOCART has been improved by the incorporation of the NOAA/ARL Fengsha dust model. GEFS-Aerosols predictions demonstrate a substantial improvement in both composition and variability of aerosol distributions, however, emissions from large fires were often overestimated. Assimilation of aerosol optical depth (AOD) observations is under development to constrain aerosol distribution. NASA and NOAA are currently working to unify global aerosol model processes using a consistent strategy for aerosol and weather interactions.
Regionally, Integration of air quality predictions into the UFS includes testing of the Community Multiscale Air Quality modeling system (CMAQ v5.3.1) predictions driven by the operational version of the FV3 Global Forecast System v16 (GFS) with raw and bias corrected forecasts of ozone and PM2.5 extend to 72 hrs twice per day. GEFS-Aerosols provides boundary conditions while anthropogenic emissions are updated to NEI2016.
Examination of Chemistry-Aerosol Combinations in the US Next Generation of the National Air Quality Forecast Capability
Presenter: Yang Zhang, Northeastern University
Presentation Description: Atmospheric composition forecast is highly sensitive to gas-phase chemistry and aerosol modules used in operational air quality forecasting models. In this work, several combinations of gas-phase mechanisms and aerosol modules available in the US NOAA’s next generation of National Air Quality Forecast Capability (NAQFC) are examined to (1) identify the best combination with optimal forecasting skill, and (2) further improve the skill of NAQFC in forecasting fine particulate matter and its composition, in particular, organic aerosol. The NAQFC is based on the offline-coupled Finite Volume Cube-Sphere Global Forecast System (FV3GFS) and the Community Multiscale Air Quality (CMAQ) modeling system version 5.3.1 (referred to as FV3GFS-CMAQ hereafter). FV3GFS-CMAQ simulations with the same aerosol module but different gas-phases chemistry (e.g., CB6r3-AERO6 vs. RACM2-AERO6 vs. SAPRC07tic-AERO6) and with the same gas-phase chemistry but different aerosol modules (e.g., SAPRC07tic-AERO6 vs. SAPRC07tic-AERO7) are conducted over contiguous US for several months in 2019. These results are intercompared and evaluated against AIRNow dataset. The discrepancies among results with various chemistry-aerosol combinations and between simulations and observations as well as underlying causes are analyzed. Recommendations will be made on the optimal chemistry-aerosol combination and areas of further improvement for the NAQFC.
Experimental air quality forecasting with the Rapid-Refresh model coupled to chemistry (RAP-Chem)
Poster Presenter: Jordan Schnell, CIRES/NOAA GSL
Lightning Presentation Description: Experimental air quality forecasts with the Rapid-Refresh model coupled to chemistry (RAP- Chem) at NOAA ESRL began in July 2020 in an effort to capture changing atmospheric composition due to the emissions reductions associated with the COVID-19 lockdowns. The full gas-phase and aerosol chemistry mechanism used in the RAP-Chem and proposed for transition into the Unified Forecast System (UFS) offers a potential lower computational cost alternative compared to mechanisms used in similarly capable operational models. Additionally, the RAP-Chem includes wildfire emissions of gases and aerosols, natural emissions of biogenic gases, dust, and sea salt, and simulates aerosol feedback to atmospheric physics allowing evaluation of the impact of changes in atmospheric composition on atmospheric conditions. Here we will show results over the August-September 2020 wildfire season and introduce key model features and developments in the RAP-chem slated for potential implementation into the UFS; specifically, we will demonstrate the suitability of a reduced complexity gas-phase chemical mechanism, the improvements associated with the use of the MYNN PBL scheme to perform non-local mixing of chemical species, as well as the use of the full TUV photolysis model with aerosol feedback combined with assimilated ozone from the Global Forecast System to more accurately capture the impacts of variable total column ozone on surface photochemistry.
An Improved National Air Quality Forecasting Capability Using the NOAA Global Forecast System. Part II: Science Advancements and Evaluations
Presenter: Youhua Tang, NOAA Air Resources Laboratory
Presentation Description: The existing National Air Quality Forecasting Capability (NAQFC) is using CMAQ 5.0.2 with cb05 chemical mechanism driven by the North American Mesoscale Forecast System (NAM). As one of our efforts to upgrade this system, the Part I described the overview of the next-generation upgrade that uses the CMAQv5.3.1 model, driven by the NOAA operational forecast based on the Finite Volume Cubed-Sphere (FV3)-Global Forecast System (GFS), version 16. The meteorological preprocessor for CMAQ was also overhauled based on the seminal version of the NOAA-ARL Atmosphere-Chemistry Coupler (NACC). Differing from the normal WRF-ARW/CMAQ system, the interpolation-based coupler can use various meteorological output to drive CMAQ even they are in different grids. Here in Part II, we evaluate NACC-CMAQ against observations over the contiguous United States (CONUS) in summer 2019, with focus on the comparison between WRF/CMAQ and NACC-CMAQ. During this period, the FV3GFS tends to have stronger PBL diurnal variation (higher during daytime, and lower at night) than those of the WRF. The FV3GFS also predicted higher daytime surface wind. In summer 2019, The NACC-CMAQ prediction showed better surface O3 and PM2.5 predictions than that driven by the corresponding WRF meteorology. This result shows that using the global-model meteorology to directly drive the regional air quality model through interpolation is feasible and can yield reasonable result compared to the traditional WRF downscaling.
Machine Learning Applications for Air Quality Research
Explainable AI for the Geosciences
Presenter: Elizabeth (Libby) Barnes, Department of Atmospheric Science, Colorado State University (invited presentation)
Presentation Description:Recent advances in machine learning have yielded many breakthroughs in commercial applications, and these techniques hold enormous promise for scientific discovery. While exciting advances with these tools have already been seen in other scientific disciplines, e.g. life sciences, they have been more slowly embraced by the geoscience community. One possible explanation for this is the perceived “black box” that outputs an answer without any explanation as to “why?” or “how?”.
In this talk, I will discuss how the field can make the most of machine learning interpretation techniques (i.e. “explainable AI”) to open the black box and push the bounds of scientific discovery. This has profound implications for machine learning use in science, as it not only increases trust in the output, but also allows us to learn new science from the decision making process of the algorithm itself. I will discuss applications in climate science, including subseasonal-to-decadal prediction, the atmospheric response to climate change, and anthropogenic impacts on Earth’s surface. While these examples are focused on climate problems, the tools and the approach are widely applicable and offer an exciting path for the future of geoscientific research.
Fusing Observed, Modeled, and Satellite-Derived Concentrations to Produce Fine-Resolution Estimates of Ground-Level PM2.5 During the 2017 California Wildfires
Presenter: Stephanie Elizabeth Cleland, Department of Environmental Sciences & Engineering, Gillings School of Global Public Health, University of North Carolina at Chapel Hill (invited presentation)
Presentation Description:Exposure to wildfire smoke poses a significant health risk, suggesting the importance of accurately estimating smoke concentrations. While chemical transport models (CTMs) and the spatial interpolation of observations are often used to assess smoke exposure, geostatistical methods can combine surface observations with modeled and/or satellite-derived concentrations to produce more accurate estimates of exposure. Here we estimate ground-level PM2.5 at a 1-km resolution during the 2017 California wildfires, using the Constant Air Quality Model Performance (CAMP) and Bayesian Maximum Entropy (BME) methods to bias-correct and fuse three concentration datasets: permanent and temporary monitoring stations, a CTM, and satellite observations. Four different BME space/time kriging and data fusion methods were evaluated for accuracy. All BME methods produce more accurate estimates than the standalone CTM and satellite products. Performing a non-linear bias-correction on the modeled concentrations, via the CAMP method, improves accuracy, with a 9% increase in R2. Adding temporary station data to the BME estimation increases the R2 by 36%. The data fusion of observations with the CAMP-corrected CTM and satellite-derived concentrations provides the best estimate (R2 = 0.713) in fire-affected regions, emphasizing the importance of combining multiple datasets. We estimate that approximately 65,000 people were exposed to very unhealthy air (daily average PM2.5 ≥150.5 µg/m3) during the fires.
Forecasting PM10 Using Deep Learning Methods
Presenter: Hande Öcba, ITU Eurasia Institute of Earth Sciences
Presentation Description: Computational science is important for providing reliable weather and climate forecasts.Machine learning(ML) techniques have been widely used in climate services and numerical prediction, since the non-linear, complex behavior can be learned from the climate data using ML concepts. ML is a field of artificial intelligence that gives computer algorithms the ability to improve automatically through learning from data without being explicitly programmed. In this perspective, we will present forecast models built using deep learning techniques; specifically CNNs, RNNs and LSTMs. Aim of this project is to forecast hourly PM10 concentration. Previous measurements are fed to the model as input and prediction for the following hour’s PM10 concentration is provided as an output. In this project, hourly data for 10 years from a station in Turkey is used. Minimum, maximum and average values for this dataset are respectively 0 mcg/m3, 2252.18 mcg/m3, and 58.56 mcg/m3. At first, this presentation will provide an overview of the role of deep learning in time-series forecasting. Then, the experimented deep learning methods and comparison of their prediction performance will be presented. The results of these experiments look promising with a relatively small error value. Finally, possible forecast ML solutions will be discussed with the help of measurement data obtained from nearby stations or data of other climate components which are in relation with PM10 concentration.
Reduced order modeling of organic aerosol tracers in LOTOS-EUROS using machine learning
Presenter: Obin Sturm, TU Delft, TNO
Presentation Description: The LOTOS-EUROS chemical transport model represents organic aerosol (OA) using a volatility basis set (VBS) approach. However, inclusion of the VBS slows down the overall model. A benchmarking analysis finds the slowdown to be exacerbated when running in parallel over 24 processors, though the VBS doesn’t require parallel communication operations. A sequential run with the VBS is slowed down by nearly 60%, and the parallel runtime with VBS is doubled, though time spent performing the VBS calculations is only 0.3% to 0.4% of the overall wall time. Further diagnosis shows that other operators are slowed down significantly, some of which do require parallel overhead, including advection and deposition. These operators must handle the 58 tracers contributed to the organic aerosols and vapors involved in the VBS, on top of the 64 tracers already included in LOTOS-EUROS. The slowdown caused by additional tracers motivates reduced order modeling. We use machine learning methods to compress the OA tracers to a latent space for other operators, while allowing for decompression into the VBS space for granular OA modeling. As operators will be resolved using the latent space, it must retain physically interpretable properties such as non-negativity, for use in gradient calculations in advection. Online non-negative matrix factorization and an autoencoder with layer constraints are explored. Though the OA tracers are the case study, this approach has potential to be generalized.
Deep learning to evaluate US NOx emissions using surface ozone predictions
Presenter: Tai-Long He, Atmospheric Physics and Composition Modelling Group, University of Toronto
Presentation Description: We use a hybrid deep learning model to predict June-July-August (JJA) daily maximum 8-h averaged (MDA8) surface ozone concentrations in the United States. A set of meteorological fields from the ERA-Interim reanalysis as well as monthly emissions of nitrogen oxides (NOx) from the Community Emissions Data System (CEDS) inventory are selected as predictors. Ozone measurements from the US Environmental Protection Agency (EPA) Air Quality System (AQS) from 1980 to 2009 are used to train the model, whereas data from 2010 to 2014 are used to evaluate the performance of the model. The model captures well daily and interannual variability in MDA8 ozone across the United States. We use the model to assess the consistency between the reported NOx trends and observations of surface ozone. Satellite observations of NO2 suggest that there was a slowdown in the rate of decline in emissions of NOx in the United States after 2010. It was also suggested that the satellite-inferred slowdown reflects the influence of background NOx. We find that the trend from satellite-derived emission estimates best reproduces ozone in low NOx emission (background) regions. In contrast, the trend from the US Environmental Protection Agency (EPA) inventory results in an underestimate of ozone. Our results confirm that the satellite-derived trends reflect anthropogenic and background influences and that the EPA inventory is overestimating recent reductions in NOx emissions.
Application of geostationary satellite and high-resolution meteorology data in estimating hourly PM2.5 levels in California
Presenter: Yang Liu, Rollins School of Public Health, Emory University (invited presentation)
Presentation Description: Wildland fire smoke contains large amounts of PM2.5 that can traverse tens to hundreds of kilometers, resulting in significant deterioration of air quality and excess mortality and morbidity in downwind regions. Estimating PM2.5 levels while considering the impact of wildfire smoke has been challenging due to the lack of ground monitoring coverage near the smoke plumes. We aim to estimate total PM2.5 concentration during the Camp Fire episode, the deadliest wildland fire in California history. Our random forest (RF) model combines calibrated low-cost sensor data with regulatory monitor measurements to bolster ground observations, GOES-16’s high temporal resolution to achieve hourly predictions, and oversampling techniques to reduce model underestimation at high PM2.5 levels. In addition, meteorological fields at 3 km resolution from the High-Resolution Rapid Refresh model and land use variables were also included in the model. Our AQS-only model achieved an out of bag (OOB) R2 (RMSE) of 0.70 (13.93 μg/m3) and cross-validation (CV) R2 (RMSE) of 0.71 (17.40 μg/m3), our combined AQS and PurpleAir weighted model achieved OOB R2 (RMSE) of 0.87 (9.20 μg/m3) and CV R2 (RMSE) of 0.76 (12.89 μg/m3), and our RF+SMOTE model achieved OOB R2 (RMSE) of 0.92 (10.37 μg/m3) and CV R2 (RMSE) of 0.81 (17.90 μg/m3). Hourly predictions from our model may aid in epidemiological investigations of intense and acute exposure to wildland fire PM2.5.
Combining mobile air quality sensor data and machine learning for more fine-grained air quality assessments in urban areas
Presenter: Valerio Panzica La Manna, IMEC
Presentation Description: As conventional air quality monitoring networks are sparse, and recent advances in sensor and IoT technologies have revolutionized air quality monitoring applications for more-fine grained air quality mapping, more accurate personal exposure assessments, ... This presentation provides an overview of different mobile sensor testbeds deployed in Antwerp (BE), Utrecht (NL) and Oakland (US), used to train two different machine learning models; i.e. a Variational Graph Auto Encoder (AVGAE) and Geographical Random Forest (GRF) model with the aim of inferring the mobile data in both space and time. Moreover, we validated the prediction performance of the considered models at different regulatory station locations following the EU FAIRMODE protocol. Combining real-time air quality sensor data with data-driven modelling for fine-grained mapping of air quality in heterogeneous urban environments. The data-driven models show to perform while needing much lower resources, computational power, infrastructure and processing Time, when compared to the state-of-the-art physical models. Moreover, all Considered context information in this study is openly available and, therefore, scalable to any city worldwide.
Openly accessible low-cost measurements for PM2.5 exposure modeling: guidance for monitor deployment with a similarity metric
Presenter: Jianzhao Bi, University of Washington
Presentation Description: High-resolution exposure modeling is critical for assessing the health effects of PM2.5, while sparse ground PM2.5 measurements may result in potential issues. This study aimed to use measurements from a low-cost PM network, PurpleAir, with an external validation dataset at residential locations of participants from the Adult Changes in Thought - Air Pollution study to improve the accuracy of exposure prediction at cohort locations and propose a metric assessing the similarity between the monitor and cohort locations to guide future monitor deployment. We used a spatiotemporal modeling framework to incorporate PM2.5 measurements from 51 “gold-standard” monitors and 58 PurpleAir monitors in Puget Sound, Washington into high-resolution exposure assessment at the two-week level from June 2017 to March 2019. We proposed a similarity metric based on principal component analysis - PCA distance - to assess PurpleAir monitors’ representativeness of the cohort locations. After including calibrated PurpleAir measurements as part of the dependent variable, the spatiotemporal validation R2 and root-mean-square error, RMSE, improved from 0.84 and 2.22 μg/m3 to 0.92 and 1.63 μg/m3. The spatial validation R2 and RMSE improved from 0.72 and 1.01 μg/m3 to 0.79 and 0.88 μg/m3. We found the PurpleAir monitors with shorter PCA distances could improve the model’s prediction accuracy more substantially than monitors with longer PCA distances, indicating the reasonability of this similarity metric.