1. Introduction
Interactions of atmospheric aerosols with clouds and radiation are the largest source of uncertainty in modeling efforts to quantify current climate and predict climate change (IPCC 2021). Their influence depends in part on the concentration, composition, sizes, optical properties, and vertical distribution of the aerosol, which are influenced by emission, deposition, chemical reactions, and transport. Tropospheric aerosols arise from natural sources, such as wind erosion dust, biomass burning aerosols, sea spray, and volcanoes; and from anthropogenic activities, such as combustion of fuels, power plants, and residencies. Anthropogenic emissions leading to atmospheric aerosol have increased dramatically over the past century and have been implicated in human health effects (Kelly and Fussell 2015), in visibility reduction in urban areas, in acid deposition, and perturbing Earth’s radiation balance (Seinfeld and Pandis 1998). Previous studies have unveiled the importance of aerosol research and demonstrated the direct and indirect contributions of aerosols on climate change from regional to global scale (Twomey 1974; Chylek and Wong 1995; Menon et al. 2002; Huang et al. 2014). Despite many aerosol studies, estimates of aerosol radiative forcing are still one of the largest uncertainties in model simulations for global climate projection (Boucher et al. 2013; Bellouin et al. 2020). There is large temporal and spatial variability in global aerosol composition and aerosol sources, which is one of the key factors resulting in such large uncertainty in model radiative forcing and its climatic effect (Myhre et al. 2013).
In parallel to aerosol simulation in climate models, the weather effects of aerosols have been relatively less considered in weather prediction models because weather forecasting prioritizes issues like cloud and moisture for fast and accurate predictions, whereas simulating prognostic aerosols requires considerable computational resources and challenging data assimilation of aerosols fields. Improvements in numerical weather prediction (NWP) were reported by Tompkins et al. (2005) who found that an updated dust climatology led to a northward shift of the southern African easterly jets (AEJ-S) in the European Centre for Medium-Range Weather Forecasts (ECMWF) NWP model. Results from NASA’s Goddard Earth Observing System (GEOS-5) forecasting system showed that the net impact of the interactive aerosol associated with a strong Saharan dust outbreak resulted in a temperature enhancement at the lofted dust level and a reduction near the surface levels, which improved forecasts of the AEJ (Reale et al. 2011). Grell et al. (2011) showed that coupling aerosols to radiation and microphysics schemes in high-resolution weather forecasting models may improve forecasts of temperature and wind during a significant wildfire event in Alaska. Toll et al. (2016) showed considerable improvement in forecasts of near-surface conditions during Russian wildfires in the summer of 2010 by including the direct radiative effect of realistic aerosol distributions. Considering the significant influence of the aerosol–radiation interaction on weather meteorological forecasts as illustrated in above studies and other studies (Zhang et al. 2010; Mulcahy et al. 2014). Several weather forecast centers have started to facilitate the inclusion of aerosol predictions in operational NWP models before including the aerosol feedback. NASA GEOS-5 started to provide near-real-time forecasts of aerosols and atmospheric compositions (Rienecker et al. 2008; Molod et al. 2015) and the ECMWF began to include global aerosol forecasts since 2008 (Hollingsworth et al. 2008; Morcrette et al. 2008; Benedetti et al. 2011).Three different gas–aerosol chemistry schemes were implemented in a two-way fully inline coupled global weather-chemistry prediction model (FIM-Chem) developed at NOAA Global Systems Laboratory (GSL), which showed good performance in forecasting the chemical composition for both aerosol and gas-phase species when compared with the observations without turning on the aerosols feedback (Zhang et al. 2022b).
The Global Forecast System (GFS) is the cornerstone of the operational production suite used for numerical guidance at National Centers for Environmental Prediction (NCEP). In collaboration with NASA/Goddard Space Flight Centre (GSFC), NCEP developed NOAA Environmental Modeling System (NEMS) GFS Aerosol Component (NGAC) for predicting the distribution of global atmospheric aerosols. The model became operational in 2016 with dust-only forecasting capability (NGACv1) (Lu et al. 2016) and all species in 2018 (NGACv2) (Wang et al. 2018). NGAC was built upon Earth System Modeling Framework (ESMF) and used the NEMS Global Spectral Model (GSM) as the atmospheric model. Since the 2019 implementation, the atmospheric forecast model used in the GFS consists of the Geophysical Fluid Dynamics Laboratory (GFDL) Finite Volume Cubed-Sphere dynamical core (FV3) and several physics updates associated with it. An aerosol model component was coupled online with the FV3 Global Forecast System (FV3GFS) for global aerosol prediction at the NCEP since September 2020 as one of the ensemble members of the Global Ensemble Forecast System (GEFS), named as GEFS-Aerosols v1 (Zhang et al. 2022a). However, in GFS, the aerosol attenuation coefficients are still determined from prescribed aerosol distributions based on a global climatological aerosol database (Hess et al. 1998) and aerosol indirect effects on clouds and precipitation formation are not accounted for. NOAA’s National Weather Service (NWS) is on its way to deploy various operational prediction applications using the Unified Forecast System (UFS, https://ufscommunity.org/), a community-based coupled, comprehensive Earth modeling system. Including the prognostic aerosols from online coupled chemical model to represent real-time predicted direct and semi-direct aerosol radiative effects is the target of the next-generation of global modeling systems developing at EMC and NCEP. This requires more realistic forecasting of aerosol optical properties to generate a more realistic forecast of aerosol feedback. Better understanding of the model performance in the current operational global aerosol forecast system, GEFS-Aerosols, would help to provide more information for potential improvements. A broader and more detailed evaluation from different aspects can help to improve the model performance significantly before including the aerosol feedback in the next step. This in turn will also help us to understand the possible impact of aerosol on weather in the future.
The paper is organized as follows. Section 2 provides a general description of GEFS-Aerosols, its available aerosol products, and data that are used in this study. Section 3 presents satellite, ground station, and analysis datasets that are used to validate the model forecasts. The evaluation of AOD against observational datasets followed by some case studies are described in section 4. Finally, section 5 includes a discussion and the conclusions.
2. Model description
NCEP has partnered with NOAA/Earth System Research Laboratories (ESRL) Global Systems Laboratory (GSL), Chemical Sciences Laboratory (CSL), NOAA/Oceanic and Atmospheric Research (OAR) Air Resources Laboratory (ARL), the NOAA National Environmental Satellite, Data, and Information Service (NESDIS) Center for Satellite Applications and Research (STAR) and NASA GSFC to develop a global aerosol model that replaced the operational NEMS GFS Aerosol Component version 2 (NGACv2) (Wang et al. 2018). The major difference between GEFS-Aerosols model and NGACv2 is not only in the chemical model part. The dynamical core, model physics, resolution, microphysics scheme, land surface model are completely different between these two models. In Zhang et al. (2022a), Table 2 summarizes the comparison of model configurations between GEFS-Aerosols and NGACv2. The new model was included as a single member named GEFS-Aerosols in Version 12 of the Global Ensemble Forecast System (GEFS) (Zhou et al. 2022). The meteorology of this new model is based on version 15 of the Global Forecast System (GFS v15) with 64 vertical levels while the aerosol modules are based on the Weather Research and Forecasting model with Chemistry (WRF-Chem) version of the Goddard Chemistry Aerosol Radiation and Transport (GOCART) model (Chin et al. 2002; Colarco et al. 2010) GEFS-Aerosols treats the sources, sinks, and chemistry of 15 externally mixed aerosol species: dust (five non-interacting size bins), sea salt (five non-interacting size bins), hydrophobic and hydrophilic black and organic carbon (BC and OC, respectively; four tracers), and sulfate (SO4). Dust and sea salt have emissions dependent on frictional velocity and wind speed, respectively. Primary sulfate and carbonaceous aerosol species incorporate emissions principally from anthropogenic and biomass burning sources, with additional biogenic sources of organic carbon. Secondary sources of sulfate include chemical oxidation of sulfur dioxide gas (SO2) and dimethyl sulfide (DMS). Loss processes for all aerosols include dry deposition (with gravitational settling), large-scale wet removal, and convective wet scavenging. Recent updates and additions include a biomass burning plume rise module; tracer convective transport and wet scavenging implemented in the simplified Arakawa–Schubert (SAS) convection scheme (Han and Pan 2011); the FENGSHA dust scheme implemented and developed at ARL (Dong et al. 2016). The FENGSHA dust scheme in GEFS-Aerosols uses three land-use types (barren land, shrub–grassland, and cropland) as potential erodible dust sources instead of dust source maps, and the dust vertical flux was calculated according to a modified Owen’s equation (Owen 1964) when the friction velocity (
GEFS-Aerosols provides a 5-day forecast of total aerosol as well as dust, OC, BC, sea salt, and sulfate aerosol at C384 grid, which is converted to ∼0.25° × 0.25° by the Unified Post Processor (UPP) 4 times per day (at 0000, 0600, 1200, and 1800 UTC). The GFS data are used as the meteorological initialization in each cycle (24 h). GEFS-Aerosols does not include aerosol data assimilation, so the chemical tracers in the restart files are used as the chemical initial condition for the next forecast. It provides AOD at seven different wavelengths such as 340, 440, 550, 650, 860, 1120, and 1640 nm. It also provides particulate matter (PM2.5 and PM10) forecasts along with three-dimensional mixing ratios of aerosol species at 64 model vertical levels. Fire emissions are updated every 24 h for each day of the run, while monthly CEDS emissions are included from the 2014 inventory. A 13-month retrospective run (only 0000 UTC cycle) has been conducted using GEFS-Aerosols to provide multispecies forecasts of aerosol optical depth (AOD) and other aerosol properties (such as single scattering albedo, angstrom exponent at different wavelengths). In this study we have used daily AOD forecast (from August 2019 to August 2020) at 550 nm (AOD550) from GEFS-Aerosols and evaluated model results against a set of observational datasets. Moreover, we have also evaluated 6-hourly forecasts of GEFS-Aerosols total and individual species (dust, OC, BC, sulfate, and sea salt).
3. Observation, reanalysis, and other model data
We have used the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA2) (Gelaro et al. 2017) aerosol dataset to evaluate GEFS-Aerosols total AOD forecast. The dataset was produced using version 5.12.4 of the Goddard Earth Observing System Model (GEOS) Data Assimilation System (DAS). Gridded data are released at a 0.625° longitude × 0.5° latitude resolution on 72 sigma–pressure hybrid layers between the surface and 0.01 hPa (Buchard et al. 2017). MERRA-2 uses the Goddard Earth Observing System, version 5 (GEOS-5) Earth system model (Rienecker et al. 2008; Molod et al. 2015) and the three-dimensional variational data assimilation (3DVar) Gridpoint Statistical Interpolation analysis system (GSI) (Wu et al. 2002; Kleist et al. 2009). The GEOS-5 model is radiatively coupled to the GOCART aerosol module. GEOS-5 is driven by daily biomass burning emissions derived from Moderate Resolution Imaging Spectroradiometer (MODIS) FRP retrievals using the Quick Fire Emission Dataset (QFED) emissions (Darmenov and da Silva 2015). In MERRA-2, aerosol and meteorological observations are jointly assimilated within GEOS-5. The assimilation of AOD involves very careful cloud screening and homogenization of the observing system by means of a neural network based scheme that translates MODIS and Advanced Very High Resolution Radiometer (AVHRR) radiances into Aerosol Robotic Network (AERONET)-calibrated AOD (550 nm). The system also assimilates (non-bias-corrected) Multiangle Imaging SpectroRadiometer (MISR) 550-nm AOD over bright surfaces (albedo > 0.15) and surface-based AERONET AOD observations at 550 nm (Buchard et al. 2017). Although MERRA2 assimilates AOD from AERONET and other satellite sources, it still underestimates AOD over some regions (Frey et al. 2021). The underestimation is likely caused by cloud contamination, using older anthropogenic emission database, no treatment for nitrate particles in GOCART (Buchard et al. 2017). We have used MERRA2 AOD from NASA Global Modeling and Assimilation Office (GMAO) generated product “M2T1NXAER v5.12.4” (GMAO 2015; Bosilovich et al. 2016). We have obtained daily hourly MERRA2 analyzed AOD550 fields.
AERONET (http://aeronet.gsfc.nasa.gov) is a ground-based and global-scale sun photometer network, which has been providing high-accuracy measurements of aerosol properties since 1990 (Holben et al. 1998). It is often used as the primary standard for validating satellite products and model simulations (e.g., Colarco et al. 2010; Levy et al. 2013). For this study, we use the quality-assured AERONET Version 3 level 1.5 product, which has better cloud-screening and better preservation of high AOD values that were often discarded in previous versions. The complete set of Version 3 cloud-screening and quality assurance algorithms and comparisons of the Version 3 product to Version 2 are provided in Sinyuk et al. (2020). For this analysis, we selected a number of AERONET sites based on the availability of contiguous data records from August 2019 to August 2020 and dominated by dust influence. Since AERONET instruments do not measure AOD550 directly, we have used AOD at 440 and 675 nm that are linearly interpolated on a log–log scale to provide 550-nm AOD. All AERONET data are sampled temporally at ±1 h of the daily GEFS-Aerosols forecasts (e.g., at any given location, AERONET measurements between 1100 and 1300 UTC are averaged to compare them to the 1200 UTC model forecast). The 2-h time window is created to ensure comparable temporal and spatial representativeness of the AEROENT data and model data.
The Visible Infrared Imaging Radiometer Suite (VIIRS) sensor onboard the Suomi National Polar-Orbiting Partnership (SNPP and NOAA-20) satellite provides AOD550 at 0.25° × 0.25° horizontal resolution. In this study we have used daily gridded enterprise processing system (EPS) VIIRS AOD (Liu et al. 2014). We have also used daily 3-hourly VIIRS gridded data (at 0.25°) from NESDIS that are at GEFS-Aerosols model forecast lead times (e.g., 0, 3, 6, 9, …, 21 h) as well for case studies.
We have used MODIS Collection 6.1 Level-3 (daily) AOD dataset (Levy et al. 2013) from the Aqua satellite to verify model forecast for case studies. MODIS data include AOD data based on refined retrieval algorithms, in particular, the expanded Deep Blue (DB) algorithm (Hsu et al. 2013; Sayer et al. 2013). We have used a merged AOD product, which combines retrievals from the Dark Target (DT) and DB algorithms to produce a consistent dataset covering a multitude of surface types ranging from oceans to bright deserts (Sayer et al. 2014).
We have included NOAA’s previous operational global aerosol forecast model NGACv2 (Wang et al. 2018) AOD forecast in all our evaluations for comparisons against the new operational model. NGACv2 output AOD at T126 grid, which roughly corresponds to ∼1° × 1° resolution and became operational in May 2017 using the spectral version of GFS (v14). It uses an earlier version of the GOCART aerosol module, the relaxed Arakawa–Schubert (RAS) scheme, the original GOCART dust scheme and puts biomass burning emissions at the lowest model layer (compared to use of plume rise in GEFS-Aerosols).
For quantitative comparison, model performances against observation dataset are evaluated using the Development Testbed Center (DTC) Model Evaluation Tools (MET) software package (https://www.dtcenter.org/community-code/model-evaluation-tools-met) (Brown et al. 2021). We use this tool to compute statistics for both point observation and gridded observations (reanalysis and satellite data). For gridded statistics (called “grid-stat” in MET) it accumulates AOD statistics separately for each horizontal grid location over time on matched forecast and observation pairs and computes both categorical and continuous type statistics. Continuous statistics are computed over the raw matched pair AOD values whereas categorical statistics are generally calculated by applying an AOD threshold to the forecast and observation values (Murphy and Winkler 1987). In this study, we have used three continuous statistics metrics: mean error or bias (Bias), root-mean-square-error (RMSE), and correlation coefficient (R) and three categorical statistics metrics: critical success index (CSI), false alarm ratio (FAR), and probability of detection (POD). Definitions of these terms are provided in appendix A.
4. Results
a. Seasonal mean and bias of AOD
We have divided the 13 months (August 2019–August 2020) of daily model forecasts used in this study into four seasons to understand aerosol characteristics in terms of its sources and transport from comparisons with observational data. Both GEFS-Aerosols and NGACv2 daily model forecasts of total AOD valid at (0000, 0600, 1200, and 1800 UTC) are averaged to compute daily means for each of the season’s analyses. We also averaged MERRA2 data at 6-h intervals to compute the daily mean for each season. VIIRS AOD data are daily gridded data with no retrievals of AOD outside of 65°S–65°N. Figure 1 shows day-1 forecasts of GEFS-Aerosols total AOD averaged over August–September–October (ASO) 2019 against MERRA2, NGACv2 and VIIRS. The model shows significant improvements compared to NGACv2 which underpredicts AOD from forest fire burning regions (mostly contributed by OC AOD in southern Africa, South America, Southeast Asia), anthropogenic emissions from East Asia (mostly contributed by sulfate and OC AOD) (Fig. 1b). Also, NGACv2 largely overpredicts dust from the North Africa source region compared to both reanalysis and satellite observation. GEFS-Aerosols is able to capture all the major aerosol events during this time period (Fig. 1a) that also include a large forest fire event in Siberia and dust transport from Taklamakan region. GEFS-Aerosols also simulates high AOD over northern India (AOD ∼ 0.5), which is largely contributed from post-monsoon agricultural fires in the northwestern part of the Indian subcontinent and transports further east (Sahu et al. 2021). Both MERRA2 and VIIRS show similar high AOD over northern India as well. However, we have found large overprediction biases of AOD over southern Africa in GEFS-Aerosols, which may be due to convection related transport of species and subsequent slow removal process.
Seasonal mean (August–October 2019) of day-1 total AOD forecast of (a) GEFS-Aerosols and (b) NGACv2 compared against (c) MERRA2 and (d) VIIRS. The white color in (d) is due to missing high-latitude data (north of 65°N and south of 65°S).
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
Figure 2 shows 4-month averages of AOD from November 2019 through February 2020 of day-1 model forecasts compared against observations. High AOD values (>0.4) on the coast of Africa contributed by both dust and biomass burning (OC AOD) are captured by GEFS-Aerosols and closely match with observations. The model also simulates high AOD (∼0.5) over the Indian subcontinent, especially south of the Himalayas, which is absent in NGACv2 (∼0.1). Dust aerosols are found in the main dust emission source regions which have been mentioned previously, such as North Africa and East Asia. The dust plumes are found in the downwind direction of the dust source regions, indicating that dust aerosols transport to the mid‐Atlantic and western Pacific and even across the ocean to the South American continent. High AOD over eastern Asia (>1.1) is mainly contributed by anthropogenic emission (mainly sulfate). GEFS-Aerosols also shows high AOD over China, which over predicts MERRA2 and VIIRS data and is mostly missed in NGACv2. GEFS-Aerosols uses CEDS2014 anthropogenic emissions, which do not take into account the widespread reduction in anthropogenic emissions due to the coronavirus (COVID-19) pandemic over the region (Zheng et al. 2021), thus generating large overprediction. Figure 2a shows high AOD (>1.1) over the Australian coast, which is due to large bush fires that occurred between December 2019 and January 2020. The unusually severe fires in the Southeastern part of Australia in 2019/20 showed record-breaking levels of AOD in the Southern Hemisphere (SH) (Boer et al. 2020; Khaykin et al. 2020). During the wildfires, an enormous volume of biomass-burning aerosols was emitted and transported to surrounding downwind areas. The prevailing westerly wind carries the smoke plume toward South America (not shown). Ohneiser et al. (2022) study shows lidar station in southern Chile detected smoke layers from the Australian fires in January 2020. NGACv2 largely under predicted both the smoke event near the source and transportation over the ocean.
As in Fig. 1, but for November 2019–February 2020.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
Figure 3 shows the seasonal average of AOD for March–April–May (MAM) of 2020. GEFS-Aerosols is able to capture forest fire events over Southeast Asia and Central America better than NGACv2 (Figs. 3a,b). However, GEFS-Aerosols shows AOD of >0.7 over the southern Indian peninsula and it is higher than both MERRA2 and VIIRS. This may be attributed to excessive biomass burning over the Indian subcontinent during spring season and transport of pollutants from upwind region like the Middle East (Nandi et al. 2020). Large overprediction over both India and East Asia in GEFS-Aerosols can be mostly attributed to OC/BC and sulfate AOD due to the older anthropogenic emission inventory included in the model. Dust source and transport from the Taklamakan desert continued to be under predicted in the model.
As in Fig. 1, but for March–May 2020.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
Figure 4 shows the 3-month average of AOD between June and August (JJA) of 2020. GEFS-Aerosols simulate AOD from biomass burning events over South America, southern Africa, and Siberia showing great improvements over NGACv2, and is close to both MERRA2 and VIIRS (Fig. 4a). However, over southern Africa AOD continued to be too strong around biomass burning areas (as seen in Fig. 1). Total AOD values of 0.7 in the Middle East observed in both satellite and reanalysis data are mainly from the contributions of dust and sea salt (Fig. 4a), which is also observed by both satellite and reanalysis. In June 2020, a massive African dust intrusion into the Caribbean Basin and southern United States happened and was named the “Godzilla” dust plume (Yu et al. 2021). It also has been captured by the GEFS-Aerosols, which shows slightly underprediction as that of MERRA-2 and VIIRS (Fig. 4). Both GEFS-Aerosols and VIIRS data show high AOD buildup south of the Himalayas (Figs. 4a,d), which is linked to dust and anthropogenic emission.
As in Fig. 1, but for June–August 2020.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
Table 1 summarizes seasonal mean and standard deviation of total AOD for all seasons. VIIRS-estimated AOD values cover the latitudes between 60°N and 60°S, which gives a slightly higher global mean than the model and reanalysis. However, model data are non-collocated against VIIRS AOD, which may lead to some bias against VIIRS. Overall GEFS-Aerosols showed much improved global mean AOD compared to its predecessor, i.e., NGACv2, in all four seasons analyzed in this study. AOD improvements in terms of seasonal mean are nearly 37% in ASO and 33% in NDJF and ∼20% in the MAM and JJA seasons. Some of these improvements are generated by better representation of forest fires and related AOD in GEFS-Aerosols. Figures 1–4 show that GEFS-Aerosols is able to simulate both seasonal forest fires and the subsequent transport of smoke plumes in South America, southern Africa, Southeast Asia, and Central America as well large, unprecedented fire events over Australia and Siberia, all of which are largely under predicted by NGACv2. GEFS-Aerosols is also able to simulate the mixed-aerosol region over India where AOD from anthropogenic and transported biomass burning dominate in the springtime (MAM) along with dust AOD in the northern part of India. At the same time, some of the fire events were largely overpredicted (southern Africa, Australia, and North America) by the model while dust events in Asia were mostly unpredicted (dust originated and transported from the Taklamakan desert).
Seasonal mean and standard deviation of day-1 total AOD forecast by of GEFS-Aerosols and NGACv2 compared against MERRA2 and VIIRS.
We have extended our study to evaluate AOD from individual AOD species. Figures 5 and 6 shows dust, OC, sulfate, and sea salt AOD bias against MERRA2 in four seasons for GEFS-Aerosols and NGACv2, respectively. Using DTC-MET we have computed day-1 forecast bias (mean error) from 6-hourly model forecasts compared against MERRA2. As the resolution of the model output and MERRA2 is different, MET interpolates the results to the coarser MERRA2 grid (similarly performed for NGACv2). In GEFS-Aerosols, dust AOD shows slightly positive bias ∼0.1 (overprediction) near the source region over northern Africa but largely negative bias (underprediction) in the transport of dust plume over the Atlantic Ocean (Figs. 5a–d). Similar biases have also been found in other studies (Nowottnick et al. 2011). However, the area of overprediction in the African dust source region is mainly during JJA and ASO season but it decreases during NDJF and MAM. However, we have found underprediction of Asian dust in the Middle East and the Gobi Desert region (∼0.2) from GEFS-Aerosols that persisted throughout the study period. NGACv2 also underpredicts Asian dust near the source region (Gobi Desert) but shows large-scale overprediction of dust transport from Asia (Figs. 6a–d).
Day-1 forecast bias of four species of AOD from GEFS-Aerosols compared against MERRA2 for all four seasons. (a)–(d) Dust AOD bias, (e)–(g) OC AOD, (i)–(l) sulfate AOD, and (m)–(p) sea salt AOD.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
As in Fig. 5, but for NGACv2.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
We have found substantial improvement of OC AOD bias with the model update, as large underprediction of OC AOD related to fire events in NGACv2 (Figs. 6e–h) is changed to some overprediction of OC near the fire sources in GEFS-Aerosols (Figs. 5e–h) and some underprediction of transported plumes. We have found similar improvement in sulfate AOD over East Asia (anthropogenic source) from NGACv2 (large under prediction) to GEFS-Aerosols (slight over prediction). Sea salt AOD bias is greatly improved in GEFS-Aerosols compared to NGACv2 (Figs. 5m–p and 6m–p), which is mostly due to the updated sea salt module included from a more recent GOCART version. Table 2 summarizes seasonal RMSE between these two models in terms of species as compared to MERRA2. It shows reduction of RMSE for dust AOD in Northern Hemisphere (NH) summer months (JJA and ASO) in GEFS-Aerosols due to improved northern African dust predictions over NGACv2. The error is largely contributed by Middle East and East Asian dust sources in GEFS-Aerosols in NH spring months (MAM). OC AOD in GEFS-Aerosols shows smaller RMSE in all seasons compared to NGACv2. Sulfate AOD produced by biogenic and anthropogenic sources show improved RMSE in all seasons.
Global average day-1 RMSE for all aerosol species of GEFS-Aerosols and NGACv2 compared against MERRA2 in all four seasons.
Figure 7 shows the contingency statistics of CSI, FAR and POD, of both GEFS-Aerosols and NGACv2 as a function of AOD thresholds for all four seasons. Critical success index (CSI or threat score) measures fraction of observed and/or forecast events that are predicted correctly. False alarm rate (FAR) is the ratio of the number of false alarms to the number of forecasts made and POD is the ratio of the number of correct forecasts to the number of observed events (descriptions of these terms are in appendix A). We have selected five AOD categories (≥0.1, ≥0.2, ≥0.4, ≥0.6, and ≥0.8) and computed categorical statistics for both models against MERRA2 using DTC MET. GEFS-Aerosols have higher POD and CSI than NGACv2 in all seasons which is consistent with our results discussed before. Both CSI and POD values show maximum at lowest AOD threshold for GEFS-Aerosols (∼0.5 for POD and 0.4 for CSI), which is nearly 2 times as high as in NGACv2 (∼0.2 POD and CSI). In GEFS-Aerosols FAR values show beneficial reduction across all thresholds than NGACv2.
Contingency metrics of CSI, FAR, and POD for GEFS-Aerosols and NGACv2 over the globe. The results shown are at five AOD intervals used in this study for four different seasons.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
b. Correlation over different regions
We selected key aerosol regions over the land and ocean and compared model forecast of AOD against reanalysis over those regions. We have used model forecasts at every 6 h at days 1, 3, and 5 of both GEFS-Aerosols and NGACv2 for the entire 13-month period from August 2019 to August 2020. The two ocean regions include the North and South Atlantic Oceans, which are major long-range aerosol transport pathways for dust, smoke, and sulfate. Latitude–longitude bounds of the ocean areas are the following: North Atlantic Ocean (0°–35°N, 10°–80°W) and South Atlantic Ocean (0°–35°S, 40°W–20°E). The land regions include two dust source regions (North Africa and the Middle East), two biomass burning regions (South America and southern Africa), three regions over North America (the eastern United States, western United States, and Canada), and three major dust and pollution source regions (Middle East, India, and East Asia). Latitude–longitude bounds of the land areas are as follows: North Africa (0°–30°N, 18°W–30°E), South Africa (0°–30°S, 8°–35°E), eastern United States (25°–48°N, 68°–95°W), western United States (25°–48°N, 95°–125°W), Canada (48°–70°N, 60°–160°W), South America (10°–35°S, 35°–80°W), Middle East (10°–32°N, 30°–70°E), East Asia (20°–48°N, 100°–140°E), and India (8°–35°N, 68°–95°E). Figure 8 shows correlation coefficients and RMSEs for GEFS-Aerosols and NGACv2 and MERRA2 for the time period at different forecast days. We used Taylor diagrams to summarize model performance in different seasons over the same regions described. Taylor diagrams (Taylor 2001) provide a statistical summary of comparisons between both models against MERRA2 in terms of their spatial correlation coefficients and the ratio of spatial standard deviations of the model and observations over the twelve predefined regions. The spatial correlation coefficient is the quantity that measures the degree of agreement of two fields, and standard deviations are normalized by the corresponding observations. In general, GEFS-Aerosols records high correlations and low variance over all the regions in all forecast lead days against NGACv2 (Fig. 8). It records lowest correlations (∼0.5) over India and Northern Canada, but the correlations improved compared to NGACv2 (∼0.2). GEFS-Aerosols forecasts have high correlation (>0.7) over major aerosol source and downwind areas (northern Africa and Atlantic Ocean for dust and southern Africa and the surrounding ocean for biomass transport). We have found high variance over the western U.S. region with GEFS-Aerosols, and it increases from day 1 to day 5 (Fig. 8). It is largely contributed by overprediction of summer fire event over California (in terms of species OC AOD) in 2020 and also associated RMSE (not shown). However, in the western United States, many model forecasts are challenged by complex atmospheric physics and air pollution characteristics (Li et al. 2015; Loría-Salazar et al. 2017). Presence of high temperatures, earlier snowmelt, less rainfall, and drought conditions, are primary reasons for generation of multiple fires and causing huge amounts of aerosol pollution (Bian et al. 2020; Brewer and Clements 2020). The smoke plumes produced by those fires undergo complex aerosol transport processes typical of the western United States, leading to a potential disconnect between aerosol measurements at the surface (i.e., PM2.5 concentrations) and the aerosol concentrations aloft (Wilkins et al. 2020). Smoke plume injection, boundary layer mixing, and entrainment aloft can disperse the smoke plumes into the free troposphere. Due to unavoidable conditions in the operation, it should be noted that GEFS-Aerosols is initialized each day using ∼2-day temporal lag fire emissions for each cycle and such temporal inconsistency would definitely cause biases at the first few days of fire events. Also, there is not diurnal fire profiles being applied into the fire emission, a daily snapshot of the fire emission has been used to each time step of the predicted periods, which may also cause large uncertainties, especially at the beginning and the end of the fire event, also for those fires catching at the maximum. Future work will try to explore the use of diurnal fire profiles based on historic Geostationary Operational Environmental Satellites (GOES) fires products, which are applied to estimate fire emission evolution to enhance forecast behavior. Additionally, a parameterization based on fire weather index (FWI) to estimate fire emissions on longer temporal scales may help to improve and extend the forecast of fire impacts. In general, low correlation over different regions as simulated by the previous NGACv2 may be due to aerosol removal from the atmosphere due to high aerosol scavenging by clouds and precipitation.
Taylor diagram for (a) day-1, (b) day-3, and (c) day-5 forecasts of total AOD of GEFS-Aerosols and NGACv2 over 12 different regions. On the Taylor diagrams, angular axes show spatial correlations between modeled and observed fields; radial axes show spatial standard deviation normalized against that of the observations. Each dot represents a region identified by its number on the top: NATL: North Atlantic Ocean; SATL: South Atlantic Ocean, NAFR: northern African region; SAFR: southern Africa region; SAMR: South America region; NCAN: Canada region; EUSA: eastern U.S. region; WUSA: western U.S. region; EURP: Europe region; IND: India; MEST: Middle East; EASIA: East Asia region.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
c. Evaluation with AERONET observation
We have compared AOD550 forecast from GEFS-Aerosols for the 13-month period with corresponding AERONET AOD measurements. Figures 9a and 9b shows correlation coefficients (R) for GEFS-Aerosols and NGACv2 compared to AERONET-derived AOD for day-1 forecast during the entire time period. (In appendix B, Table B1 summarizes the names of the AERONET sites along with R, RMSE and the number of paired observation points for the 51 stations used in Fig. 9.) To facilitate comparison between ground-based AERONET observations and gridded model output, the 0.25° × 0.25° GEFS-Aerosols model grid within which the AEORNET level 1.5 data fall are first identified and model AOD is sampled from the identified grid. A similar methodology is used for NGACv2 (which has resolution of 1° × 1°). We will discuss model performance over AERONET stations that are mostly influenced by dust first, followed by station locations dominated by other aerosol species.
Correlation coefficient (R) map of total AOD at 550 nm (a) between GEFS-Aerosols and the AERONET sites and (b) between NGACv2 and the AERONET sites. Approximate location of AERONET centers in the map are represented as filled circles. Names and locations of these sites are listed in Table B1.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
We used four different hourly forecast values from GEFS-Aerosols that cover four days of AOD550 values, i.e., 12, 36, 60 and 84 forecast hours from each day in the study period for dust-dominated AERONET stations. We have identified 11 such AERONET stations that are located mostly in northern Africa and near the Saharan Desert source (stations are Dakar, Capo Verde, Banizoumbu, Tamanrassett, Zinder Airport, Ilorin, Tenerife, Saada, Ben Salem, Quena, and Sede Boker). We have also included two downwind dust locations (Ragged Point and San Juan) across the Atlantic Ocean that are influenced by long-range Saharan dust transport. Also included are two other AERONET stations in Asia (Dalanzadgaad, influenced by transport of dust from Gobi Desert, and Dushanbe, Tajikistan, influenced by Central Asia desert dust transport). The selection of these 15 dust stations is based on available data for this study period and numerous references to them in previous studies. Tables 3 and 4 list GEFS-Aerosols and NGACv2 model correlation (R) and RMSE, respectively, for each of the four forecast lead hours for the period.
Correlation of GEFS-Aerosols and NGACv2 at dust dominated AERONET stations.
RMSE of GEFS-Aerosols and NGACv2 at dust-dominated AERONET stations.
Dakar and Capo Verde are located in the Atlantic Ocean, on the main path of Saharan dust transport and record high dust load throughout the year. GEFS-Aerosols shows good agreement in terms of R (above 0.5) in both stations compared to NGACv2 (∼0.35) in all forecast hours (Table 3). However, RMSE values are high (above 0.25) at both locations for both models indicating failure to capture intensity of dust properly. Tamanrassett station, located in southern Algeria, is in the heart of the Sahara Desert. GEFS-Aerosols showed big improvement over NGACv2 in terms of R values for this location (0.67 for 12-h forecast compared to 0.32) and also some decrease in RMSE (0.20 for 12-h forecast compared to 0.25). Three stations (Zinder Airport, Ilorin, and Banozoumbu) are located south of the desert regions, within important pathways for Saharan dust transport. GEFS-Aerosols showed better correlation and RMSE than NGACv2 at these locations. Tenerife and Saada are located northwest of the Sahara Desert, and Ben Salem is north of the desert. All three record very high AOD in summer months. In all these dust outflow regions, GEFS-Aerosols show significant improvements in terms of lower RMSE and higher correlation than NGACv2.
For eastern Sahara AERONET stations (Quena and Sede Boker) we found huge improvement in GEFS-Aerosols over NGACv2 (Tables 3 and 4). In Dushanbe, high RMSE (but still better than NGAC) is associated with GEFS-Aerosols at all four forecast hours and we have found consistent underprediction throughout the study period. Dalanzadgad is in the territory of eastern part of Gobi Desert, which is an important site for Asian dust monitoring especially in spring and summer months. GEFS-Aerosols perform better than NGACv2 at this location. GEFS-Aerosols outperforms NGACv2 at two stations located on the western Atlantic (Ragged Point and Camaguey) indicating reasonable dust transport over the Atlantic Ocean in the model.
Stations 15–24 are located in various parts of the globe and dominated by mixed aerosols (dust, OC/BC, and sulfate) (based on Table B1). We found that GEFS-Aerosols showed improved correlation and lower RMSE than NGACv2 for all these stations. Only exception is at AERONET site in Tumbarumba (station 18 in Table B1) which is located in the east coast of Australia and largely impacted by large over prediction of GEFS-Aerosol model during the Australia fire event in December of 2019. Two stations over India show correlation below 0.5 associated with higher RMSE. The largest contribution from aerosol load over India comes from the anthropogenic component (dominated by sulfate, OC, and BC) and by dust blown from the Middle East and western India during May–July. We found some of the poor correlation at these stations to be caused by representation of dust events in the summer months (Figs. 5a–d). Sites 25–29, located in equatorial and southern Africa, are influenced mainly by biomass burning. Biomass burning activity peaks during August–September at these sites and the magnitude of the maximum AOD at the three southern African sites is overestimated by GEFS-Aerosols by a factor of almost 2–3 (also high RMSE in Table B1). A similar improvement of AOD is also observed over five of the South American sites (30–34 in Fig. 9). Model-simulated AOD correlates well (0.59) at site 30, which is located near a biomass burning region. However, the model simulates low AOD at these sites during the non-biomass burning seasons but correlates well with observed AOD between September and November during the biomass burning season in Brazil. Sites 36–48 and 50 are located in North America (Figs. 9a,b) and GEFS-Aerosols shows considerable improvement in terms of R and RMSE over NGACv2.
d. Case studies
In the following section we will describe three case studies assessing GEFS-Aerosols performance in high aerosol events.
1) Dust event over northwest Africa
Dust from North Africa is mostly exported to the Atlantic, resulting in frequent dust concentrations with magnitudes in the thousands of micrograms per cubic meter (μg m−3) in western North Africa and from tens to hundreds (μg m−3) in the Canary Islands and Cape Verde (Rodríguez et al. 2015; Almeida-Silva et al. 2013). Between 22 and 24 February 2020, a severe episode of Saharan dust occurred, deteriorating air quality to unprecedented levels throughout the Canary Islands (NASA EOS imagery 2020). Recorded data from representative measurement sites in Santa Cruz de Tenerife (coast of Morocco) show PM10 and PM2.5 concentrations values of 1940 and 350 μg m−3, respectively, during the first peak (at 1900 UTC 22 February) and 2930 and 1060 μg m−3 on 23 February during the second peak (1800 UTC) (WMO 2021). The 22–24 February 2020 dust outbreak has been the most widely reported dust episode in the Spanish media in its history (WMO 2021). The severe impact on the operation of the eight Canarian airports caused by this dust outbreak, which were closed for hours, and its concurrence during the famous Carnivals of Santa Cruz de Tenerife and Las Palmas de Gran Canaria, with their huge influx of national and foreign tourists to the archipelago, explains the extensive publicity given to it by the media from all over the world (WMO 2021).
Figure 10 shows GEFS-Aerosols, NGACv2 24-h forecast and MERRA2 reanalysis of total AOD over the African coast between 21 and 29 February 2020. MODIS AOD (combined land and ocean) is also shown in the figure for the same days. GEFS-Aerosols forecast matches closely with both MODIS and MERRA2 forecast qualitatively (in terms of location and transport of dust from northwest of Africa to over the Atlantic Ocean). GEFS-Aerosols accurately shows the plume pathway over the Canary Islands on 23 February, when most of the ground stations record high surface dust concentrations. NGACv2 on the other hand misses the dust outflow pathway and predicts a lower intensity plume. Figure 11 shows four AERONET stations that are located around northwest Africa (Dakar, Cape Verde, Tenerife and Graciosa, marked in Fig. 10 map plot) recorded high AOD peaks during this event. We have added day-1 AOD forecasts from GEFS-Aerosols and NGACv2, and MERRA2 analysis for the same ground stations. Both GEFS-Aerosols and MERRA2 datasets show high AOD on the 23 and 24 February over the Tenerife location (Fig. 11c) despite no AOD being recorded by the ground station during this high dust event. Dakar and Cape Verde show two dust plumes after 25 February, which correlates well with GEFS-Aerosols and MERRA2. ARM-Graciosa is the station further away from the continent and also recording a high AOD around 27 February which again matches well with GEFS-Aerosols.
Comparison of total AOD forecast of GEFS-Aerosols and NGACv2 against MERRA2 and MODIS for selected days of February 2020 along the northwest coast of Africa. For both models, the 24-h forecast of model AOD is used against daily satellite observations of MODIS. Satellite data have data gaps that are white in color.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
Comparison of 550-nm AOD forecast between 15 and 29 Feb 2020 at four AERONET locations. The map plot shows the average of total AOD of MODIS data (between 15 and 20 Feb) and locations of the ground stations.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
2) Agricultural fire over North India
Aerosol emissions in northern India are mainly from biomass burning and anthropogenic sources (Gani et al. 2019). Both satellite and ground-based monitors have detected enhanced aerosol-loading downwind of smoke plumes from agricultural fires across northern India and Pakistan in recent years (Kaskaoutis et al. 2014; Liu et al. 2018; Jethva et al. 2018). Crop residue burning, which is a by-product from harvesting and processing of crops, is one of the major sources of ambient PM2.5 concentration during the post-monsoon season (October–November) in the Indo-Gangetic Plain (IGP) of India. Every year, the National Capital Region (NCR) in and around Delhi, India, faces acute air pollution levels in terms of fine particulate matter (PM2.5) especially from October to January (PM2.5 reaching 600 μg m−3) (Cusworth et al. 2018; Kulkarni and Sreekanth 2020). Figure 12 shows one such fire event that started around the middle of October 2019 in northwestern India and later on the transported smoke plume PM2.5 over rest of the northern India in IGP. Both MERRA2 and VIIRS show AOD values well above 2 over the rest of northern part of India, including the Delhi region. GEFS-Aerosols 24-h forecast for each of the four days (15, 18, 20, and 24 October 2019) in Fig. 12 is compared against NGACv2. GEFS-Aerosols underpredicts the initial fire event on 15 October (Fig. 11a) as MERRA2 and VIIRS both show very high AOD values (>3) in the Northwestern part of the country. On 18 October, the pollution plume spread over much of Northern India and GEFS-Aerosols matches closely with both MERRA2 and VIIRS. NGACv2, which used the same GBBEPx fire emissions, largely underpredicted the entire fire event.
Comparison of total AOD forecast of GEFS-Aerosols and NGACv2 against MERRA2 and VIIRS for selected days of October 2019 over northern India. For both models, the 24-h forecast of model AOD is used against daily satellite observations of VIIRS. Satellite data have data gaps that are white in color.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
We have found three AERONET sites that are located in IGP and recorded this high AOD event. They are Lahore, Amity, and Lumbini and are shown in the Fig. 13 map plot. At Lahore, the westernmost station in the figure, GEFS-Aerosols largely underpredicts the high fire event on 17 October (peak AOD intensity remains below 1, whereas the ground station record values > 2; Fig. 13a). However, the GEFS-Aerosols forecast closely matches with AERONET peaks at Amity and Lumbini, which record the transported pollution plume around 19 and 25 October, respectively (Figs. 13b,c). NGACv2 overall largely missed this event with peak AOD intensity never exceeding 0.4.
Comparison of 550-nm AOD forecast for the month of October 2019 at three AERONET locations over northern India. The map plot shows the average of total AOD of MODIS data (between 15 and 21 Oct 2020) and locations of the ground stations.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
In Table 2 of Zhang et al. (2022), it shows the detailed comparisons between the GEFS-Aerosols and NGACv2, including the fire emission. Even both of them are using the GBBPEx fire emission data, but the ones used in the GEFS-Aerosols are an updated version of GBBPEx v3, while the NGACv2 is using GBBEPx (Zhang et al. 2012). The other major difference is the biomass burning plume rise module adapted from High-Resolution Rapid Refresh (HRRR)-Smoke based on WRF-Chem has been implemented into GEFS-Aerosols, which has the capabilities of using the GBBEPx v3 and FRP data. Zhang et al. 2022a also pointed out that NGACv2 results are quite different from the satellite observations and MERRA2 analysis, underestimating the AOD more than 50%–90% percent over the southern Africa fire source region and showing little obvious enhancement. And the updates in fire emission using GBBEPx v3 emission and FRP by applying the 1D plume rise scheme in GEFS-Aerosols model show great improvements in the AOD forecast during the fire events compared to the NGACv2. Wildfire plumes significantly impact regions near the fire source, they can reach high altitudes above the mixing layer and can be transported long distances (Nisantzi et al. 2014) to affect global atmospheric chemistry and climate (Peterson et al. 2018; Das et al. 2021). Though the evaluated periods and locations are different to this study, both of them shows consistent conclusions that the GEFS-Aerosols using plume rise module with updated GBBEPx v3 and FRP data show significant improvements in fire forecast compared to that of NGACv2.
3) August Complex Fire, Northern California
Massive wildfires and extreme fire events are becoming more frequent across the western United States, creating a need to better understand how megafire impacts air quality from local to continental scale. (Dennison et al. 2014; Fu et al. 2021; Zhuang et al. 2021). A series of lightning strikes started hundreds of fires across Northern California in August 2020, dubbed the August Complex Fire. They are the largest fires in California’s history, together burning 1.03 million acres in seven counties and continuing into November (NIFC 2020). Another fire, the Santa Clara Unit (SCU) Lightning Complex Fire, also started in the same month, is the third largest fire on record in the state, burning almost 400 000 acres over five Northern California counties near San Francisco (Cal Fire 2020). Figure 14 shows four days (18, 20, 22 and 24 August 2020) of 24-h AOD forecast from GEFS-Aerosols and NGACv2 compared against VIIRS and MERRA2. Some of the fires already started on 18 August and rapidly expanded to much bigger areas as seen on 24 August (Fig. 14). We have used VIIRS observation that occurred at 2100 UTC for each of these days to compare against model results. GEFS-Aerosols captured rapidly growing fire sources remarkably well as compared to MERRA2 in the initial stage, but largely overpredicted AOD as time progressed. Fire emissions are based on real-time satellite observations in GBBEPx data. However, GEFS-Aerosols is initialized each day using 2-day old fire emissions for the each cycle and such temporal inconsistency may cause the low positive bias in AOD predictions over the fire source region at the first time when the fire event started and may cause high bias in AOD predictions over the downwind areas after the fire event started.
Comparison of total AOD forecast of GEFS-Aerosols and NGACv2 against MERRA2 and VIIRS for selected days of August 2020 over the western coast of North America. For both models, the 24-h forecast of model AOD is used against VIIRS data at 2100 UTC that day. Satellite data have data gaps that are white in color. VIIRS AOD data range between 0 and 5.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
We identified four AERONET stations close to the actual fire event and all of them are on the downwind side of the plume. Figure 15 shows the stations locations of the stations (Meridian and Taylor Ranch in Idaho, University of Reno in Nevada, and PNNL in Washington state) and AOD time series for the entire month of August based on daily forecast lead times beginning 21 August, GEFS-Aerosols forecast very high AOD exceeding 5 over Meridian and Taylor Ranch (Figs. 15a,b) followed by another peak of around 3 after 26 August. Both MERRA2 and AERONET peak values never exceeded 2 in those locations. University of Reno, which is closest to the fire, recorded high forecast AOD (>3) from both MERRA2 and GEFS-Aerosols after 16 August, but both underpredict ground station records of the high AOD event (∼4) on 20 August (Fig. 15c). GEFS-Aerosols largely overpredicts the northernmost station, PNNL, between 22 and 26 August. NGACv2 peak intensities match closely with GEFS-Aerosols at Meridian and Taylor Ranch but it remains largely inconsistent at the other two stations. Combining Figs. 14 and 15, we have found GEFS-Aerosols AOD forecasts consistent with MERRA2 around the southern outflow region of the fire plume (University of Reno station) at the beginning of the historic fire event and consistently overpredict at the two more eastern AERONET sites in Idaho. However, we found as the fire progressed that GEFS-Aerosols predicted stronger plumes in the northern part, which is largely an overprediction. Also, the uncertainties of fire emission data (GBBEPx) may contribute to the large over prediction in the model forecast.
Comparison of 550-nm AOD forecast for the month of August 2020 at four AERONET locations in the western United States. The map plot shows VIIRS total AOD on 18 Aug 2020 and locations of the ground stations.
Citation: Weather and Forecasting 38, 2; 10.1175/WAF-D-22-0083.1
5. Conclusions
This paper presents an evaluation of AOD from NOAA’s latest operational aerosol forecast model GEFS-Aerosols. The model was implemented as one member of GEFS version 12 into operations in September 2020 and replaced the previous operational global aerosol prediction system NGACv2 at NCEP. GEFS-Aerosols forecasts five species of aerosol (dust, sea salt, BC, OC, and sulfate) and total AOD at every 6 h out to 5 days, four times per day (0000, 0600, 1200, and 1800 UTC) on a global 0.25° × 0.25° horizontal grid. We extensively evaluated 13 months of retrospective AOD forecast from the model (only 0000 UTC), both temporally and spatially against the previous operational model, NGACv2, MERRA2 reanalysis, and satellite (MODIS, VIIRS) data. We have used both continuous and categorical statistics in our AOD evaluations. We also compared model results with more than 50 AERONET station observations, which are spread globally and represent different aerosol regimes. We have also used three high aerosol events that occurred in the study period as case studies to critically evaluate GEFS-Aerosols performance.
The GEFS-Aerosols AOD forecast shows great improvement from its predecessor in all four seasons of the study period. The model reproduces the prominent temporal and geographical features of AODs as observed by MERRA2 and satellites, such as dust plumes over northern Africa and the Arabian Peninsula, biomass burning plumes in southern Africa and northern Canada, and high-altitude sea salt bands. In terms of aerosol species bias, GEFS-Aerosols greatly reduced high overprediction of dust compared to NGACv2 over northern Africa and improved dust underprediction over the Gobi Desert. In our case study, GEFS-Aerosols qualitatively and quantitatively matched dust intensity of surface AOD measurement for high dust events over northwestern Africa. For OC AOD, though better physics and using a plume rise algorithm, GEFS-Aerosols was able to capture some of the biggest biomass burning events (e.g., southern Africa, South America, Siberia, North America) compared to NGACv2, and improved underprediction of AOD forecasts near fires and downwind locations. Also, GEFS-Aerosols simulated both agricultural fires over India, as shown in the second case study, and also over Mexico, which take place in March–April (not shown), greatly improving the aerosol forecasts over those regions. We have found improvement in sulfate AOD underprediction over NGACv2 (from both anthropogenic and biomass-related origins) over East Asia and Europe. Large overprediction of sea salt in the remote Southern Ocean as simulated by NGACv2 is greatly reduced in GEFS-Aerosols. We have found with GEFS-Aerosols AOD forecast correlation of above 0.6 over major regions of the globe alongside with smaller RMSE for days 1, 3, and 5 and both correlation and RMSE greatly improved from NGACv2. The comparisons of model forecasts with surface point locations show results similar to our comparisons with MERRA2 in larger gridded domains. The model reproduces the seasonal variations at most of the sites, especially those sites where dust and biomass plumes dominate. The model also captures dust and smoke outflow from Africa at AERONET locations that are present in the Atlantic Ocean (Cape Verde, Ascension Island) even though the magnitudes do not match with these point observations. Model AOD captures two other dust regions (the Arabian Peninsula and Asian dust near the source region) but underestimates them quantitatively as these dust plumes undergo long-range transport over Asia, which shows improvement over NGACv2.
Our study also found that GEFS-Aerosols largely overpredicted two historical fire events that happened during the study period: coast of Australia fire in December 2019 (not shown) and western United States fire in the summer of 2020. As shown in the final case study, model AOD forecast qualitatively matches with ground observations but the intensity is very high (reaching AOD above 5) at some downwind ground stations.
We also found that the AOD forecast caused by the U.S. fire events is challenging in model prediction due to the uncertainties in fire emission and modeled scheme diversities. For instance, though the GEFS-Aerosols AOD predictions look much stronger than that of the ICAP-MME analysis and GEOS-5 model, it is quite comparable to that of the ECMWF forecast (figure not shown here). Several factors can lead to very high AOD values such as 1) using 2-day temporal lag fire emission in the model, which impacts simulation of some of big aerosol events, especially near the end of the fire events over the downwind areas of fire source regions, 2) uncertainty in GBBEPx fire emission itself and the validation data of satellite observation and reanalysis data, 3) use of plume rise 1D model inside GEFS-Aerosols so that the injection height is highly depending on the vertical change of meteorological fields (e.g., convection), 4) absence of wet removal process that removes OC/BC adequately as the event goes on, and 5) lack of any data assimilation in the model to improve initial conditions of AOD species. Without real-time assimilation, the initial aerosol fields may contain errors that propagate to all forecast hours. Aerosol data assimilation using VIIRS is under development and expected to be implemented in the near future. The future direction for GEFS-Aerosols will be focused on 1) implementing aerosol data assimilation to improve aerosol forecasts, 2) reducing the time lag in fire emission input and applying diurnal variability and better speciation, and 3) improving the representation of aerosol–radiation–cloud interaction in the atmosphere model toward improving weather forecast and climate prediction.
Acknowledgments.
We thank Atmosphere Archive and Distribution System (LAADS) Distributed Active Archive Center (DAAC), located in the Goddard Space Flight Center in Greenbelt, Maryland, for providing MODIS Level-3 data. We acknowledge the SNPP/VIIRS science team for the high-quality products. We acknowledge the AERONET team for the production of the data used in this work. The authors thank the two anonymous EMC internal reviewers for their valuable suggestions and comments. Li Zhang was supported by NOAA Cooperative Agreement NA22OAR4320151.
Data availability statement.
NCEP operational products are accessible to general users, free of charge in real time at NOAA Operational Model Archive and Distribution System (NOMADS). The NCEP Central Operations (NCO) ftp site provides the source code, relevant run scripts, and fixed fields files. Model outputs are in Gridded Binary Version 2 (GRIB2) format on a grid, with 3-hourly output up to 120 h. Operational GEFS-Aerosols products from NOMADS are available at http://nomads.ncep.noaa.gov/pub/data/nccf/com/gens/prod (last access: March 2022). The NCAR Command Language (NCL) program is used to generate all the figures in this paper (https://www.ncl.ucar.edu/, last access: March 2022).
APPENDIX A
Verification Measures
APPENDIX B
REFERENCES
Ahmadov, R., and Coauthors, 2017: Using VIIRS fire radiative power data to simulate biomass burning emissions, plume rise and smoke transport in a real-time air quality modeling system. IEEE Int. Geoscience and Remote Sensing Symp. (IGARSS), Fort Worth, TX, Institute of Electrical and Electronics Engineers, 2806–2808, https://doi.org/10.1109/IGARSS.2017.8127581.
Almeida-Silva, M., S. M. Almeida, M. C. Freitas, C. A. Pio, T. Nunes, and J. Cardoso, 2013: Impact of Sahara dust transport on Cape Verde atmospheric element particles. J. Toxicol. Environ. Health, 76, 240–251, https://doi.org/10.1080/15287394.2013.757200.
Bellouin, N., and Coauthors, 2020: Bounding global aerosol radiative forcing of climate change. Rev. Geophys., 58, e2019RG000660, https://doi.org/10.1029/2019RG000660.
Benedetti, A., J. S. Reid, and P. R. Colarco, 2011: International cooperative for aerosol prediction workshop on aerosol forecast verification. Bull. Amer. Meteor. Soc., 92, ES48–ES53, https://doi.org/10.1175/BAMS-D-11-00105.1.
Bian, Q., B. Ford, J. R. Pierce, and S. M. Kreidenweis, 2020: A decadal climatology of chemical, physical, and optical properties of ambient smoke in the western and southeastern United States. J. Geophys. Res. Atmos., 125, e2019JD031372, https://doi.org/10.1029/2019JD031372.
Boer, M. M., V. Resco de Dios, and R. A. Bradstock, 2020: Unprecedented burn area of Australian mega forest fires. Nat. Climate Change, 10, 171–172, https://doi.org/10.1038/s41558-020-0716-1.
Bosilovich, M. G., R. Lucchesi, and M. Suarez, 2016: MERRA-2: File specification. GMAO Office Note No. 9 (version 1.1), 73 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Bosilovich785.pdf.
Boucher, O., and Coauthors, 2013: Clouds and aerosols. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 571–657.
Brewer, M. J., and C. B. Clements, 2020: The 2018 camp fire: Meteorological analysis using in situ observations and numerical simulations. Atmosphere, 11, 47, https://doi.org/10.3390/atmos11010047.
Brown, B., and Coauthors, 2021: The Model Evaluation Tools (MET): More than a decade of community-supported forecast verification. Bull. Amer. Meteor. Soc., 102, E782–E807, https://doi.org/10.1175/BAMS-D-19-0093.1.
Buchard, V., and Coauthors, 2017: The MERRA-2 aerosol reanalysis, 1980 onward. Part II: Evaluation and case studies. J. Climate, 30, 6851–6872, https://doi.org/10.1175/JCLI-D-16-0613.1.
CAL FIRE, 2020: Wildfire activity statistics. California Department of Forestry and Fire Protection, 53 pp., https://www.fire.ca.gov/media/0fdfj2h1/2020_redbook_final.pdf.
Chin, M., and Coauthors, 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements. J. Atmos. Sci., 59, 461–483, https://doi.org/10.1175/1520-0469(2002)059<0461:TAOTFT>2.0.CO;2.
Chylek, P., and J. Wong, 1995: Effect of absorbing aerosols on global radiation budget. Geophys. Res. Lett., 22, 929–931, https://doi.org/10.1029/95GL00800.
Colarco, P., A. da Silva, M. Chin, and T. Diehl, 2010: Online simulations of global aerosol distributions in the NASA GEOS-4 model and comparisons to satellite and ground-based aerosol optical depth. J. Geophys. Res., 115, D14207, https://doi.org/10.1029/2009JD012820.
Cusworth, D. H., L. J. Mickley, M. P. Sulprizio, T. Liu, M. E. Marlier, R. S. DeFries, S. K. Guttikinds, and P. Gupta, 2018: Quantifying the influence of agricultural fires in northwest India on urban air pollution in Delhi, India. Environ. Res. Lett., 13, 044018, https://doi.org/10.1088/1748-9326/aab303.
Darmenov, A. S., and A. da Silva, 2015: The Quick Fire Emissions Dataset (QFED)—Documentation of versions 2.1, 2.2, and 2.4. Tech. Rep. NASA/TM-2015-104606/Vol. 38, NASA, 212 pp.
Das, S., P. R. Colarco, L. D. Oman, G. Taha, and O. Torres, 2021: The long-term transport and radiative impacts of the 2017 British Columbia pyrocumulonimbus smoke aerosols in the stratosphere. Atmos. Chem. Phys., 21, 12 069–12 090, https://doi.org/10.5194/acp-21-12069-2021.
Dennison, P. E., S. C. Brewer, J. D. Arnold, and M. A. Moritz, 2014: Large wildfire trends in the western United States 1984–2011. Geophys. Res. Lett., 41, 2928–2933, https://doi.org/10.1002/2014GL059576.
Dong, X., J. S. Fu, K. Huang, D. Tong, and G. Zhuang, 2016: Model development of dust emission and heterogeneous chemistry within the community multiscale air quality modeling system and its application over East Asia. Atmos. Chem. Phys., 16, 8157–8180, https://doi.org/10.5194/acp-16-8157-2016.
Freitas, S. R., and Coauthors, 2007: Including the sub-grid scale plume rise of vegetation fires in low resolution atmospheric transport models. Atmos. Chem. Phys., 7, 3385–3398, https://doi.org/10.5194/acp-7-3385-2007.
Freitas, S. R., K. M. Longo, J. Trentmann, and D. Latham, 2010: Technical note: Sensitivity of 1-D smoke plume rise models to the inclusion of environmental wind drag. Atmos. Chem. Phys., 10, 585–594, https://doi.org/10.5194/acp-10-585-2010.
Frey, L., F. Höpner, A. Kirkevåg, and F. A.-M. Bender, 2021: Absorbing aerosols over Asia—An inter-model and model-observation comparison study using CAM5.3-Oslo. Tellus, 73B, 1–25, https://doi.org/10.1080/16000889.2021.1909815.
Fu, R., A. Hoell, J. Mankin, A. Sheffield, and I. Simpson, 2021: Tackling challenges of a drier, hotter, more fire-prone future. Eos, Trans. Amer. Geophys. Union, 102, https://doi.org/10.1029/2021EO156650.
Gani, S., and Coauthors, 2019: Submicron aerosol composition in the world’s most polluted megacity: The Delhi aerosol supersite study. Atmos. Chem. Phys., 19, 6843–6859, https://doi.org/10.5194/acp-19-6843-2019.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
GMAO, 2015: MERRA-2 tavg1_2d_aer_Nx: 2d, 1-Hourly, Time-averaged, Single-Level, Assimilation, Aerosol Diagnostics V5.12.4 (M2T1NXAER). Goddard Earth Sciences Data and Information Services Center (GES DISC), accessed 7 October 2022, https://doi.org/10.5067/KLICLTZ8EM9D.
Grell, G., S. R. Freitas, M. Steufer, and J. Fast, 2011: Inclusion of biomass burning in WRF-CHEM: Impact of wildfires on weather forecasts. Atmos. Chem. Phys., 11, 5289–5303, https://doi.org/10.5194/acp-11-5289-2011.
Han, J., and H.-L. Pan, 2011: Revision of convection and vertical diffusion schemes in the NCEP Global Forecast System. Wea. Forecasting, 26, 520–533, https://doi.org/10.1175/WAF-D-10-05038.1.
Hess, M., P. Koepke, and I. Schult, 1998: Optical properties of aerosols and clouds: The software package OPAC. Bull. Amer. Meteor. Soc., 79, 831–844, https://doi.org/10.1175/1520-0477(1998)079<0831:OPOAAC>2.0.CO;2.
Hoesly, R. M., and Coauthors, 2018: Historical (1750–2014) anthropogenic emissions of reactive gases and aerosols from the Community Emissions Data System (CEDS). Geosci. Model. Dev., 11, 369–408, https://doi.org/10.5194/gmd-11-369-2018.
Holben, B. N., and Coauthors, 1998: AERONET—A federated instrument network and data archive for aerosol characterization. Remote Sens. Environ., 66, 1–16, https://doi.org/10.1016/S0034-4257(98)00031-5.
Hollingsworth, A., and Coauthors, 2008: Toward a monitoring and forecasting system for atmospheric composition: The GEMS Project. Bull. Amer. Meteor. Soc., 89, 1147–1164, https://doi.org/10.1175/2008BAMS2355.1.
Hsu, N. C., M.-J. Jeong, C. Bettenhausen, A. M. Sayer, R. Hansell, C. S. Seftor, J. Huang, and S.-C. Tsay, 2013: Enhanced deep blue aerosol retrieval algorithm: The second generation. J. Geophys. Res. Atmos., 118, 9296–9315, https://doi.org/10.1002/jgrd.50712.
Huang, J., T. Wang, W. Wang, Z. Li, and H. Yan, 2014: Climate effects of dust aerosols over East Asian arid and semiarid regions. J. Geophys. Res. Atmos., 119, 11 398–11 416, https://doi.org/10.1002/2014JD021796.
IPCC, 2021: Climate Change 2021: The Physical Science Basis. Cambridge University Press, 2391 pp.
Jethva, H., and Coauthors, 2018: Agricultural burning and air quality over Northern India: A synergistic analysis using NASA’s A-train satellite data and ground measurements. Aerosol Air Qual. Res., 18, 1756–1773, https://doi.org/10.4209/aaqr.2017.12.0583.
Kaskaoutis, D. G., and Coauthors, 2014: Effects of crop residue burning on aerosol properties, plume characteristics, and long-range transport over northern India. J. Geophys. Res. Atmos., 119, 5424–5444, https://doi.org/10.1002/2013JD021357.
Kelly, F. J., and F. J. Fussell, 2015: Air pollution and public health: Emerging hazards and improved understanding of risk. Environ. Geochem. Health, 37, 631–649, https://doi.org/10.1007/s10653-015-9720-1.
Khaykin, S., and Coauthors, 2020: The 2019/20 Australian wildfires generated a persistent smoke-charged vortex rising up to 35 km altitude. Commun. Earth Environ., 1, 22, https://doi.org/10.1038/s43247-020-00022-5.
Kleist, D. T., D. F. Parrish, J. C. Derber, R. Treadon, W.-S. Wu, and S. Lord, 2009: Introduction of the GSI into the NCEP global data assimilation system. Wea. Forecasting, 24, 1691–1705, https://doi.org/10.1175/2009WAF2222201.1.
Kulkarni, P., and V. Sreekanth, 2020: Spaceborne lidar retrieved composite and speciated aerosol extinction profiles and optical depths over India: A decade of observations. Atmos. Pollut. Res., 11, 946–962, https://doi.org/10.1016/j.apr.2020.02.007.
Levy, R. C., S. Mattoo, L. A. Munchak, L. A. Remer, A. M. Sayer, F. Patadia, and N.C. Hsu, 2013: The collection 6 MODIS aerosol products over land and ocean. Atmos. Meas. Tech., 6, 2989–3034, https://doi.org/10.5194/amt-6-2989-2013.
Li, J., B. E. Carlson, and A. A. Lacis, 2015: How well do satellite AOD observations represent the spatial and temporal variability of PM2.5 concentration for the United States? Atmos. Environ., 102, 260–273, https://doi.org/10.1016/j.atmosenv.2014.12.010.
Liu, H., L. A. Remer, J. Huang, H.-C. Huang, S. Kondragunta, I. Laszlo, M. Oo, and J. M. Jackson, 2014: Preliminary evaluation of S-NPP VIIRS aerosol optical thickness. J. Geophys. Res. Atmos., 119, 3942–3962, https://doi.org/10.1002/2013JD020360.
Liu, T., and Coauthors, 2018: Seasonal impact of regional outdoor biomass burning on air pollution in three Indian cities: Delhi, Bengaluru, and Pune. Atmos. Environ., 172, 83–92, https://doi.org/10.1016/j.atmosenv.2017.10.024.
Loría-Salazar, S. M., A. Panorska, W. P. Arnott, J. C. Barnard, J. M. Boehmler, and H. A. Holmes, 2017: Toward understanding atmospheric physics impacting the relationship between columnar aerosol optical depth and near-surface PM2.5 mass concentrations in Nevada and California, U.S.A., during 2013. Atmos. Environ., 171, 289–300, https://doi.org/10.1016/j.atmosenv.2017.10.023.
Lu, C.-H., and Coauthors, 2016: The implementation of NEMS GFS aerosol component (NGAC) version 1.0 for global dust forecasting at NOAA/NCEP. Geosci. Model Dev., 9, 1905–1919, https://doi.org/10.5194/gmd-9-1905-2016.
Menon, S., J. Hansen, L. Nazarenko, and Y. Luo, 2002: Climate effects of black carbon aerosols in China and India. Science, 297, 2250–2253, https://doi.org/10.1126/science.1075159.
Molod, A., L. Takas, M. Suarez, and J. Bacmeister, 2015: Development of the GEOS-5 atmospheric general circulation model: Evolution from MERRA to MERRA-2. Geosci. Model Dev., 8, 1339–1356, https://doi.org/10.5194/gmd-8-1339-2015.
Morcrette, J.-J., A. Beljaars, A. Benedetti, L. Jones, and O. Boucher, 2008: Sea-salt and dust aerosols in the ECMWF IFS model. Geophys. Res. Lett., 35, L24813, https://doi.org/10.1029/2008GL036041.
Mulcahy, J. P., D. N. Walters, N. Bellouin, and S. F. Milton, 2014: Impacts of increasing the aerosol complexity in the Met Office global numerical weather prediction model. Atmos. Chem. Phys., 14, 4749–4778, https://doi.org/10.5194/acp-14-4749-2014.
Murphy, A. H., and R. L. Winkler, 1987: A general framework for forecast verification. Mon. Wea. Rev., 115, 1330–1338, https://doi.org/10.1175/1520-0493(1987)115<1330:AGFFFV>2.0.CO;2.
Myhre, G., and Coauthors, 2013: Radiative forcing of the direct aerosol effect from AeroCom Phase II simulations. Atmos. Chem. Phys., 13, 1853–1877, https://doi.org/10.5194/acp-13-1853-2013.
Nandi, I, S. Srivastava, Y. Yarragunta, R. Kumar, and D. Mitra, 2020: Distribution of surface carbon monoxide over the Indian subcontinent: Investigation of source contributions using WRF-Chem. Atmos. Environ., 243, 117838, https://doi.org/10.1016/j.atmosenv.2020.117838.
NASA EOS imagery, 2020: Dust blankets the Canary Islands. NASA, accessed 20 March 2022, https://earthobservatory.nasa.gov/images/146337/dust-blankets-the-canary-islands.
NIFC, 2020: National large incident year-to-date report. NIFC, 66 pp., https://web.archive.org/web/20201229021815/https://gacc.nifc.gov/sacc/predictive/intelligence/NationalLargeIncidentYTDReport.pdf.
Nisantzi, A., R. E. Mamouri, A. Ansmann, and D. Hadjimitsis, 2014: Injection of mineral dust into the free troposphere during fire events observed with polarization Lidar at Limassol, Cyprus. Atmos. Chem. Phys., 14, 12 155–12 165, https://doi.org/10.5194/acp-14-12155-2014.
Nowottnick, E., P. Colarco, A. da Silva, D. Hlavka, and M. McGill, 2011: The fate of Saharan dust across the Atlantic and implications for a Central American dust barrier. Atmos. Chem. Phys., 11, 8415–8431, https://doi.org/10.5194/acp-11-8415-2011.
Ohneiser, K., and Coauthors, 2022: Australian wildfire smoke in the stratosphere: The decay phase in 2020/21 and impact on ozone depletion. Atmos. Chem. Phys., 22, 7417–7442, https://doi.org/10.5194/acp-22-7417-2022.
Owen, P. R., 1964: Saltation of uniform grains in air. J. Fluid Mech., 20, 225–242, https://doi.org/10.1017/S0022112064001173.
Peterson, D. A., J. R. Campbell, E. J. Hyer, M. D. Fromm, G. P. Kablick, J. H. Cossuth, and M. T. DeLand, 2018: Wildfire-driven thunderstorms cause a volcano-like stratospheric injection of smoke. npj Climate Atmos. Sci., 1, 30, https://doi.org/10.1038/s41612-018-0039-3.
Reale, O., K. M. Lau, and A. da Silva, 2011: Impact of interactive aerosol on the African easterly jet in the NASA GEOS-5 global forecasting system. Wea. Forecasting, 26, 504–519, https://doi.org/10.1175/WAF-D-10-05025.1.
Rienecker, M. M., and Coauthors, 2008: The GEOS-5 Data Assimilation System—Documentation of versions 5.0.1, 5.1.0, and 5.2.0. Tech. Memo. NASA/TM-2008-104606/Vol. 27, 101 pp., https://sciences.gsfc.nasa.gov/earth/gmao/content/uploadFiles/highlight_files/GEOS5_104606-Vol27.pdf.
Rodríguez, S., E. Cuevas, J. M. Prospero, A. Alastuey, X. Querol, J. López-Solano, M. I. García, and S. Alonso-Pérez, 2015: Modulation of Saharan dust export by the North African dipole. Atmos. Chem. Phys., 15, 7471–7486, https://doi.org/10.5194/acp-15-7471-2015.
Sahu, S. K., P. Mangaraj, G. Beig, A. Samal, C. Pradhan, S. Dash, and B. Tyagi, 2021: Quantifying the high resolution seasonal emission of air pollutants from crop residue burning in India. Environ. Pollut., 286, 117165, https://doi.org/10.1016/j.envpol.2021.117165.
Sayer, A. M., N. C. Hsu, C. Bettenhausen, and M.-J. Jeong, 2013: Validation and uncertainty estimates for MODIS Collection 6 “Deep Blue” aerosol data. J. Geophys. Res. Atmos., 118, 7864–7872, https://doi.org/10.1002/jgrd.50600.
Sayer, A. M., L. A. Munchak, N. C. Hsu, R. C. Levy, C. Bettenhausen, and M.-J. Jeong, 2014: MODIS Collection 6 aerosol products: Comparison between Aqua’s e-Deep blue, dark target and “merged” data sets and usage recommendations. J. Geophys. Res. Atmos., 119, 13 965–13 989, https://doi.org/10.1002/2014JD022453.
Seinfeld, J., and S. Pandis, 1998: Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. 1st ed. John Wiley and Sons, 1326 pp.
Sinyuk, A., and Coauthors, 2020: The AERONET version 3 aerosol retrieval algorithm, associated uncertainties and comparisons to version 2. Atmos. Meas. Tech., 13, 3375–3411, https://doi.org/10.5194/amt-13-3375-2020.
Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 7183–7192, https://doi.org/10.1029/2000JD900719.
Toll, V., E. Gleeson, K. P. Nielsen, A. Männik, J. Mašek, L. Rontu, and P. Post, 2016: Impacts of the direct radiative effect of aerosols in numerical weather prediction over Europe using the ALADIN-HIRLAM NWP system. Atmos. Res., 172–173, 163–173, https://doi.org/10.1016/j.atmosres.2016.01.003.
Tompkins, A. M., C. Cardinali, J.-J. Morcrette, and M. Rodwell, 2005: Influence of aerosol climatology on forecasts of the African easterly jet. Geophys. Res. Lett., 32, L10801, https://doi.org/10.1029/2004GL022189.
Twomey, S., 1974: Pollution and the planetary albedo. Atmos. Environ., 8, 1251–1256, https://doi.org/10.1016/0004-6981(74)90004-3.
Wang, J., and Coauthors, 2018: The implementation of NEMS GFS aerosol component (NGAC) version 2.0 for global multispecies forecasting at NOAA/NCEP—Part 1: Model descriptions. Geosci. Model Dev., 11, 2315–2332, https://doi.org/10.5194/gmd-11-2315-2018.
Wilkins, J. L., B. de Foy, A. M. Thompson, D. A. Peterson, E. J. Hyer, C. Graves, J. Fishman, and G. A. Morris, 2020: Evaluation of stratospheric intrusions and biomass burning plumes on the vertical distribution of tropospheric ozone over the Midwestern United States. J. Geophys. Res. Atmos., 125, e2020JD032454, https://doi.org/10.1029/2020JD032454.
WMO, 2021: Desert dust outbreak in the Canary Islands (February 2020): Assessment and impacts. World Meteorological Organization GAW Rep. 259, 123 pp., https://library.wmo.int/index.php?lvl=notice_display&id=21852#.YaZPA7pOnIU.
Wu, W.-S., R. J. Purser, and D. F. Parrish, 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 2905–2916, https://doi.org/10.1175/1520-0493(2002)130<2905:TDVAWS>2.0.CO;2.
Yu, H., and Coauthors, 2021: Observation and modeling of the historic “Godzilla” African dust intrusion into the Caribbean Basin and the southern U.S. in June 2020. Atmos. Chem. Phys., 21, 12 359–12 383, https://doi.org/10.5194/acp-21-12359-2021.
Zhang, L., and Coauthors, 2022a: Development and evaluation of the aerosol forecast member in the National Center for Environment Prediction (NCEP)’s Global Ensemble Forecast System (GEFS-aerosols v1). Geosci. Model Dev., 15, 5337–5369, https://doi.org/10.5194/gmd-15-5337-2022.
Zhang, L., G. A. Grell, S. A. McKeen, R. Ahmadov, K. D. Froyd, and D. Murphy, 2022b: Inline coupling of simple and complex chemistry modules within the global forecast model FIM (FIM-Chem v1). Geosci. Model. Dev., 15, 467–491, https://doi.org/10.5194/gmd-15-467-2022.
Zhang, X., S. Kondragunta, J. Ram, C. Schmidt, and H.-C. Huang, 2012: Near real time global biomass burning emissions product from geostationary satellite constellation. J. Geophys. Res., 117, D14201, https://doi.org/10.1029/2012JD017459.
Zhang, Y., X.-Y. Wen, and C.-J. Jang, 2010: Simulating chemistry-aerosol-cloud-radiation-climate feedbacks over the continental U.S. using the online-coupled Weather Research Forecasting model with Chemistry (WRF-Chem). Atmos. Environ., 44, 3568–3582, https://doi.org/10.1016/j.atmosenv.2010.05.056.
Zheng, B., Q. Zhang, G. Geng, C. Chen, Q. Shi, M. Cui, Y. Lei, and K. He, 2021: Changes in China’s anthropogenic emissions and air quality during the COVID-19 pandemic in 2020. Earth Syst. Sci. Data, 13, 2895–2907, https://doi.org/10.5194/essd-13-2895-2021.
Zhou, X., and Coauthors, 2022: The development of the NCEP Global Ensemble Forecast System version 12. Wea. Forecasting, 37, 1069–1084, https://doi.org/10.1175/WAF-D-21-0112.1.
Zhuang, Y., R. Fu, B. D. Santer, R. E. Dickinson and A. Hall, 2021: Quantifying contributions of natural variability and anthropogenic forcings on increased fire weather risk over the western United States. Proc. Natl. Acad. Sci. USA, 118, e2111875118, https://doi.org/10.1073/pnas.2111875118.