Alaska WEP/AOI and PSC Analysis

December 17, 2020

An Area of Influence (AOI) and Weighted Emissions Potential (WEP) analysis was performed for the IMPROVE sites in Alaska to identify the anthropogenic sources of emissions within and nearby the state that had the potential to contribute the most to visibility impairment on the Most Impaired Days (MID) from 2014 to 2018 at Class I Areas (CIA) in the state. A Potential Source Contributions (PSC) analysis was also performed to characterize the relative potential contributions of natural (e.g., volcano) and anthropogenic (e.g., on-road mobile sources) emission sources groups to the ammonium sulfate (Amm_SO4) extinction on the MID. The input data, methods, and resulting data products for the WEP/AOI and PSC analyses are described separately in the following sections.  Although the procedures used to conduct the Alaska WEP/AOI analysis of anthropogenic emissions and PSC analysis of natural and anthropogenic SOx emissions are similar, they are very different analysis and need to be viewed separately.

The CIA and representative IMPROVE monitors included in the analyses are provided in Table 1. No IMPROVE monitoring data is available for the Bering Sea Wilderness CIA, and so it was not included in the analysis. The Tuxedni IMPROVE site (TUXE1) stopped operating in 2014 so the MID from 2012 to 2014 were used. The Kenai Peninsula Borough (KPBO1) site was added across the Cook Inlet from TUXE1 in 2016 to replace TUXE1, but KPBO1 could not be included in the WEP/AOI nor PSC analysis as no MID metric data is available for the site. Instead, an AOI and WEP analysis was performed for the 20% highest measured visibility extinction for ammonium sulfate (Amm_SO4), ammonium nitrate (Amm_NO3), and coarse mass (CM) at TUXE1 and KPBO1 for the 3 most recent years of available data (2012 to 2014 and 2016 to 2018, respectively).

Weighted Emissions Potential (WEP)/Area of Influence (AOI) Analysis

Input Data and Analysis Domain

IMPROVE Data for Most Impaired Days

The MID at DENA1, TRCR1, SIME1, and TUXE1 were identified as those assigned to impairment group 90 in the IMPROVE Impairment Daily Budgets dataset. The impairment group metric includes days in which missing data was patched using statistical approaches. The 20% highest Amm_SO4, Amm_NO3, and CM days at TUXE1 and KPBO1 were identified using the IMPROVE Daily Budgets dataset. Days with patched or substituted data for the species being examined were not included when determining the 20% highest Amm_SO4, Amm_NO3 and CM extinction days.

HYSPLIT Back Trajectory Modeling

The Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Stein et al., 2015; Rolph et al., 2017) was used to calculate 72-hour back trajectories arriving on each of the MID at four times per day (6:00, 12:00, 18:00, 24:00 local standard time) and at four heights above the ground (100 meter (m), 200 m, 500 m and 1,000 m). The archived NAM hybrid sigma-pressure gridded (NAMS) meteorological data for Alaska was downloaded for 2012 to 2018 from the NOAA Air Resources Laboratory FTP server for use in HYSPLIT model. The NAMS data is output hourly and covers Alaska at 12 km resolution. The HYPSLIT model was configured to provide trajectory endpoints every 10 minutes of the simulation period.

Analysis Domain

The analysis performed using the 9-kilometer (km) domain of EPA’s Regional Haze Modeling platform for Alaska aggregated to 27-km resolution. Trajectory endpoints located outside the EPA 9-km domain were dropped from the analysis. The analysis domain and IMPROVE sites are shown in Figure 1.

Emissions

The WEP analysis was performed using both gridded emissions from the EPA’s 2016 Alaska modeling platform and 2014 and 2017 facility-level National Emissions Inventory data provided by ADEC in a May 20, 2020 e-mail from Molly Birnbaum. The EPA’s gridded emissions for 2016 were aggregated into the following source sectors for the WEP analysis:

• TOTAL_ANTHRO – All anthropogenic emissions
• PT_EGU – Electric generating unit emissions
• PT_NON-EGU – Point source emissions from industrial activities
• OG_AREA_POINT – Oil and Gas area and point sources (Upstream and Midstream)
• NON-POINT – Low-level area source emissions including non-point, agricultural, residential wood combustion, and fugitive dust emissions
• ON-ROAD – On-road mobile source emissions
• NON-ROAD – Off highway mobile source emissions including non-road, airport, commercial marine (C1, C2, and C3), and rail sources

For the MID analysis, EPA 2016 gridded emissions of nitrogen oxides (NOx), sulfur oxides (SOx), primary organic aerosol (POA), and primary elemental carbon (PEC) were used for the analysis of Amm_NO3, Amm_SO4, OC, and EC, respectively. The analysis of the facility-level WEP on the MID used point source emissions of NOx and SO₂.

The gridded WEP and AOI analysis of the 20% highest extinction days of Amm_NO3, Amm_SO4, and CM at TUXE1 and KPBO1 was conducted in the same manner as for the MID described above only using EPA 2016 emissions of NOx, SOx, and coarse primary mass (CPRM), respectively.  The facility level emissions analysis was based on 2014 and 2017 point source emissions of NOx, SO₂, and PM₁₀ for the TUXE1 and KPBO1 sites, respectively.

WEP/AOI Products

The WEP/AOI analysis products are provided for the 100 m and 1000 m trajectory heights and for a combined analysis in which data from all four trajectory heights are aggregated (All). The products include:

• Plots of residence time (RT), extinction weighted residence time (EWRT), and WEP for each CIA
• Plots of the gridded 2016 EPA emissions used in the WEP analysis for each of the source sectors described above (EMISSIONS)
• Excel spreadsheets of facility-level 2014 and 2017 precursor emissions and the corresponding WEP at each CIA (RANK_POINT)

The RT, EWRT, WEP, and EMISSIONS plots are provided at the following link using the directory structure shown below with TUXE1 as an example. The analysis plots for the 20% highest Amm_SO4 (20pctHigh_SO4) and CM (20pctHigh_CM) at TUXE1 and KPBO1 are provided using the same directory structure as 20% highest nitrate extinction days (20pctHigh_NO3) shown below. Descriptions of each analysis product are provided in the following sections.

AOI, WEP, and Gridded Emissions Plots

Shapefiles of the RT, EWRT, and WEP results by source sector are provided for the MID and 20 percent highest Amm_NO3, Amm_SO4 and CM days at the link below (WEP_SHP). In each CIA subdirectory there is a zip file for each source sector containing the ESRI shapefile (.shp) and related files (.cpg, .dbf, .prj, .shx). The RT, EWRT, and WEP results data are available within the attribute tables of the shapefiles provided which can be accessed in ArcMap or ArcCatalog. The attribute tables can be exported to csv, then opened in Excel. Files are provided for the 100m and 1000m trajectory heights and for a combined analysis in which data from all four trajectory heights are aggregated (All). The directory structure for the WEP_SHP is shown below using the TUXE1 results aggregated across all trajectory heights as an example. The shapefiles for 20pctHigh_SO4 and 20pctHigh_CM at TUXE1 and KPBO1 are provided using the same directory structure as the 20pctHigh_NO3.

WEP/AOI Shapefiles

The facility-level RANK_POINT results for the MID and 20 percent highest Amm_NO3, Amm_SO4 and CM days are provided at the link below. The results are provided in Excel spreadsheets with the 2014 and 2017 results provided in separate tabs. The directory structure is shown below using TUXE1 as an example.

RANK_POINT Spreadsheets

Residence Time (RT):  The RT folder contains plots showing the AOI that back trajectories of air parcels traveling from a given location arrived at the IMPROVE monitor on the 2014-2018 MID.  HYSPLIT 72-hour back trajectories are calculated to arrive on each of the MID four times a day (6:00, 12:00, 18:00, 24:00 local standard time) and at four heights above the ground (100 m, 200 m, 500 m and 1,000 m). Plots are provided for the 100m and 1000m heights and for a combined analysis in which data from all trajectory heights are aggregated (All). RT plots for the MID at DENA1 are provided as an example in Figure 2.

Extinction Weighted Residence Time (EWRT):  EWRT are calculated by weighting the HYSPLIT trajectories by the monitored extinction at the IMPROVE site on each MID. Plots of EWRT for Amm_SO4, Amm_NO3, OA and EC are provided for the MID analysis, while plots of EWRT for Amm_SO4, Amm_NO3, and CM are provided for the 20% highest extinction day analysis of TUXE1 and KPBO1.  Figure 3 shows the EWRT plots for sulfate and nitrate at DENA1 on the MID using the aggregated trajectory height analysis (All) as an example.

Weighted Emissions Potential (WEP):  WEP is obtained by overlaying the EWRT results with the EPA 2016 emissions of light extinction precursors (e.g., NOx emissions for ammonium nitrate extinction) in each grid cell divided by the distance of the source to the IMPROVE monitor. The gridded WEP values for each source sector are then normalized by the sum of the WEP for the total anthropogenic emissions (TOTAL_ANTHRO) across all grid cells. Using the sum of the total anthropogenic values as a common denominator allows for the WEP results to be compared across source sectors. The WEP folder for the MID contains four subfolders corresponding to the precursor emissions for four major components of light extinction: EC (Elemental Carbon), NOx (Ammonium Nitrate), POA (Organic Aerosol) and SOx (Ammonium Sulfate).  The WEP folder for the 20% percent highest extinction days for TUXE1 and KPBO1 contain subfolders for NOx, SOx, and coarse primary mass (CPRM). Each precursor species subfolder contains 21 plots in which the EWRT at three heights above ground are overlaid with 2016 emissions from 7 gridded Source Sectors (21 = 3 x 7). The source sectors are described in the Emissions section above, and example plots for DENA1 on MID are shown in Figure 4.  The dark green and light green isopleths in the WEP plots correspond to the, respectively, 0.5 and 0.1 percent frequency from the corresponding EWRT results.

As described above, shapefiles of the WEP results (along with RT and EWRT) for each CIA and source sector are also provided. The columns of the attribute tables of these files are described in Table 2 below.

Rank Point (RANK_POINT):  The RANK_POINT spreadsheets consist of facility level 2014 and 2017 precursor emissions overlaid with the corresponding EWRT for 3 trajectory height scenarios (100m, 1000m and All).  There is a separate directory for each scenario (MID, 20pctHigh_NO3, 20pctHigh_SO4, 20pctHigh_CM) containing the various RANK_POINT Excel spreadsheets. The results for 2014 and 2017 are provided in separate tabs.

These results can be used to assess the potential contributions of specific facilities to visibility impairment at each Class I Area in various ways. We recommend sorting the facilities by the WEP metrics (e.g. WEP_NO3 column for Amm_NO3 extinction). These metrics account the air parcel trajectories and extinction on the MID, facility precursor emissions, and facility distance from the IMPROVE monitor. The columns of the RANK_POINT sheets are described in Table 3, and the other metrics provided in the RANK_POINT files are described below.

The QD columns provide the simple emissions over distance metric (Q/D) using the 2014 and 2017 facility-level precursor emissions.  While the Q/D metric can be used to screen the potential contributions of sources to visibility impairment at a given Class I area, it is simply based on the emission rate and distance from the IMPROVE monitor and does not account for the air parcel trajectories on the MID.

The EWRTxQ and WEP metrics do account for the air parcel trajectories on MID as they are calculated by multiplying the EWRT of the grid cell of the facility with the facility’s emission rate (Q) and Q/D, respectively.

Table 4 shows a subset of the results for 2014 NOx sources at DENA1 on the MID aggregated across all trajectory heights. The sources are ranked by [WEP_NO3] (last column) for ammonium nitrate, and the top 30 facilities whose NOx emissions potentially contribute to visibility impairment on the 2014-2018 IMPROVE MID at DENA1 are shown.

Potential Source Contributions (PSC) Analysis

A PSC analysis was performed to assess the relative potential contributions of anthropogenic and natural emission source groups within the EPA 27-km Alaska modeling domain to Amm_SO4 extinction on the MID at each CIA. PSC was calculated by integrating (i.e., summing) the WEP across the modeling domain for each source group.

Unlike the WEP analysis, which only considered anthropogenic emission sources, the PSC analysis also included volcanic emissions of sulfur dioxide (SO₂) and oceanic emissions of dimethyl sulfide (DMS). An analysis of 2014 GEOS-Chem emissions for a region essentially equivalent to EPA’s CMAQ Alaska 27-km domain found that ~70% of the reactive sulfur emissions were from volcano degassing and DMS (Guo and Morris, 2020). Including these sources in the PSC allows for characterization of potential natural contributions to visibility impairment on the MID and provides context for the potential anthropogenic source contributions.

Input Data and Analysis Domain

IMPROVE Data for Most Impaired Days

The MID at DENA1, TRCR1, SIME1, and TUXE1 for the site-specific analysis periods provided in Table 1 were identified as those assigned to impairment group 90 in the IMPROVE Impairment Daily Budgets dataset. The impairment group metric includes days in which missing data was patched using the statistical approaches. KPBO1 could not be included in the PSC analysis as no MID impairment metric data is available for the site.

Analysis Domain

The PSC analysis was performed using the 27-kilometer (km) domain of EPA’s Regional Haze Modeling platform. Trajectory endpoints located outside the EPA 27-km domain were dropped from the analysis. This is different than the WEP/AOI analysis which used the extent of the EPA 9-km domain at 27-km resolution. Figure 5 displays the 27-km and 9-km domains used in EPA’s CMAQ modeling and the, respectively, PSC and WEP/AOI analysis.

HYSPLIT Back Trajectory Modeling

The HYSPLIT back trajectories generated for the WEP/AOI analysis were used to calculate PSC. The HYSPLIT configuration and meteorology data used are document in the WEP/AOI section above.

Emissions

The PSC analysis for Amm_SO4 included both anthropogenic and natural emission sources of sulfur oxides (SOx). Anthropogenic emissions are from the EPA’s 2016 27-km modeling platform and were aggregated into the following source sectors for the PSC analysis:

• PT_EGU – Electric generating unit emissions
• PT_NON-EGU – Point source emissions from industrial activities
• OG_AREA_POINT – Oil and Gas area and point sources (Upstream and Midstream)
• NON-POINT – Low-level area source emissions including non-point, agricultural, residential wood combustion, and fugitive dust emissions
• ON-ROAD – On-road mobile source emissions
• NON-ROAD – Off highway mobile source emissions including non-road, airport, commercial marine (C1, C2, and C3), and rail sources
• International – point, area, on road and fugitive dust sources from Canada. No emissions from Russia were included in the 2016 EPA modeling platform

In reviewing the results of the gridded WEP/AOI analysis ADEC noticed that the EPA modeling platform did not include emissions from the Healy Power Plant. For the PSC analysis, ADEC provided the reported SO₂ emissions for Healy in 2016 (427.2 tons per year) and these emissions were added to the PT_EGU sector.

Annual SO₂ emissions from volcanoes within the analysis domain were estimated using NASA’s satellite derived volcanic emissions inventory. The annual emissions were averaged across the analysis period for each IMPROVE site (see Table 1) for the PSC analysis. DMS emissions were estimated for the year 2016 using monthly climatologies of surface ocean DMS concentration and sea-to-air emission flux. The DMS emissions were scaled by a 0.6 factor to account for the amount of DMS that is likely ultimately oxidized to SO2/SO4 based on the work of Chen et al., (2018) who estimate that approximately 75% of the DMS is converted to SO2 and Veres et al., (2020) who found a new DMS product species (HPMTF) that results in approximately 25% of the DMS oxidation products in the Gulf of Alaska (0.6 ~ 0.75 x 0.75). Table 5 summarizes the total SO2 or SO2 equivalent (i.e., DMS) emissions within the 27-km domain for the various source sectors.  The anthropogenic emissions are from EPA’s 2016 CMAQ modeling, DMS was calculated using 2016 meteorology and volcanic emissions were based on satellite inventories for 2014-2018.  The volcanic and DMS natural emissions contribute 83% of the SO2 emissions within the 27-km CMAQ domain.  This is higher percentage of natural SO2 emissions than the 71% contribution estimated analyzing 2014 GEOS-Chem inventories for a similar size domain as documented in a June 30, 2020 Memorandum to the Alaska Department of Environmental Conservation. These differences are due in part to the CMAQ 2016 modeling not including emissions from Russia as a large portion of Russia is included in the 27-km domain (see Figure 5), although uncertainties in calculating volcanic and DMS emissions may also have contributed to the differences.

Differences in the WEP/AOI and PSC analyses

There are several differences between the WEP/AOI and PSC analyses:

• WEP/AOI was based on the EPA’s 9-km modeling domain extent, while PSC used the larger 27-km domain. Both analyses were performed at 27-km grid resolution (see Figure 5).
• The PSC analysis was only calculated for Amm_SO4, while the WEP/AOI analysis also included Amm_NO3, EC and OC.
• Emissions from the Healy Power Plant were included in the PSC analysis, but were missing in the gridded WEP/AOI analysis (although Healy was included in the 2014 and 2017 RANK_POINT results).
• Volcanic and DMS emissions were included in the PSC analysis but not in the WEP/AOI.

PSC Products

Just like the WEP/AOI, plots of the gridded residence time (RT), extinction weighted residence time (EWRT), and WEP for each CIA are provided along with plots of the emissions for each source group. Each of these metrics is described in detail in the WEP/AOI section above. In addition, pie charts showing the PSC for each source group as a percentage of the total are provided for each CIA. The plots are provided at the following link using the directory structure shown below with TUXE1 as an example.

Potential Source Contribution Plots

The pie charts of PSC for SO2 emission source sector potential contributions from sources within the 27-km domain to AmmSO4 extinction at the DENA1, TRCR1, TUXE1 and SIME1 CIAs on the MID are shown in Figures 6 through 9. A significant fraction of the PSC for DENA1 and TRCR1 were from anthropogenic emission sources (approximately 83% and 27%, respectively), while the PSC for TUXE1 and SIME1 were dominated by DMS and volcanic emissions (approximately 3% and 2% from the anthropogenic emission sources, respectively). DMS constitutes a significant fraction (8-23%) of the PSC at all four IMPROVE sites. Volcanic emissions also constitute a significant fraction at all sites but were the dominant source at TRCR1, TUXE1 and SIME1. The volcanic contribution increased with proximity to the Alaska Peninsula and Aleutian Islands.

It is important to note that the PSC is not source apportionment for many reasons, not the least of which is that source apportionment needs to account for contributions from all sources and the PSC just integrates the SO2 emissions WEP within the 27-km modeling domain. There are contributions to sulfate extinction from outside the domain used for the PSC analysis so the PSC does not include all sources. In addition, the PSC analysis does not consider chemical transformation, dispersion or deposition and has a very simplified representation of transport. However, it can be useful to provide a rough approximation of the ranking of potential contributions of in-domain SO2 emissions to sulfate extinction at the IMPROVE sites on the MID.