3

Sources of PM₁₀ Emissions in the Owens Valley Planning Area

As described in the previous chapters, both on-lake and off-lake sources in the Owens Valley Planning Area (OVPA) continue to cause exceedances in particulate matter with an aerodynamic diameter of 10 micrometers or less, known as PM₁₀ (Chapter 1). Potentially emissive sources include dunes, sand sheets, alluvial washes and fans, and sandy to silty flood deposits, in addition to on-lake features, which were more the focus of the prior National Academies of Sciences, Engineering, and Medicine report (see Chapter 2; NASEM 2020). This chapter evaluates the impacts of current and future emissions of off-lake sources to PM₁₀ exceedances, the distribution of these sources in the OVPA, and how they might be better characterized and monitored.

Currently, there are two different indicator pollutants for particulate matter: PM_2.5, particles with an aerodynamic diameter of 2.5 micrometers or less, and PM₁₀, which by definition includes PM_2.5. The national health-based standard for PM₁₀ is 150 micrograms per cubic meter (µg/m³) averaged over 24 hours or 1 day. A PM₁₀ National Ambient Air Quality Standard (NAAQS) exceedance occurs when an air quality monitoring station records a 24-hr average PM₁₀ level over 150 µg/m³. An exceedance day in the OVPA represents a day when one or more NAAQS PM₁₀ monitors in the OVPA have an exceedance. Attainment with the PM₁₀ NAAQS is achieved by having no more than one expected exceedance¹ per year at each monitor, averaged over 3 years (40 C.F.R. § 50.6[a]). Thus, four or more exceedances (at any level) at a single monitor in the OVPA within 3 years leads to nonattainment of the PM₁₀ NAAQS. This means that it is not the level by which a measurement surpasses the standard that is important for an exceedance, but simply that it exceeds the standard. As of the end of 2024, the OVPA was in nonattainment of the federal PM₁₀ NAAQS. Furthermore, OVPA is in nonattainment for the California state 24-hr average PM₁₀ standard of 50 µg/m³. However, as described in Chapter 1, the focus of this report is on the 24-hr NAAQS standard. Unlike PM₁₀, PM_2.5 has two different primary standards as defined by the NAAQS: a 24-hour average of 35 µg/m³ (assessed as the 98th percentile of these daily values for each year and then averaged over 3 years) and an annual average of 9 µg/m³ for PM_2.5 (averaged over 3 years). Currently, most of the particulate matter monitors in the OVPA measure PM₁₀, not PM_2.5. It is generally understood that as dust-rich PM₁₀ is controlled, the mineral fraction of PM_2.5 levels will be reduced as well, since during dust events a fraction (<20 percent) of PM₁₀ is PM_2.5.

___________________

¹ Expected exceedance totals account for periods of time with missing data. As defined in 40 C.F.R. § 50.6(a), “In the simplest case, the number of expected exceedances at a site is determined by recording the number of exceedances in each calendar year and then averaging them over the past 3 calendar years. When data for a year are incomplete, it is necessary to compute an estimated number of exceedances for that year by adjusting the observed number of exceedances.”

TRENDS IN OWENS LAKE AIR QUALITY OVER TIME

Since 2000, PM₁₀ concentrations have been measured at nine shoreline and community sites around the lake (Figure 3-1), and these monitors are used to assess attainment with the PM₁₀ NAAQS. Additional PM₁₀ monitors are operated by the Fort Independence Indian Community of Paiute Indians and the Lone Pine Paiute Shoshone Tribe (both within the OVPA) but are not used to assess the OVPA’s compliance with the PM₁₀ NAAQS.

___________________

² There is a statistically significant decreasing trend in the total number of exceedance days (p-value <0.01) and the 24-hr average PM₁₀ exceedance value (p-value <0.01) shown in Table 3-1 using the Mann-Kendall test.

2021–2023, the average number of PM₁₀ exceedance days per year decreased to 16 (±5) with almost a factor of four decrease in the average 24-hr PM₁₀ exceedance value of 295 (±40) µg/m³. Although in recent years the total number of days when the PM₁₀ NAAQS was exceeded and the intensity of the exceedances have decreased sharply, the OVPA remains in nonattainment. Only two of the nine monitors—Lone Pine and Bill Stanley (hereafter, Stanley)—have consistently met the standard of no more than one exceedance per year, averaged over 3 years (Table 3-2).

By combining data on wind direction, wind speed, and the PM₁₀ concentrations measured at the different monitoring sites on the lake, the Great Basin Unified Air Pollution Control District (the District) determined that by 2016 the contribution of on-lake sources to PM₁₀ had decreased, and off-lake sources represented substantial relative contributions to PM₁₀ exceedances (Figure 1-3; GBUAPCD 2016). However, in the 2016 State Implementation Plan (SIP) the District noted that emissions from off-lake sources were expected to decrease in the future as aeolian processes steadily removed dust from areas above the regulatory shoreline, which had historically originated on the lakebed (GBUAPCD 2016). The importance of this process, referred to as “winnowing,” to future off-lake PM₁₀ emissions is discussed further in Chapter 4. The following sections discuss the data and tools for understanding the trends in PM₁₀ emissions from off-lake sources.

CURRENT TOOLS AND METHODS FOR PM₁₀ SOURCE ATTRIBUTION

At Owens Lake, the District has used meteorological and particulate matter data as well as CALPUFF modeling for over two decades to identify the contributions of different dust source regions to observed PM₁₀ concentrations (Ann Logan, GBUAPCD, personal communication, July 2023). Since 2017, a more systematic approach has been recorded in an exceedance database to track and categorize the sources for each exceedance, with there being five different established categories: 1) dust—primarily on-lake sources, 2) dust—primarily local off-lake sources, 3) dust—primarily regional event, 4) wildfire smoke, and 5) mixed—dust and wildfire sources. The classification of each event into the above categories is a nonquantitative best estimate and is based on the following data and components.

• PM₁₀ and meteorological data analysis. PM₁₀ pollution roses and time series of PM₁₀ and wind speed for a few hours before and after an exceedance (Figure 3-2) are generated and analyzed. Hourly average wind speed and direction are computed from 5-second readings.

TABLE 3-2 Exceedances Per Year at Each PM₁₀ Monitor in the OVPA Excluding Measurements with Exceptional Event Flags due to Wildfire Events

NOTES: Table does not include two PM₁₀ exceedances in 2020 and 14 exceedances in 2021, which were excluded due to wildfire smoke events. The Keeler site has three PM₁₀ monitors; the highest annual exceedance count among the three monitors is listed.
SOURCE: Ann Logan, GBUAPCD, personal communication, May 2024; C. Howard, GBUAPCD, personal communication, April 2025.

• CALPUFF modeling. Hourly dispersion of on-lake PM₁₀ plumes is modeled using CALPUFF, driven by CALMET-derived meteorological (e.g., wind) fields, and estimates of sand flux. CALPUFF is a Lagrangian, non-steady state, source-oriented model that follows air parcels emitted from the modeled source regions as they undergo horizontal and vertical diffusion during transport (Scire, Strimaitis, and Yamartino 2000). The District uses CALMET (Scire et al. 2000) to provide meteorological inputs to CALPUFF on an hourly basis. CALMET is a diagnostic meteorological model that develops meteorological fields (e.g., wind, temperature, mixing height, Monin Obukhov length, friction velocity, and others) using meteorological measurements, orography, and land use data. Sand flux estimates generated from sand motion monitoring (Gillette, Ono, and Richmond 2004; Ono 2006) on the lakebed and Keeler Dunes are used together with finer-scale wind data across the lake and PM₁₀ measurements at upwind and downwind sites to identify the predominant area(s) responsible for the exceedance.
• Review of photographic evidence. If the exceedance occurs during the daytime, time-lapse photos from 20 cameras sited in 11 locations distributed around the lake and in the nearby foothills are reviewed to assess whether the source of dust is regional or more localized.
• PM_2.5. Using PM_2.5 data from the Keeler monitoring station, ratios of PM_2.5 to PM₁₀ can be used to determine the influence of wildfire smoke on elevated PM₁₀. Research has shown that wildfire smoke has a larger fraction of aerosols in the PM_2.5 fraction (Schweizer, Cisneros, and Buhler 2019), and therefore exceedance events with a very high PM_2.5 fraction (>70 percent) are determined to be driven by a wildfire instead of wind-blown dust. Because wildfire events deteriorate air quality on a regional scale, this source attribution at the Keeler monitoring station is assumed to apply to all other monitoring locations in the OVPA.
• Observation and news. Finally, notes from site operators and field inspectors along with weather reports on larger-scale synoptic events or media reports of dust storms are considered to qualify the exceedance event as regional, on-lake, or local off-lake and are also used to identify local anthropogenic sources (e.g., construction; Ann Logan, GBUAPCD, personal communication, July 2024).

Because exceedances are rare events, they may not be robust indicators of the emission potentials of sources. However, this approach of exceedance source attribution is systematic and detailed enough to parse out events as having on- or off-lake sources and differentiate the primary source of local off-lake sources (e.g., flood deposits, landfills, regional dust events, wildfires, dunes; see Box 3-1). Some ambiguity remains when several sources line up with the dominant wind direction or if the wind direction is near the angle used to separate on- and off-lake

source attribution. Additionally, categorization is somewhat uncertain when multiple sources of dust are suspected to contribute to an event—for example, when exceedance events driven primarily by on-lake sources are also impacted by some amount of PM₁₀ emission from off-lake sources. For example, in their modeling of PM₁₀ events driven by on-lake sources, the District includes a fixed background concentration in their base model runs or a variable background in their hybrid approach to account for contributions from up-valley or other off-lake sources.

IMPACTS OF LOCAL OFF-LAKE SOURCES TO PM₁₀ EXCEEDANCES

In this section, the panel examines data from individual air quality monitoring stations in the OVPA to identify the most frequent local off-lake sources of PM₁₀. In recent years, five monitoring stations show a large number of PM₁₀ exceedances from local off-lake sources—Dirty Socks, Keeler, Mill, Olancha, and Shell Cut (Figures 3-3 and 3-4). Stanley, Lizard Tail, and Lone Pine sites have had fewer PM₁₀ exceedances due to local off-lake sources. For this reason, Stanley and Lizard Tail will not be analyzed further in this chapter. However, Lone Pine is analyzed in addition to data from Fort Independence to examine the potential influence of local off-lake sources from the valley north of Owens Lake. The following sections discuss the data available for the remaining seven monitoring sites (in order from south to north) with an emphasis on exceedances and potential local off-lake sources for each of these sites.

__________________

^a Design values are used to determine the needed level of control to demonstrate attainment of the PM₁₀ NAAQS. The design value is determined by calculating the expected exceedances each year, for 3 calendar years according to the approach discussed in 40 C.F.R. § 50, appendix K. This calculation adjusts for the number of days in the year without PM₁₀ data.

California (e.g., Imperial Dunes, Oceano Dunes) that OHV activities contribute to declines in vegetation cover, disturbance of surface conditions, and elevated exceedances of PM₁₀ standards. Although research on these associations is lacking at the Olancha Dunes, it is reasonable to assume that areas in the Owens Valley subject to OHV activities could also contribute to elevated PM₁₀ emissions in the region. Increased monitoring and baseline research would help identify the nature and extent of land use change and potential contributions of OHV activity to dust emissions at Olancha Dunes.

Dirty Socks Monitoring Site

The Dirty Socks monitor is located along the southeastern edge of Owens Lake (Figure 3-8A). During 2017–2024, of all the monitors in the OVPA, Dirty Socks had the largest number of exceedances attributed to local off-lake sources (Figure 3-4). As shown in an analysis using consistent exceedance calculations over time, on-lake sources dominated at the Dirty Socks monitoring site until 2012, but since 2015 (when monitoring at the site resumed), the frequency of consistently calculated PM₁₀ exceedances due to off-lake sources have been greater than or equal to on-lake sources (Figure 3-9). Overall, there is a statistically significant decreasing trend (Mann-Kendall test p-value <0.01) in the number of consistently calculated PM₁₀ exceedances from on-lake sources while there is no significant trend in the number of calculated off-lake exceedances at Dirty Socks (Mann-Kendall test p-value >0.05). Analysis of wind direction and hourly PM₁₀ data from 2021–2023 (Figure 3-8B) reveal off-lake sources are mostly to the south of the monitor toward the Vermillion Canyon Alluvial Fan Complex (with flood deposits) and Olancha Dunes. Furthermore, hourly PM₁₀ concentrations during the largest off-lake dust events are slightly lower than the largest on-lake events, but events with hourly PM₁₀ greater than 1,000 µg/m³ occurred more frequently from off-lake sources than on-lake.

Among the 29 PM₁₀ exceedances from local off-lake sources during 2017–2024 reported in the District’s exceedance database, 59 percent were at least partially attributed to flood deposits, 48 percent to Olancha Dunes, 10 percent to alluvial fans, and 17 percent to other sources (Figure 3-10).³ As discussed in the previous section,

___________________

³ Total exceeds 100 percent because a single exceedance may be attributed to more than one source.

there is not sufficient evidence to suggest how much material is being sourced from the OHV and dispersed camping recreational area that makes up approximately 36 percent of the Olancha Dunes area.

Shell Cut Monitoring Site

The Shell Cut monitoring site is located roughly at the midpoint along the southeastern edge of Owens Lake (Figure 3-11A). An analysis using consistent exceedance metrics over time shows that from 2001 to 2009, consistently calculated PM₁₀ exceedances were predominantly due to dust events from on-lake sources, but a majority of calculated PM₁₀ exceedances since 2011 have been attributed to off-lake sources (Figure 3-12). Overall there is a statistically significant decreasing trend (Mann-Kendall test p-value <0.01) in the number of consistently calculated PM₁₀ exceedances from on-lake sources while there is no significant trend in the number of calculated off-lake exceedances at Shell Cut (Mann-Kendall test p-value >0.05). Analysis of wind direction and hourly PM₁₀ data from 2021 to 2023 (Figure 3-11B) reveal off-lake sources to the south towards Coso Wash Alluvial Fan Complex. Extensive flood deposits from the remnants of Hurricane Kay in 2022 have been mapped near the Centennial Wash, but these do not appear as major sources on the rose diagram (Figure 3-11B). PM₁₀ hourly concentrations during 2021–2023 events were generally comparable for on-lake and off-lake sources, except for one on-lake dust event that resulted in the hourly concentration of approximately 10,000 µg/m³ PM₁₀ (Figure 3-11B).

Of a total of 21 local off-lake PM₁₀ exceedances that were reported in 2017–2024 in the District’s exceedance database, 90 percent of these had at least partial contributions from flood deposits, 20 percent from alluvial fans, and 20 percent from other sources (e.g., shoreline deposits; Figure 3-13).

Mill Monitoring Site

The Mill monitoring site is located along the eastern edge of the Owens Lake bed (Figure 3-14A). The number of exceedances at both on- and off-lake sites are generally low (Figures 3-2 and 3-3), and an analysis of consistent exceedance metrics shows no clear trend over time (Mann-Kendall test p-value >0.05; Figure 3-15). Off-lake dust events affecting the Mill site are most often associated with winds from the southeast, with hourly PM₁₀

concentrations between 500 and 1000 µg/m³. This direction is consistent with the location of recent flood deposits, including those mapped from the remnants of Hurricane Kay in 2022. In fact, for the seven PM₁₀ exceedances attributed to local off-lake sources in 2017–2024 in the District’s exceedance database, 57 percent had at least partial contributions from flood deposits, 29 percent from Keeler Dunes, 14 percent from up-valley sources, and 29 percent from other sources (Figure 3-16).

Keeler Monitoring Site

The Keeler monitoring site, located east-northeast of Owens Lake (Figure 3-17A), has long been a concern for its high levels of PM₁₀ emissions from off-lake sources. As shown in an analysis of data from 1993–2023 based on consistent exceedance metrics (Figure 3-18), there is a statistically significant decreasing trend (Mann-Kendall test p-value <0.01) in the number of consistently calculated PM₁₀ exceedances from on-lake sources while there is no significant trend in the number of calculated off-lake exceedances at Keeler over the full monitoring period (Mann-Kendall test p-value >0.05). Recently, a few years have had historically low numbers of consistently calculated exceedances after implementation of the Keeler Dunes Dust Control Project (Box 6-1) project, although the apparent downward trend in calculated exceedances since 2014 has not been sustained in 2022 and 2023 (Figure 3-18).

Many elevated hourly PM₁₀ concentrations in 2021–2023 are associated with winds from the direction of Keeler Dunes or winds from the southeast in the direction of some flood deposits that have been diverted by runoff berms above Highway 136 (Figure 3-17B). The hourly PM₁₀ values from 2021–2023 under the influence of winds from the direction of Keeler Dunes were significantly higher than values observed under the influence of wind from other on-lake or off-lake directions, approaching and occasionally exceeding thousands of µg/m³ (Figure 3-17).⁴

___________________

⁴ Based on the Student t-test, committee analysis showed that hourly PM₁₀ values in 2021–2023 under the influence of winds from the direction of Keeler Dunes were significantly higher (average value of 64.6±244 µg/m³) than the values observed under the influence of wind from other off-lake (not including the Keeler Dunes) or on-lake directions (average value of 17±33 µg/m³).

Among the 19 local off-lake PM₁₀ exceedances in 2017–2024 reported in the District’s exceedance database, 79 percent had at least partial contributions from Keeler Dunes, 21 percent from flood deposits, 16 percent from up-valley sources, and 21 percent from other sources (including 1 from Swansea Dunes; Figure 3-19). Since Keeler Dunes, Swansea Dunes, and flood deposits around the dunes are all in the same general direction as the Keeler monitoring site, distinguishing the influence of each of these sources to a PM₁₀ exceedance event at Keeler is challenging and remains uncertain.

Continuous PM_2.5 measurements have been conducted at the Keeler monitoring site since 2009. These data can be useful to identify cases where the OVPA is under the influence of regional wildfires since the coarse fraction of PM₁₀ (i.e., [PM₁₀-PM_2.5]/PM₁₀) is expected to be less than 30 percent under these conditions (Schweizer, Cisneros, and Buhler 2019). One may also expect different PM coarse fractions from different dust sources of PM₁₀. Upon committee examination of the fractional distribution of PM coarse fractions observed at the Keeler site from 2015–2023 under the influence of air masses from on-lake sources, off-lake (not including Keeler-Dune sources), and off-lake from Keeler Dunes, it is apparent that there is an increased contribution of coarse particles to PM₁₀ with Keeler Dunes’ emissions. However, such high coarse fractions are not unique to Keeler Dunes because similar values are occasionally also observed under the influence of other on-lake or off-lake sources (Figure 3-20). This observation makes it difficult to use measurements of PM_2.5 to unambiguously distinguish between on-lake and off-lake dust sources of PM₁₀.

North Beach Monitoring Site

At the North Beach monitor, located at the northern end of Owens Lake (Figure 3-21A), the majority of exceedances since 2017 have been attributed to sources other than the lakebed (Figure 3-22). Based on an analysis of data using consistent exceedance metrics over time, there is a statistically significant decreasing trend (Mann-Kendall test p-value <0.01) in the number of consistently calculated PM₁₀ exceedances from on-lake sources while there is no significant trend in the number of calculated off-lake exceedances at North Beach (Mann-Kendall

test p-value >0.05). Furthermore, higher concentrations of hourly PM₁₀ were observed from the off-lake direction (Figure 3-21B).

Although the exceedance database attributes some off-lake exceedances to human disturbances (e.g., county landfill, roads), other exceedances have different local and regional off-lake sources. All of the regional exceedance days at North Beach from 2017 to 2024 were sourced from a northerly wind event. Although the primary origin of these regional events was determined to be outside the OVPA, the exceedance database noted the potential for additional PM₁₀ sources in the valley during some regional events (e.g., “picking up additional sources as the front traveled down the valley” [10/11/2021] and “augmented by local sources north of North Beach between the monitor and the Lone Pine landfill” [05/11/2018]). These exceedances suggest that there may be some occasional, local off-lake sources that lie to the north of North Beach, besides the landfill.

Lone Pine Monitoring Site and Evidence for Other Northern Local Off-Lake Sources

At Lone Pine, a small residential community north of Owens Lake (Figure 3-23A), two to five PM₁₀ exceedances occurred in most years between 1993 and 2003 (P. Kidoo, GBUAPCD, personal communication, May 29, 2024). Exceedances in this period were dominated by on-lake sources in some years and by off-lake sources in others. In an analysis of long-term trends using consistent metrics for calculating exceedances (Figure 3-24), there is a statistically significant decreasing trend (Mann-Kendall test p-value <0.01) in the number of

calculated PM₁₀ exceedances from on-lake sources while there is no significant trend in the number of calculated off-lake exceedances at Lone Pine (Mann-Kendall test p-value >0.05). Based on source attribution in the District’s exceedance database, from 2017 to 2024, in addition to regional and wildfire events, there were two exceedances attributed to on-lake sources, one exceedance attributed to off-lake sources (a landfill between Lone Pine and the lake), and one exceedance with mixed sources (Chris Howard, GBUAPCD, personal communication, April 2025).

An additional PM₁₀ tapered element oscillating microbalance (TEOM) monitor within the OVPA (Figure 3-25) is located at Fort Independence and is run by the Fort Independence Indian Community of Paiute Indians. There have been 12 daily averages at Fort Independence since 2010 that have been greater than 150 µg/m³, all of which have been since 2017. By comparing PM₁₀ concentrations at Fort Independence with the monitors on the north side of the lake (Lone Pine, North Beach, and Lizard Tail) and the monitor to the north of the OVPA (Bishop), along with wind direction and information from the exceedance database, the panel estimated that approximately 5 of these 12 events (10/20/2017, 7/20/2022, 1/3/2023, 9/1/2024, and 9/2/2024) could be reasonably attributed to sources local to Fort Independence (Table 3-3). However, there is not enough information to attribute these to a specific source like construction, roads, fallow agricultural fields, open desert, alkali meadow, or other source(s). Of note, the Fort Independence Tribe began several large projects in the fall of 2021, including construction for a Grinding Rock Aggregates facility, construction of several homes, and other developments (Kimberley Mitchell, GBUAPCD, personal communication, March 2025; Sean Scruggs, Fort Independence Indian Reservation, personal communication, January 2025). Thus, while the PM₁₀ data suggests that there may be sources near Fort Independence that may be primary causes of some high PM₁₀ concentrations observed at the Fort Independence monitor, it is unclear if these sources are derived from active disturbances of the surface (e.g., construction, or highway traffic), or if they are from other off-lake sources of dust that are the focus of this report (Table 1-1).

Overarching Assessment of Local Off-Lake PM₁₀ Exceedance Sources

Figure 3-26 summarizes the sources of local off-lake PM₁₀ exceedances at all monitoring sites from 2017 to 2024, based on the District’s exceedance database (Chris Howard, GBUAPCD, personal communication, April 2025). This analysis suggests that flood deposits, Olancha dunes, and Keeler Dunes are dominant sources of local off-lake exceedances. Human-disturbed surfaces (e.g., landfills, construction sites, roads), alluvial fans,

TABLE 3-3 Dates Since 2010 with 24-hour Average PM₁₀ Concentrations Greater Than 150 µg/m³ at the Fort Independence Monitor, Along with Lower Associated 24-hour Average PM₁₀ Concentrations at Other Monitors North of the Lake, Indicating a Local Source

NOTES: Bishop monitoring station is located to the north, outside of the OVPA. Note that a power outage across Owens Lake monitoring sites during peak wind speeds on 10/20/17 likely skewed those 24-hour averages.
SOURCE: Data from Chris Howard, GBUAPCD, personal communication, March 2025.

and up-valley sources are also important, but individually each contributes to less than 13 percent of documented exceedances. Contribution of “other” sources include open areas, deposits around the shoreline, Swansea Dunes, and regional sources, etc.

TEMPORAL EMISSION TRENDS

An alternative approach to assess the relative role of on- and off-lake sources is to estimate relative emission fluxes and their trends with time. This approach is based on the idea that the product of the concentration and the wind speed measured downwind of an area source is a measure of the emission flux, the emission of PM₁₀ per unit area, of the upwind source affecting a monitor. The committee conducted its own analysis of data (methods detailed in Appendix A) that links observed concentrations to emissions accounting for the occurrence of meteorological variations. This approach allows one to compare the impact of on- and off-lake emissions at a given site and examine trends in emission fluxes over time. The empirical relationship between the number of exceedances and emission flux allows us to estimate the expected number of exceedances associated with a given magnitude of emission flux. The results of the modeling are emission fluxes normalized by the mean of the fluxes at Keeler from 2000 to 2023. This normalization isolates the processes that govern fluxes without specifying their uncertain absolute magnitudes.

An analysis of the estimated emissions trends was completed for the monitoring sites that the panel had consistent hourly wind and PM₁₀ data since the early 2000s—Dirty Socks and Keeler. Analyses of data from Dirty Socks (Figure 3-27) shows no statistically significant trend for off-lake sources (p-value of 0.96), while estimated on-lake source emissions have had a statistically significant decline over time (p-value <0.01). Keeler (Figure 3-28) has a steep, statistically significant downward trend in on-lake source estimated emissions (p-value <0.01), along

with a shallower but statistically significant decreasing trend in off-lake estimated emissions (p-value <0.01), consistent with recent efforts at Keeler Dunes to control off-lake PM₁₀ emissions (see Chapter 6 and Box 6-1).

Overall, the analysis indicates that the two monitors with consistent, long-term data show statistically significant declines for on-lake estimated emissions over time, while off-lake emissions either have no trend or are decreasing at a slower rate. At these levels of emission fluxes (see Appendix A, Figure A-1), off-lake emissions will continue to lead to PM₁₀ exceedances at monitors in the future if not controlled. Further details are found in Appendix A.

ADDITIONAL MONITORING AND MODELING APPROACHES

Monitoring and modeling are often linked, as air quality models are commonly used to interpret observations or compare modeled predictions with measured observations. There are a number of potentially useful monitoring and modeling approaches that could be conducted to enhance the understanding of which off-lake sources are contributing to continued high PM₁₀ levels observed at monitors around or near Owens Lake. This would be most useful for sites that are impacted by emissions from multiple off-lake sources during an exceedance (e.g., Dirty Socks or Keeler). Because current monitors are placed in locations more suited to capture emissions from on-lake sources as opposed to off-lake sources, the panel discussed additional monitoring to better capture and link off-lake emissions to models. Additionally, two general categories of air quality models—receptor vs. source-oriented models—are also discussed in this report.

Additional Imagery

Several cameras around the lake have been useful in the attribution of sources (Figure 3-29), but additional cameras could help distinguish the different sources that contribute to a single exceedance. For example, more cameras southwest of Dirty Socks could distinguish Olancha Dunes emissions from flood deposits emissions. Wider-view cameras north of Lone Pine could better identify sources between Lone Pine and Fort Independence.

Satellite Imagery

Satellite imagery may also be useful in identifying PM₁₀ source areas and their evolution over time. Although the OVPA is unusual in that there are a large number of PM₁₀ monitoring stations in a relatively small area, they are generally concentrated along the 3,600-ft-elevation regulatory shoreline. In most dust source regions, air quality measurement stations are sparse and located at great distances from each other, leaving vast areas without dust detection or measurement. During the last several decades, remote sensing from orbital platforms have been used to help characterize dust sources. Instrumentation on the satellites measure discrete wavelengths of reflected solar radiation (Ciren, Kondragunta, and Huff 2024) to estimate dust loading in the atmosphere. A limitation is that dust estimates are only available when the satellite is in a position to observe the location, and when the surface is illuminated and without cloud cover. Another limitation of satellite observations is that they are not a direct measure of dust concentrations at the surface but are instead an integrated measure over an atmospheric column.

Some satellites that can provide information on dust are in geosynchronous orbit, such as the Geostationary Operational Environmental Satellites (GOES), the latest version being the GOES-R series. The instruments on these satellites focus on one area of Earth 24 hours a day, but remaining geosynchronous requires an orbit almost 22,236 miles (35,786 km) above Earth, limiting their spatial detail. For most GOES-R wavelengths, the pixel size is about 4 square kilometers. Historically, many estimates of atmospheric dust have been made with the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument package carried on the Terra and Aqua satellites (Baddock et al. 2016; Eibedingil et al. 2024; Kandakji, Gill, and Lee 2020, 2021; Webb and Pierre 2018). Having a non-geosynchronous orbit, Terra and Aqua are in an orbital band of 438 miles (705 km) above the Earth, 1/50th the distance to Earth’s surface and thus have greater spatial detail with pixel sizes as small as 0.024 square miles (0.0625 km²). These estimates of Aerosol Optical Depth (AOD) are comparable to ground-based estimates from the Aerosol Robotic Network (AERONET) at low and middling latitudes, but MODIS-derived AOD estimates are

better at high latitudes (Eibedingil et al. 2021). Since the satellites are not geosynchronous, dust outbreaks may be missed when the satellite is not over the dust source.

The MODIS instrument is past its design life and is being replaced by the Visual Infrared Imaging Radiometer Suite (VIIRS) instrument package carried on the Suomi National Polar-Orbiting Partnership spacecraft NOAA-20 and NOAA-21 (NOAA 2024b). VIIRS has a 0.024 square mile (0.0625 km²) pixel size and 22 wavelength bands (NOAA 2024a), allowing for good dust detection when the instrument is over a cloud-free source region during daylight hours. The National Aeronautics and Space Administration (NASA) is also planning to launch the Multi-Angle Imager for Aerosols (MAIA) instrument in 2026 aboard the PLATiNO-2 satellite (NASA 2024) The PLATiNO-2 completes an orbit every 100 minutes (NASA 2024). As its name suggests, it is specifically aimed at providing

more detailed information on aerosols, including an ability to help identify composition, thus providing a linkage to its source (e.g., dust, fire, secondary). MAIA will allow for spectrum analysis of wavelengths bands from 300 to 2,300 nanometers (spanning ultraviolet [UV] to short infrared) and 14 filters (NASA 2024). UV wavelengths detect certain mineral particulates and organic matter, whereas the shorter visible wavelengths detect very small particles according to their diameters, and the short infrared wavelengths are useful for coarse aerosols such as dust and volcanic wash (NASA 2024). MAIA is planned to have a resolution of about 984 ft (300 m; Liu and Diner 2017).

CubeSats are very small (0.3-ft cubes alone or joined in multiples) satellites that orbit at about the same distance from Earth’s surface as the satellites carrying the MODIS and VIIRS instrument packages. They are increasing in number and with sufficient numbers flying in formation or constellations, they will be capable of continual coverage of the surface of Earth. High-resolution cameras used in CubeSats have pixel sizes of 10.7 square feet (2.25 m²) from a 310-mile (500 km) orbit. The Starlink constellations launched by SpaceX are visible satellite constellations that orbit at only 280 miles (450 km) above the surface, 1/80th the altitude of GOES-R. The PlanetScope visual product developed from a constellation of about 130 satellites has been combined with machine learning to estimate ground level PM_2.5 (not PM₁₀) at a 656-ft (200 m) resolution, with the results suggesting that it can be used to identify areas with elevated PM levels linked to high emissions (Zheng et al. 2020).

NASA’s Earth Surface Mineral Dust Source Investigation (EMIT) mission may provide an additional resource for understanding dust sources in the OVPA (NASA n.d.). EMIT, launched in 2022 for deployment on the International Space Station, is a relatively new mission aimed at characterizing the composition of surface dust at a relatively fine resolution (approximately 200 ft or 60 m) and can be teamed with other satellite- and ground-based observations (e.g., from GOES, VIIRS or MAIA in the future) and modeling to characterize OVPA off-lake source regions. Satellite imagery may also play a role in assessing dune movement (Donnallan, Hallet, and Leprince 2015).

The growing information from satellite-based instruments provides a potential source for the District and other stakeholders to identify major dust source areas, and there is little doubt that such capabilities will continue to improve. Furthermore, their use in tandem with ground-based observations and air quality models has proven helpful in characterizing emissions of pollutants (Qu et al. 2022; Wang et al. 2012). A similar approach (e.g., integrating satellite observations of source locations, ground-based observations, and adjoint modeling) could be done to help quantify the contributions of on- and off-lake sources to measured exceedances.

Additional Monitoring

Currently, the OVPA has a relatively large number of PM₁₀ monitors given the size of the planning area. However, most of these monitors are located near the historic lake shoreline to capture air masses impacted by on-lake emissions. More detailed identification of off-lake emissions of concern and quantification of their emission strengths can be enhanced by additional monitoring, similar to what was done around Keeler Dunes when a temporary TEOM (a Federal Equivalent Method [FEM] instrument) was installed at the northwestern edge of the Keeler community from March 2015 to February 2020 (Grace Holder and Chris Howard, GBUAPCD, personal communication, January 2025).

While Keeler Dunes has been more intensely studied with near source PM₁₀ monitoring and dust emission potential, other off-lake sources such as Olancha Dunes and areas north of the lake are less well characterized. Additional PM₁₀ monitoring can help fill that knowledge gap, especially if the observations are linked with air quality modeling analyses. In the last OLSAP report, the panel recommended using lower cost monitoring (NASEM 2020). These lower cost monitors include both very inexpensive monitors (e.g., less than $1,000) and monitors that can be more expensive while still remaining less than the cost of FEM and Federal Reference Method (FRM) monitors to purchase and operate. Lower cost sensors of PM typically run on the principles of PM light scattering, by either nephelometry or optical particle counting techniques (Hagan and Kroll 2020). Studies have found that in dusty areas with significant contribution from coarse particles (i.e., particles between 2.5 and 10 µm), nephelometry-based sensors perform poorly when compared with FRM or FEM measurements of PM₁₀ while performance of some of the optical particle counters has been shown to be promising (Alfano et al. 2020; Hagan 2022; Kaur and Kelly 2023; Kuula et al. 2020; Molina Rueda et al. 2023; Ouimette et al. 2022; South Coast Air Quality Management District 2024). For example, during a month-long field study at three sites in Salt Lake Valley, which included sampling

during five dust events, optical particle counting-based PM₁₀ measurements correlated very well (r² >0.87) with FEM measurements of PM₁₀ by Met-One E-BAM PLUS, with a slope of 0.92–1.39 and root mean squared error (RMSE) of 12–18 µg/m³ (Kaur and Kelly 2023). On the other hand, the nephelometry-based data were poorly correlated with E-BAM (r² <0.49), with RMSE approximately 35–45 µg/m³ and slopes of less than 0.099 (Kaur and Kelly 2023). However, in-field calibration of low-performing units improves accuracy and reduces bias to some extent (Alfano et al. 2020; Kaur and Kelly 2023). In many cases, nephelometry-based sensors also perform poorly in high velocity winds, due to the dynamics of the air intake ports. Ideally, air intake ports should be isokinetic and pointed into the wind to draw air at the same velocity as the ambient flow field. These deal situations are difficult to achieve in a turbulent flow field. Another solution that is often employed is to place the sensor in an aspirated chamber where the stilling volume of the chamber permits air intakes that are more closely isokinetic than could be attained in the high velocity flow field. While the data collected by low-cost sensors would not be used to determine compliance with the NAAQS, it can be used to help characterize source areas and for model evaluation, accounting for instrument accuracy. This evidence supports the use of lower cost monitors for characterizing source regions that are less well-studied, particularly if they can be solar powered, rather than using more costly regulatory monitors that would be difficult to use in these conditions (e.g., Chauhan et al. 2022; Riter et al. 2023).

Another type of monitoring that may assist in better defining the distribution of sources are Portable In-Situ Wind Erosion Lab (PI-SWERL) analyses. PI-SWERLs have been used in the OVPA in the past to help quantify PM₁₀ emission potential in off-lake areas (Kolesar et al. 2022b). These devices use a rotating annular blade 6 cm from the surface of interest to mimic wind across the surface and create shear stress to generate emissions of PM₁₀. The threshold friction velocity of wind that initiates particulate movement and PM₁₀ emissions is then determined (Finnigan 1988; Raupach 1992). Replicate or transect-based testing is necessary because PI-SWERL captures emissions on a small area of the surface.

One advantage of PI-SWERL measurements of surface dust emission potentials is that a range of surface conditions at a single site may be considered. For instance, within a small area of just 2–6 square feet, replicate tests of dust emission potentials from undisturbed or crusted surface conditions may be compared with the dust emission potentials of the same surfaces when disturbed to different extents. Transects could be especially useful in areas like Olancha Dunes to better distinguish the impact of recreational activity on dust emission potentials (Gillies et al. 2022). In addition to disturbance effects, temporal or seasonal factors of dust emission potentials may be measured. Thus, as emissivity varies in response to meteorology, modelling can be used to yield annual predictions based on the expected erosivity of the seasonal winds. As shown in the past applications in the OVPA, linking PI-SWERL measurements, other observations, and modeling results can better quantify the impact of emissions source regions on air quality.

PM₁₀ Composition and Receptor Modeling

Having detailed information on the composition of the particulate matter (i.e., the elemental abundances) could also facilitate more detailed source identification. For example, while the Owens Lake bed is a texturally varied mixture of fine clay particles, sodium carbonate, sodium sulfate, and soluble salts (Gill 1995; House, Buck, and Ramelli 2010; Reheis 1997; Tyler et al. 1997), off-lake dune sources are distinctly different in that they are predominately quartz, plagioclase, potassium feldspar with minor amounts of calcite, and other minerals (Lancaster and Bacon 2012; Lancaster et al. 2015). There is evidence of gypsum present in passive dust collectors downwind of the Owens Lake bed (Reheis 1997); however, the source of this gypsum is unknown as it is not present in the surface crusts of playas (Gill 1995). Higher concentrations of heavy metals such as arsenic and antimony are found in Owens Valley alluvium and lake-marginal deposits farther away from the dry bed of the Owens Lake, possibly due to the proximity to naturally occurring minerals in the Inyo Mountains, including the Cerro Gordo Mining District (Reheis 1997; Reheis, Budahn, and Lamothe 2002). Elevated zinc and lead concentrations are also found in dust resuspended from alluvial sediment near Keeler on the northeast side of Owens Lake (Barone et al. 1979; Barone et al. 1981; Cahill et al. 1994). These compositional differences between on- and off-lake sources highlight that chemical compositions could potentially provide some insight into the proportion of on-lake and off-lake dust sources for individual exceedance events. Although it would require high sensitivity, it may also be possible to use

the compositional differences between various off-lake sources to better attribute an exceedance event to one or multiple off-lake sources (e.g., contribution from Olancha Dunes vs. flood deposits to exceedances at Dirty Socks).

Elemental analysis of dust from events that are shown to be dominated by different sources can be interpreted using data analysis techniques, often referred to as source apportionment models, discussed below. This will require chemical speciation of the airborne dust, as well as well-defined chemistry of the surrounding source areas (e.g., Frie et al. 2017; 2019; Wang et al. 2023). Chemical speciation of airborne dust can be gained by collecting PM₁₀ on filters and then using methods such as X-ray absorption near edge structure (XANES) for metal speciation, inductively coupled plasma mass spectrometry (ICPMS) for elemental measurement, and multi-collector ICPMS analysis for isotopic characterization of the particulate matter on the filters. Filters from the Partisol Sequential Sampler that the District has deployed can be used for such analysis.

The measured composition of PM₁₀ (including the elemental composition described above) can be used in receptor models to quantify the contribution of sources with unique chemical signatures to the observed PM₁₀. Receptor models are observation-driven, using the measurements at one or more receptors to estimate how different sources are impacting concentrations at that receptor (Watson 1984). Example receptor models that are commonly used include the Chemical Mass Balance method and the Positive Matrix Factorization or Non-Negative Matrix Factorization (Coulter 2004; Friedlander 1973; Paatero and Tapper 1994; Watson, Cooper, and Huntzicker 1984). Concentrations of individual ions (e.g., sulfate, nitrate, chloride, ammonium, sodium; often measured from filters using ion chromatography), elements (e.g., metals; often measured using x-ray fluorescence or mass spectrometry) and elemental and organic carbon are typical species used in source apportionment analyses. Using the composition of emissions from specific source regions or types, the amount of PM₁₀ contributed from the modeled sources (e.g., on-lake, specific dunes, regional transport, and flood deposits) can be quantified. While the Chemical Mass Balance method requires specific knowledge of source composition profiles, Positive Matrix Factorization or Non-Negative Matrix Factorization can simultaneously provide a source composition profile and strength based on its input data of the PM₁₀ composition. Ideally, hourly composition data can capture the relatively fast shifts in source areas with changes in wind direction. With enough composition data of PM₁₀, receptor models are generally applied in readily available computer environments, as the models are limited by the availability of the speciated measurements.

Receptor modeling has been conducted at the Oceano Dunes area using a weight of evidence approach to quantitively estimate contributions of sources to exceedances (Wang et al. 2023). This approach involved relationships between combinations of specific PM₁₀ species and source emissions. For example, sodium, chloride, magnesium, potassium and sulfate were used to estimate fresh sea salt (FS):

FS = fsNa⁺ + Cl- + ssMg²⁺ + ssK⁺ + ssCa²⁺ + ssSO₄^2-

In this example, fsNa⁺ is the fresh sea salt fraction of sodium ion; Cl- is the concentration of chloride ion; ssMg²⁺ is the sea salt magnesium; ssK⁺ is the sea salt potassium; ssCa²⁺ is the sea salt calcium ion; and ssSO₄^2- is the sea salt sulfate ion. Similar relationships were developed for aged sea salt, mineral dust, and other sources. The source relationships developed at Oceano Dunes are unique to the dominant sources surrounding those dunes; however, a similar approach could be applied to Owens Lake once chemical signatures of the sources in Owens Valley are quantified. This approach is very similar to the Chemical Mass Balance method in that it uses known chemical compositions of suspected sources.

Source-Oriented Modeling

Source-oriented models simulate the transport and transformation of pollutants from their emission as they evolve in the atmosphere, typically based on computationally solving the equations governing pollutant dynamics. Those models are based on accurately describing the important physical (e.g., wind velocity, turbulent transport) and chemical processes impacting pollutant concentrations from emissions (which are often uncertain) to their fate. An important use of source-oriented models is to assess our understanding of source emissions strengths to prioritize emission control strategies and better predict emissions and impacts from a specific source on PM₁₀ concentrations. If a modeling system performs well under a variety of conditions, that indicates the model is cap-

turing the important sources and processes prevalent in the atmosphere. Lack of agreement, conversely, suggests that there are potentially large gaps in our understanding, which limits the accuracy of the modeling system. For example, if the model results are biased low in comparison to observations, that would suggest that the estimated model source strengths are low or there are missing sources not accounted for in the modeling. If the model simulations are biased high, that would suggest that the source strength estimates are too large. Biases can also be introduced by errors in the meteorological inputs. Diagnosing model errors helps identify errors in our understanding of model processes. Many different source-oriented models have been developed for applications from very local scales (such as the Owens Lake area) to global scales (EPA 2024a; Mejia et al. 2019; Pennington et al. 2024; Vohra et al. 2021).

As described earlier, the District’s current modeling approach is to use CALPUFF, along with CALMET as the source of its meteorological inputs to model primarily on-lake source regions. Keeler Dunes is the sole off-lake source currently treated in the District’s modeling. With the 2017 revisions to the Guideline on Air Quality Models (Appendix W to 40 C.F.R. § 51), CALPUFF is no longer an Environmental Protection Agency (EPA) preferred model. One concern that arises from the panel’s review of the modeling conducted by the District is that the model results can differ substantially from the observations (Figure 3-30). The District developed a hybrid approach using a time-varying background concentration based on monitored off-lake measurements (GBUAPCD 2016), which does improve model performance. However, there are still distinct differences between model results and observations, particularly on days when the observed PM₁₀ concentrations are low.

A second concern with the District’s current modeling approach is that by only modeling the downwind air quality impacts from on-lake emissions and Keeler Dunes, the ability to quantitatively analyze the impact of other off-lake sources on PM₁₀ exceedances and emissions rates is impaired. This information would be valuable for identifying the most effective control program to mitigate future exceedances and in exceptional event analyses. Such information would also reduce bias in the current modeling result, as there are a number of cases where the simulated levels are much less than observed, suggesting that other sources not accounted for in the modeling may be contributing to the exceedances.

Alternative Approaches

One of the most widely used dispersion models in the United States is AERMOD (Chen et al. 2009; Perry et al. 2005). AERMOD is a steady-state plume model and is an EPA-preferred model for dispersion modeling (EPA 2024b). It has a more limited description of atmospheric chemistry than CALPUFF, but this is not a significant issue when modeling PM₁₀ mass concentrations. AERMOD is often driven by AERMET, a meteorological preprocessor that utilizes observations to develop meteorological inputs to AERMOD, analogous to CALPUFF being driven by CALMET. AERMOD is computationally fast compared to most three-dimensional models described below or other models that include more complex chemistry. AERMOD has been used to assess source impacts on particulate matter in a variety of applications (Batterman et al. 2014; Chen et al. 2009; Colledge et al. 2015; Jittra et al. 2015; Özkaynak et al. 2013; Perry et al. 2005) including carbon monoxide, nitrogen oxides, particulate matter less than 2.5 μm in diameter, and diesel exhaust emissions, have been associated with adverse human health effects, especially in areas near major roads. In addition to emissions from vehicles, ambient concentrations of air pollutants include contributions from stationary sources and background (or regional.

Other plume models, include the LSPDM (Mejia et al. 2019), LAPMOD (Bellasio et al. 2017; Graff, Strimaitis, and Yamartino 1998), and KSP (Graff, Strimaitis, and Yamartino 1998) space-time varying meteorological conditions, and the desirability of having a model which can yield the probability distribution function (PDF, but it is not apparent if these models have significant advantages over CALPUFF. At Oceano Dunes, the LSPDM is tied to a very fine resolution emissions model based on extensive dust emissions data collected using the PI-SWERL (Etyemezian et al. 2007). At present, there are insufficient PI-SWERL observations in Owens Valley to drive a similar fine-scale emissions estimation approach for the off-lake areas other than Keeler Dunes. However, PI-SWERL testing can be relatively rapid and four replicate tests can be performed in an hour. Thus, it could be possible to obtain the level of spatial detail necessary to model the complex landscapes in Owens Valley at different scales.

The Weather Research and Forecasting (WRF) model is a prognostic meteorological model that solves the basic equations of fluid and energy transport in the atmosphere to provide important fluid-mechanical properties affecting pollutant transport, including wind velocities and turbulent diffusion. CALPUFF can use meteorological outputs from the WRF (Skamarock et al. 2007) and the District’s model performance might be improved by including some of these properties. The WRF with chemistry (WRF-Chem) model also uses meteorological parameters from WRF to solve the equations governing pollutant transport from one fixed grid to another, vertically and horizontally (Grell et al. 2005). Contractors for the LADWP used WRF-Chem to conduct preliminary modeling of the region with additional off-lake sources. WRF-Chem is a Eulerian three-dimensional model that differs from plume models in that it is grid-based with a fixed spatial coordinate system. Similar Eulerian models that are widely used include CAMx and the Community Multiscale Air Quality (CMAQ) model.

The use of WRF-Chem or other three-dimensional Eulerian pollutant transport models instead of a plume model has some advantages and disadvantages. Disadvantages include that WRF is computationally more demanding than CALMET, and WRF-Chem (or CAMx and CMAQ) are more computationally demanding than CALPUFF. Another concern is that WRF-generated wind fields can differ substantially from observations (EPA 2019c; Parajuli and Zender 2018). Such differences can alter near-source dispersion calculations, though it is not known how this might manifest in the Owens Lake area. Some advantages include that the Eulerian grid model results are spatially complete, that multiple sources and source areas are treated simultaneously, that the computational time does not increase markedly as new sources are added (although input preparation will), and that the simulations can be used to carry out source-apportionment calculations across the modeling domain, not just along specific trajectories. Treating multiple source areas simultaneously can be particularly important when assessing the cumulative impacts from all the source areas on- and off-lake, though plume models can do so by using plumes from each source area. Some grid-based models have been specifically instrumented to follow emissions from specific sources that can be used to quantify the impacts from on- and off-lake sources at all the monitoring sites in the area (EPA 2025b; Ramboll Americas Engineering Solutions, Inc. 2020). Another advantage of grid-based models is that they can capture the importance of chemical interactions of pollutants from different sources. However, the contributions of atmospheric chemistry to PM₁₀ exceedances around Owens Lake is likely small given the short transport times, lack of major local sources of secondary particulate matter formation, and the dominance of direct PM₁₀ primary emissions.

Inverse modeling of source-oriented models can be used to better estimate source emissions by adjusting source strengths to better capture observed concentrations. Inverse modeling is typically done by applying a model with the estimated emissions as the input and comparing the model results to the observations, then adjusting the emissions to improve model results. Given multiple observations over multiple times, the estimated emissions are optimized to best capture observations. Zhang (2024) used AERMOD along with multiple linear regression to develop better emissions estimates at a fine scale, as might be done for Owens Valley. More complex approaches include Kalman filtering (Carmichael et al. 2008; Napelenok et al. 2008) and adjoint modeling (Chen et al. 2021; Hakami et al. 2005; Kaiser et al. 2018; Stavrakou and Müller 2006; Zhang et al. 2009).

Value of Including Off-Lake Sources in Models

As noted above, the District includes on-lake source areas and the Keeler Dunes in their CALPUFF modeling. The WRF-Chem model applied by LADWP contractors included some additional off-lake source regions. However, neither appear to consider the broad range of potential off-lake sources contributing to PM₁₀ exceedances in the Owens Lake area. Including additional off-lake source areas in air quality analyses of the Owens Lake area would improve the identification and characterization of sources impacting PM₁₀ exceedances, provide a platform for air quality planning, and advance assessments of the effectiveness of dust controls.

Historically, the District has used their CALPUFF-based modeling system to demonstrate how controls to on-lake sources and Keeler Dunes will impact air quality (GBUAPCD 2016). This exemplifies a major use of air quality models: to identify effective strategies to meet the region’s air quality goals. If additional off-lake controls were to be considered to reduce PM₁₀ from off-lake sources, it would be important to include these off-lake areas in the modeling effort. Models can also be used to perform scenario analyses to provide expected source impacts

under a range of conditions and controls across the region. Another reason to include a more comprehensive set of off-lake sources in the modeling is to more formally treat the background used in prior CALPUFF modeling. The Hybrid Modeling Approach used by the District includes a time varying background to account for off-lake sources. Inclusion of those sources in the model application would more directly incorporate off-lake source impact on modeled concentrations.

Finally, as discussed in Chapter 5, exceptional event demonstrations are based on thorough analyses of the sources leading to an observed exceedance and an assessment of whether the sources and conditions satisfy the criteria to be considered exceptional events. An important component of an exceptional event demonstration could be using a well-evaluated modeling system to link the observed high levels to specific sources.

CONCLUSIONS AND RECOMMENDATION

Dust control measures have made substantial progress toward reducing the frequency and intensity of on-lake exceedances, but both on-lake and off-lake sources continue to cause PM₁₀ exceedances in the OVPA. The relatively consistent number of PM₁₀ exceedances that the District has attributed to off-lake sources over the last 25 years, despite trends indicating a declining number of exceedances from on-lake sources, demonstrates the importance of these off-lake sources and suggests that these sources could hinder attainment with the PM₁₀ NAAQS in the region.

Since 2017, the District has used additional information to attribute PM₁₀ exceedances to specific sources within the OVPA. This information includes particulate and meteorological data, modeling, cameras, field observations, and media reports of dust storms. These data are compiled into the District’s exceedance database. Based on these data, the District classifies each exceedance as one of the following: 1) dust—primarily on-lake sources, 2) dust—primarily local off-lake sources, 3) dust—primarily regional event, 4) wildfire smoke, and 5) mixed—dust and wildfire sources; and provides detailed comments on likely source areas. The panel supports the District’s general approach to source apportionment and use of this information to identify a few specific local off-lake sources that cause a disproportionate impact on the PM₁₀ exceedances in the OVPA. These sources include flood deposits (including channelized, sheet/overland flow, and impounded flood deposits), Keeler Dunes, Olancha Dunes, alluvial fans, up-valley sources, and anthropogenic disturbances.

This District’s method for source attribution is useful for assessing broad trends, but the classifications are nonquantitative, and some uncertainties remain in the identification of specific off-lake source areas. For example, the current methodology does not allow for the quantification of PM₁₀ contributions from different sources for a single exceedance, and data are often not collected in locations that are ideal to capture detailed information about off-lake sources. Additional measurements and modeling would enable the District to more definitively identify how specific sources have contributed to exceedances and support future air quality management decisions.