Nitrogen Isotopes Reveal High NOx Emissions from Arid Agricultural Soils in the Salton Sea Air Basin

Air quality management commonly aims to mitigate emissions of oxides of nitrogen (NOx) from combustion, reducing ozone and particulate matter pollution. Despite such efforts, regulations have recently proven ineffective in rural areas like the Salton Sea Air Basin of Southern California, which routinely violates air quality standards. With $2 billion in annual agricultural sales and low population density, air quality in the region is likely influenced by year-round farming. We conducted NOx source apportionment using nitrogen stable isotopes of ambient NO2, which indicate a substantial contribution of soil-emitted NOx. The soil source strength was estimated based on the mean δ15N-NOx from each emission category in the California Air Resources Board’s NOx inventory. Our annual average soil emission estimate for the air basin was 11.4 ± 4 tons/d, representing ~30% of the extant NOx inventory, 10× larger than the state’s inventory. Therefore, the impact of soil NOx in agricultural regions must be re-evaluated.


Introduction
Nitrogen oxides (NOx = NO + NO2) are a precursor to ozone (O3) and particulate matter (PM), both of which are federally regulated criteria pollutants, known to negatively impact human health and the environment. 1NOx is produced through natural, biogenic, and anthropogenic processes, although a critical distinction is necessary.Natural sources of NOx occur without influence from human activity, such as lightning or weather-induced wildfires.
Biogenic NOx is naturally occurring and is produced by living organisms through microbial nitrification/denitrification cycles in soils.Further, anthropogenic NOx results from human activities such as fossil fuel combustion in vehicles and power plants.However, these natural processes can be affected by anthropogenic activity, such as biomass burning in agricultural systems, human-ignited wildfires in poorly managed landscapes, and microbial activity in soils of fertilized croplands.3][4][5][6][7][8] These practices have largely reduced precursor emissions, yet several of the United States' worst-air quality districts remain in rural regions, 9,10 including the Imperial and Coachella Valleys of Southern California.These valleys together constitute the Salton Sea Air Basin (SSAB, Fig. 1), which is currently nonattainment for the O3, PM2.5, and PM10 National Ambient Air Quality Standards (NAAQS).Interregional transport from larger metropolitan areas is often blamed, [11][12][13][14][15][16] hewing to the traditional paradigm of combustion dominant sources.This issue is a serious matter of environmental injustice because residents of Imperial County are 86% Latino and 21% of residents live below the poverty line. 17Additional burdens these communities face include insufficient housing availability and poor infrastructure, language barriers, asthma, and impaired water quality. 18,19These factors put residents at risk of respiratory damage and diseases like asthma and COVID-19. 20In fact, Imperial County held both the highest infection and mortality rates per capita in California from COVID-19 throughout the height of the pandemic. 21,22Members of the rural Eastern Coachella Valley also suffer from similar socioeconomic and environmental injustices. 19e SSAB is an agriculturally active desert responsible for over $2 billion in annual agricultural sales due to mild temperatures and abundant sunshine in the winter, allowing for year-round production. 23However, growing crops in sandy soil is challenging and can require anthropogenic land modifications such as carbon-based amendments, heavy application of inorganic fertilizers, and intensive irrigation.This is primarily due to the soil's texture, with larger particle diameters and lower organic carbon composition, resulting in poor retainment of water and nutrients necessary for plant growth. 245][26] Hard-setting soils have high soil strength that is not suitable for root growth or shoot development.Consequently, this requires frequent soil-wetting through irrigation practices, especially in summer months when temperatures regularly exceed 35°C; these dry, hot conditions and changes in volumetric soil moisture (VSM) may lead to large, transient soil NOx emissions (Figs. 1 & S1). 10,24,27n addition, inorganic fertilizer amendments are not regulated, leading to overapplication and nutrient leaching into the surrounding environment, such as the groundwater, local water sources, atmosphere, and soils. 24,28The objective of this work is to understand the implications of agricultural practices in arid agroecosystems of the SSAB on regional air quality.
Observations and modeling studies show a massive expansion of arid lands globally due to climate change, ranging from 2.5 -6.5 million km 2 by the end of the century, especially in large croplands in the northern hemisphere. 29Consequently, agricultural production in arid ecosystems will become more prominent, 30 therefore the implications of these practices need to be better understood.Microbes in soils naturally produce NO and nitrous oxide (N2O) gases (and nitrous acid gas, HONO) by the same processes, but their relative amounts are highly uncertain.2][33][34] In dry, well-aerated soils like in the Salton Sea Air Basin, nitrification is likely the dominant process resulting in soil NO emissions, which can diffuse out of the soil before being consumed in further biogeochemical cycling.However, in wet soils, denitrification dominates, and NO is mostly consumed before leaving the soil, resulting in pulses of N2O.In soils that are more wet and highly anaerobic, N2O is further reduced to N2 (Firestone &   Davidson, 1989).][36] Soil pH is noteworthy because high pH generally factors microbial ecosystems that promote nitrification and NO production, while acidic fertilized soils produce HONO through chemical equilibrium with nitrite in water. 34However, Oswald et al. (2013) found comparable quantities of HONO and NO emitted from nonacidic soils, suggesting that HONO may be an unaccounted, yet significant amount of reactive nitrogen released from soil worldwide, particularly dominant in arid and arable environments like the SSAB.This is crucial as the daytime lifetime of HONO with respect to photolysis of NO + OH is around 15 minutes, causing soil emissions to rapidly cycle into the photostationary state of O3-NO-NO2, becoming an additional source of NOx to the atmosphere. 37 NOx emissions from combustion sources decline under ongoing air quality regulations, it becomes crucial to comprehend the impact of anthropogenically induced biogenic NOx production, particularly from agricultural soils.The nitrogen stable isotope composition (δ 15 N) is proposed as a valuable tool for NOx source apportionment due to differences in mean δ 15 N-NOx values between anthropogenic and biogenic emission sources. 38- 40 mbient δ 15 N-NO2 is recognized as an effective means of source apportionment, benefiting from well-quantified isotope fractionation factors and offsetting isotope effects resulting from the interplay between isotope equilibrium and photochemical fractionation 41 Previous ambient δ 15 N-NO2 studies have successfully tracked the influence of highly variable biogenic soil NO emissions impacting the NOx budget in a small Midwestern city during the summer, 42 suggesting it could be a powerful tool for constraining biogenic soil emissions in the SSAB.
In this study, δ 15 N was quantified from NO2 (and total nitrate, tNO3) actively collected on a Thermo Scientific TM ChemComb® Speciation Cartridge (CCSC).Ambient NO2 samples were collected in two locations in the SSAB for 10 months and δ 15 N was analyzed to quantify the soil source contributions to the region's NOx inventories.Using our field measured δ 15 N-NO2 values and accounting for nitrogen isotopic fractionation to adjust these values to δ 15 N-NOx, we used a mixing model approach to estimate the soil source strength of the field observations based on the mean δ 15 N-NOx from each emission source and from the a priori source apportionment reported in CARB's California Emissions Projection Analysis Model (CEPAM) inventory.Because nitrate is influenced by long-range transport, we do not use tNO3 as a marker for source apportionment.This study aims to improve our understanding of the sources of persistent air pollution in the SSAB by assessing the current NOx inventory's estimation of soil NOx emissions.

Sample Collection, Extraction, and Isotopic Analysis
Ambient NO2 and total particulate nitrate (tNO3) were collected and quantitatively speciated using a denuder-filter pack active sampling device known as a ChemComb Speciation Cartridge (CCSC), as previously described. 43Briefly, the body of the CCSC contains an inlet with an impactor plate for the removal of coarse particles, a glass transition piece to induce laminar flow, two honeycomb denuders to collect ambient NO2 gas, and a downstream filter pack. 44nuders were soaked in 10% (v/v) hydrochloric acid (HCl) for 24 hours, then rinsed in triplicate with 18.2 MΩ Millipore water and air-dried prior to field preparation.6][47] This mixture selectively binds NO2 as nitrite (NO2 -).Following the denuders is a single Nylon filter (Measurement Technologies Laboratory NY47P, 47 mm and 1 μm pore size) for the collection of tNO3.Ambient air was sampled at a flow rate of 10 LPM using a mass-flow controller (MKS IE50A008304SBVP20) and vacuum pump.Details about chemicals and materials used can be found in Supplementary Table S1.Field sampling was performed from June 2022 -April 2023 at two sampling sites in the SSAB (Fig. 1), one in the Imperial Valley (Calipatria High School in Calipatria, CA) and one in the Coachella Valley (Torres-Martinez air station in Thermal, CA).This project incorporated community engagement and citizen science; Comité Cívico del Valle (CCV), a grassroots community organization based in Imperial County that was founded on the principle that "Informed People Build Healthy Communities", served as our community partners.Trained air monitoring technicians from CCV oversaw field set-up and break-down.Sampling was performed for 3-7 days, approximately monthly over a period of ten months.
The extraction of NO2 and tNO3 was performed with 30-mL and 20-mL of MQ water for the front denuder and nylon filter, respectively, then vacuum filtered through a Whatman grade 1 filter.Back denuders were used to check for breakthrough and were not used for isotopic analysis.Field blanks were collected for method validation, however, blanks contained negligible concentrations of NO2 and tNO3.Nitrate (NO3 -) and nitrite concentrations in aqueous solution were determined on a UV-Vis spectrophotometer using a colorimetric method developed by Doane and Horwath (2003) (extraction methods in Text S2). 48NO2 denuder extracts were neutralized with 0.1 M HCl, then isotopic analysis was performed at the UC Davis Stable Isotope Facility by the denitrifier method on a GasBench-PreCon-Isotopic Ratio Mass Spectrometer to determine δ 15 N-NO2. 49,50The pooled standard deviation for the 15 N reference was 0.11‰.
The lifetime of aerosols in the atmosphere is typically 5-10 days, meaning the N in particulate nitrate is derived from a large integrated area that spans several air basins across California.Although there was fair correlation between NO2 and particulate NO3 δ 15 N values across the experiment (Pearson correlation coefficient of ~0.45), the average for the NO3 samples was ~10 ‰ heavier than for NO2, indicating that the aerosol nitrate was likely more strongly influenced by distant, traditional fossil-fuel combustion sources of NOx.Additionally, quantifying isotope effects associated with tNO3 formation is more difficult. 51Uncertainty in knowing exactly where the tNO3 was produced along its long-range trajectory makes it a much more challenging marker for source apportionment studies, therefore we do not use the tNO3 data (but it is reported in Table 3).

Evaluation of δ 15 N References
The isotopic signature of nitrogen is expressed in terms of its  15 N, where: In this equation, R represents the 15 N/ 14 N ratio.The standard is atmospheric N2, which is considered to have a globally uniform isotopic composition. 52Studies of the natural variations in 15 N/ 14 N ratios from different sources are used to apportion sources of NOx pollution.
A literature review was conducted to evaluate published stable 15 N isotope ratios for various sources of NOx, and relevant δ 15 N-NOx values are summarized in Table 1.On average, biogenic soil sources have been observed to have a significantly larger ratio of light 14 N isotopes than combustion sources.This phenomenon can be elucidated through the "leaky pipe" analogy, a concept derived from the hole-in-the-pipe model proposed by Davidson and Firestone. 31In the "leaky pipe" scenario, soil NO-producing processes, such as nitrification, denitrification, and chemodenitrification, act as microbial conduits where microbes preferentially utilize the lighter 14 N isotope.This preference arises from kinetic factors that favor the incorporation of the lighter N isotope, leading to a more depleted δ 15 N signature. 45,52,53The "leaky pipe" effect in soils contributes to the observed larger variations reflected in the standard deviations reported in the literature (Table 1 and Supplementary Fig. S4).
It is important to consider the uncertainties of the isotopic signatures for each source due to variations in measurement techniques found in the literature.For this study, passive sampling techniques [54][55][56] were excluded because the measurement periods were much longer, could allow for a mixture of NOx sources, and are more difficult to reproduce, ultimately introducing a bias in the δ 15 N-NOx. 57Additionally, studies on vehicles without catalytic converters were excluded 58 2).Further, coal burning is not a method of energy production in the Salton Sea Air Basin or surrounding areas, therefore, studies that measured the isotopic signature of NOx produced through coal combustion have been excluded from our stationary source measurements. 58,61See Section 2.3 for uncertainty propagation.

Soil Source Strength Mixing Model Estimation and Propagation of Uncertainty
A best estimate of the sources of NOx in the SSAB is compiled in the California Air Resources Board's California Emissions Projection Analysis Model (CEPAM) shown in Table 2.
CEPAM was used over the EPA's National Emissions Inventory (NEI) because CEPAM is broken into county and air basin, and due to its California focus, should in principle be more accurate.
Because chemical and physical processing can induce isotopic fractionation such that δ 15 N may not be conserved, isotopic fractionation was calculated to determine δ 15 N-NOx (see Text S3 and Eq.S1-3).
Equation1 assumes that the current CEPAM inventory is correct aside from the soil source because the observed δ 15 N was (usually) lighter than expected; without adjusting the soil source signature, the CEPAM inventory would indicate a mean δ 15 N-NOx of -5.04‰ and -3.11‰ for Imperial County and the Coachella Valley, respectively.Therefore, we solve for the magnitude of the soil source such that the observed δ 15 N signature can be explained by the sum of the inventory sources plus the revised soil emissions.This procedure further assumes that the sampled NO2 is entirely from NOx (or HONO) emitted from sources within the air basin.
Figure 1 shows annual NO2 column data from the European Space Agency's TROPOspheric Monitoring Instrument (TROPOMI) for the year of our sampling.
We propagate the various uncertainties associated with these measurements of sources from the literature to estimate the overall error associated with our calculated soil emissions.
Equation 2 comes from the propagation of error for the final soil source strength (   ).The analytical uncertainty δobs (on average 0.11‰) was ignored because it is an order of magnitude smaller than the standard deviations of the δ 15 N-NOx measurements for each source (Table 1 where  = (  − ∑      ,, )

Results and Discussion
Results of collected δ 15 N-NO2 signatures are shown in Table 3.The quantitative basis of our estimates relies on the assumption that the ambient NO2 at the sampling site comes from sources entirely within the SSAB.Figures 1 (and Supplementary Fig. S1) show the overall pattern of column NO2 observed by TROPOMI during its overpass time near 13:30 local time.The satellite data illustrates two important aspects of the regional NOx.
First, although there is a flow connection between the Coachella Valley and upwind urban sources in the LA basin (Supplementary Fig. S2), the valley plume is distinct and does not appear to simply represent a decaying tail (especially noticeable in the Fall and Winter maps of Supplementary Fig. S1).The Imperial Valley, on the other hand, is only downwind of Mexicali during July/August (Supplementary Fig. S3) and yet it exhibits a broad amorphous shape that deviates from the circular symmetry of a concentration field falling off with distance from an urban core.Second, the basin-wide concentrations are greatest during spring/summer when the photochemical lifetime of NOx is shortest, supporting our finding that agricultural soil sources are significant, especially during the warmest part of the growing season.Regardless of the details of source apportionment for the sampling, Equation 1shows that any influence from urban sources upwind (with their heavier δ 15 N fractions) would effectively increase our soil emission estimates, meaning that the values reported here represent lower limits.Further details of the regional flow and NOx advection are discussed in Text S1.

Environmental Influences on Soil NOx Production
To better understand the environmental influences that control the production of soil NOx, we look at the leading parameters used in modeling estimates; namely nutrient availability, temperature, and soil moisture. 36,63Temperature is known to strongly influence soil NOx production, although the exact dependence at higher temperatures is still debated, 35 and has been observed by satellite over croplands. 27Nevertheless, no clear temperature dependence was observed in this study (Figs. 2 & 3) despite observations ranging from 17-44°C (Supplementary Tables S3, S4, and Fig. S6).As an agriculturally active desert, NOx production is likely limited not principally by temperature as in unmanaged landscapes but by nutrient abundance and soil moisture determined by agricultural activity.
To further understand the influence of these parameters, we looked at volumetric soil moisture content using NASA's Soil Moisture Active Passive (SMAP) satellite data.Soil moisture tends to be higher on average in the Imperial Valley agricultural lands than in any other region in the air basin (Fig. 4).This is because of the region's appropriation of approximately 3 billion m 3 of water from the Colorado River, used primarily for frequent irrigation of its ~200,000 ha of arable soil.It is equivalent to ~150 cm of water over the croplands whereas the climatological rainfall observed usually only amounts to < 5-10 cm (Supplementary Tables S5 and S6).
Additionally, δ 15 N outliers were observed for October in Calipatria and for November and January in Thermal, yielding estimated soil emissions that were effectively zero within the errors of our measurement technique (Fig. 2).The outliers measured in October (Calipatria) and November (Thermal) correspond to precipitation events during our measurement periods that likely suppressed NOx emissions during the short intervals of sampling by wetting the soil to excess temporarily (Fig. 5). 36During our sampling interval in January (Thermal), precipitation was negligible, but examining HYSPLIT back trajectories for the interval, it appeared that there was a high-pressure system over the region that may have forced subsidence of cleaner tropospheric air into the Coachella Valley, which is consistent with the sampled NOx being lower than the monthly average by 5.5 ppb (Supplementary Tables S4 and S6).However, this circulation did not persist throughout the entire 5-day sampling period, and the February sample was also collected when NOx was 5 ppb lower than average, so ultimately, we are uncertain as to the exact cause of this near zero soil emissions measured at Thermal in January.
Although Figure 3 could be interpreted as exhibiting a crude correspondence between soil NOx emissions and VSM, temperature only appears to be weakly related, which likely indicates that nutrient availability is the dominant factor controlling soil NOx production in this heavily agricultural region.According to the California Department of Food and Agriculture (CDFA), N fertilizer sales in Imperial County have increased by 137% since 1991, 23,64 and an estimated 18-fold since 1930, 65 much greater than the national average sevenfold increase.In addition, despite being ranked as the 9 th highest in county agricultural sales in the state, Imperial County used more N fertilizer than any other county during 2022 (8% of California total), including the top three crop producing counties (Fig. 6), likely a result of the insufficient nutrient retainment in sandy soils.
Spatially and temporally relevant fertilization data are not currently publicly available, therefore it is difficult to assess the direct influence of this leading parameter on the soil NOx emission variance observed in this study.In Figure 6, we used data from the Trajectories Nutrient Dataset for Nitrogen (TREND-nitrogen) to look at historical N fertilizer applications and legacy N concentrations in Imperial County.N surplus was calculated using TREND-nitrogen, where the county-scale surplus or "legacy" nitrogen (kg N ha -1 y -1 ) is calculated by mass balance from the inputs of atmospheric deposition, fertilizer application, biological N fixation, and human N waste, minus the crop and pastureland N uptake. 65It is apparent that N fertilizer usage and surplus N in Imperial County has been steadily increasing from 1920 to the present day (Fig. 6).Our estimates are based on a small fraction of days in the Salton Sea Air Basin; in our study, we sampled ~39 days in Calipatria and ~43 days in Thermal, accounting for about 11% of the year.Therefore, our estimates are based on a small fraction of days in the SSAB.
Inspecting the conditions of our 19 sampling intervals with respect to the annual values for VSM and air temperature (Supplementary Fig. S5 and Table S7), they compare fairly well.
Nevertheless, without knowing the exact timing of N amendments to all the cropped area, it is quite possible that while our estimates of annual average VSM and temperature are not statistically different from the average, the soil emissions, predominately influenced by the fertilizer inputs, may be much higher than our estimates having missed many periods of high fertilization in our sporadic sampling.
To understand the overall influence of fertilizer inputs on atmospheric N, we calculated the NOx flux as a proportion of fertilizer consumption using the net fertilizer usage recorded. 23perial County purchased 57,630 tons of all Nitrogen fertilizer, minerals, and compost in 2022. 23Our annual average soil emission of 6.7 tons NOx/d in Imperial County implies an average flux of 1.3% of the applied N being released as NO or HONO from agricultural fertilizer inputs.This is squarely within the 0.3-2.5% range reported by Jaeglé et al. (2005).Note that we were unable to quantify this for the Coachella Valley because it is part of the much larger Riverside County and fertilizer sales are only reported county-wide.

Comparison Between Studies
To further assess the validity of our results for Imperial Valley, we compared our measurements to Oikawa et al. (2015), which found some of the highest soil NOx emissions on record from chamber studies across the growing season.They further tested their findings by using WRF-Chem to compare against ambient NOx measurements from the air quality network.
The default soil emissions in the model needed to be enhanced by an order of magnitude to match their chamber measurements more closely, and the enhancement was found to minimize the model's root-mean-square error relative to ambient measurements for one week in September 2012.To compare our integrated valley emissions to the modeled surface fluxes, we used the area of agricultural land in the Imperial Valley to be 270,500 ha based on MODIS satellite land surface characterization. 27The integrated emissions for their one week in September (the month of our highest estimate of 10.1  4 tons/d) amounts to 17.2 tons/d (Table 4).That study also attempted to match agreement with satellite NO2 measurements and found that the emission rate that corresponded best was more like 111.2 tons/day.Their study also found that such soil NOx emission increases led to concomitant rises in surface level ozone of 2-9 ppb, highlighting the importance of this source to air quality and the observed disappearance of ozone abatement in the region. 9maraz et al. to the state NOx inventory was 1.1%. 67However, they did point out a region of anomalously high emissions in the Imperial Valley, with an annual flux of 0.58 kg N ha -1 yr -1 .
While there is considerable variability between the various estimates presented (Table 4), undoubtedly in part due to the highly episodic and transient nature of soil NOx emissions, the CEPAM inventory estimate is consistently much lower than the other models and observations.We expect that the dominant control of the instantaneous emission rates is the N amendments made to the agricultural soils daily, so the wide range presented above is likely consistent with such inherent variability.Therefore, to obtain an accurate annual average emission rate from soils will take much more comprehensive understanding of N availability in the soil across these highly variable croplands.More work is needed to better constrain the impact of N availability on soil NOx emissions, as well as the precise soil conditions that control microbial activity and gas-exchange.As of 2022, the Sustainable Groundwater Management Act was modified to require farmers in California to report their daily N inputs for consequences of excess N leaching into groundwater.This emerging data set will be crucial for future studies to investigate soil NOx emissions in arid agricultural environments and accurately assess its impacts on air quality in rural communities.

Figure 2
displays the soil emission estimates for this field campaign at both sites, comparing the monthly magnitudes to the overall annual CEPAM NOx inventory.It is apparent that the CEPAM NOx budget underestimates the soil contribution to ambient NOx (and consequently to PM2.5 and O3) in both Calipatria (Imperial Valley) and Thermal (Coachella Valley).Soil NOx contributed 0.3 -10.1 (mean: 6.7 ± 3) tons/d and 0 -13.8 (mean: 4.7 ± 4) tons/d throughout the duration of the field sampling for the Calipatria and Thermal sites, respectively.These soil emissions amount to on average 11.4 ± 4 tons/d throughout the entire SSAB, an order of magnitude larger than what is represented in the inventory (1.0 ton/d).
abbreviated as FEST-C*66 Data from their 2017 model for Imperial County indicates an annual

Figure 2 .
Figure 2. The NOx budget (in tons/day) based on CEPAM2019v1.03for the year 2022 is shown, which is split into

Figure 3 .
Figure 3. Soil NOx estimates from our field campaign compared to the average VSM and maximum temperature

Figure 5 .
Figure 5. Monthly average Volumetric Soil Moisture, VSM (2022-2023, shown as a line plot), and the standard

Figure 6 .
Figure 6.Imperial County N fertilizer usage (red) and legacy N history (blue) was prepared using the Trajectories in the Salton Sea Air Basin.Additionally, both studies were performed in the northeastern US in large metropolitan areas where biogenic sources are believed to be negligible.The reported mean from Miller et al. (2017) was then averaged with the mean of warm and/or driven vehicles from Walters, Goodwin et al. (2015) to represent mobile source δ 15 N-NOx (Table 15,60 catalytic converters are mandatory and regulated through California smog checks.For these reasons, only two publications were used to represent mobile source emissions: Miller et al. (2017) and Walters, Goodwin et al. (2015). 59,60measured values reported in Walters, Goodwin et al. (2015) were used because some measurements were taken from engines that were cold and/or in neutral.Catalytic converters take a few minutes to warm up and work efficiently, and efficient catalytic converters preferentially remove lighter14N isotopes,59,60leading to heavier15N emissions.Therefore, cold engines are biased with more negative isotopic signatures and likely do not represent background mobile source emissions measured 15,62These δ15N-NOx values were then used to calculate the soil source emission strength (Es) in Equation 1 based on the average δ 15 N-NOx of each emission source based on the a priori source apportionment reported in the CEPAM NOx inventory, which is used for regulatory modeling in California. Here,bs represents the observed δ 15 N-NOx, i represents the source contributes to the overall CEPAM NOx inventory, δi represents the literature derived δ 15 N-NOx for each source (Table2), and Einv represents the total NOx budgeted in the CEPAM inventory (in tons/d).
main source types accounted for in the CEPAM NOx inventory (where a, b, c, and s represent mobile, biomass burning, stationary, and soil sources, respectively), αi represents the proportion each ).

Table 1 .
15mmary of previously reported δ15N-NOx values sorted by emission source type.
*Only a portion of the measurements from this paper were used since they were measured with a cold, neutral engine, which wouldn't be relevant for ambient temperatures.

Table 2 .
The 2022 California Emissions Projection Analysis Model (CEPAM) NOx emission inventory and our estimated soil NOx adjustments for Imperial County and the Coachella Valley, as well as for the combined SSAB, are shown.The CEPAM inventory is based on an annual average, so we averaged our soil source estimates for our sampling duration.Also shown are the adjustments to the NOx inventory based on the average results from our field sampling campaign.The EPA's National Emissions Inventory (NEI) for 2020 is also shown for comparison for Imperial County only.

Table 3 .
Field sampling results for δ 15 N-NO2 and estimated δ 15 N-NOx based on isotopic fractionation calculations, as well as the results for δ 15 N-NO3 for Calipatria and Thermal.

Table 4 .
Comparing our average soil NOx flux estimate for Imperial County to CARB and other studies.