Abstract

The development of a new emissions inventory of ammonia volatilization from native soil sources (excluding direct emissions from fertilizer application sources) for the state of California is discussed. Because a comprehensive measurement dataset is currently lacking upon which to build a reliable emissions inventory for NH3 from native soil sources in California, an ecosystem modeling approach that is based on satellite remote sensing and other geographic datasets was used to develop a new estimate of statewide biogenic emissions rates of N-NH3. The NASA–Ames version of the Carnegie–Ames–Stanford Approach (CASA) ecosystem model is applied for soil nitrogen gas emissions at 8-km spatial resolution. The NASA–CASA model estimates seasonal patterns in carbon fixation, nutrient allocation, litterfall, soil nitrogen mineralization, and soil NH3 emissions. The general conditions and spatial patterns favoring soil NH3 volatilization from soils (high pH, low moisture) have been integrated in the NASA–CASA formulation. Based on the modeling inventory estimate discussed here, statewide emissions of NH3 from native soil N sources could range from 12 to 57 × 106 kg N-NH3 annually, depending on the importance of soil pH on emissions rates.

The most important land cover types in terms of contributions to this statewide emissions inventory are the croplands and semiagricultural ecosystems that cover about 20% of the total area of the state, but which make up one-third to one-half of the total soil N sources for NH3 emissions annually. Other native soil source areas that contribute substantially to the statewide emissions inventory for emissions of NH3 are soils of evergreen needleleaf forests, woodland, and wooded grassland ecosystems, mainly on the basis of their large area coverage of the state's natural areas. The model predicts that October is the peak month overall for NH3 emissions from native soils in California. When totaled for the entire region, native soil sources of NH3 predicted for Central Valley counties are highest from July through January. This seasonal pattern in predicted native soil NH3 emissions is fairly consistent with observed seasonality in PM2.5 levels in the San Joaquin Valley Air Basin.

It is hypothesized that the combination of productive vegetation communities growing on (even slightly) alkaline soils results in the largest annual emissions of NH3 from native soil N sources. However, a comparison is presented for these native soil sources of NH3 to a regional model budget for potential foliar absorption fluxes of ammonia, which implies that vegetation cover on a statewide basis could actually make ecosystems a strong net sink for locally emitted NH3 sources.

1. Introduction

Ecosystem sources of ammonia contribute to NH3 being a dominant gaseous base in the atmosphere and a principal neutralizing agent for atmospheric acids. Continental sources of NH3 are normally much larger than marine sources (Dentener and Crutzen, 1994). The supply of alkaline soil dust and gaseous NH3 available in the atmosphere may control the acidity of precipitation. Volatilized NH3 may react to form ammonium nitrate or ammonium sulfate and thereby contribute to airborne particulate matter (PM). National standards in the United States for PM apply to the mass concentrations of particles with aerodynamic diameters less than 2.5 μm (PM2.5) and less than 10 μm (PM10). Recent measurements of PM2.5 in California have shown that on average, the highest 24-h concentrations in 1999 occurred in November–December–January while the lowest concentrations occurred between March and August (Chow et al., 1992). Seasonality is found to be most pronounced in the San Joaquin Valley Air Basin, where the November–December–January concentrations were on the order of 4–5 times greater than those for March through August. Estimated patterns of nitrogen deposition suggest that, for California locations close to photochemical smog source areas, concentrations of oxidized forms of N (NO2, HNO3, PAN) dominate, while in areas near agricultural activities the importance of reduced N forms (NH3 and NH4+) may increase significantly (Bytnerowicz and Fenn, 1996).

Nevertheless, NH3 remains one of the most poorly characterized atmospheric trace compounds in terms of overall sources. This situation persists as a result of several factors: experimental difficulties associated with NH3 measurements, rapid gas-to-particle conversion of NH3 in the atmosphere, the capacity of organic matter and vegetation to act as both sources and sinks for atmospheric NH3, and variability in soil properties related to NH3 emissions (Langford et al., 1992; Bouwman et al., 1997). Consequently, there is a limited amount of published information from which to develop direct emissions estimates of NH3 for the state of California in general, and the state's Central Valley in particular. Preliminary measurements of NH3 background concentrations in the San Joaquin Valley by Fitz et al. (Fitz et al., 1997) estimated February levels of 3–16 μg m−3 around open alfalfa fields. However, until now, the magnitude and distribution (both regionally and seasonally) of current NH3 emissions from fertilized and managed ecosystems are still largely undetermined for the state of California and many other large regions where agriculture is a major land use (Matthews, 1994).

The main objective of this study was to estimate rates of NH3 emissions from all native soil sources (i.e., those excluding direct emissions sources from fertilizer applications) in California. An ecosystem modeling approach, one that utilized satellite remote sensing and other geographic datasets for land surface properties, was used to develop a new inventory of statewide biogenic emissions rates of native soil N-NH3. Our modeling method for biogenic trace gas budgets is designed to include more climatic controls and ecological process information than conventional estimates based on the emissions factor approach (Warn et al., 1990; Pain et al., 1998).

This new NH3 inventory information will ultimately assist the state in evaluating the important, but sometimes conflicting needs of maintaining both good air quality and a vital California agricultural industry. We note that these model estimates of NH3 emissions from native soil sources do not include any direct emissions sources from fertilizer application events in California croplands, which we have reported on in a separate study by the California Air Resources Board (CARB, 2001). Because some fraction of the NH3 emitted from soil sources could be recaptured through leaf absorption into vegetation growing at a site, we have included a statewide estimate of the foliar absorption sink flux as well, which permits calculation of the net ecosystem emissions of native soil N-NH3 sources.

2. Statewide geographic information system development

For this regional model application to the state of California, we have used the same National Aeronautics and Space Administration–Carnegie–Ames–Stanford Approach (NASA–CASA) ecosystem model algorithms as for our previous global simulation studies (Potter, 1999; Potter and Klooster, 1998). However, in the place of global 1° inputs, regional datasets (8-km resolution) from a statewide geographic information system (GIS) were used as model drivers and land surface parameter files. Following the same general approach reported by Davidson et al. (Davidson et al., 1998) for a regional application of NASA–CASA soil nitrogen model component in the southeastern United States, we assembled a complete set of regional GIS raster coverages to serve as model-compatible inputs, including monthly rainfall and surface air temperature, surface solar radiation, soil texture, land cover type, and satellite vegetation index for the state of California and, in many cases, for the larger western U.S. region.

All raster maps were gridded at 8-km spatial resolution in an equal area projection. In terms of single grid cell size (64 km2), this produces an improvement in spatial resolution of more than 150 times, compared to the global 1° (∼104 km2 cell size) data drivers for the model. The coastal boundary line file used as a base to georeference the 8-km map set was taken from the Digital Chart of the World (DCW, 1993).

2.1. Satellite vegetation index

In order to estimate ammonia emissions from native soil sources, it is necessary to identify the seasonal patterns of vegetation types throughout California. To do this, we obtained the monthly 8-km composites for the years 1982–94 of the Normalized Difference Vegetation Index (NDVI) from the Advanced Very High Resolution Radiometer (AVHRR), which is available from the (National Oceanic and Atmospheric Administration) NOAA–NASA Pathfinder AVHRR Land (PAL) program at the NASA Goddard Space Flight Center (GSFC) Distributed Active Archive Center (DAAC). Complete AVHRR datasets are produced from NOAA Global Area Coverage (GAC) Level-1B data, and consist of reflectances and brightness temperatures derived from the five-channel cross-track scanning AVHRR aboard the NOAA polar orbiter “afternoon” satellites (NOAA-7, -9, and -11). DAAC references by Agbu and James (Agbu and James, 1994) and Kidwell (Kidwell, 1991) provide more information on the derivation and potential use of these NDVI products.

Monthly composite datasets are designed to remove much of the contamination due to cloud cover present in the daily AVHRR datasets (Holben, 1986). To generate a composite dataset, 8–11 consecutive days of data are combined, taking the observation for each 8-km bin from the date with the highest NDVI value. Only data within 42° of nadir are used in the composite to minimize spatial distortion and bidirectional effect biases at the edge of a scan. A Rayleigh correction is calculated and applied using a standard radiative transfer equation and methodology, which follows the work of Gordon et al. (Gordon et al., 1988).

Having obtained original PAL files for NDVI, we ran a low-pass filter over the data to remove several narrow lines of anomalous high values that are presumably a result of compositing. The filter routine computes the mean of six nearby grid cell values (located two rows and three columns above, i.e., north of, each cell location) and compares this average to the actual cell value. If the difference between the original cell value and the average of the values above that cell was greater than 200 units, then the value of that cell is replaced by the average of the nearby cell values. Typically, less than 25% of the NDVI values over the entire regional land area required modification with this filtering step.

Although PAL composite datasets are produced expressly for studies of temporal and interannual behavior of surface vegetation, subsequent processing is recommended if a more complete cloud-free signal is required. Consequently, we applied solar angle corrections (S) and Fourier smoothing algorithms (FA) developed by Los et al. (Los et al., 1994) for AVHRR datasets to further remove anomalous NDVI signals, due presumably to remaining cloud cover interference. Settings for this FA correction include three temporal harmonics and a weighted Fourier transform; that is, values that fall above the Fourier curve are given more weight than values below the curve. This assumes that higher NDVI values are more likely to be correct than low NDVI values that could occur during periods of cloud or smoke formation. The FA algorithm modified mean annual NDVI values by more than +10% of their original values in approximately 4 out of every 10 grid cells in the region. This is largely due to the persistent ground fog and marine layer presence throughout the year in different parts of California.

From these 8-km monthly FAS–NDVI datasets, we applied empirical algorithms described by Potter et al. (Potter et al., 1993) to compute second-level model drivers for the fraction of intercepted photosynthetically active radiation (FPAR) and leaf area index (LAI).

2.2. Land cover type

The land area within the state of California covers approximately 400,000 km2. For land cover characterization at the 8-km resolution, we used the classification scheme of DeFries et al. (DeFries et al., 1995), which was generated from analysis of 1-km AVHRR–NDVI patterns over the year. Ten general classes are represented in this global land cover map at 8-km grid cell resolution (Figure 1). In California, evergreen needleleaf forest is the most common (31% of the total area), followed by open shrub land and deserts (24%), cropland and semiagricultural lands (21%), woodlands (17%), and then all other cover types combined, including wetlands, river ways, and bare ground (8%). The combination of semiagricultural lands, such as residential lawns, golf courses, parks, and marginal/fallow lands, together with the actively cultivated farming areas generally covered by the California Department of Water Resources'(DWR, 1993–1998) crop coverage for each county of the state, makes the cropped land cover category from DeFries et al. (DeFries et al., 1995) about twice the size of the DWR crop coverage maps used for estimating emissions from active fertilizer application (CARB, 2001).

Figure 1.

Satellite-based map of major land cover classes in California at 8-km grid cell resolution

Figure 1.

Satellite-based map of major land cover classes in California at 8-km grid cell resolution

Model parameters that are assigned according to the 10 general (DeFries et al., 1995) land cover types include leaf litter nitrogen (Table 1) and lignin content (Potter, 1999), decomposition rates of soil carbon in cultivated soils, and plant rooting depth. For the forest classes, rooting depth is set uniformly to 2 m, whereas in the other land cover classes, it is set uniformly to 1 m. With respect to this general application of the model algorithm coefficients, we did not attempt to distinguish between primary, secondary, or recently cleared forest types. In summary, we applied the same algorithm coefficients developed from the global version of the NASA–CASA model (Potter, 1999) for relatively undisturbed forests or grasslands to their respective disturbed or converted cover categories. This means that any differences in model results reported for different forest cover types would not be attributable to internal model settings in the general land cover groups defined above, but instead to patterns in NDVI, climate, or soil inputs to the model.

Table 1.

Foliar nitrogen concentration setting in the NASA–CASA model for vegetation types in California (Source: Potter, 1999)

Foliar nitrogen concentration setting in the NASA–CASA model for vegetation types in California (Source: Potter, 1999)
Foliar nitrogen concentration setting in the NASA–CASA model for vegetation types in California (Source: Potter, 1999)

2.3. Soil attributes

State Soil Geographic (STATSGO) soils data for the state of California were obtained from the National Resources Conservation Service. Soil pH (Figure 2) and average clay content of the surface horizon was calculated for each map unit from the weighted average of all soil series indicated within the map unit, using methods described in detail by Davidson and Lefebvre (Davidson and Lefebvre, 1993). The 1-km STATSGO soils dataset was subsequently aggregated to 8-km grid cells.

Figure 2.

STATSGO map of soil pH in California at 8-km grid cell resolution

Figure 2.

STATSGO map of soil pH in California at 8-km grid cell resolution

Soil map units were assigned to the four aggregated texture classes determined by the Food and Agricultural Organization (FAO, 1971), according to their clay content (0%–5%, 5%–15%, 15%–30%, and >30% clay, respectively). The most common texture class in the state is the coarse-medium 5%–15% clay, which covers 30% of the area. Coarse (0%–5% clay) and medium (15%–30% clay) texture classes each make up about 22% of the state's soils. This spatial information on soil texture is used in the NASA–CASA model to define regional patterns of soil moisture holding capacity and rates of soil organic matter accumulation.

STATSGO data at the level of soil order in the U.S. classification system were also used to define three general soil fertility classes: low, medium, and high (Birkeland, 1974). On low-fertility soils, an adjustment (+10%) is favored that allocates increasing root biomass from Net Primary Productivity (NPP) for the acquisition of soil nutrients (Potter, 1999). On medium-to-high fertility soils, a similar adjustment is favored that allocates increasing stem and leaf biomass from NPP for light harvesting functions in the canopy.

2.4. Monthly precipitation and surface air temperature

Monthly mean climate maps for California were obtained from ZedX, Inc. (Boalsburg, Pennsylvania). We regridded the original files from 1-km spatial resolution to our nominal 8-km cell resolution. These average climate datasets are generated based on long-term (1961–90) records from weather stations in California, which are part of the Global Historical Climatology Network (GHCN) at the Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory (Vose et al., 1992). The spatial interpolation is done by three-dimensional linear regression. Accuracies of the resulting climate maps are checked by percent absolute difference analysis with comparison to original station values. Accuracy of the data in comparison with actual stations was approximately ±0.6°C for temperature and ±0.1 cm for precipitation.

2.5. Monthly solar surface radiation

Surface solar radiation flux (Srad) gridded for the state of California was not in the same format as other monthly mean climate maps. Therefore, we estimated Srad in monthly average units of watts per squared meters, based on diurnal temperature range (DTR) data reported in GHCN minimum and maximum temperature records. Our estimation technique for Srad is founded on the atmospheric transmittance theory from Bristow and Campbell (Bristow and Campbell, 1984), who reported that the difference between maximum and minimum daily temperatures is closely correlated with the amount of solar radiation received. At times when the net flux of solar radiation at the Earth's surface is low (e.g., during overcast sky conditions), the difference in surface temperature extremes is also generally low. The opposite is true for clear sky conditions. Thus, measurements of daily temperature extremes should be related to the atmospheric transmittance for solar radiation flux using relatively simple least squares regression functions. This method to estimate regional patterns in Srad has been demonstrated for the continental United States as part of the Vegetation/Ecosystem Modeling and Analysis Project (VEMAP) ecosystem simulation experiment (Kittel et al., 1995).

For our modeling purposes, a set of global regression functions was developed for the prediction of monthly mean Srad from DTR measurements. Calibration datasets for DTR were obtained as monthly mean values from the Climatic Research Unit (CRU), University of East Anglia, Norwich, United Kingdom, gridded originally to 0.5° resolution (New et al., 2000). Multiyear DTR data (1983–91) from the CRU dataset were used to develop third-order polynomial equations, which predict monthly mean Srad within four global latitude zones: > 50°N, 50°N–0°, 10°N–20°S, and >20°S. The measured Srad data for these regressions were obtained from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) radiation flux estimates from Bishop and Rossow (Bishop and Rossow, 1991), which are derived as a product of the International Satellite Cloud Climatology Project (ISCCP) and gridded originally at a spatial resolution of 2.5° for the period July 1983–June 1991. Starting from weather station locations listed in the GHCN (Vose et al., 1992), we subsampled from the CRU globally distributed DTR coverage across each latitude zone on the basis of those locations producing the highest R2 regression coefficients with Srad as the dependent variable. This subset of regression curves was selected to represent the four latitude zones. In cases of extreme DTR values beyond the applicable range of the regression curves (either lower or higher than the independent variable bounds), we set minimum and maximum Srad values predicted at the extremes of the calibration DTR values. The resulting Srad regression equation for California has an R2 of 0.54 (p<0.05), applicable over the DTR of 3°–16°C.

3. Emissions model description and implementation

The NH3 emissions rates were computed in this study using an ecosystem model that takes into account spatial patterns in water balance, snowmelt dynamics, soil pH, and soil moisture. Our model, the NASA–Ames version of the CASA model for soil nitrogen gas emissions, has been documented previously in Potter (Potter, 1999) and Frolking et al. (Frolking et al., 1998). The model simulates seasonal patterns in carbon fixation, nutrient allocation, litterfall, and soil nitrogen mineralization, and soil ammonia emissions (Figure 3). Several other soil trace gas fluxes (i.e., N2O, NO, CH4, and CO uptake) are simulated with NASA–CASA (Potter and Klooster, 1998). The use of remote sensing drivers in NASA–CASA (including NDVI) has clear advantages for scaling up to regional estimates of vegetation production that can drive trace gas fluxes. For regional studies, the net fixation of CO2 by vegetation, also called NPP, and the plant uptake of soil N is constrained in our model by satellite NDVI (Potter, 1999). NPP is calculated in NASA–CASA as a product of the fraction of intercepted photosynthetically active radiation (FPAR, from NDVI), surface solar irradiance, and an optimal light utilization efficiency term (emax), modified by surface temperature and soil moisture estimates.

Figure 3.

NASA–CASA model framework for predicting ammonia emissions from soils

Figure 3.

NASA–CASA model framework for predicting ammonia emissions from soils

The soil component of the NASA–CASA model simulates carbon–nitrogen (C–N) cycling and associated flux of trace gases using a set of soil organic matter pools with a structure comparable to the CENTURY ecosystem model (Parton et al., 1992). First-order equations simulate exchanges of decomposing plant residue (metabolic and structural fractions) at the soil surface, together with surface soil organic matter (SOM) fractions that presumably vary in age and chemical composition. Active (microbial biomass and labile substrates), slow (chemically protected), and passive (physically protected) fractions of the SOM are represented. The model computes the fraction of water-filled soil pore space (WFPS) in the various layers in order to estimate scalars that represent the effect of soil moisture on organic matter turnover and CO2 emissions rates, which couple to N mineralization fluxes and N trace gas emissions.

For the monthly time step simulations in this study, the model was brought to initial state (for 31 December) with respect to soil moisture, litter inputs, soil C–N pool sizes and turnover times using monthly climate, together with remote sensing drivers from the CASA–Biosphere version (Potter, 1999), in a 100-yr initialization simulation run. Carbon entering the ecosystem yearly as NPP (defined as net fixation of CO2 by vegetation) and annual litterfall return of C and N to the soil for mineralization was estimated on a pixel-by-pixel basis (Figure 4), using the 100-yr “spinup” simulation run with long-term mean climate and NDVI to approximate reported carbon and nitrogen content of surface soils. The NASA–CASA soil carbon pools were initialized to represent accumulation conditions in the near–steady state \[annual (NEP) less than 0.5% of annual NPP\] with respect to the mean climate inputs. An atmospheric CO2 effect on NPP is not included in CASA. The global model settings for nitrogen content of plant leaf litter are provided in Table 1 according to vegetation cover class. Further details on the model algorithms for plant water use, snow dynamics in cold winter areas, and a review of limiting factorson fertilizer ammonia emissions are provided in the sections that follow.

Figure 4.

Statewide NASA–CASA estimates of annual NPP and soil nitrogen mineralization at 8-km grid cell resolution

Figure 4.

Statewide NASA–CASA estimates of annual NPP and soil nitrogen mineralization at 8-km grid cell resolution

3.1. Water balance equations

An empirically based potential evapotranspiration (PET) algorithm in this model is based on a modified formulation of the Priestly and Taylor (Priestly and Taylor, 1972) method described by Campbell (Campbell, 1977) and Bonan (Bonan, 1989):

 
formula

where PET is potential evapotranspiration (cal cm−1 day−1), Ta is mean air temperature (°C), Rs is mean surface solar radiation (cal cm−1 day−1), and a and b are empirical constants (set as functions of saturation vapor pressure) derived by Jensen and Haise (Jensen and Haise, 1963) and Jensen (Jensen, 1973). Conversion of PET to units of centimeters per day is made by dividing by the latent heat of vaporization.

Estimated evapotranspiration flux (ET) for the stand is calculated by comparing PET to the multilayer model estimate for soil moisture content. The soil profile is treated as a series of three layers: M1 is the surface organic layer, M2 is the surface layer rooting zone, and M3 is the mineral subsoil (Figure 3). These layers can differ by ecosystem and crop cover type in terms of bulk density, moisture holding capacity, texture, and carbon–nitrogen storage. Where drainage is impeded, water can accumulate upwardly in a ponded layer (M0) above the surface layer. Where drainage is unimpeded, excess water percolates through to lower layers and may eventually leave the system as runoff.

Water balance in each of the organic and mineral soil layers is modeled as the difference between net inputs of precipitation (PPT), plus, in the case of lower soil layers, addition of volumetric percolation inputs, and outputs of ET, followed by drainage for each profile layer.

3.2. Snowmelt

Snow dynamics algorithms from the Regional Hydroecological Simulation System (RHESSys) developed by Coughlan and Running (Coughlan and Running, 1997) have been added to the NASA–CASA model to improve predictions of snow accumulation rate, and the timing and flow rates of spring snowmelt at high-altitude sites. These snow algorithms were developed to improve estimates of annual forest snow hydrology for point and regional calculations of annual productivity. Model algorithms depend upon surface air temperature, solar insolation, precipitation inputs, and canopy leaf area to compute snowpack water equivalent, snow thermal content, albedo, and albation from snowmelt and sublimation fluxes. Snow accumulation rates are dependent on estimated nighttime air temperatures. A heat summation function is used for estimation of snow thermal content to determine when the snowpack is isothermal. The RHESSys snow model has been successfully tested at 10 snow telemetry (SNOTEL) stations in the western United States (Coughlan and Running, 1997). Comparisons of simulation results to published snow depletion dates have shown that the snow model accurately predicts the relative ranking and magnitude of depletion for different combinations of land cover, elevation, and aspect.

4. Ammonia emissions controllers

For the NASA–CASA component for N trace gas emissions, processes of ammonification and nitrification are lumped into combined mineralization fluxes from litter, microbial, and soil organic matter pools to a common mineral N pool. This design is intended to make the model general enough to be driven by organic matter inputs derived from remote sensing observations.

The general conditions and spatial patterns favoring soil NH3 emissions from soils and chemical fertilizers (many of which have been documented in this section) are represented in the NASA–CASA formulation. Using results from N fertilizer applications to cropped soils to form a conceptual model, our generic NH3 emissions algorithm is built as a function of fertilizer type (FT) and application method (AP) of the applied nitrogen concentration (Napp; g N m−2). In croplands, a correction is made to soil N pool inputs by removing harvested biomass sources (assumed to be 50% of annual NPP for all crop types) from the computation of N available for possible NH3 emissions loss. In the following analysis of native soil emissions and land cover types (i.e., which excludes direct fertilizer emissions sources of NH3), the FT and AP terms are not considered, and Napp is estimated as the monthly mineralization rate of soil nitrogen.

The available mineral N substrate for NH3 emissions is potentially modified by scalars (multipliers ranging from 0 to 1) for soil surface temperature (Ts , °C), pH-dependent response (Figure 5) and a constant term (c), and a soil moisture scalar (Ms). Equation (2) is derived from reports by Denmead et al. (Denmead et al., 1982) and the National Research Council Subcommittee on Ammonia (National Research Council, 1979):

 
formula

The NH3 emissions Equation (2) includes a scalar term for a maximum rate of volatilization (kA), which was set at a value of 0.25 (van der Weerden and Jarvis, 1997; Ryden and McNeill, 1984), pending the availability of more specific measurements from California native soil sources.

Figure 5.

Temperature-dependent scalar for the potential effect of soil pH on volatilization of ammonia. The constant value (c) in Equation (2) is equal to 1.3. Values for the dotted, dashed, and solid lines are 30°, 20°, and 10°C, respectively. After Denmead et al. (Denmead et al., 1982)

Figure 5.

Temperature-dependent scalar for the potential effect of soil pH on volatilization of ammonia. The constant value (c) in Equation (2) is equal to 1.3. Values for the dotted, dashed, and solid lines are 30°, 20°, and 10°C, respectively. After Denmead et al. (Denmead et al., 1982)

4.1. Soil pH effects

As a first approximation, high NH3 volatilization can be strongly affected by relatively high soil pH (7–9) levels, as represented in Figure 5. However, under special circumstances observed (so far) mainly in cropped areas, volatilization losses can occur from acid as well as from alkaline soils. This is due to elevated pH and NH4+ concentrations at wet “microsites” where, for instance, surface-applied urea (CO\[NH2\]2) particles may dissolve and hydrolyze (Fenn and Richards, 1986). Microsite formation of ammonium carbonate (NH4+ HCO3-) by the soil microbial enzyme urease can promote high NH3 volatilization losses, well after urea is incorporated into the soil organic matter. Therefore, in the absence of more definitive observational information on the magnitude of soil pH in controlling NH3 volatilization fluxes from native soils in California, we have estimated statewide NH3 emissions with both a moderate pH effect (version A; c = 1.3; Figure 5) and with a minimal pH effect (version B; c = 10). The value of c = 1.3 in Equation (2) for version A is fairly consistent with soil pH effects observed in recent field measurements of NH3 volatilization losses directly from fertilizer application sources in agricultural soils of California's Central Valley (CARB, 2001), whereby the most rapid increases in NH3 volatilization losses were detected between soil pH levels of 7 and 8. This modeling approach will result in a fairly wide range of potential NH3 volatilization rates, which can nonetheless begin to put rational boundaries on the native soil source of NH3 emissions for the state. Verification of the actual form of the soil pH effect represented in Figure 5 will depend on the analysis of new measurements of NH3 volatilization rates under carefully selected field conditions that are designed to control for native soil pH effects on gas emissions fluxes.

4.2. Moisture effects

Soil-wetting patterns can strongly influence NH3 losses. Generally, moist soils emit less NH3 than drier soils, owing to the high solubility of NH3 and the lower gas diffusivity in wetter soils. To account for an effect of limited diffusion of NH3 gas through relatively moist soil layers in our model, a scalar term (Ms) is defined below as a function of percent volumetric soil moisture content (θ; m m−1) and its soil texture dependence. In this context, the Ms scalar is intended to represent the damping influence of elevated soil moisture conditions on NH3 gas movement through the soil:

 
formula

where a and b are soil texture–dependent empirical coefficients (defined as in Saxton et al., 1986). According to this equation, heavy clay soils will retain moisture longer than sandy soils, thereby reducing emissions of NH3 in clays under any given conditions of water supply.

It is worth mentioning, however, that under certain nitrogen application conditions, even moist soils have been shown to emit higher amounts of applied fertilizer NH3 than drier soils. As previously cited, hydrolysis of urea is promoted under conditions of elevated soil moisture, which can then enhance evaporation losses as NH3 and CO2. Volatilization rates are typically diminished when, for example, urea can be rapidly transported to deeper soil layers following heavy irrigation (Fenn and Miyamoto, 1981). Field studies suggest that merely delaying urea application for a few hours after irrigation to avoid accumulation at wet soil surfaces may be a practical way to reduce NH3 volatilization in humid areas (Priebe and Blackmer, 1989). High temperatures and strong winds may interact with humidity and soil moisture to promote higher volatilization losses. However, in the winter, natural snow cover and cold temperatures can decrease airborne soil dust and possibly the evolution of NH3 from soils (Munger, 1982).

5. Statewide ammonia emissions estimates from native soil sources

5.1. Emissions modeling results

Based on our NASA–CASA model inventory estimate, statewide emissions of NH3 from native soil N sources could range from 12 to 57 × 106 kg N-NH3 annually (Table 2). The low end of this range is computed with Equation (2) version A, using the moderate pH effect on soil emissions of NH3 (Figure 4), while the high-end estimate is derived from the model version B operating with minimal pH effect. Using version A for Equation (2), the model predicts that cropland and semiagricultural soils generate 60% of the statewide NH3 emissions total (Figure 6), whereas version B predicts that these same areas generate 33% of statewide NH3 emissions from soils (Figure 7). This difference derives from the pattern of relatively large coverage by high pH soils (6.5–8) in the large agricultural counties like Fresno, Kern, Kings, and Imperial, which in version A elevates the importance of predicted NH3 emissions in croplands, while severely depressing emissions in areas of evergreen needleleaf forest and other woodlands with relatively lower soil pH.

Table 2.

Estimated emissions of N-NH3 from native soil sources in California using the NASA–CASA model [Equation (2)]

Estimated emissions of N-NH3 from native soil sources in California using the NASA–CASA model [Equation (2)]
Estimated emissions of N-NH3 from native soil sources in California using the NASA–CASA model [Equation (2)]
Figure 6.

NASA–CASA model results [Equation (2) version A] for statewide ammonia emissions from native soil sources

Figure 6.

NASA–CASA model results [Equation (2) version A] for statewide ammonia emissions from native soil sources

Figure 7.

NASA–CASA model results [Equation (2) version B] for statewide ammonia emissions from native soil sources

Figure 7.

NASA–CASA model results [Equation (2) version B] for statewide ammonia emissions from native soil sources

The majority of model results subsequently reported in this study will be based upon the high-end estimate from version B, which would appear to be a more conservative approach (particularly in a state with such large cropland areas), and is justified until the time when enough flux measurements are collected to determine the actual magnitude of soil pH effects on NH3 emissions. As indicated above, the single most important land cover type in terms of contributions to the statewide emissions inventory for emissions of NH3 is evergreen needleleaf forest (36%), followed by cropland and semiagricultural lands (Figure 7). Croplands make up one-third of the total native soil N sources for NH3 emissions annually. This results from the high nitrogen levels in the soils of these cropland areas that are periodically fertilized for crop production purposes. Nitrogen that remains in the surface soils and dead root systems each year can mineralize, and potentially on the alkaline soil areas (Figure 2), volatilize under optimal conditions of moisture and temperature to generate elevated emissions of NH3 gas.

Other native source areas that contribute substantially (22%) to the statewide emissions inventory for emissions of NH3 are soils of woodlands and wooded grassland ecosystems (Figure 7), mainly on the basis of their large area coverage of the state's natural areas. The combination of productive vegetation communities growing on even slightly alkaline soils (pH>5.5) results in the largest annual emissions of NH3 from native soil N sources outside the cropland and semiagricultural land areas.

On the basis of Equation (2) version-B estimates of NH3 emissions from native soil sources, there appear to be several areas in the state with notably high annual NH3 fluxes (> 0.4 g N m−2 yr−1). Within the major valley cropland areas of the state, these include locations in the general vicinity of the geographic coordinates and cities listed in Table 3. The model also predicts that there are several areas of high annual NH3 fluxes that are classified with a land cover of evergreen needleleaf forest, wooded grassland, or annual grassland (Figure 1). All of these locations listed in Table 3 are associated with relatively high soil pH input values of 6 or greater.

Table 3.

Locations of elevated NH3 emissions from native soil sources in California, predicted by the NASA–CASA model [Equation (2) version B]

Locations of elevated NH3 emissions from native soil sources in California, predicted by the NASA–CASA model [Equation (2) version B]
Locations of elevated NH3 emissions from native soil sources in California, predicted by the NASA–CASA model [Equation (2) version B]

Our model estimates suggest that Mendocino, San Luis Obispo, and Shasta counties, in addition to San Joaquin Valley counties like Fresno and Kern, are among the largest contributors of soil NH3 fluxes as single-county fractions of the statewide total (in 106 kg yr−1), but this is mainly due to extensive area coverage, including 40%–50% cropland soils in Fresno and Kern counties. On a per-squared-meter basis, Sacramento Valley counties, including Yuba and Yolo, as well as Lake, Santa Barbara, and Sonoma counties are estimated to emit among the highest annual flux rates (g N m−2 yr−1) of NH3 from native soil sources in the state. This is mainly due to the comparatively productive soils of these county areas (Figure 4) and the high predicted rates of mineral nitrogen cycling on an annual basis.

When viewed in more detail for several selected areas in the state (i.e., those with notably high annual fluxes; Table 3), seasonal flux patterns for NH3 emissions from native soil sources appear to be influenced in the model by a combination of surface air temperature and moisture patterns (Figure 8). Throughout the state, air temperature and precipitation patterns show a reverse seasonal relationship: rainfall is highest from November to March, whereas surface temperatures reach a maximum in July or August. In response, the NASA–CASA model generally predicts that spring (March–May) is the season with the lowest predicted NH3 emissions, since native soils are predicted to remain relatively cool and moist until the summer months. Peak estimated emissions of soil NH3 come in midsummer to early fall (July–October), when native soils are predicted to become relatively warm and dry. A pulse of mineralized nitrogen for NH3 volatilization is predicted in October when foliage in the model is shed from vegetation, and this relatively labile organic matter begins to decompose with wetting from early season rainfall.

Figure 8.

NASA–CASA model results [Equation (2) version B] for seasonal NH3 emission rates (g N m−2 month−1) from native soil sources at four selected locations from Table 3. Cropland soils: (a) Orland, Glenn County, and (b) Lindsay, Tulare County. Coniferous forest soil: (c) Wynton, Shasta County. Wooded grassland soil: (d) Cuyama, Santa Barbara County. Temperatures in degrees Celsius, precipitation in centimeters per month

Figure 8.

NASA–CASA model results [Equation (2) version B] for seasonal NH3 emission rates (g N m−2 month−1) from native soil sources at four selected locations from Table 3. Cropland soils: (a) Orland, Glenn County, and (b) Lindsay, Tulare County. Coniferous forest soil: (c) Wynton, Shasta County. Wooded grassland soil: (d) Cuyama, Santa Barbara County. Temperatures in degrees Celsius, precipitation in centimeters per month

These predicted seasonal patterns for NH3 soil emissions for contiguous 8-km zones covering the entire state are shown in Figure 9. The model predicts that October is the peak month overall for NH3 emissions from native soils in California. When totaled for the entire region, native soil sources of NH3 predicted for Central Valley counties are consistently high from July through January. Zones of consistently high seasonal emissions (July–January) are predicted in the Sacramento Valley between 40° and 38°15′N, in the San Joaquin and Salinas Valleys between 37°30′ and 36°15′N, and in the Central Coast Valleys between 36° and 35°N latitude. This seasonal pattern in predicted soil NH3 emissions is fairly consistent with observed seasonality in PM2.5 levels for the San Joaquin Valley Air Basin.

Figure 9.

NASA–CASA model results [Equation (2) version B] for total seasonal NH3 emissions rates from native soil sources. Emissions fluxes are summed across 8-km latitude zones for each month

Figure 9.

NASA–CASA model results [Equation (2) version B] for total seasonal NH3 emissions rates from native soil sources. Emissions fluxes are summed across 8-km latitude zones for each month

5.2. Potential foliar absorption fluxes for ammonia

Our NASA–CASA model results for NH3 emissions are designed to represent gas flux at the soil surface, as might be measured using a small closed chamber for field sampling. Nonetheless, when considering the whole ecosystem effect on net fluxes of this biogenic gas, it is important to note that some fraction of the NH3 emitted from native soil sources could be recaptured through leaf absorption into actively growing vegetation on that soil (Sutton et al., 1995). Experimental work suggests that exchange through leaf stomata is the major process for NH3 gas uptake by plants (Husted and Schjøerring, 1996). In contrast to this process of stomatal NH3 absorption into foliar tissues, we do not, however, consider atmospheric deposition of NH3 onto canopy surfaces as a (biogenically) comparable net ecosystem recycling pathway for local soil NH3 sources. While dry deposition is recognized as a potentially important regional N input (Erisman et al., 1998), particularly for areas located in proximity to livestock enclosures or large fertilizer application activities, we would categorize the atmospheric dry deposition flux as a relatively passive process, compared to active biological emission or absorption of NH3 by living components of the ecosystem.

To date, there are few reported assessments of canopy NH3 absorption that can be used to estimate the large-scale (i.e., regional) processes of foliar sink fluxes of locally emitted (soil) NH3 sources. Therefore, a more theoretical approach is adopted for our analysis, based on the concept of NH3 compensation point (Farquhar et al., 1980) and temperature dependence of NH3 absorption and emissions by plant leaves. The following leaf exchange rate functions for NH3 are derived from linear approximation of NH3 compensation point results from Schjøerring et al. (Schjøerring et al., 1998):

 
formula

where Ea and Ef are absorption and emissions by plant leaves, respectively, in units of micrograms NH3-N m−2s−1, and T is air surface temperature (°C) as a surrogate for leaf temperature.

Statewide application of these foliar absorption and emissions equations (4) and (5), using the same 8-km mean monthly temperature maps that were generated to drive our NASA–CASA model for native soil NH3 emissions, results in a potential absorption flux of nearly 206 × 106 kg NH3-N yr−1 by vegetation in California (Table 4). This statewide total for foliar absorption of NH3 must be treated as a potential sink flux only, because it is based on the assumption that foliar density of the vegetation cover throughout the state, on a year-round basis, is high enough and continually in a nondormant state (with respect to NH3 absorption capacity estimated by Schjøerring et al., 1998), so as to maintain a consistent sink for atmospheric NH3 pools. While this assumption is subject to verification from extensive year-round field measurements, the potential fluxes shown in Table 4 imply that absorption by vegetation cover could make ecosystems a strong net sink for locally emitted NH3 sources.

Table 4.

Potential foliar absorption of NH3 estimated for vegetation classes in California

Potential foliar absorption of NH3 estimated for vegetation classes in California
Potential foliar absorption of NH3 estimated for vegetation classes in California

Regional patterns of the potential vegetation sink fluxes (Table 4) suggest that seasonal temperature conditions are most favorable for foliar NH3 absorption year-round primarily along the Pacific coast in the central and southern portions of the state, as well as in the Central Valley areas during the months of May–October. The desert areas in the southeastern portion of the state are potential emissions zones for foliar NH3, provided that vegetation cover there is dense enough and actively growing during the period of May–September.

5.3. Comparison of model to emissions measurements

To our knowledge, all soil NH3 emissions rates reported in the literature (e.g., Bouwman et al., 1997) have not been measured outside of California. Hence, these published data cannot be used directly for validation of our predicted emissions rates from native soils in the state. Moreover, measured trace gas emissions rates from soils are highly variable in space and time, which makes comparisons to average emissions estimates less meaningful and underscores the critical need to characterize local emissions conditions for accurate flux inventories in any region. Consequently, the data summarized in this section are intended mainly to furnish guidance for new measurements and regional model development on the potential scope of NH3 emissions rates expected in environments similar to those found in California. In this manner, new measurements of NH3 emissions from ecosystem sources in California can be better placed into the larger context of emissions nationwide.

Reported NH3 emissions rates were extracted from literature and entered into a database for synthesis of standardized flux measurements. Whenever possible, we grouped the emissions data according to natural vegetation types of California (Mayer and Laudenslayer, 1988). The overall availability of emissions measurements for native soils reveals many missing data values for major vegetation types common to California (Table 5). Only shrub lands and rangelands are represented with more than a few published emission measurements. Nonetheless, the NASA–CASA model predicts that most natural (noncropland) ecosystems in California emit NH3 from soils at rates between 5 and 25 μg N m−2 hr−1 (based on Figure 6 data values), which is well within the range of measured emissions rates reported by the sources listed in Table 5. In a few cases, measured loss rates may be over 20% of mineralized soil nitrogen, although there is little knowledge of the potential timing and extent of such high NH3 emissions from natural rangelands.

Table 5.

Soil ammonia emissions rates reported for natural vegetation types found in California

Soil ammonia emissions rates reported for natural vegetation types found in California
Soil ammonia emissions rates reported for natural vegetation types found in California

According to our survey of the literature, there are no emissions measurements for soil NH3 sources currently available for oak woodland and chaparral ecosystems, which cover a substantial portion of California's wild lands. Information on net NH3 fluxes in conifer forests of the Sierra Nevada mountain regions is also lacking in the available literature. To fill these gaps in soil NH3 measurements and validation datasets for our emissions model, we are currently carrying out field measurement campaigns to determine seasonal flux rates and emissions factors for ammonia emissions from selected native soils in California. Sites have been selected for sampling in Sierran mixed conifer (mainly Ponderosa pine) forest, foothill blue oak/pine woodland, sagebrush, mixed chaparral, desert scrub, and annual grassland.

6. Summary and conclusions

Statewide emissions of NH3 from soil sources are predicted to range from 12 to 57 × 106 kg N annually, with nearly one-third to one-half from the croplands and semiagricultural ecosystems that cover only 20% of the total area of the state. In comparison, we have estimated in separate field measurement studies (CARB, 2001) that emissions of NH3 directly from chemical fertilizer applications in California total nearly 12 × 106 kg N annually. In a statewide budget, the CARB has estimated that livestock emissions sources of NH3 could exceed 120 × 106 kg N annually (Gaffney and Shimp, 1999).

Outside of croplands, other native soil areas that contribute substantially to the statewide emissions inventory for emissions of NH3 are the evergreen needleleaf forests, woodland, and wooded grassland ecosystems, mainly on the basis of their large area coverage of the state's natural areas. The combination of productive vegetation communities growing on (even slightly) alkaline soils results in the largest annual emissions of NH3 from native soil N sources.

The NASA–CASA model predicts that October is the peak month overall for NH3 emissions from native soils in California. When totaled for the entire region, native soil sources of NH3 predicted for Central Valley counties are highest from July through January.

A preliminary (subject to field validation studies) regional model budget for potential foliar absorption fluxes of ammonia implies that vegetation cover on a statewide basis could actually make ecosystems a strong net sink for locally emitted NH3 sources.

The spatial approach for ecosystem modeling presented in this study is extensible and flexible enough to allow inclusion of new data of all types as they become available. Nevertheless, this study marks the first instance in which the state agency charged with emissions inventory assessments is able to spatially allocate NH3 fluxes statewide for all ecosystem classes and native soil types.

Key to land cover classes

Key to land cover classes
Key to land cover classes

Acknowledgments

This research was funded by California Air Resources Board Contract No. 98–716 to California State University, Fresno, and to California State University, Monterey Bay, and by NASA Ames Research Center. The authors acknowledge the contributions of the following individuals and institutions: Michael Benjamin and Patrick Gaffney of the California Air Resources Board; Vanessa Brooks-Genovese and Alicia Torregrosa from NASA Ames Research Center and California State University, Monterey Bay, who supported all the spatial data analysis and quality control for statewide datasets; and Casey Walsh Cady of the California Dept. of Food and Agriculture.

References

References
Agbu
,
P. A.
and
M. E.
James
.
1994
.
NOAA/NASA Pathfinder AVHRR Land Data Set User's Manual.
Goddard Distributed Active Archive Center, NASA Goddard Space Flight Center, Greenbelt, MD, 28 pp
.
Birkeland
,
P. W.
1974
.
Pedology, Weathering and Geomorphological Research.
Oxford University Press, New York, 285 pp
.
Bishop
,
J. K. B.
and
W. B.
Rossow
.
1991
.
Spatial and temporal variability of global surface solar irradiance.
J. Geophys. Res.
96
:
16839
16858
.
Bonan
,
G. B.
1989
.
A computer model of the solar radiation, soil moisture and soil thermal regimes in boreal forests.
Ecol. Modell.
45
:
275
306
.
Bouwman
,
A. F.
,
D. S.
Lee
,
W. A. H.
Asman
,
F. J.
Dentener
,
K.
van der Hoek
, and
J. G. J.
Oliver
.
1997
.
A global high resolution emission inventory for ammonia.
Global Biogeochem. Cycle
11
:
561
587
.
Bristow
,
K. L.
and
G. S.
Campbell
.
1984
.
On the relationship between incoming solar radiation and daily maximum and minimum temperature.
Agric. For. Meteorol.
31
:
159
166
.
Burch
,
J. A.
and
R. H.
Fox
.
1989
.
The effect of temperature and initial soil moisture content on the volatilization of ammonia from surface-applied urea.
Soil Sci.
147
:
311
318
.
Bytnerowicz
,
A.
and
M. E.
Fenn
.
1996
.
Nitrogen deposition in California forests: A review.
Environ. Pollut.
92
:
127
146
.
CARB,
2001
.
Statewide inventory estimates of ammonia emissions from native soils and chemical fertilizers in California.
CARB Contract Number ID 98-716, California State University, Fresno, CA, 95 pp
.
Camire
,
C.
and
B.
Bernier
.
1981
.
Retention de l'azote et evolution des proprietes d'un humus brut de station de pin gris (Pinus banksiana Lamb.) apres application d'engrais azotes.
Can. J. For. Res.
11
:
51
61
.
Campbell
,
G. S.
1977
.
An Introduction to Environmental Biophysics.
Springer, New York, 159 pp
.
Chow
,
J. C.
,
J. G.
Watson
,
D. H.
Lowenthal
,
P. A.
Solomon
,
K.
Magliano
,
S.
Ziman
, and
L. W.
Richards
.
1992
.
PM10 source apportionment in California's San Joaquin Valley.
Atmos. Environ., 26A, 3335–3354
.
Coughlan
,
J. C.
and
S. W.
Running
.
1997
.
Regional ecosystem simulation: A general model for simulating snow accumulation and melt in mountainous terrain.
Land. Ecol.
12
:
119
136
.
Davidson
,
E. A.
and
P. A.
Lefebvre
.
1993
.
Estimating regional carbon stocks and spatially covarying edaphic factors using soil maps at three scales.
Biogeochemistry
22
:
107
131
.
Davidson
,
E. A.
,
C. S.
Potter
,
P.
Schlesinger
, and
S. A.
Klooster
.
1998
.
A regional model of NO emission from soils.
Ecol. Appl.
8
:
748
759
.
DCW,
1993
.
Digital Chart of the World: For Use with ARC/INFO Software.
Environmental Systems Research Institute, Inc., Redlands, CA
.
DeFries
,
R.
,
M.
Hansen
, and
J.
Townshend
.
1995
.
Global discrimination of land cover types from metrics derived from AVHRR Pathfinder data.
Remote Sens. Environ.
54
:
209
222
.
Denmead
,
O. T.
,
J. R.
Freney
, and
J. R.
Simpson
.
1976
.
A closed ammonia cycle within a plant canopy.
Soil Biol. Biochem.
8
:
161
164
.
Denmead
,
O. T.
,
R.
Nulsen
, and
G. W.
Thurtell
.
1978
.
Ammonia exchange over a corn crop.
Soil Sci. Soc. Am. J.
42
:
840
842
.
Denmead
,
O. T.
,
J. R.
Freney
, and
J. R.
Simpson
.
1982
.
Dynamics of ammonia volatilization during furrow irrigation of maize.
Soil Sci. Soc. Am. J.
46
:
149
155
.
Dentener
,
F. J.
and
P. J.
Crutzen
.
1994
.
A three-dimensional model of the global ammonia cycle.
J. Atmos. Chem.
19
:
331
369
.
DWR,
1998
.
cited 1993–1998: California Department of Water Resources.
Standard Land Use Legend. Division of Planning, Sacramento, CA. \[Available online at http://wwwdpla.water.ca.gov/cd/landwateruse/1998legend.html.\]
.
Erisman
,
J. W.
,
A.
Bleeker
, and
J. A.
van Jaarsveld
.
1998
.
Atmospheric deposition of ammonia to semi-natural vegetation in the Netherlands: Methods for mapping and evaluation.
Atmos. Environ.
32
:
481
490
.
Farquhar
,
G. D.
,
P. M.
Firth
,
R.
Wetselaar
, and
B.
Weir
.
1980
.
On the gaseous exchange of ammonia between leaves and the environment: Determination of the ammonia compensation point.
Plant Physiol.
66
:
710
714
.
Fenn
,
L. B.
and
S.
Miyamoto
.
1981
.
Ammonia loss and associated reactions of urea in calcareous soils.
Soil Sci. Soc. Am. J.
45
:
537
540
.
Fenn
,
L. B.
and
J.
Richards
.
1986
.
Ammonia loss from surface applied urea-acid products.
Fert. Res.
9
:
265
275
.
Fitz
,
D.
,
D.
Pankratz
,
J.
Zwicker
,
E.
Ringler
,
T.
Waldron
, and
D.
Coe
.
1997
.
Evaluation and improvement of methods for determining ammonia emissions in the San Joaquin Valley.
Field data report prepared for California Regional PM10 and PM2.5 Air Quality Study, Sacramento, CA, by the Center for Enviromental Research and Technology, College of Engineering, Riverside, CA, Rep. 97-AP-RT37-002-FR, 54 pp
.
FAO,
1971
.
Soil Map of the World: 1:5,000,000.
United Nations Educational, Scientific, and Cultural Organization, Paris, France
.
Frolking
,
S. E.
et al.,
.
1998
.
Comparison of N2O emissions from soils at three temperate agricultural sites: Simulations of year-round measurements by four models.
Nutr. Cycle Agroecosyst.
55
:
77
105
.
Gaffney
,
P.
and
D.
Shimp
.
1999
.
Ammonia emission inventory development: Needs, limitations, and what is available now.
First Particulate Matter Emission Inventory Research Workshop, California Air Resources Board, Sacramento, CA, 22 October 1999
.
Gordon
,
H. R.
,
J. W.
Brown
, and
R. H.
Evans
.
1988
.
Exact Rayleigh scattering calculations for use with the Nimbus-7 coastal zone color scanner.
Appl. Opt.
27
:
2111
2122
.
Holben
,
B. N.
1986
.
Characteristics of maximum-value composite images from temporal AVHRR data.
Int. J. Remote Sens.
7
:
1417
1434
.
Husted
,
S.
and
J. K.
Schjøerring
.
1996
.
Ammonia fluxes between Brassica napus plants and the atmosphere in response to changes in leaf temperature, light intensity and air humidity. Interactions with stomatal conductance and apoplastic NH4+ and H+ concentrations.
Plant Physiol.
112
:
67
74
.
Jensen
,
M. E.
1973
.
Consumptive Use of Water and Irrigation Water Requirements.
American Society of Civil Engineers, New York, 215 pp
.
Jensen
,
M. E.
and
H. R.
Haise
.
1963
.
Estimating evapotranspiration from solar radiation.
J. Irrig. Drain. Div. Am. Soc. Civ. Eng.
89
:
15
41
.
Kidwell
,
K.
1991
.
NOAA Polar Orbiter Data User's Guide.
NCDC/SDSD. National Climatic Data. Center, Washington, DC, 13 pp
.
Kittel
,
T. G. F.
,
N. A.
Rosenbloom
,
D. S.
Schimel
,
T. H.
Painter
,
H. H.
Fisher
, and
A.
Grimsdell
.
VEMAP Members,
.
cited 1995
.
The VEMAP United States Database.
\[Available online at http://www.cgd.ucar.edu/vemap/.\]
.
Langford
,
A. O.
and
F. C.
Fehsenfeld
.
1992
.
Natural vegetation as a source or sink for atmospheric ammonia: A case study.
Science
255
:
581
583
.
Langford
,
A. O.
,
F. C.
Fehsenfeld
,
J.
Zachariassen
, and
D. S.
Schimel
.
1992
.
Gaseous ammonia fluxes and background concentrations in terrestrial ecosystems of the United States.
Global Biogeochem. Cycle
6
:
459
483
.
Los
,
S. O.
,
C. O.
Justice
, and
C. J.
Tucker
.
1994
.
A global 1×1 NDVI data set for climate studies derived from the GIMMS continental NDVI data.
Int. J. Remote Sens.
15
:
3493
3518
.
Matthews
,
E.
1994
.
Nitrogen fertilizer consumption: Global distribution and associated emissions of nitrous oxide and ammonia.
Global Biogeochem. Cycle
8
:
411
440
.
Mayer
,
K. E.
and
W. F.
Laudenslayer
Jr.
.
1988
.
A Guide to Wildlife Habitats of California.
California Department of Forestry and Fire Protection, Sacramento, CA, 166 pp
.
McInnes
,
K. J.
,
R. B.
Ferguson
,
D. E.
Kissel
, and
E. T.
Kanemasu
.
1986
.
Field measurements of ammonia loss from applications of urea solution to bare soil.
Agron. J.
78
:
192
196
.
Munger
,
J. W.
1982
.
Chemistry of atmospheric precipitation in the north central USA: Influence of sulfate nitrate ammonia and calcareous soil particulates.
Atmos. Environ.
16
:
1633
1646
.
National Research Council,
1979
.
Committee on Medical and Biologic Effects of Environmental Pollutants, Subcommittee on Ammonia.
University Park Press, Baltimore, MD
.
New
,
M.
,
M.
Hulme
, and
P.
Jones
.
2000
.
Representing twentieth century space–time climate variability. Part II: Development of 1901–96 monthly grids of terrestrial surface climate.
J. Clim.
13
:
2217
2238
.
Pain
,
B. F.
,
T. J.
Van Deerden
,
B. J.
Chambers
,
V. R.
Phillips
, and
C.
Jarvis
.
1998
.
A new inventory for ammonia emissions from UK agriculture.
Atmos. Environ.
32
:
(3)
.
309
313
.
Parton
,
W. J.
,
B.
McKeown
,
V.
Kirchner
, and
D.
Ojima
.
1992
.
CENTURY User's Manual.
Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO, 35 pp
.
Potter
,
C. S.
1999
.
Terrestrial biomass and the effects of deforestation on the global carbon cycle.
BioScience
49
:
769
778
.
Potter
,
C. S.
and
S. A.
Klooster
.
1998
.
Interannual variability in soil trace gas (CO2, N2O, NO) fluxes and analysis of controllers on regional to global scales.
Global Biogeochem. Cycle
12
:
621
637
.
Potter
,
C. S.
,
J. T.
Randerson
,
C. B.
Field
,
P. A.
Matson
,
P. M.
Vitousek
,
H. A.
Mooney
, and
S. A.
Klooster
.
1993
.
Terrestrial ecosystem production: A process model based on global satellite and surface data.
Global Biogeochem. Cycle
7
:
811
841
.
Priebe
,
D. L.
and
A. M.
Blackmer
.
1989
.
Soil moisture content at time of application as a factor affecting losses of nitrogen from surface-applied urea.
J. Fert. Issues
6
:
(3)
.
62
67
.
Priestly
,
C. H. B.
and
R. J.
Taylor
.
1972
.
On the assessment of surface heat flux and evaporation using large-scale parameters.
Mon. Weather Rev.
100
:
81
92
.
Ryden
,
J. C.
and
J. E.
McNeill
.
1984
.
Application of the micrometeorological mass balance method to the determination of ammonia loss from a grazed sward.
Food Agric.
35
:
1297
1310
.
Saxton
,
K. E.
,
W. J.
Rawls
,
J. S.
Romberger
, and
R. I.
Papendick
.
1986
.
Estimating generalized soil-water characteristics from texture.
Soil Sci. Soc. Am. J.
50
:
1031
1036
.
Schimel
,
D. S.
,
W. J.
Parton
,
F. J.
Adamsen
,
R. G.
Woodmansee
,
R. L.
Senft
, and
M. A.
Stillwell
.
1986
.
The role of cattle in the volatile loss of nitrogen from a shortgrass steppe.
Biogeochemistry
2
:
39
52
.
Schjøerring
,
J. K.
,
S.
Husted
, and
M.
Mattsson
.
1998
.
Physiological parameters controlling plant–atmosphere ammonia exchange.
Atmos. Environ.
32
:
491
498
.
Schlesinger
,
W. H.
and
W. T.
Peterjohn
.
1991
.
Processes controlling ammonia volatilization from Chihuahuan desert soils.
Soil Biol. Biochem.
23
:
637
642
.
Sutton
,
M. A.
,
J. K.
Schjøerring
, and
G. P.
Wyers
.
1995
.
Plant–atmosphere exchange of ammonia.
Philos. Trans. R. Soc. London
351
:
261
278
.
van der Weerden
,
T. J.
and
S. C.
Jarvis
.
1997
.
Ammonia emission factors for N fertilizers applied to two contrasting grassland soils.
Environ. Pollut.
95
:
205
211
.
Vose
,
R. S.
,
R. L.
Schmoyer
,
P. M.
Steurer
,
T. C.
Peterson
,
R.
Heim
,
T. R.
Karl
, and
J.
Eischeid
.
1992
.
The Global Historical Climatology Network: Long-term monthly temperature, precipitation, sea level pressure, and station pressure data.
ORNL/CDIAC-53, NDP-041, Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, TN
.
Warn
,
T. E.
,
S.
Zelmanowitz
, and
M.
Saeger
.
1990
.
Development and selection of ammonia emission factors for the 1985 NAPAP Emissions Inventory. EPA-600/7-90-014,.
U.S. Environmental Protection Agency, Research Triangle Park, NC, 30 pp
.
West
,
N. E.
and
J.
Skujins
.
1977
.
The nitrogen cycle of North American cold-winter semi-desert ecosystems.
Oecologia Plantarum
12
:
45
53
.
Woodmansee
,
R. G.
1978
.
Additions and losses of nitrogen in grassland ecosystems.
Bioscience
28
:
448
453
.

Footnotes

* Corresponding author address: C. Potter, Ecosystem Science and Technology Branch, NASA Ames Research Center, Moffett Field, CA 94035. cpotter@mail.arc.nasa.gov