Sullivan Environmental and our research partner Ajwa Analytical Laboratories have conducted over 50 studies to characterize emission rates as a function of time from the application and subsequent off-gassing of agricultural fumigants.  Many insights gathered from this research have broader applications to other complex area sources of air pollutants.

The first question:  what is an agricultural fumigant?    When you go to the store and see high-quality produce, much of what you see was produced with the benefit of agricultural fumigants.  Fumigants are generally applied as a liquid or gas and injected into the ground prior to planting, fumigants create an environment that is more conducive to the growth of young crops by suppressing weeds, disease, and nematodes (small worm-like creatures that nibble at the tender roots).    Applications general are made by injecting a fumigant(s) into the soil, or via drip irrigation, hand-line, or center pivot application.    The chemicals break down in the soil (biodegrade) but some of the liquid volatilizes and is released into the atmosphere.  We measure how much is released across test fields as a function of time, generally for 4 days to 14 days after application.  We typically find daytime and nighttime differences in emission rates with a steadily decreasing trend.  By two weeks, the chemicals typically present are in negligible quantities in the soil and off-gassing effectively ceases in most cases.

How would the lessons learned from agricultural fumigants be useful for other sources such as landfills, waste lagoons, oil & gas fields, or complex chemical operations?  The methods used and refined for agricultural fumigants can be directly applied in most cases or adapted in others to refine the estimation of emission rates.  Why is this important?  In air quality analysis, as in most analyses involving health and safety, conservative assumptions are used to simplify the analysis when more specific data are not available.  One approach to refine, and in many cases lower the emission rates beyond default treatments, is to empirically measure the emissions in representative circumstances.

We use two methods that can be applied to most area source assessments:  (1) profile sampling in conjunction with the integrated horizontal flux (IHF) method, and (2) the back-calculation method whereby air concentrations are measured all around an area source and dispersion modeling is used to compute the emission rate based on regression methods.  We have found the following refinements to be important:

The IHF Method is Generally Preferred as the More Cost-Effective and Efficient Way to Conduct the Research

• If only the average emission rate across the area source, one profile is generally sufficient.
• If there is a need to evaluate the variability in emission rates across an area source, establishing a profile in each quadrant and in the center of the area source provides a basis to evaluate both composite emissions and spatial variability.
• Five-level profiles are generally required by regulatory agencies for agricultural fumigants. We find five-level profiles helpful if there is a need to quantify uncertainty.  On the other hand,  with three-level profiles we essentially get the same straight line of concentration or wind speed as a function of the natural logarithm of height (at least within the first 3 m of the atmosphere where most profiles are established).
• If an active source, there may be a need to either wear a respirator and other personal protective equipment or to have the respirator available as needed.
• We rely on solar-powered and shielded environments for our sampling pumps. Changing batteries at 3:00 AM or having sampling pumps buffeted by rain and wind obviously are problems to be avoided.  The shelters and equipment need to be durable. Murphy’s Law must be respected.  The wind will blow 50 mph (which has happened) and there will be torrential rains.  The samplers must continue working.
• 2D sonic anemometer profiles are preferred. While cup and vane systems are much more interesting to watch as compared to a sonic anemometer with no moving parts, the only monitors that are reliable in all wind conditions are sonic.  If you do not know wind speed (generally because it is below the threshold) you cannot compute flux with either the IHF or back-calculation method.  It simply is not worth the risk of using the less costly standard cup and vane wind sensors.

The Back-Calculation Method is Preferred in Certain Circumstances

This method requires full coverage around the area source with air quality monitors and one representative source of wind data.  Generally, eight monitors are spaced approximately uniformly around the compass at a distance of 10-25 m from the edge of the area source.   It is never a good idea to try to second guess the wind and focus the monitors downwind of the prevailing flow.  Murphy’s Law will demonstrate the problem.    The method is simple in concept.  A dispersion model, such as AERMOD, is run with a normalized emission rate of 1 μg m^-2 sec^-1, and period-average concentrations are computed for each measured time block for each monitoring site.  Regression is then used to compute the emission rate.

The back-calculation method is ideally suited to circumstances where there is an access complication or a lot of complexity within the source.  Examples would be:

• A lagoon where establishing a profile is problematic (although profiles could be established near the edge at all four quadrants).
• A complex chemical facility or other locations where the emissions are not uniformly distributed.
• Sources where there are major obstructions to flow, such as buildings, trees, or crops that would adversely affect in-field profiles.

Atmospheric Dilution Rates Over the Area Source Can be Different from Standard Modeling Assumptions

When using the back-calculation method, or when modeling exposures based on flux data computed by any method, it is important to accurately represent the rate of dispersion of airborne emissions.  This is especially true when there is a need to model off-source exposures in the near-field, such as within ~ 100 m of a source.    Dispersion models assume dilution rates based on meteorological conditions.  During sunny summer afternoons with light winds, vigorous atmospheric dilution will be assumed.  On the other hand, during nighttime conditions with light winds and clear skies, standard model assumptions will treat dilution as substantially suppressed, i.e. inversion conditions.  One complication for area sources is that the standard assumption may not apply to the particular source at hand.  Agricultural fumigants are an excellent example.  Irrigated and tarped bedded fields experience dilution conditions that are more consistent with neutral atmospheric conditions (moderate dilution) than stable, inversion conditions.  This condition has been measured many times with mid-field profiles.  Similarly, during daytime conditions that would be characterized as unstable with vigorous mixing, the mid-field temperature profiles show trending more towards neutral conditions.  Other area sources could experience similar observations, such as a waste water lagoon.  Measuring data to characterize atmospheric stability, such as temperature profiles and co-variance monitoring of sensible and latent heat flux over the source in question can refine back-calculated emission rates and subsequent near-field modeling.

Avoid Modeling Results from Area Sources Associated with Intermittent Sources as Though Always Operating at the Maximum Emission rates

Intermittent sources are problematic for standard modeling analysis.  As an example, consider a chemical plant that makes a special batch of a product a few times per year.  When the process for the particular batch is in operation, emissions occur in a predictable manner until completed, and then the batch ceases to operate.  The same intermittent process occurs for agricultural fumigants where perhaps once every three years a field will be fumigated.  Should a modeler assume that the process is always in operation?  Well, that is one valid way to conservatively approach the problem.  This approach will overstate the impacts and provide a way to confidently regulate operations.  The alternative approach is to treat the emissions in a Monte Carlo manner and “turn on” the source on a random basis consistent with the operational frequency and seasonality of the source.  The Monte Carlo approach will provide an unbiased estimate of the expected distribution of exposures.

In the 1990s, Sullivan Environmental developed the TOXST model for the U.S. Environmental Protection Agency which provides a means to address intermittent sources in a Monte Carlo manner.  Although developed for ISCST3 dispersion modeling, the same basic methodology can be used with the primary dispersion model in use at this time, i.e. AERMOD, through the proper use of hourly emission files and post-processing.  A specialized model, FEMS, provides a Monte Carlo solution for agricultural fumigants.  The FEMS approach also could be used for other complex areas source with substantial variability in emissions.