Pictures of chickens, flowers, wheat, a boy looking through a magnifying glass, irrigation pipe, soybean pods, and fruits and vegetables.

Cooperative Extension Service

Search | Publications | Jobs | Personnel Directory | Links
County Offices | Departments

About Us

Find Us

For the Media

Agriculture

Aquaculture
       & Fisheries
Beef
Beekeeping
Corn
Cotton
Dairy
Forage/Pasture
Forestry
Grain Sorghum
Horses
Horticulture
      Commercial
Poultry
Rice
Soybean
Specialty Agriculture
Swine
Wheat

Links
Newsletters

Business & Communities

Families & Consumers

Health & Nutrition

Home & Garden

Natural Resources

4-H Youth Development

Public Policy Center

For Faculty & Staff

Giving

Dale Bumpers College
of Agricultural, Food &
Life Sciences

Division Home

Agricultural Experiment
      Station Home

Cooperative Extension
      Service Home

Animal Manure and Mortality Management
Arkansas Swine Odor Survey

Executive Summary

At the request of the Arkansas Pork Producers Association, a swine odor survey was conducted. The primary objective was to provide an unbiased assessment of odor levels typically associated with Arkansas swine production. The collected information would also be available as a guide for future research and educational efforts.

Odor measurements were made on 36 farms in 7 counties from September 1996 to June 1997. The farms were typical of the facilities and environmental conditions under which swine are raised. Seven odor panels, consisting of 42 individuals from Cooperative Extension Service (C E S), Natural Resources Conservation Service (N R C S), Tyson, Cargill, and the general public, made 1,157 odor measurements. Forty-five percent of the measurements were made within 1/10 mile from the odor source. Ninety-nine percent were made within 1/2 mile of the odor source. Less than one percent were made at a distance of one mile. Most of the odors were identified as originating from the facility or manure application sites.

Odor measurements were collected using two different ranking procedures. The first was a Nasal ranking system where the odors were ranked from 1 (Non-Detectable) to 4 (Strongly Offensive). For comparison with the Nasal ranks, one panel member used a Scentometer to assign odor ranks. The Scentometer ranks ranged from 1 (Non-Detectable) to 7 (Strongly Offensive). When adjusted for scale, both sets of measurements were sufficiently similar to be combined.

Of the 1,157 measurements made, 57% were classified as Non-Detectable, 28% Detectable But Non-Offensive, 13% Mildly Offensive, and 2% Strongly Offensive. Odor levels recorded were correlated to measurement distances. As distances increased, the average odor values decreased. The maximum distance an offensive odor was measured was 3/10 of a mile for swine facilities and 1/2 mile for manure applications sites.

There were significant differences between individuals in how they ranked the odors. However, when the individuals were grouped by their affiliations (Agency, Industry, or General Public), there were no significant differences in how the groups ranked odors.

There were also significant differences in odor levels between the farms. All measurements on the farm with the lowest mean odor value were classified as Non-Detectable. The farm with the highest mean odor value had 28% of the measurements classified as Non-Detectable odors, 33% Detectable But Non-Offensive, 36% Mildly Offensive, and 3% Strongly Offensive.

The facility design and manure handling system, farm appearance, and the history of complaints all appeared to influence odor levels. The effects of temperature, humidity, and wind speed were minor to non-significant. In many cases trends were difficult to identify, probably due to the complex interactions of the many factors that influence the generation and transmission of odors.

This project was able to develop a method to survey odor levels on swine farms, quantify existing odor levels, and to establish some general trends. The findings indicate that the offensive odors associated with swine production tend to occur relatively infrequently. The survey also demonstrates that many factors will need to be considered in future efforts to improve odor management. This information will prove valuable in future efforts to address odor concerns.

Project History

In June 1996, the Arkansas Pork Producers Association (A P P A) requested that a committee composed of individuals from the University of Arkansas Cooperative Extension Service (C E S), the Natural Resources Conservation Service (N R C S), the Arkansas Department of Pollution Control and Ecology (A D P C & E), the Arkansas Soil and Water Conservation Commission (A S W C C), and the University of Arkansas (U of A) Animal Science Department develop a proposal for a swine odor survey project. As a result, a steering committee was formed and a proposal developed, approved by A P P A, and implemented.

The objectives of this project were to:

• Develop a procedure for the impartial quantification of odors associated with swine production;

• Collect baseline information on the current status of odors associated with swine production;

• Document the survey information to provide a guide for future efforts to address odor concerns.

This report records the completion of the first two items and documents the survey findings for future use as a reference source.

Explanation of Report Format

Since this report will serve as a guide for future efforts to address odor concerns, a large number of graphs, tables, and supporting documentation are included. To improve readability, they have been placed in the appendices. To simplify finding the appropriate information, the sections of the appendices parallel those of the body of the report. Since the body and appendices have parallel formats and most sections in the body have supporting information in the appendices, the report has few referrals to the appendices. Instead, the reader should assume that there is supporting information in the appropriate appendices.

Data Collection and Analysis Procedures

 Since an objective of this project was to quantify the odor levels on existing farms, this project is a survey and not a controlled experiment. As a result, the odor measurements were collected on a wide variety of farms under different environmental conditions. Under these parameters, variability in data and interactions between the various odor-influencing factors are to be expected and unavoidable. However, this type of "real world" data is needed to quantify current odor levels and help establish trend relationships between odors and influencing factors.

Another issue associated with this project was the limited availability of man-hours and funds to support the project. The Arkansas and National pork industry were able to supply $10,000 for the purchase of equipment. The personnel involved in the project added their time on this project to their existing workload.

The data collection and analysis portion of the project consisted of farm selection, participant training, data collection, and data analysis.

Farm Selection

To ensure the information collected was representative of the types of farms and conditions typically found in Arkansas, it was determined to include about 40 farms (a little less than 10% of the permitted swine farms) in the project. Eight counties in northwest, westcentral, and southwest Arkansas were selected. During farm selection, an effort was made to ensure that all the commonly used production types, manure storage facilities, and manure application systems were included. To help ensure an unbiased farm selection, a list of randomly selected farms was made for each county. These farms were then contacted to confirm their interest in participating in the project.

Participant Training

One participant training meeting was held where the agency and industry personnel in the project counties were informed of the project, its goals, and data collection procedures.

Data Collection

Since an objective of the project was to collect impartial odor measurements, a panel approach was used. Panel members included county Extension agents, county NRCS employees, integrator field men, and individuals not associated with swine production. Ideally the panels consisted of at least four individuals to independently rank the odors. An odor panel was assigned for each county.

To maximize the value of the odor information collected, a significant amount of additional information was collected. The additional information was needed to help assess the factors that influence odor levels at the various measurement locations. The additional information consisted of both farm and environmental information. The farm information included such items as the size and number of animals typically on the farm, as well as manure handling facilities and practices. The environmental information included not only the odor levels but also such items as the distance from the odor source and environmental conditions.

A panel member in a one-time interview collected the basic farm information with the producer. To facilitate this data collection, prior to the interview, a standard interview form was developed and partially completed for each farm. During the interview the supplied information was verified and additional information collected.

To help quantify the environmental information, the A P P A purchased $10,000 worth of equipment. Most of the funds were used to purchase eight Scentometers. Scentometers are portable air dilution chambers that allow control of the ratio of purified air to odorous air. This control enables identification of the Dilutions to Threshold (D/T) ratio at which an odor is detectable. For the purposes of this project, the six standard D/T ratios for the Scentometers were assigned ranks from 2 to 7. A rank of 1 was assigned when no odor was detectable. Equipment to measure air temperature, relative humidity, and wind speed was also purchased.

To collect the odor information, each farm was visited about every other week, and odor measurements were taken at several locations downwind from the potential odor source. One panel member would use the Scentometer, while the remaining panel members assigned Nasal ranks. This approach was taken to maximize the number of odor measurements made while avoiding the prohibitive expense of supplying a Scentometer to each panel member.

At each sampling location, the odor rank, the distance from the potential odor source, temperature, relative humidity, wind speed, and any written observations were recorded. Each panel member independently recorded his or her results. For the purpose of tracking the data, each time a panel member recorded their results a measurement took place. A site was defined as the combination of farm, date, and distance from the potential odor source at which a measurement was made. As a result, each site was associated with several measurements, one for each panel member.

The data collection procedure was to work upwind toward the potential odor source, making measurements at 1/2 mile, 1/4 mile, 1/8 mile, and 200 ft. Closer measurements were encouraged. However, due to bio-security concerns, the industry representatives made the determination if closer measurements were acceptable. Due to such conditions as topography, ground cover, and wind direction, the recommended distances were to be treated as suggestions and modified as conditions warranted. Also, as the purchased equipment did not include equipment to measure distances, the panel members estimated the distance values recorded.

Data Analysis

As the data collection forms were completed, they were sent to the C E S state office for analysis. The data was coded and transcribed into an electronic spreadsheet. Then several standard sets of calculations were made as part of the data analysis.

Odor Level Comparisons by Class

The data analysis was performed by subdividing the data into classes, then making comparisons between classes. For example, to investigate trends between the type of storage units on the farms and odor levels, the data was divided into classes (groups) based on the type of storage units and calculations made to summarize the odor levels for each class. The results were then compared to draw any warranted conclusions. This process was repeated until all the factors desired were investigated.

This type of analysis ignores possible interactions of the many factors that may influence odors. When the odor values are grouped by one set of classes, it is possible for the other factors to mask or override trends that may exist. The likelihood of this taking place is largely a matter of chance based on the combination of influencing factors occurring when the odor measurements are made. Because of this, the determination of whether the mean (or variation of the values) of one class is the same or different from another becomes not the matter of absolutes but rather of probability.

The determination that the means of two compared classes are indeed different has a risk that they are not really different, but only appear so due to chance. The risk of determining there was a difference, when there was not a true difference, was set to be 5% (a=.05).

As a result, when a significant difference is found, there is a 95% probably that there truly is a difference. Conversely, if a significant difference is not found, that does not mean there is not a difference. It is possible that there is a difference, just that the probability of there being a difference is less than 95%.

The number of observations, the means, and the variation in the data play significant roles in determining if the means are significantly different or not. Small numbers of observations, closely spaced means, and wide variations all make the determination of a significant difference less likely.

Two examples of standard comparison tables can be found in the appendices.

Measuring Central Tendency

An issue of concern is how to best describe the central tendency for the distribution of odor rank values. The most commonly used measure is the mean or average value. However, the mode and median are also standard measures of central tendency. The median is the middle value of a list of values in sequential order. The mode is the most frequently occurring value. For a perfectly symmetrical distribution of values, the mean, median, and mode have very similar values. However, for non-symmetrical distributions each of the measures of central tendency will often give different values. For non-symmetrical distributions the mean will always be skewed in the direction of the tail more than the median or mode. The difference between the mean and the median and mode is influenced by both the variation in the data and the magnitude of the values.

For the ranking scales used in this project, the non-symmetrical distributions ensure that the mean will always have a greater value than the median or mode. For example, for all the Nasal ranks recorded in this study the mean value is 1.57, while both the median and mode are 1. This does not imply that a mean of 1.57 is wrong, just that it places the central point of the Nasal ranks at a slightly higher value than the mode and median. Due to the common usage of the mean, it will be used as the primary measure of central tendency. However, one should remember that for the Nasal and Scentometer ranks, mean values tend to give consistently greater results than the mode and median values.

For the 1 to 4 range of Nasal ranks, the amount of skew would probably not cause misleading perceptions. While for the 1 to 7 range of the Scentometer ranks, misleading perceptions may be more likely to occur. As mentioned above, the mean, median, and mode for all the Nasal samples are 1.57, 1, and 1, respectively. At first impression, a mean value of 1.57 would likely be considered a "Non-Detectable" odor, the same as for the median and mode values of 1. However, the mean, median, and mode values for all the Scentometer samples are 2.11, 1, and 1. In this case, the mean indicates that the distribution of the Scentometer samples is centered about a "Detectable But Non-Offensive" odor level. In contrast, the median and mode indicate that the distribution of Scentometer ranks is centered at the "Non-Detectable" odor level.

Linear Regression Analysis

Linear regression analysis was performed to investigate several relationships. Regression is the mathematical process of determining the linear equation that "best" predicts the value of the dependent variable based on the value of the independent variable. For example, a regression equation was developed that describes odor levels as a function of distance from the odor source.

Regression equations are useful in determining trends between independent and dependent variables. However, the precision of a regression equation in prediction is affected by the variability of the data used to develop the equation. For this reason, the concepts of probability and the associated risks of erroneous conclusions discussed earlier are also present in regression analysis. Refer to the appendices for details on the regression calculations and interpretation of the values.

Survey Descriptive Statistics

Odor measurements were collected on 36 farms in 7 counties between September and November of 1996 and March and June of 1997. A total of 1,157 odor measurements were made at 253 different measurement sites. Nasal ranking accounted for 907 measurements, while Scentometer ranking accounted for 250 measurements. For both scales, the odor measurements ranged full scale from Non-Detectable to Strongly Offensive.

Measurements were made between 8 a.m. and 4 p.m. and were identified as coming from the facilities (1,035 values), land application (83), and dead animal disposal units (11). In addition, odor measurements (28) were also made on odors that either came from off-farm or from non-swine related sources. The downwind distances at which the odors were measured ranged from 5 feet to 1 mile. During the odor measurements, temperatures ranged from 48ºF to 88ºF, relative humidity from 25% to 100%, and wind speed from calm to 13 m p h. Comments on the data forms indicated that very often the wind was very gusty in nature.

Forty-two individuals served as odor panel members, of whom 24 worked for either C E S or N R C S, 14 for Tyson or Cargill, and 4 were from the general public. By class, there were 651 agency measurements, 430 industry measurements, and 44 general public measurements. Twenty-two measurements were made by either an unidentified individual or by a panel that reported one measurement for each site. These measurements are identified as being made by observer 43, 44, or 45.

The appendices provide additional information on the survey descriptive statistics.

Survey Results and Implications

General Findings

Overall Odor Level Findings

To better understand the odor values recorded in the survey, it is appropriate to apply descriptors to the odor rank values and to correlate the Scentometer and Nasal rank ranges. According to the Scentometer documentation, odors with dilutions to thresholds (D/T) above 7 start to cause complaints while odors with D/T values above 31 are usually considered a serious nuisance. Based on this information and the ranking systems used in this survey, the rank descriptors and correlations shown in the table below were applied. 

The first finding of interest is that the majority of the odor measurements had ranks of 1. For 56% of the Nasal samples and 60% of the Scentometer samples, no odors were detected. The Nasal Rank (N R) and Scentometer Rank (S R) charts reveal that as odors became detectable and more intense, the percentage of the odors at each odor level decreases. Thirty-one percent of the Scentometer samples and 17% of the Nasal samples were classified as non-offensive odor ranks. Only 6% of the Scentometer samples and less than 1% of Nasal samples indicated the presence of strongly offensive odors. It is important to remember that at this point these are summary values for the entire survey, so important factors such as distance and odor source are not considered.

The skewed distribution of odor ranks observed for all of the Nasal and Scentometer ranks is very typical of the distribution of values when the data is subdivided into various classes. This skew of the odor ranks to the "Non-Detectable" level with few measurements at a "Strongly Offensive" level indicates that a significant portion of the swine industry's odor problems is due to relatively infrequent occurrence of offensive odors rather than consistent state of offensive odors.

Comparisons of Nasal and Scentometer Measurements

Of interest is the correlation between the Nasal and Scentometer measurements. While they have different ranges and means, both have median and mode values of 1. The Scentometer ranks appear to be slightly more likely to have Non-Detectable scores than Nasal ranks. However, when a detectable odor is present, the Scentometer ranks have a tendency to indicate stronger odors than the Nasal ranks.

Determination of the correlation between Scentometer and Nasal rank scores is important because the "subjective" aspect of Nasal ranks raises questions about the "reliability" of nasal-based measurements. The Scentometer by nature of its design and history raises fewer questions concerning "reliability." In addition, closely related scores would help to increase confidence in the accuracy of all the odor measurements collected during this project.

To enable additional comparisons between the Nasal ranks (N R) and Scentometer ranks (S R), the Scentometer samples were scaled to a 1 to 4 range based on the theoretical relationship, S R=2 H N R !1. The standard data comparison calculations discussed in the data analysis section above for N R, S R, and scaled Scentometer ranks (S R 4) show the means of the NR and S R 4 values to be statistically the same. However, the variances are statistically different. Inspection of the percent frequency information indicates there is still a tendency for the S R 4 values to rank detectable odors higher than the N R values.

Regression analysis was used to further investigate the relationship between the Nasal and Scentometer measurements. To get the required data pairs for the regression, it was necessary to average the Nasal ranks for each site where a Scentometer rank was available. The data was then plotted and a regression analysis performed.

As the plot indicates, there is quite a bit of variability in the data. The R Squared value indicates that only about 48% of the variability in the S R values is associated with the variability of the N R values. However, the regression results indicate that the relationship between the S R and N R values is statistically significant. In addition, both of the regression coefficients are also significant. It should also be noted that the 99% upper and lower boundaries for the regression coefficients include the theoretical coefficient values of 2 and 1.

From this information it is apparent that there is a significant statistical relationship between the S R and N R values. Also, the statistical relationship is very similar to the theoretical one of S R=2 H N R ! 1. 

While a higher R Squared value would be desirable, a value this low is not surprising. Measuring and quantifying odors is a complex process that involves not only the odor level at a given location but also the individual's physical ability to detect the odor and the individual's subjective perception of the strength and offensiveness of the odor. Variability in the odor measurements is to be expected in a data set collected by 42 individuals under different environmental conditions.

Due to the strong relationship between S R and N R values, all further odor trends were investigated using a general Odor Rank (OR) value. The OR values ranged from 1 to 4. An OR value is equivalent to an N R value. S R values are scaled to OR values by the equation OR=(S R+1)/2.

Of interest is the fact that while the SR values have some influence in the distribution of the OR values, the distribution of the N R and OR values is statistically the same. This is not surprising since there were over three times as many N R values as S R values, and statistically, the scaled S R and N R values had the same mean.

Observer Effects

In the previous section, some of the variation in the odor measurements was attributed to the individuals making the measurements. To compare the odor ranks recorded by the individuals, the OR values were grouped by the individuals making the measurements.

Because all observers were not on the same farms on the same dates, differences in the means and variances of the values are not going to be due solely to differences in individuals, but will also be influenced by farm-based factors and weather conditions.

The calculated results show that many of the individual-based measurement sets have statistically different means and variances. Of the 42 sets, the lowest has mean, median, and mode values of 1.17, 1, and 1, while the values for the highest are 2.23, 2, and 1. The individual associated with the lowest mean odor rank was an individual from the general public. The highest mean odor rank was associated with an individual that worked for one of the government agencies.

In conclusion, there is strong evidence that there are differences in how individuals evaluated odor intensities.

Observer Affiliation Effects

To determine if a correlation existed between how individuals evaluated odor levels and their affiliation to the swine industry, the odor measurements were grouped by affiliation (Agency, Industry, and Public). The comparative tests failed to find statistical differences between the odor values when classed by Agency, Industry, and the Public. All the median and mode values for the odor measurements were 1. The means for odor measurements ranged between 1.5 and 1.6.

From this study, how severely an odor is ranked is influenced by who is making the evaluation. However, this influence appears to be based on individual differences, not industry affiliations. There is no statistically significant trend that any affiliation group ranked the odors lower or higher than the other groups.

Farm-Based Findings

Odor Source Effects

The comparative calculations reveal that there are statistical differences in how odors downwind of facilities, manure applications, and mortality disposal units are ranked. The odor measurements associated with manure application and mortality disposal were statistically the same with median and mode values of 2. The mean values were 2.39 for land application and 2.27 for animal mortality. The median, mode, and mean associated with facilities are 1, 1, and 1.51, respectfully. As the odors classified "other" came from off-farm or from non-swine related sources, they will not be discussed.

Only on 2 of 253 measurement sites were odors attributed to swine mortality. Since every farm had disposal units, one can conclude that most of the time the odors coming from disposal units were either undetectable or undistinguishable from the odors coming from the facilities. Therefore, any conclusions concerning odors from disposal units should be based on the absence of odor measurements as well as on the values recorded. From this study, it is apparent that it is possible for the odors from the disposal units to be comparable with land application odors. However, most of the odors coming from disposal units are either Non-Detectable or undistinguishable from the general facility odors.

Individual Farm Effects

There were significant differences in the means and variances when the odor measurements were grouped by farm. The median and mode values ranged from 1 to 2. While the mean odor values ranged from 1 to 2.04. Almost all the farms had odor measurements that were ranked as 3 and above. For the farm with the lowest mean odor value, all of the odor values were ranked as a 1. For the farm with the highest mean odor value, 28% of the measurements were ranked as a 1, 33% as a 2, 36% as a 3, and 3% as a 4.

From this it is apparent that while offensive odors were occasionally found on these farms, most of the measurements found non-offensive odors. This finding strengthens an earlier conclusion that a significant portion of the industry's odor problems is associated with infrequent offensive odors rather than constant odors.

Type of Swine Production Effects

Some statistical differences were found between the odors generated by the various types of swine production. Placing the facility types in ascending order of means results in: Farrowing Through Weaning (1.18), Farrowing Through Nursery (1.44), Finishing (1.58), Nursery (1.63), and Nucleus Purebred (1.77). The general trend appears to be that more and larger animals resulted in higher odor levels. However, other factors also appear to be influencing the odor levels, as one would expect the odors from Nursery and Nucleus Purebred operations to be lower than for Finishing operations. This is likely to be the case since the survey included only 2 Nursery and Nucleus Purebred operations and 19 Finishing operations. This small number of representative systems makes it easier for other factors such as weather or topography to obscure any existing odor trends between wet and dry systems.

Wet Vs. Dry Manure Management Effects

Grouping the measurements according to whether a facility had a liquid or a dry manure management system showed that there were no significant differences between the two. However, the means of 1.51 and 1.39 indicate that the dry systems had lower (albeit insignificantly lower) odor levels. This trend is further supported by the fact that the maximum odor value on the dry farm was ranked as a 3. It should also be noted the medians and modes associated with both types of systems were 1. The lack of significant differences may be due to only three farms having dry systems. This small number of representative systems makes it easier for other factors such as weather or topography to obscure any existing odor trends between wet and dry systems.

Manure Storage Unit Effects

Significant differences in means and variances occurred when the odor measurements were grouped by the type of manure storage units. However, there were no easily identifiable trends. This lack of clear trends was probably due to two related factors. First, the odor measurements were for the odors coming from both the houses and storage units, not for just the storage units. Also, there were only 1 Lagoon, 1 Settling Basin and Lagoon, 2 Fresh-Water Holding Ponds, and 3 Dry Bedding Systems compared to 7 In-House Pits, 9 Holding Ponds, and 13 Settling Basin and Holding Ponds in the project. The relatively small number of some types of systems makes it easier for other factors such as odors from the houses, weather, or topography to obscure the effects of the storage unit on the odor levels found on the farm.

Appearance Effects

To investigate any potential trends between farm appearance and odor ranks, the initial farm interview for each farm was used to classify the overall appearance of the farm as Poor, Fair, Good, or Very Good. Since these assignments were made secondhand and based on the interview form, some precision was lost. However, statistical differences occurred when the odor measurements were grouped by these classes. The facilities classified as Poor had a mean odor value of 1.85 while the mean odor ranks for the other classes were 1.53 or less. The Poor class also had a median value of 2 while the others had values of 1. However, the maximum odor rank of the Poor class was 3 while the maximum rank on two of the other classes was 4. This apparent discrepancy is explained by the fact that for the Poor class the frequency of values was fairly uniform for the odor ranks of 1, 2, and 3. While for the other appearance classes, over half of the odor values had ranks of 1.

The relationship of poor farm appearance and higher odor rank raises the question of causality. Did farm appearance influence perceptions, or are higher odor ranks the result of a lower level of overall farm management? Probably both management and perceptions played a role. This study was not designed to address this question. Whatever the cause, a link exists between farm appearance and the odor levels. Farms with better appearance rankings were associated with lower odor measurements.

Complaint History Effects

To investigate any potential trends between a history of complaints against the farms and odor ranks, the initial farm interview form for each farm was used to classify the overall complaint history of the farm as None, Very Few, Few, and Several. Since these assignments were made secondhand and based on the interview form, some precision was lost.

When the odor measurements were grouped by these classes, there were some statistical differences in the means. The statistical differences occurred between the Few (1.60), None (1.41), and Very Few (1.35) classes. The percent frequency information supports the trend that a history of complaints was related to higher odor ranks. The strength of this relationship is tempered somewhat by the fact that all the median and mode values are 1, except for the Few class which has a median value of 1.25. Some of the "uncertainty" in the trend may be explained by the fact that in several instances the farmer indicated that some complaints had as much to do with personality conflicts as with odor.

Even given the questions of whether the complaints were justified and the possible lack of precision in assigning the farms to the classes, a history of complaints was associated with higher odor levels.

Odor Control Additives Effects

Seven farms were either using, or had used, some kind of additive to help control odors. Nine farms had indicated they had never used any odor control additives. The history of the use of odor control additives is unknown on the remaining farms.

To investigate any effects additives might have on odor levels, two sets of comparisons were attempted. One comparison looked at the odor levels where the source of the odors was the facilities. The other looked at the odor levels where the source of the odors was the land application of manure.

The effect of additives on odors associated with land application of manure was not possible because the survey did not sample any application events on the additives farms. The comparison with the facilities as the odor source was possible, but no significant differences in either the means or variances were found. The median and mode values in both cases were all equal to 1, further indicating no discernable differences in odor levels with the use of additives.

There are several possible reasons for the lack of significant reductions in odor levels with the use of additives. In this project the "use of additives" classification was very broad. Any farm that had ever used additives (not just those using additives now) was classified as using additives. Also, several different brands of additives had been used, raising the question of which, if any, of the additives were effective. Finally, since the odors were from both the houses and the ponds, house odors could have masked any trends.

Due to the intrinsic nature of statistics and the structure of this survey, to conclude that additives are ineffective at controlling odors is incorrect. Rather, the results conclude that the data fails to prove that they had a significant effect on odor levels at an ? level of .05. At an ? level of .1 the difference in odor levels is significant. (Refer back to the data analysis section for questions regarding ?.)

Environment-Based Findings

To this point in the investigation of the potential odor relationships, the data has been grouped into classes and comparative calculations have been used to draw conclusions. For the remaining trend investigations, graphs and regression analyses will also be used. The factors that will be investigated are distance, temperature, relative humidity, and wind speed. For each case, the effects of the factor on facility odors, application odors and the combined facility, application, and mortality odors will be presented. The mortality odors are included in the combined odor data because they were measurements recorded during the study. However, since only 11 measurements at two sites were attributed to mortality, separate graphical and regression analysis will not be reported for mortality-based odor measurements.

Distance Effects

The analysis revealed that there is a relationship between distance and odor. The general trend is for odor levels to decrease as distance from the source increases. In addition, facility odors are lower and decrease more rapidly with distance compared to manure application odors. For the measurements made within 528 feet (1/10 mile) of the source, the mean, median, and mode values were 1.76, 2, and 1 for the facilities and 2.58, 2.5, and 2 for the manure applications. The maximum distance that an offensive facility odor was recorded was 1,584 feet (3/10 mile) compared to 2,640 feet (1/2 mile) for a manure application.

All of the regression relationships were statistically significant. All of the regression coefficients were also statistically significant with the exception of the distance coefficient for the application-only odors. This may be due to the 6 odor measurements with a rank of 2 made at a distance of 1 mile. These measurements were made at one site and reflect the odors encountered from the land application of manure from an in-house pit. The measurements were made at 12:30 p.m. and the weather conditions included an 80ºF temperature, 35% relative humidity, and a 3 m p h wind, a set of conditions that should be fairly conducive to the release and movement of odors. While these 6 odor measurements do not completely follow the trend of the other values, it is important to note that with a value of 2 they were ranked as Detectable But Non-Offensive.

These 6 values point out that while there is a valid statistical relationship between distance and odors, the variability in the data indicates that there are also other factors which play a role in the transmission of odors. The regression analyses indicate that distance accounts for less than 20% of the variability in the odor measurements.

Temperature Effects

There appears to be no consistent odor trend associated with temperature. For anaerobic digestion in liquid manures it is commonly held that increases in temperature are associated with increases in odor. This relationship is attributed to increased biological activity, which in the process of breaking down the organic material in the odor source increases the amounts of odorous compounds.

The distribution and mean values information from this study fail to show this trend. Instead, the temperature coefficients for the combined odors and facility-only odors are not statistically significant in explaining the variability in the odor measurements. In addition, while the temperature coefficient for the application odor data is significant, the regression analysis indicates that the application odors decreased with increasing temperatures for this study.

Apparently other factors are overriding any effects temperature may have. This lack of influence may be due in part to the fact that the maximum temperature recorded during the project was only 88ºF. Higher temperatures may result in higher odor levels. In addition, the facilities-based odors came not only from the manure storage units but also from the houses.

Relative Humidity Effects

There is a slight inverse relationship between relative humidity and odor. Increases in relative humidity are associated with slight decreases in odor levels. The R Squared values indicate that in this study the relative humidity is associated with less than 2% of the variability in the odor measurements.

Wind Speed Effects

For odors originating from facilities, wind speed appears to have a positive relationship. As wind speed increased, the odor level increased slightly. A similar trend was also found, but was not statistically significant, for the odors associated with the land application of manure. In all cases, wind speed had only a slight affect on odor levels. At best, wind speed accounts for less than 2% of the variation in odor values.

Acknowledgements

The development and implementation of this project required the cooperative efforts of many individuals and organizations. A special word of appreciation goes to the Arkansas Pork Producers Association who requested and funded the project, the producers whose farms were surveyed, and the team members who collected the information. This project would not have been possible without their willingness to participate.

Project Manager: Dr. Karl VanDevender, Extension Agricultural Engineer, Cooperative Extension Service, University of Arkansas.

Survey Participants

• Arkansas Pork Producers Association

• National Pork Producers Council

• Cooperative Extension Service, University of Arkansas

• USDA, Natural Resources Conservation Service

• Arkansas Department of Pollution Control and Ecology

• Arkansas Soil and Water Conservation Commission

• University of Arkansas Animal Science Department

• Tyson Foods Inc., Swine Division

• Cargill Pork

• Arkansas Swine Producers

• Individuals from the General Public

References 

Kennedy, John B., and Adam M. Neville. Basic Statistical Methods for Engineers & Scientists. 2nd edition. Harper & Row Publishers. 1976.

Mendenhall, William, Richard L. Scheaffer, and Dennis D. Wackerly. Mathematical Statistics with Applications. 2nd edition. Duxbury Press. Boston, Massachusetts. 1985.

Neter, John, William Wasserman, and Michael H. Kutner. Applied Linear Statistical Models. 2nd edition. Richard D. Irwin, Inc., Homewood, Illinois. 1985.

Scentometer: An Instrument for Field Odor Measurement. Barnebey & Sutcliffe Corporation, P. O. Box 2526, Columbus, Ohio 43216.¹

Steel, Robert G.D., and James H. Torrie. Principles and Procedures of Statistics, A Biometrical Approach. 2nd edition. McGraw-Hill Book Company. 1980.

¹ The mention of products and trade names in this publication does not signify that these products are endorsed or approved to the exclusion of comparable products.

Back to Animal Manure and Mortality Management 

© 2006
University of Arkansas
Division of Agriculture
All rights reserved.
Last Date Modified 02/02/2012
Webmaster

University of Arkansas • Division of Agriculture
Cooperative Extension Service
2301 South University Avenue
Little Rock, Arkansas 72204 • USA
Phone (501) 671-2000 • Fax (501) 671-2209
 

Mission • Disclaimer • EEO •
Privacy • FOI