Example Case Study
Based on the Report*:
“Hydrogen Sulfide and Methane in Colorado”, by Joanna Gordon, Kelsi Middleton, John Ortega, Ricardo Piedrahita, Jana B. Milford, and Michael P. Hannigan (Department of Mechanical Engineering, University of Colorado, Boulder CO)
*(Disclaimer: this example case study utilizes methods and data from the report, but it does not directly reflect the report’s objectives, findings, or conclusions – it is available for educational purposes only)
Research Question:
Given that different activities and sources that can contribute to elevated levels of hydrogen sulfide (H2S) in the air, how do concentrations of this compound vary in different regions of Colorado?
Background Information:
- What is H2S?
o A colorless gas, known for its ‘rotten egg’ smell
o Potentially harmful to human health, depending on the concentration and length of exposure time (can be deadly at very high levels)
o Flammability presents another danger
- Where does it come from?
o H2S occurs both naturally and as the result of human activity; when it occurs naturally it is sometimes a by-product of bacteria breaking down organic material
o Examples of natural occurrences: wetlands, volcanoes and geothermal areas
o Examples of H2S used for, or produced as a result of human activity: oil and gas refining, mining, pulp and paper processing, and tanning
o Other places H2S can occur(typically as the by-product of bacteria breaking down organic material): sewers, manure pits, well water, oil and gas wells
- How can we detect it?
o Humans can detect this gas by smell, if concentrations are above approximately 5 ppb, depending largely on the individual
o Additionally gases tend to absorb light as very specific wavelengths, and by measuring the transmittance of light at a specific wavelength (e.g., using a laser) through a space we can determine the amount of the gas present
o Some lower-cost sensors are also available to measure H2S, these rely on chemical interactions with H2S and the sensor then essentially measures the extent of the interactions to determine concentrations
- For more information:
o https://www.osha.gov/SLTC/hydrogensulfide/
o http://www.idph.state.il.us/envhealth/factsheets/hydrogensulfide.htm
o http://www.epa.gov/iris/toxreviews/0061tr.pdf
Methods, Planning, and Additional Considerations:
In order to compare H2S levels in different areas of Colorado, a mobile sampling campaign was undertaken. In this case researchers wanted to make a spatial comparison of pollutant levels. Conversely, they could have chosen to look at how the concentrations change over time, in which case a stationary sampling plan would have been chosen. Instruments were mounted in a car that was able to collect data while driving. This particular study used a Picarro G2204 Cavity Ring-Down Spectrometer; a device that takes advantage of the optical properties of H2S mentioned above in the background section. Along with hydrogen sulfide concentrations, it was necessary to collect supporting data such as GPS information, and parameters specific to the Picarro monitor that ensured it was functioning properly (e.g., internal temperature).
The Picarro draws a sample into the device using a pump; then it illuminates the sample with a laser whose wavelength falls within the absorption spectrum of H2S. The laser is switched off and the instrument measures the absolute optical extinction of the light. This data is then used to compute the concentration of hydrogen sulfide in the sample. Having a good understanding of the instrument you intend to use in a study is vital, including its limits and uncertainty. The Picarro for example, has a lower detection limit of 3.0 ppb, with an operating range of 0 to 300 ppm. This means the device cannot reliably differentiate between 1.0 and 2.0 ppb, but if it is reading between 3.0 and 300 ppb it can be ‘trusted’. In addition to understanding the limits of your measurement device, you should also understand its accuracy and precision. The diagram below illustrates the difference between these two concepts.
http://extensionengine.com/accuracy-precision/
If you test your device in the lab, before taking it into the field and discover that it displays high precision, but not high accuracy, you can typically find a way to correct for this inaccuracy using laboratory data. Conversely, knowing that your device displays low precision, but high accuracy data, is also important. Knowing the accuracy, allows you to determine whether or not your results are statistically significant.
Beyond knowing what instruments you will use to collect the data required to answer your research question, you must also consider the logistics of your sampling plan. For example, when you will sample, where you will sample, for how long, etc… Sometimes this includes practical considerations, such as getting permission to sample on private land, or working with members of a community. It is also useful to consider what you can do if there is a problem with the equipment, or if the data quality is not adequate (e.g., you might need to build in extra time in case more data collection is required). Below is a picture of the Picarro mounted in the back of the car along with the power supply required for mobile sampling.
Results and Discussion:
Statistics and data visualization are extremely useful tools in understanding your data. Useful statistics include simple summary statistics, for example average, minimum, maximum, and median values. Time series can provide quick summaries of changing pollutant levels over time, while scatter plots can describe correlations between pollutants. If two pollutants are extremely well correlated (e.g., when one is high the other is high, and they change in relatively the same way) this can sometimes imply a common source. The scatter plot examples below illustrate no relationship (on the left) and a linear relationship (on the right), potentially indicating a common source.
Below is an example of a time series plot displaying Picarro data. One important observation from this data might have to do with the extreme variance in the data; or sharp changes. An additional note regarding data spikes is that sometimes they can obscure observations that exist on a smaller scale. For this reason you should visualize your data in a variety of ways to get an idea of the whole picture. Since we know this is mobile data, it seems to indicate drastic differences between physical locations. A next step would be comparing the days with where the data was collected.
The next two figures provide examples of how data may be spatially plotted. In the first, hydrogen sulfide is plotted vs. latitude and longitude with color indicating concentration (red – higher, blue – lower). The second plots relative concentrations on a recognizable map, the area is Commerce City in Denver CO. This plot might suggest that the elevated levels of H2S result from the variety of industrial activity taking place. Both plots help to illustrate where the high concentrations occur spatially; however, in addition to considering the sources and concentrations, you would also need to take into account transport (possibly using wind speed and direction) in order to fully understand the possible origins of the H2S. For reference, the levels of H2S shown below and in the previous chart are not high enough concentrations, or present for long enough durations to present a health concern.
*(note, peak concentration = 138 ppb)
Conclusions:
At the completion of a project, it is important to consider what you learned and what future steps would be useful in order to verify and build on your conclusions. For example, the study referenced above provided information on where elevated levels of hydrogen sulfide occur in Colorado, from there the researchers might be able to draw inferences regarding the impact of different types of industry and their level of impact. Next steps might involve collecting more data to verify claims regarding H2S emission sources. In addition to summarizing the findings of the study, you should consider the following questions:
- What is the relevancy of these findings in a larger scientific context?
- Who might be affected by this information?
- Do these findings bring new questions to light?
Comments (0)
You don't have permission to comment on this page.