Blog originally posted on The GLOBE Scientists' Blog: http://blog.globe.gov/sciblog/2013/06/19/when-bad-things-happen-to-good-experiments/
The National Aeronautics and Space Administration, NASA, is famous for studying stars and planets and galaxies in outer space. But did you know it also has excellent programs that study planet Earth? The GLOBE Program is one of these programs. And over the past few years, I have been lucky enough to be a part of The Earth Observing System. More specifically, I work on an experiment called HIRDLS, which stands for High Resolution Dynamics Limb Sounder. HIRDLS was designed to measure the temperature and aerosols in Earth's atmosphere.
HIRDLS was launched on the Aura satellite on July 15, 2004. Unfortunately, during the launch, we believe something went wrong. It appeared that some of the insulating material inside the satellite fell on top of the HIRDLS instrument, which blocked it. When the HIRDLS team started to receive data, we found that about 70% of HIRDLS was blocked. This was not a good development. But, never fear, the HIRDLS scientists persevered! How? We figured out what the systematic error was. [caption id="attachment_1745" align="aligncenter" width="300"]
Usually, the word "error" is associated with negative connotations. Most people think an error is a mistake. But to scientists, "error" is a tool we can use to characterize data. The goal is for the truth to be contained within the experimental uncertainty. There are two general types of error that make up experimental uncertainty: random error and systematic error. A random error has no pattern and can be attributed to chance. A systematic error is a persistent error that often has a pattern and is not due to chance. The insulation covering our machine caused a systematic error. If we could understand and describe it, we could account for it and we could correct for it. Thus, over the course of many months, we studied how the HIRDLS data differed from what we expected and eventually, we figured out what this systematic error was. To get the good data, we subtracted off the systematic error from what we were receiving from the satellite. With these adjustments, HIRDLS is able to provide important support toward reaching the goals of the Aura mission: monitoring the complex interactions that affect the globe.
Suggested activity: Have you ever seen a random error in one of your research projects? What about a systematic error? Let us know how you could design an experiment to study these types of errors by leaving a comment, sending us an email at firstname.lastname@example.org or telling us about it on our Facebook page.