There are potentially a large number of errors that can occur in almost all applications of radiation thermometry, particularly in industrial furnaces. In this chapter, we concentrate specifically on the errors that occur with spectral-band thermometers, since the large majority of instruments used in furnace applications are of the spectral-band type and they are generally the best suited.
Errors in spectral-band thermometry can be separated into two groups: those associated with the instrument itself and those related to the target and its surroundings (the environment). The most important errors in industrial applications are environmental, with the largest errors being associated with the emissivity of the target surface, reflected radiation originating from surrounding objects, and in some cases atmospheric absorption and emission. The instrumental errors are usually negligible in comparison. In fact, a leading radiation thermometrist  once stated, "It can be safely said, for most applications, that the quality of commercial radiation thermometer instrumentation exceeds our capability to use it intelligently." While this statement was made some twenty years ago, it is still largely true today and reflects the fact that the largest errors are environmental, rather than instrumental, in nature. The environmental effects may be different each time we use a radiation thermometer, so it is vitally important to understand these effects to achieve reliable and accurate measurements.
There are several approaches to managing the errors. First, by adopting careful measurement practice we can eliminate or minimize many of the errors. In these cases there may be a residual error, which can be treated as an uncertainty in the measured temperature. Secondly, where the errors are large and often unavoidable, we can apply corrections to the measurements based on auxiliary measurements of influencing factors. Uncertainties are also associated with corrected measurements. Lastly, when only limited information is available about an error, we can estimate its range of likely values and treat this range as an uncertainty. In this chapter we focus on the errors themselves, while methods for handling and propagating uncertainties are treated in Chapter 5.
Online access to SPIE eBooks is limited to subscribing institutions.