Abstract
The metrology community provides careful attention to the uncertainty in their measurements. The National Institutes of Standards and Technology developed a Technical Note on Measurement Uncertainty NIST Tech Note 1297 which serves as the “Gold Standard” for measurement error, oops actually uncertainty. The Tech Note advocates treating uncertainty as Type A (those which are improved with statistical methods) and Type B (those which are improved with other methods). We also have familiarity with terms such as accuracy and precision, bias, and reproducibility of the measurement results. For those of us that make our career by measuring quantities, we may put nearly as much of our personal energy into identification of the uncertainty of the measurement as we put into determining the actual reported value for the measurement.
At some level, these approaches are very admirable. And I want to turn our attention toward measurement uncertainty by turning our attention away from measurement uncertainty. Folks making the Standards and establishing the true quantitative measured value must by nature of their requirements adhere strictly to measurement uncertainty protocol. And many of us are making measurements for “less noble” purposes, for the purpose of using that measurement to explain something other than say a fundamental physical quantity. Well, for an application of the measurement to a scientific problem (our most familiar problems in the Earth sciences are environmental problems), just what benefits derive from fastidious attention to uncertainty protocol? Where may we draw a line where actions beyond that demarcation for uncertainty analysis simply are not of enough value to make the added work worthy? And so what, perhaps, if we may over-perform in uncertainty analysis beyond our data user needs.
This presentation will serve as in introduction to the following papers in this Session on several approaches to identification of measurement uncertainty, and provide some insight into relationships between documented uncertainty of measurements and data reprocessing in Earth science satellite data sets.
Uncertainty & Precision: Conversations about Data Set Reprocessing and Utility of the Data We Produce
The metrology community provides careful attention to the uncertainty in their measurements. The National Institutes of Standards and Technology developed a Technical Note on Measurement Uncertainty NIST Tech Note 1297 which serves as the “Gold Standard” for measurement error, oops actually uncertainty. The Tech Note advocates treating uncertainty as Type A (those which are improved with statistical methods) and Type B (those which are improved with other methods). We also have familiarity with terms such as accuracy and precision, bias, and reproducibility of the measurement results. For those of us that make our career by measuring quantities, we may put nearly as much of our personal energy into identification of the uncertainty of the measurement as we put into determining the actual reported value for the measurement.
At some level, these approaches are very admirable. And I want to turn our attention toward measurement uncertainty by turning our attention away from measurement uncertainty. Folks making the Standards and establishing the true quantitative measured value must by nature of their requirements adhere strictly to measurement uncertainty protocol. And many of us are making measurements for “less noble” purposes, for the purpose of using that measurement to explain something other than say a fundamental physical quantity. Well, for an application of the measurement to a scientific problem (our most familiar problems in the Earth sciences are environmental problems), just what benefits derive from fastidious attention to uncertainty protocol? Where may we draw a line where actions beyond that demarcation for uncertainty analysis simply are not of enough value to make the added work worthy? And so what, perhaps, if we may over-perform in uncertainty analysis beyond our data user needs.
This presentation will serve as in introduction to the following papers in this Session on several approaches to identification of measurement uncertainty, and provide some insight into relationships between documented uncertainty of measurements and data reprocessing in Earth science satellite data sets.