Uncertainty and propagation analysis on DIMITRI

Abstract

DIMITRI is a tool that has been developed for monitoring the radiometric performance of optical instruments onboard Earth observation satellites. It includes a reflectance model through which a calibration coefficient is computed which is used for performance monitoring. In this thesis, the uncertainty that is associated with this coefficient is researched and a classification in random and systematic uncertainty components is attempted. Throughout this research it became clear that such a classification is somewhat ambiguous and can be misleading because systematic uncertainties are not always constant. In order to identify such uncertainties, the entire modelling / measurement process with respect to the input variables of the reflectance model, and the algorithm of the reflectance model itself, have been researched for potential uncertainty sources. What became clear is that input variable uncertainties are of both nature; random and systematic. In addition, a sensitivity analysis has been conducted to assess how the calibration coefficient depends on changes in the input variables. Out of these, the Chlorophyll variable causes the highest deviations due to the large uncertainties associated with this parameter. This research has been able to identify uncertainty sources but has not been able to assign any value to those sources. Any subsequent research could use the identified uncertainty sources described here and attempt to quantify their value to yield a total uncertainty that can be associated to the calibration coefficient. This will increase the usability of this performance monitoring coefficient which means that the user will obtain more information regarding the accuracy of the performance degradation of the optical instrument onboard Earth observation satellites.