Attribute uncertainty modeling in lunar GIS data

A novel methodology to evaluate uncertainties in lunar element abundances is presented. Contrary to most terrestrial applications, lunar GIS data cannot be verified by in-situ measurements due to the limited number of ground control points and their reduced spatial extend. This investigation evaluates the uncertainty in lunar element abundance measurements without the use of groundchecks but by statistical evaluation and comparison of datasets. We find that major elements (Oxygen, Iron, Aluminium, Titanium, Silicon and Magnesium) show distinct correlations between each other. This allows calculating the abundance of an element by deriving its value through a correlation law with another element. By using this method, we can verify remote measurements of the above mentioned geochemical components, and identify regions on the Moon where these correlation laws do not apply. These derivations can be explained by i) an erroneous measurement or ii) by an exotic mixture of elements in the lunar soil. Based on these considerations, conclusions can be drawn about the attribute uncertainty of geochemical measurements in the lunar soil. A special observation of this work was that most theoretically obtained values fit well to the measured ones. High derivations however appear mainly in the Near Side lunar mare regions where the correlation model does not fit.