Comparing thematic maps based on map value
When comparing maps to decide which map is better for a given application, the measure on which the comparison is based should represent the value or utility of the map to the intended application. Several map value measures are presented which employ user-selected weights to reflect the utility of the map to project objectives. Map value measures may be derived from either user's or producer's accuracy, or from the information in the full error matrix. Some commonly used map comparison measures such as the overall proportion of area correctly classified, average user's accuracy, and the tau coefficient of agreement are shown to be interpretable as measures of map value, thus illustrating that comparisons based on these measures also employ a weighting scheme. Although not interpretable as a map value measure, the kappa coefficient, is also evaluated for use in comparing maps.