A parametric model for estimating relations between unprecisely located field measurements and remotely sensed data

Mislocations in the acquisition of field measurements may affect estimates of functional relations between field variables and their remotely sensed r adiometric responses. Under such field errors least squares (LS) estimators perform poorly because they present significant systematic biases. New sta tistics have been developed to achieve better estimates of parameters of th e functional model when such errors occur. An example used thematic mapper data and field measurements from a forest inventory. Bias and consistency o f new statistics are analyzed with extensive simulations, showing their hig her reliability compared with LS statistics. Finally, the effects of non-sp atially and spatially related field errors are compared. (C) Elsevier Scien ce Inc., 1998.