CROP AREA ESTIMATES THROUGH REMOTE-SENSING - STABILITY OF THE REGRESSION CORRECTION

Classified high-resolution satellite images are useful to improve the precision of crop area estimates in an area frame survey through the r egression estimator. This is the method used for the Regional Inventor ies of the MARS Project of the IRSA that is briefly described. Some as pects concerning stability are discussed. Regression results seem to b e more reliable if training pixels are chosen at random in a random su bset of segments (elements of the area sampling frame). Some risk of b ias is observed if segments with training pixels are included to compu te the regression parameters.