Abstract: An improved fourcomponent decomposition scheme for analyzing polarimetric synthetic aperture radar（POLSAR)data is proposed in this paper.To solve the problem of volume scattering overestimation，the improved decomposition utilizes the new doublebounce model of distributed dielectric dihedral corner reflectors instead of the original one adopted by the Yamaguchi fourcomponent decomposition.To be performed，the improved decomposition introduces normalize circularpol correlation coefficients in the decomposition process.An airborne L band quadpol data acquired by DLR ESAR sensor over Oberpfaffenhoffen area in German and a spaceborne L band quadpol data over Beijing acquired by the Phased Array Lband SAR sensor of Advanced Land Observing Satellite (ALOS PALSAR )are applied to the improved decomposition scheme.Three comparisons have been made to demonstrate the effectiveness of the improved decomposition， including the number of neTgative power pixels，the magnitude of each scattering mechanism，and performance between the decomposition results and optical image.The improved decomposition can reduce the pixels with negative powers and obtain correct scattering mechanisms of nonreflection structures not paralleled to the flight path.