提出了一种对合成孔径雷达（SAR）原始数据在载体上进行压缩的方法．原始数据先用分块自适应量化（BAQ）减小数据动态范围，但不减少量化位数．然后对数据进行小波包正交小波变换．根据频率包络的能量变化自适应地进行数据压缩，即高能量区域用较多位来表示，低能量区用少的位数来表示，处理带宽以外的区域就不再进行编码．用BAQ编码法对块内频域中的数据进行编码．实验表明所提出的WT－BAQ算法优于FFT－BAQ和BAQ算法． A method of compressing original data of Synthetic Aperture Radar (SAR) on a carrier is proposed. Raw data first uses sub-block adaptive quantization (BAQ) to reduce the data dynamic range, but without reducing the number of bits to be quantized. Then the data wavelet packet orthogonal wavelet transform. The data compression is adaptively performed according to the energy variation of the frequency envelope, that is, the high-energy region is represented by more bits, the low-energy region is represented by fewer bits, and the regions other than the processing bandwidth are not encoded any more. The data in the frequency domain of the block is encoded by the BAQ coding method. Experiments show that the proposed WT-BAQ algorithm outperforms the FFT-BAQ and BAQ algorithms.