Compression process in remote sensing confronts the problem of errors encountered in digital data transfer, when reconstructing the original image. A study on DCT bases for JPEG compression process is carried out on a sample image and a typically remote sensed image. A formula for quantization matrix as a function in one variable is proposed. Effective compression ratio (ECR) is investigated versus the variable of this formula. In ideal case of data transfer, ECR can get a value of 18 with root mean square error (RMSE) <6%. The peak signal to noise ratio (PSNR) is used as a quantitative figure of merit. It's found to be >55 for ECR≤19 for the two selected images. Data may encounter errors in its transfer process. Therefore, error rate and level are proposed as 2 independent variables to analyze image quality after reconstruction. Error rate is taken to be one faulty bit or a burst of bits for every 8, 16, 32, 64, 96, and 128 Kbit of transferred data. It is found that the quality of reconstructed image is the same like that of ideal data transfer at low error rate ≤1/100K. This quality starts to degrade dramatically if the error rate grows beyond 1/64K as a threshold value, especially at compression ratios ECR<11. The influence of error rate is only investigated for low compression ratios. The errors in transfer of data are hidden underneath the encountered errors by high compression ratios.