Many satellite-derived products such as the atmospheric wind vector depend their accuracy on the accuracy of the estimated cloud top altitude. The uncertainty in the derived cloud top altitude occurs mainly when there is thin semitransparent cloud where the cloud radiation is contaminated by radiation from the surface and low cloud. Further, validation of the derived cloud top altitude is not easy task, simply due to lack of truth data. Here, we use ground based rawinsonde, radar, and lidar data for the validation of the cloud top altitude derived from GOES-9 satellite data. The preliminary results show that the infrared-water vapor method compares better than the single infrared method for all of the ground truth data.