We have adopted Polaris sighting as our azimuth observation at our observatories for a long time. This method, however, has some difficulties such as weather-dependent observation conditions. Therefore, we examined the satellite method and found that its precision was comparable to the Polaris sighting. We also found that a significant difference of about 10 arcseconds between them. This is due likely to a deflection of vertical, which is well known in the field of geodetic surveying.