Abstract
Based on KIJKKAMÄKI's temperature model t=a+bzc, the authors devised an experiment to test how KUKKAMAKI's corrections improved the discrepancy between fore and back leveling in actual leveling circumstances. Although KUKKAMAKI's c values obtained from temperature measurements were widely scattered, the correction using these c values could improve the discrepancy. On the other hand, the correction using c=-1/3 made the discrepancy larger than that without corrections. Additionally, it was found that terrain profile in a section was an important factor in applying refraction corrections.
