Airport Navigation will require more stringent localization performance requirements than in-flight navigation [1]. GNSS signals (Global Navigation Satellites Systems) can be envisaged to elaborate the aircraft estimate position on the airport surface. To improve the performance of localization on the airport, the errors on GNSS signals particular to the airport environment must be characterized. Most of these errors are well known such as ionosphere error, troposphere error, etc, and do not depend on the airport environment. But to achieve the expected sub-metric performance, it is necessary to better model multipath error for which a model already exists but is valid for operations from en-route down to CAT I only. In this paper, an analysis of real GPS measurements (using code pseudo range measurement, carrier phase measurement, Doppler measurement and the estimate C/N0 ratio measurement) during taxiing operation on the airport surface is conducted. The goal of this paper is to evaluate when multipath occurs and to compare the multipath model (elaborated from the standard deviation of the measurement errors due to multipath) based on those collected measurements in the airport with different models proposed in the literature (not necessary proposed for airport navigation).