GNSS satellite signals suffer considerable delays while travelling through the troposphere. The delay caused, it can be separated into two different parts: the effect of gases in hydrostatic equilibrium and the effect of water vapour and condensed water present in the troposphere. In safety-of-life navigation applications of GNSS (e.g. positioning and navigation of aircrafts, autonomous vehicles, etc.) not only the accuracy of the positioning needs to be known,
but the integrity of the positioning service should be evaluated, too. The integrity information means that the maximum positioning error at an extremely rare probability level (approximately 10-7), called the protection level, must be determined. The widely adopted RTCA (Radio Technical Commission for Aeronautics) recommends the minimum operational performance standard (MOPS) for GNSS systems used in the aeronautics. According to this recommendation, the maximum total tropospheric delay error in the zenith direction is 0.12 m in terms of standard deviation. However, this model neglects both the geographical and seasonal variation of the accuracy performance of the troposphe ... mehrric delay models. Our study focuses on the theoretical background of the assessment of tropospheric delay model performances under worst-case scenarios. The developed computational strategy is capable to estimate the magnitude of extremely rare tropospheric delay error and takes into consideration not only the geographical but also the seasonal variation of model performance. The results show that the proposed methodology provides a conservative model for assessing the maximal tropospheric delay error in worst case scenarios. However, the derived model is significantly less conservative than the RTCA recommendation based on radiosonde observations obtained in Budapest.