Recent studies revealed the effect of meteorological seasons on the accuracy of GPS positioning. The existing accuracy model is weak because it is derived from a climate classification in which the earth is divided into latitudinal zones, and each climate zone in the classification does not have sufficient number of International GNSS Service (IGS) sites as representative. Furthermore, the prediction is biased due to the loading effects and tectonic motion. Here, we will derive a new accuracy model using the climate zone classification of Köppen and Geiger, which is based on actual meteorological measurements. Fifty IGS stations almost homogenously covering the Köppen-Geiger geographical regions and the globe have been selected, and the data has been analyzed using Gipsy-X software. Using the new model, the average improvement in the accuracy prediction of the north, east, and up components is 50%, 68%, and 94%, respectively.