<p>We compared and analyzed data of vertically Integrated Water Vapor (IWV) from two different re-analysis models (ERA-Interim from ECMWF and MERRA-2 from NASA's Global Modeling and Assimilation Office) with respect to IWV values from Global Navigation Satellite Systems (GNSS) at 53 stations of Central and South America during the 7-year period from January 2007 till December 2013.</p> <p>The comparison was performed taking into account the geopotential height differences between each GNSS station and the correspondent values assigned by the models. Thus, the set of GNSS stations was divided into 3 groups: Small, Large and Critical height difference stations. Moreover, the performance of the re-analysis models was also analyzed by using an additional classification of three levels according to the mean IWV (<span style="text-decoration: overline">IWV</span>) value expected at the station: <span style="text-decoration: overline">IWV</span> > 30 kg m<sup>−2</sup>, 12 kg m<sup>−2</sup> ⩽ <span style="text-decoration: overline">IWV</span> ⩽ 30 kg m<sup>−2</sup> and <span style="text-decoration: overline">IWV</span> < 12 kg m<sup>−2</sup>.</p> <p>Both models (IWV<sub>ERA-Interim</sub> and IWV<sub>MERRA-2</sub>) offered a very good representation of the IWV from GNSS values (IWV<sub>GNSS</sub>) for stations with a Small height difference (smaller than 100 meters). That is to say, the differences between the mean values of IWV from GNSS (<span style="text-decoration: overline">IWV</span><sub>GNSS</sub>) with respect to the IWV averages from both re-analysis models are always below 7 % of the <span style="text-decoration: overline">IWV</span><sub>GNSS</sub> in the worse case.</p> <p>In general, the discrepancies between the re-analysis models with respect to IWV<sub>GNSS</sub> raise as the geopotential height difference between the GNSS station and the static geopotential height interpolated from the models grows. Effectively, the difference between IWV<sub>GNSS</sub> and IWV from the re-analysis models can be as large as 10 kg m<sup>−2</sup> for stations with a critical height difference (larger than 500 meters). For this reason, we proposed a numerical correction that compensates the effect of the geopotential height difference and the results were tested with values from ERA-Interim.</p> <p>The suggested correction was successful and reduces the differences |IWV<sub>GNSS</sub> − IWV<sub>ERA-Interim</sub>| to less than a 7 % of the mean IWV<sub>GNSS</sub> values. This strategy is especially recommended for stations that were classified as Critical, most of them located in mountainous areas of South America. In the case of Large height difference stations, the correction procedure is not advisable either for a coastal station and/or stations in islands. Generally, in those cases, two or more grid point are on the water. Thus, the interpolated IWV value for the re-analysis model will be overestimated. At one hand, if the geopotential height of the model is smaller than the geopotential height of the GNSS station, the subtracting numerical correction would compensate this overestimation of IWV near the water and thus the strategy will represent an improvement. On the other hand, if the relationship between the geopotential heights is the opposite, the correction will be additive causing thus a worse agreement between both time series.</p>