Water vapor is essential in maintaining the climate balance. It is a pivotal input parameter for remote sensing applications, including correcting atmospheric error and retrieving surface temperature. Precipitable water vapor (PWV) facilitates the exchange of water and energy between the Earth's surface and the surrounding atmosphere. It reflects the amount of water vapor in the atmosphere. Since the 1990s, GNSS meteorology has been proposed as a new method for PWV retrieval owing to the merits of superior temporal resolution, excellent precision, and cost-effectiveness. GNSS-derived PWV was a ground-based reference in numerous studies but only provides observation over the station. Fortunately, with the launch of the Landsat 8 carrying the Thermal Infrared Sensor (TIRS), PWV can be retrieved with a high spatial resolution. The thermal infrared PWV retrieval methods mainly include regression analysis, split-window method. For the ocean, the PWV retrieval is mainly achieved by establishing a regression relationship between PWV and the brightness temperature. For land, an atmospheric transmittance ratio between the two thermal infrared bands is calculated using a moving window, and then the mathematical equation between PWV and the ratio is established. This algorithm is called the split-window covariance-variance ratio (SWCVR). Unfortunately, the conventional SWCVR method is susceptible to various factors, resulting in low retrieval accuracy. Moreover, the SWCVR algorithm will lose some retrieval results, seriously aggravating the unavailability of thermal infrared data. So far, rare studies have been made on the availability of the SWCVR algorithm. Therefore, we improve the traditional SWCVR model and propose a new PWV retrieval method using ensemble machine learning to address these limitations. The new method uses Gradient Boosting Decision Tree (GBDT) to establish the model between brightness temperature, GNSS-derived PWV, and related surface parameters. The results of the test set show that the improved SWCVR model has an RMSE, Bias, and availability of 0.4947 g/cm2, 0.0276 g/cm2, and 29.6%, respectively. By contrast, the GBDT model's corresponding values are 0.2870 g/cm2, -0.0094 g/cm2, and 67.9%, respectively. Compared with SWCVR, the GBDT improves the RMSE and availability by 41.99% and 38.3%, respectively. The GBDT algorithm is significantly better than the SWCVR model in low altitude and complex surface coverage areas. From a temporal perspective, the advantages of the GBDT method are more apparent in the summer. Finally, the SWCVR and GBDT models are externally validated using measured PWV data from the sun photometer and radiosonde, and the RMSE is 0.5148 g/cm2 and 0.3775 g/cm2, respectively. Based on the findings, we know that the accuracy of GBDT has significantly improved against the SWCVR.