Based on the machine learning, an application model(GBDT) of real-time fusion on temperature, precipitation, wind direction, and wind speed elements at selected locations is developed by using the professional service product (CAR) of the public meteorological service center of the China Meteorological Administration, the multi-source fusion observation analysis data (ART) of the National Meteorological Information Center, the national radar retrieval precipitation product (RAD), the Fengyun 4 satellite retrieval precipitation product (SAT), and the hourly data of the national meteorological observatory. A national and regional error analysis was conducted on the fusion products output by GBDT, CAR, and ART, respectively, with hourly observation data. The national and regional inspection results of the 15 day hourly GBDT fusion product show that: the GBDT temperature fusion product has improved compared to CAR and ART in six regions: Northeast China, North China, Northwest China, Central China, Xinjiang, and Tibet, with the most significant improvement in Tibet. GBDT is superior to ART and inferior to CAR in East and Southwest China, and GBDT error slightly increases compared to ART and CAR in South China and Inner Mongolia. The GBDT precipitation fusion product has a slight increase in error compared to ART and CAR in Inner Mongolia, where there are fewer samples, while other areas have improvements or are basically equivalent. The GBDT wind speed and direction fusion products have significant improvements compared to ART and CAR. The experimental results indicate that machine learning methods can be applied to fuse multi-source real-time analysis products and observation data to provide real-time meteorological information services for selected location temperature, precipitation, wind direction, and wind speed elements.