Abstract: Using the 2015-2017 European Center numerical forecasting product, combined with different numbers of decision trees for model training, establishes regression forecasting model based on the random forest method at Urumqi airport. Through the inspection of the prediction results of the model, we can see the average absolute temperature error is less than or equal 2 ℃, accounting for 94% of the total number of samples, the airport temperature is between -10 and 30 ℃, and the average absolute error is 1.2 ℃ . The effect is good, so you can try to use this method to produce objective product guidance products for civil aviation airports.