Background: Airborne gamma-ray spectrometry is made up of multiple NaI（Tl） detector

but the detectors have the different level of noise.Purpose: Aimed at reducing the affection of noise

a weighted least square fusion estimation algorithm is presented to extract the Spectra Characteristics.Methods: The method doesn’t need any prior knowledge on the detector

but carries on the variance estimated on-line these data and timely adjust weights of various fusion sensors in order to make the mean-square error of fusion results least all the time.Results: It is used to process the date come from the standard model

the results show that measurement error range of 40K decreased from 17% to 12%.Conclusions: The result has shown that the method can dramatically decrease the error and improve the accuracy.