Abstract: Shear wind is regarded as one of the largest threat to the safety of aircrafts when landing or taking off.Therefore，it is very important to identify the shear wind precisely.Wind shear radar is the main tool to detect and identify the shear wind.Wind shear radar processes the echo signal to estimate the wind shear risk factor.The risk factor is a key parameter in wind shear recognition，and wind speed estimation is an important part of the calculation for the risk factor.Thus，strengthening the research into the algorithms for wind speed estimation is very necessary.The wind speed estimation method based on frequency domain can obtain more spectrum information than methods based on time domain.However the classical estimation methods based on frequency domain have low resolution and large variance.In view of the problems of traditional spectrum estimation methods based on frequency domain，we present a kind of method based on the improved Burg algorithm to estimate the speed of shear wind.Burg algorithm can estimate the spectrum with higher frequency resolution than the classic spectrum estimation.While under the condition of short sequence，the Burg algorithm will cause spectrum estimation errors because of the division and migration of spectrum peaks.Analysis shows that the error of the AR model coefficients estimated by Burg algorithm is the main cause of the spectrum division and migration.The improved Burg algorithm accumulates the multistage echo data and combines the original burg algorithm to estimate AR model coefficients.The improved Burg algorithm reduces the error of AR model coefficients，so as to improve the precision of spectrum estimation.This improved algorithm inherits the advantages of the original burg algorithm to obtain high frequency resolution and the small variance.Results of the simulation show that the proposed wind speed estimator reaches the resolution of the Burg spectrum estimation and the accuracy of the FFT spectrum estimator.