Abstract: Considering the filter properties of the long horizontal line array (HLA) to the sound field modes, we established an adaptive optimal method to design HLA weights, which can change the filter pass band with the frequency. Using the normal modes theory and combining with the feature of the sound field HLA beam forming, and analyzing the differences of definition and slope in the LOFAR (Low Frequency Analysis Recording) spectrogram between the single- hydrophone and the beam former, a frequency adaptive optimal estimate was obtained by solving a quadratic optimization model with linear equality constraint. The numerical simulation and the experimental data analysis indicated that we can obtain the beam former signal with the designed optimal array weight which was made up by the Surface- Reflected Bottom-Reflected modes or Non-Surface-Reflected Bottom-Reflected modes. The slope of striations in the signal spectrogram fitted well with the theoretical analysis.