An optimal Digital Filter Initialization (DFI) technique is used in the HRM (High Resolution Model) in order to suppress spurious high frequency oscillations caused by an imbalance between initial mass and wind fields. The filter is calibrated for a large Brazilian domain and its efficiency on removing high frequency oscillations is analyzed. The DFI is compared to the nonlinear Normal-Mode Initialization (NMI) also implemented on HRM. Both methods yield to satisfactory results in suppressing noise in the forecast, but the DFI appears to be more efficient.

initialization, digital filter, HRM.