Abstract: The aerodynamic characteristics of double-deck bridge girders are highly dependent on the aerodynamic interference among their subcomponents (upper/lower decks, windward/leeward trusses). These characteristics are profoundly sensitive to turbulence parameters (turbulence intensity I, integral scale L) and exhibit differential responses to the longitudinal (u-) and vertical (w-) turbulence. However, traditional wind tunnel tests cannot independently control I and L nor decouple the differential effects of u-/w- components. This study employed an improved Consistent Discrete Random Flow Generation (CDRFG) method to generate wind fields with selectively modulated L_u or L_w parameters, enabling the isolation of their differential impacts. Key findings reveal that the aerodynamic drag is mainly borne by the windward and leeward trusses, while the drag on the upper and lower decks is relatively small. The aerodynamic lift and moment are almost entirely borne by the upper and lower decks. Specifically, increasing L_u does not significantly change mean drag but notably reduces mean lift and increases fluctuating drag. Conversely, increasing L_w significantly enhances the fluctuating values of drag, lift, and pitching moment, with a marginal effect on their mean values. Pressure coefficient analysis shows L_u increases mean negative pressure on the lower surface, while L_w enhances it on the upper surface but weakens it on the lower, with both significantly increasing fluctuating pressures. Pressure correlation analysis reveals that L_u enhances streamwise correlation on the upper surface's front section but weakens the rear, and improves overall spanwise correlation; L_w reduces overall streamwise correlation on the upper surface, but enhances spanwise correlation on both surfaces. This research provides crucial insights and theoretical support for the refined wind-resistant design of double-deck truss girders.