Gravel-covered surfaces are common in arid regions, where both natural and artificial gravels play a key role in suppressing soil wind erosion. Previous studies have often focused on specific regions or soil types, leaving the combined effect of gravel cover and soil particle characteristics on the wind erosion rate (q) underexplored. The main objective of this study was to clarify the combined effects of gravel cover and soil particle characteristics on q and to develop a predictive equation through wind tunnel experiments using four soil samples with different particle characteristics under five incoming friction wind velocity (u*infinity) for bare soil and 25 gravel-covered beds. Results showed that greater lateral coverage (lambda) significantly suppressed q, but at high friction wind velocity (u*) and low lambda, q intensified. Soil particle characteristics affect bare and gravel-covered beds differently, indicating that traditional adjustment methods based on bare soil erosion were unsuitable. Random Forest (RF) algorithm was applied as an auxiliary tool to rank variables affecting wind erosion: for bare soil beds, u*, mean particle diameter (d), density (rho p), sorting coefficient (sigma), and sphericity (Sp); for gravel-covered beds, u*, lambda, Sp, sigma, d, and rho p. The RF model achieved RMSETest of 0.010 kg center dot m- 2 center dot s- 1 for bare soil and 0.001 kg center dot m- 2 center dot s- 1 for gravel-covered beds. Based on these insights, we proposed an extended wind-blown soil flux equation incorporating lambda and Sp. This new equation achieved an RMSE of 0.004 kg center dot m- 2 center dot s- 1. This equation effectively captures the combined effects of gravel cover and soil particle characteristics.