Ensemble transform Kalman filter (ETKF) is an extension of ensemble Kalman filter (EnKF), which avoids using 'perturbed observations' to eliminate additional sampling errors. This paper demonstrates the capability of ETKF algorithm for sequentially correcting dynamically evolving fire perimeter positions at regular time intervals to enhance the prediction accuracy of wildfire spread. Forecast error covariance inflation scheme is adopted in the ETKF to address the underestimation problem of forecast error covariance of EnKF. Coupled to a widely-used fire spread simulator, FARSITE, the proposed approach is employed to a landscape with complex topography, where a fire barrier is also considered. The merits of ETKF algorithm for wildfire spread prediction are highlighted by simulation experiments using synthetically-generated observations. In order to quantitatively evaluate the prediction performance of ETKF, this paper has adopted a conservative index, Hausdorff distance, which is widely used in image processing area. This work is the first attempt of applying ETKF to wildfire spread simulation. The ETKF algorithm has been demonstrated to be more accurate than EnKF for a given ensemble size for wildfire spread simulation. The findings show that the ETKF-based data assimilation strategy is a promising tool for large-scale wildfire spread simulation.