The counter-battery radar system tracks the artillery projectile trajectory and finds the launch point (LP) to take swift protective actions against enemies' succeeding shelling. The LP estimation highly depends on the accuracy of target state estimate. However, measurements from the surveillance radar systems include clutter measurements as well as target measurements, so the data association is needed for target tracking in the cluttered environments. In this paper, the smoothing integrated probabilistic data association (sIPDA) algorithm is applied for estimation of artillery projectile trajectory in the cluttered environments and backward prediction with the constant axial force (CAF) model is carried out for launch point prediction (LPP). The proposed method significantly improves the accuracy of LPP compared with the integrated probabilistic data association (IPDA) algorithm.