Abstract The estimation of the parameters of K-distribution based on the z log(z) expectation approach is extended to zrlog(z) expectation. The parameter estimation expression of the based on zr log(z) expectation is derived. When r is taken different values, the accuracies of the estimations based on the zrlog(z) expectation are analyzed and compared with one another. Simulation results show that in the parameters range of K distribution radar spiky clutter, the estimation based on zr log(z) expectation is more accurate than z log(z) expectation and Normalised Log Estimator (NLE) approaches when r is less than unity.