Due to the diffuse scattering on sea surface and dynamic geometry of coastal GNSS- R, the altimetry performance is strong dependent on elevation angle. As elevation angle decreases, the bias in retrieved height using DER and HALF retracker increases. In this paper, an multi-parameter estimator is proposed to combine the DER retracker and several HALF retrackers to obtain unbiased height. The simulated results show that the bias of the uncalibrated height using proposed algorithm has no significant dependence on elevation angle, but the RMSE for the low elevation angle case still are larger than that of the high elevation angle. In addition, the calibrated performance depends on the used fractional values so that proper fractional values should be chosen. The data from Dongying experiment acquiring GPS L1 CA and L5 signal and Qingdao experiment for BeiDou B1I signal are used to demonstrate proposed algorithm. After calibrating using proposed algorithm, regardless of GPS L1 CA, L5 or BeiDou B1I signal, the calibrated heights all are agreement with in-situ heights. Finally, the calibrated heights from BeiDou B1I signal are used to retrieve sea surface height. The results show that retrieved sea surface heights from GEO and IGSO/MEO satellites all appear the same trend with in-situ ones, furthermore, GEO satellites can provide the better retrieved performance than IGSO/MEO satellites. Through averaging the heights from several satellites and smoothing with the half-hour span, compared to in-situ sea surface height the RMSE of 0.67 m is obtained.